á

-

IUFRO

ITE symposium no. 29ECTF symposium no. 1

Edited by

R R B Leakey and A C Newton

London: HMSO

Edinburgh Centrefor Tropical Forests

N Institute ofTerrestrial

W.7 Ecology

Tropical trees:the potential for domesticationand the rebuilding of forestresources

The proceedings of a Conference organisedby the Edinburgh Centre for Tropical Forests,held at Heriot-Watt University Edinburgh,on 23-28 August 1992,as part of the 1UFRO Centennial Year (1892-1992)

@ Crown Copyright 1994Applications for reproduction should be made to HMSOISBN 0 11 701562 8

The Institute of Terrestrial Ecology (ITE) is a component research organisation within the Natural EnvironmentResearch Council. The Institute is part of the Terrestrial and Freshwater Sciences Directorate, and was establishedin 1973 by the merger of the research stations of the Naiure Conservancy with the Institute of Tree Biology It hasbeen at the forefront of ecological research ever since. The six research stations of the Institute provide a readyaccess to sites and to environmental and ecological problems in any part of Britain. In addition to the broadenvironmental knowledge and experience expected of the modern ecologist, each station has a range of specialexpertise and facilities. Thus, the Institute is able to provide unparallelled opportunities for long-term,multidisciplinary studies of complex environmental and ecological problems.

ITE undertakes specialist ecological research on subjects ranging from micro-organisms to trees and mammals,from coastal habitats to uplands, from derelict land to air pollution. Understanding the ecology of different speciesof natural and man-made communities plays an increasingly important role in areas such as monitoring ecologicalaspects of agriculture, improving productivity in forestry controlling pests, managing and conserving wildlife,assessing the causes and effects of pollution, and rehabilitating disturbed sites.

The Institute's research is financed by the UK Government through the science budget, and by private and publicsector customers who commission or sponsor specific research programmes. ITE's expertise is also widely usedby international organisations in overseas collaborative projects.

The results of ITE research are available to those responsible for the protection, management and wise use of ournatural resources, being published in a wide range of scientific journals, and in an ITE series of publications. TheAnnual Report contains more general information.

R R B Lenkey* and A C NewtonInstitute of Terrestrial EcologyBush EstatePenicuikMidlothian EH26 OQBScotland, UK

Tel: 031 445 4343

*In November 1993, Dr Leakey was seconded to the International Centre for Researchon Agroforestry as Director of ResearchHis current address is:ICRAF, United Nations Avenue, Gigiri, PO Box 30677, Nairobi, Kenya

Contents

Preface

Domestication of 'Cinderella' species asthe start of a woody-plant revolution

The resource of useful tree species:for domestication

Amazonian tree diversity and thepotential for supply of non-timberforest products

Designing commercially promisingtropical timber species

Timber trees: architecture and ecology

Fruit trees: diversity and conservationstrategies

Trees for drugs and other extractives:future propects

Techniques to domesticate trees

In vitro systems for the conservationand utilisation of germplasm

Capture of genetic resources by in vitroculture: field evaluation andprecocious flowering

Capture of genetic resources bycollection and storage of seed: aphysiological approach

Capture of genetic variation byvegetative propagation: processesdetermining success

Selection and breeding to conserve andutilise tropical tree germplasm

Strategic concepts in the domesticationof non-industrial trees

Clonal development and deployment:strategies to enhance gain whileminimising risk

Regeneration of new forest resources

Indigenous trees in West African forestplantations: the need for domesticationby clonal techniques

The role of tree domestication inagroforestry

R R B Leakey & A C Newton

identification of priorities

G T Prance

J R Palmer

R A A Oldeman &A H M Sieben-Binnekamp

J C Okafor & A Lamb

P G Waterman

F Engelmann

S S Khuspe, R S Nadgauda,S M Jagtap, C K John &A F Mascarenhas

P B Tompsett

R R B Leakey A C Newton &J McP Dick

R D Barnes & A J Simons

A J Simons, D J MacQueen &J L Stewart

G S Foster & F L G Bertolucci

G J Lawson

F L Sinclair, I Verinumbe &J B Hall

Page nos

3

7

16

25

34

42

49

53

61

72

84

91

103

112

124

Use and potential of domesticated treesfor soil improvement

Matching trees to their light environments

Harnessing symbiotic associations:ectomycorrhizas

Harnessing symbiotic associations:vesicular—arbuscular mycorrhizas

Harnessing symbiotic associations: thepotentials of nitrogen-fixing trees

Case studies

Capturing and managing the geneticvariation in Prosopis spp. foreconomically useful characters

Domestication of lesser-known tropicaltree species: the Australian experience

Domestication of lesser-known speciesof the genus Leucaena

Domestication of Sesbania sesban foragroforestry systems in eastern andsouthern Africa

Domestication of trees in semi-arid EastAfrica: the current situation

Domestication potential of indigenousfruit trees of the miombo woodlandsof southern Africa

Amazonian fruits and nuts: potential fordomestication in variousagroecosystems

Genetic improvement of West Africantree species: past and present

Genetic improvement of trees in CentralAmerica, with particular reference toCosta Rica

Domestication of mahoganies

Production of clipterocarp planting stockby cuttings in Indonesia

Poster paper titles

List of Conference participants

E C M Fernandes, D P GarrityL T Szott & C A Palm

F E Fasehun & J Grace

F Lapeyrie & P H6gberg

P A Mason & J Wilson

J I Sprent

P Felker

T H Booth & J W Rirnbull

J L Brewbaker & C T Sorensson

F Owino, P A Oduol & F Esegu

P B Milimo, J McP Dick &R C Munro

J A Maghembe, F Kwesiga,M Ngulube, H Prins &F M Malaya

C R Clement & H Villachica

A C Newton, R R B LeakeyW Powell, K Chalmers, R Waugh,Z Tchoundjeu, P J Mathias,P G Alderson, J F Mesén,P Baker & S Ramnarine

W T M Smits, A C de Fraiture& I Yasman

137

148

158

165

176

183

189

195

205

210

220

230

D 0 Ladipo, S P K Britwum,Z Tchoundjeu, 0 Oni &R R B Leakey 239

J F Mesén, D H Boshier &J P Cornelius 249

256

267

273

275

Preface

Domestication of 'Cinderella' species as the startof a woody-plant revolution

R R B Leakey & A C NewtonInstitute of Terrestrial Ecology*, Bush Estate, Penicuik, Midlothian EH26 OQB, UK

*A component of the Edinburgh Centre for Tropical Forests

We are all aware that the pressures on tropicalecosystems are increasing daily: that human and

livestock populations are rising, and that thestock of productive land is fmite and dwindlingas a result of misuse. Deforestation is reducingspecies diversity and eroding the genetic base ofmany tropical tree species, including those used

traditionally by local people for products thatfulfil many of their domestic needs. Forestdegradation is also having negative impacts onsoil fertility and is modifying climate bylocal/regional influences on the albedo andhydrological cycle, as well as by the emission of

trace gases to the atmosphere.

Together, the above consequences ofdeforestation combine to reduce the area ofproductive land needed by growing humanpopulations, and to increase poverty and theincidence and severity of famine. At present, the

demand for farm land is met by opening upnew areas of forest to agriculture. It is thereforeimportant to increase the resource of trees indeforested areas, so as to meet the traditionalneeds of local people, including the urbanpopulations. Somehow a way has to be found to

make the land more productive and torehabilitate degraded areas in a way which will

diversify production, promote geneticconservation, enhance the development ofsustainable land use and contribute to domestic,

regional and international trade.

Great progress was made with food crops in the

l960-80s, as a result of the 'green revolution',but now there is a need to diversify into new

crops, many of which could be tree crops. Thisdiversification is desirable ecologically andeconomically to reduce the dependence oncommodity crops with falling product prices. In

addition, without genetic engineering, furtherlarge increases in the yield of the current narrow

range of crop species are unlikely

Throughout the tropics there are numerousperennial woody species that have provided

indigenous peoples with many of their needs formillennia. Many of these people have now leftthe land for the concrete jungles of urban life,

but would still like supplies of traditional food,medicines and other natural products. Thesetraditionally important woody plants are virtuallyundomesticated, ignored by the 'green revolution',

yet probably have great genetic diversity Theyalso play a key role in the biological, chemical,physical and hydrological cycles of the soil and

the atmosphere, protecting soils from damageand acting as a buffer against environmentalchange. These same plants are the ones that are

cut down indiscriminately to relinquish a fewvaluable logs per hectare, to clear land for'development' projects or to provide asubsistence living for a year or two. Theseoverlooked and often highly desirable trees arethe 'Cinderella' species of our forests.

'Cinderella' species are often ignored byagriculturalists of the developed world in favour

of a handful of exotic species which have beendomesticated and grown artificially on a largescale. The choice is often made without muchthought for the environment, and, perhaps, withthought only for relatively short-term economicgains. The same scenario has occurred inforestry While planting and domesticating a fewexotic fast-growing trees for pulp and paper, the

majority of the hardwood species forming thebasis of the timber trade have not, for a varietyof reasons, contributed significantly toreafforestation programmes. Surprisingly therefore,

the absence of even beautiful and commerciallyimportant timber trees such as mahogany fromman-made forests allows them to be classed as'Cinderella' species. The resources of these'Cinderella' tree species can be tapped by the

process of 'domestication', which has beenapplied so successfully to agricultural andhorticultural crops.

One of the objectives of this Conference is toraise the profile of the 'Cinderella' species, and,

particularly to draw attention to a growing effort

to domesticate a much wider array of treespecies and introduce them into agricultural and

forestry systems. The domestication oftraditionally utilised 'Cinderella' species providing

fuelwood, poles, timber, fruits, gums, nuts, resins,

3

fibre, pharmaceutical products, etc, can now berapidly achieved by applying some of thetechniques described in these Proceedings. Thisapproach has the added environmental benefit ofdiversifying the species grown on farm land andin forest plantations. In addition, domesticatingthese species will go some way towardsreducing economic dependence on a few'commodity' species. In this way a start is beingmade to initiate a 'woody-plant revolution', whichcould provide some cheer in the otherwisegloomy outlook for tropical forests. Opportunitiesare currently being lost because of a lack ofawareness of the potential to domesticate foresttree species for the production of timber andnon-timber products.

What are the issues that have to be resolved totrigger this new revolution? From the viewpointof a fanner, there are:

• the political and social issues, such as how toacquire the right to own and protect a pieceof land and the trees on it, and the need forincentives to plant trees;

• the economic issues, such as what is thevalue of these trees in terms of their wood,other products and environmental services;

• the biological issues, such as how to growthe trees wanted by farmers; how they canbe made more desirable and productive, tothe extent of satisfying the farmers' needsand even providing a surplus which couldbe sold to urban populations.

This Conference is about solving the biologicalissues concerned with the provision of timber,food and other domestic necessities. However,there are also direct economic and socialbenefits, such as improving community welfarethrough the market-place. To plant and growtrees on a scale necessary to have any impacton either the environmental or socialconsequences of deforestation, it is essential thatfarmers, communities and companies seefinancial and other incentives accruing from treeplanting. The incentive of greater profitability isproposed as a powerful means of promoting theplanting and growth of tree species. Geneticimprovements in yield and quality shouldincrease the value of the products of a tree crop,while increased growth rates in timber trees willallow a shortening of the rotation, and hence amore rapid return on invested capital. It shouldnot be at the expense, however, of theenvironmental benefits of tree planting.

There has been a recent awakening to thesebiological issues, and significant progress hasbeen made in the past '10-15 years. It has beenrealised that it is possible to capture the geneticvariation inherent in tropical tree populations, andthat this can be used to domesticate treesrapidly for timber as well as non-timber products

4

for sale in local, regional and even internationalmarket-places. It has also been realised thatthere is a need to improve management systemsif new cultivars are to be as productive as theycould be, and if they are to be used torehabilitate deforested land. Options includesystems which rehabilitate the environment on alocal scale, by improving soil structure andfertility Other systems may have more wide-scaleeffects, such as beneficial impacts on theregional hydrology or global climate, by thecreation of complex canopies which mimicnatural forests.

There is also a growing realisation that thedomestication of these 'Cinderella' trees involvesprotecting their genetic resources and growingthem wisely and efficiently to protect theenvironment. Perhaps a 'woody-plant revolution' ofthis sort could contribute significantly to therehabilitation of our planet and the feeding of itspopulation. Hopefully the Proceedings of thisConference will inspire fairy godmothers,particularly policy-makers and donors, tomobilise their glass carriages and get Cinderellato the ball.

Dedication

These Proceedings are dedicated to Douglas GB Leakey who as a forester in Kenya all hiscareer tried to encourage the planting of treesfor environmental protection. It was thisawareness of the importance of trees that ledhim to write a paper in 1948 for a conference inthe Belgian Congo, which subsequently inspiredProfessor Anthony Young of the InternationalCentre for Research in Agroforestry to commendhim as one of the early advocates of agroforestry

(See ICRAF Newsletter (October 1986), 18, 5.)

Domestication: a definition

The concept of a domesticated species isfamiliar to most people in terms of farm animalsand household pets, but is not so common inplants. However, it is well laiown that moderngrain cultivars have been bred over generations

- from wild grasses, and that our gardens andorchards are stocked with woody plants that havesimilarly been subjected to selection andbreeding. Foresters have been slow to see theneed to domesticate their crops. Until relativelyrecently foresters have collected seeds from the'wild and used them without selection. The use ofprovenance selection, an early stage in thedomestication process, gained internationalimportance in the 1960s, although techniquessuch as clonal selection have been used forcenturies in a limited number of species - Salix,Populus, Cryptomeria japonica, etc.

The following are defmitions which help toexplain the processes which are involved in thedomestication of a plant species.

• To domesticate is to naturalise; make fond ofhome; bring under human control, tame;civilise (Concise Oxford Dictionary)

• Domestication is closely linked to the idea ofselection, of fitness for purpose, of pushingnature into a higher gear and in a particulardirection (Palmer, pp16-24)

• Domestication is a two-stage process inplants: the bringing into cultivation of wildplants or exposing them to some form ofmanagement, and subjecting these todifferential production or selection (Janick etal. 1982)

• Domestication is human-induced change inthe genetics of a plant to conform to humandesires and agroecosystems, culminating inthe plant's loss of its ability to survive innatural ecosystems (Harlan 1975)

• The four elements of the domesticationprocess in forest trees (Libby 1973):

— the original genetic variability

— selection of the desired trees or genes

— a packaging of those genes in the plantsto be used

— converting those gene packages into thegrowing trees which are harvested as arenewable resource.

• There are several stages to the domesticationof crops (described by Clement & Villachica,pp230-238):

— managed species

— semi-domesticates

— full domesticates

At the Conference, the full 'domestication' of atree species was taken to encompass theidentification and characterisation of itsgerrnplasm resources; the capture, selection andmanagement of genetic resources; and the

Socio-economicappraisalSpecies selection

Resource identificationand characterisation

Figure I. Stages involved in the domestication of tropical trees

DOMESTICATION OF TROPICAL TREES

Development of techniques:• capture by seed

• capture by vegetative propagation

4.Application o techniques:

germplasm conservation and utilisation• selection• breeding

regeneration and sustainable cultivation of thespecies in managed ecosystems (Figure 1). Inthe papers presented, it is clear that the startingpoint can vary considerably depending on theapproach taken, which in turn seems to dependon whether the domesticator is a peasant farmeror a research scientist. Many of the speciesdescribed can, thus, at best be described assemi-domesticates.

In forestry the term 'domestication' has mostlybeen applied to the genetic improvement oftrees for industrial plantations, but one messagefrom the Conference is that the domesticationprocess in trees also has a strong socialinvolvement, seen clearly from the examples ofnon-timber species. The need to develop thissocial aspect has recently been recognised inthe strategic plan for the new Centre forInternational Forestry Research (CIFOR) preparedby the Australian Centre for InternationalAgricultural Research. CIFOR is the newestinstitute of the Consultative Group forInternational Agricultural Research.

References

Harlan, J.R. 1975. Crops and man. Madison, Wisconsin:American Society of Agronomy/Crop Science Society of

America.

janick, J., Schery, R.W, Woods, F.W. & Ruttan, V.W.1982. Plant science. San Fi-ancisco, California: Freeman.

Libby, W.J. 1973. Domestication strategies for forest trees.

Canadian Journal of Forest Research, 3, 265-276.

Sponsors

Aracruz International Ltd

Australian Centre for International AgriculturalResearch (ACIAR)

Centre Technique de Co-operation Agricole etRurale (CTA)

Soil ameliorationSilviculture

AgroforestryHarnessing symbiotic associations

Capture, se ection and Regeneration and sustainablemanagement management of forest resources

of genetic resources

5

Commission of European Communities DG XI-B4

Commonwealth Science Council (CSC)

Edinburgh Centre for Tropical Forests (ECU-)

International Union of Forestry ResearchOrganizations (RJFRO)

Shell International Petroleum Ltd

United Nations Educational, Scientific and CulturalOrganisation (UNESCO)

United Nations Environment Programme (UNEP)

Acknowledgements

We wish to thank our sponsors for their fmancialsupport, without which the Conference could nothave taken place. These funds partiallysupported the participation of the invitedspeakers. Our thanks are also due to IUFRO andECTF for their support and assistance inorganising the meeting, the ConferenceOrganising Committee and staff of the HeriotWatt Conference Centre. Particular thanks are

6

given to Mrs Carol Morris in the ConferenceSecretariat for all her hard work, Ms Lynn Davyfor proof reading and editorial assistance, andMrs Penny Ward for producing the fmal script.We are grateful to colleagues who acted asreferees of the manuscripts. Mr Bob Munro andDr Alan Longman are also thanked for their partin ensuring the smooth running of the meeting,especially the Poster Session. Finally weacknowledge the presenters of poster papers fortheir contributions to the information exchange atthe Conference, and sincerely thank all theauthors for preparing their manuscripts onschedule, and for their tolerance during theediting phase. This has allowed the Proceedingsto materialise without more delay than wasinevitable as a result of our own overseas travelcommitments.

June 1993

The resource of useful tree species:identification of priorities fordomesticationAmazonian tree diversity and the potential forsupply of non-timber forest products

G T PranceRoyal Botanic Gardens, Kew, Richmond, Surrey TW9 3AB, UK

ABSTRACTThe huge diversity of tree species found in the Amazon rainforest presents both an opportunity and a problem for the

exploitation of non-timber products. The diversity offers a vast array of potential harvestable crops but this also means that they

are sparsely distributed. A few crops, such as the Brazil nut and rubber, have been exploited for many years with little harm to

the forest, but other crops have been extracted in a destructive way such as rosewood and the various species providing balata

latex. It is not the most species-diverse forests which have the most potential for non-timber products; in certain areas under

adverse environments, such as swamp forests and transition forests where there is a dominance of one or a few species, there is

considerable opportunity to use non-timber products. The single most important plant family for such products are the palms,

which often dominate the species-poor habitats. However, there are many other plant species from the species-diverse forest with

potential for domestication; a number of these are described in this paper.

INTRODUCTION

In this paper it is intended to place theimportance of non-timber forest products (NTFPs)firmly on the agenda. Non-timber products, suchas medicines and fruits (see Waterman, pp42-48;Okafor & Lamb, pp34 41; Maghembe et al.,pp220-229; Clement & Villachica, pp230-238),are relevant to all systems of management, notjust to extraction from otherwise untouched forest,where there are severe limits to the possibilitiesof extractivism. Consequently the relevance ofNTFPs to timber plantations, to agroforestrysystems and to the domestication of new forestcrops in Amazonia will also be considered.

THE DIVERSITY OF THE AMAZONRAINFOREST

The Amazon rainforest is a potential source ofmany NTFPs because of its enormous speciesdiversity (Table 1). The highest diversity recordedis from Yanomono in Peru (Gentry 1988), where300 species of 10 cm diameter or greater wererecorded in a single hectare. In addition to treespecies, there is a wealth of herbs, shrubs,epiphytes, and parasitic species that addconsiderably to the total diversity of the forest. Inthe Rio Palenque Reserve in western Ecuador,Gentry and Dodson (1987) found 365 species ofvascular plants in a tenth of a hectare; in theforests of Costa Rica, Whitmore, Peralta andBrown (1985) found 233 species in 100 m2.These fmdings equate to a sixth of the Britishflora in an area smaller than a tennis court!Many of these species have some potential use,

as is apparent from studies of a quantitativeinventory of forest used by the Amazon Indians(Prance et al. 1987).

Boom (1985a, b) made a quantitative inventory offorest use by the Chacobo Indians of theBolivian Amazon, and found that they have a usefor 75 out of the 91 tree species (82%) on asample hectare. This included 95% of theindividual trees (619 out of a possible 649).Similarly Balée (1986) found that the Ka'aporIndians of Maranhao, Brazil, had a use for everysingle species of tree on a sample hectare. Theuses included food, fibres, fuel, material forconstruction and crafts, medicines and rubberlatex. If the Indians know of so many ways inwhich to use the forest, it would indicate thatthere are many more products to be marketedor species to be domesticated than the few

Table 1. Data on tree species diversity from inventories of

Amazon rainforest

No. of trees>10 cm

Locality diam ha-1 Source

Yanomono, Peru 300 Gentry (1988)

Mishana, Peru 289 Gentry (1988)

Yasuni, Ecuador 228 Balslev et a/. (1987)

Cocha Cashu, Peru 189 Gentry (1988)

Manaus, Brazil 179 Prance, Rodrigues &da Silva (1976)

162 Campbell et al. (1986)

157 Black, Dobzhansky &Pavan (1950)

Alto Ivon, Bolivia 91 Boom (1986)

Pantanal, Mato Grosso, Brazil 35 Prance & Schaller (1982)

Xingu, Brazil

Breves, Brazil

7

species producing timber, latex and -nutscurrently harvested from the region.

EXTRACTION OF NON-TIMBER PRODUCTS

Extraction of NTFPs for export from Amazonia isnothing new It began in the 16th and 17thcenturies when such products as cacao(Theobroma cacao), vanilla, sarsaparilla andipecac were some of the main products. Muchhas been written in recent years about theadvantages and disadvantages of creatingextractive reserves in Amazonia and elsewhere.Extractive reserves are areas of forest that,through legislation, have been set aside for thelocal population to extract NTFPs withoutclearcutting and which they manage collectivelyThrough the action of the Brazilian Council ofRubber Tappers and through the work of variousactivists and scientists in favour of the concept(eg Allegretti 1990; Fearnside 1989; Schwartzman1989),'Brazil has set aside over 3 millionhectares as extractive reserves, mainly in thestates of Acre, Amapa and Rondeinia. This in itselfis a measure of the initial success of the conceptof extractive reserves. Interest and enthusiasm forthis type of reserve were increased by variousdemonstrations that the economics of extractioncompared favourably with alternative land usessuch as cattle pasture and timber plantations(Alcorn 1989 for Mexico; Peters, Gentry &Mendelsohn 1989, Peters 1990, Prance 1989a forAmazonia; and Balick & Mendelsohn 1992 formedicinal plants in Belize). The economiccalculations of Peters, Gentry and Mendelsohn(1989) were based on the hypothetical marketvalue of a hectare of forest rather than the actualsales, and so these authors themselves pointedout the limitations of their study. Since then, alarge number of authors have discussed thedifficulties facing extractive reserves (Anderson1992; Browder 1990, 1992; Fearnside 1989;Pinedo-Vásquez et al. 1990; Vascruez & Gentry1989).

SOME LIMITATIONS OF EXTRACTIVISM

Contemporary extractivism in Amazonia hasarisen from the needs of the long-neglectedpopulation of rubber tappers, who are theremainder of the rubber boom and the debtpeonage system of aviamento (see Dean 1987;Prance 1989b; Weinstein 1983). Many rubbertappers outside extractive reserves are still indebt (Whitesell in Browder 1992). Withinextractive reserves, the economy is based largelyon two products, Brazil nuts (Bertholletia excelsa)and rubber. Until recently, in Brazil, the price ofrubber was sustained above worldwide marketvalue by subsidies. It has now droppedconsiderably; 60% of Brazilian rubber demand isnow provided by plantations outside the Amazonregion, where the rubber trees are not attacked

8

by diseases endemic to Amazonia. The price ofBrazil nuts has also dropped in the past fewyears, despite efforts by many people to expandthe market for this product. This instability of themarket price of the main extractive products isseriously straining the system, and is leading thedwellers of the reserves to fell more forest foragriculture and even to invest their meagreearnings in cattle.

The extraction of many products from forests hasled to over-exploitation rather than to thesustainable use of the resource. While theextraction for Brazil nuts and rubber has beenreasonably sustainable despite their otherproblems, there are many examples of bothtimbers and NTFPs where extraction has resultedin the virtual elimination of the species from theforest. For example, rosewood (Anibaroseaeodora) for the essential oil linalol, caucho(Castilla elastica) for rubber latex produced byfelling, and sorva (Couma utilis) for the latexused in chewing gum have all suffered fromover-exploitation, which has reduced thesespecies to near-extinction.

Other extraction products have been through this'boom and bust' cycle because of changes inthe international market. For example, barbasco(Lonchocarpus spp.) was used for the productionof insecticides until replaced by synthetics; tagua(Phytelephus macrocarpa) was used for vegetableivory until plastic replaced it; and oil producedfrom copaiba (Copaifera spp.) lost favour when itwas found to pose health hazards.

Conservationists are succeeding, to a limitedextent, in reviving markets for the latter twoproducts, but there is no doubt that the vagariesof the international market are not favourable toextraction products that are only harvested inrelatively small quantities. A further limitation toextractivism is the low population density that itcan sustain in comparison to such alternativeland uses as agroforestry (Anderson 1992;Browder 1992).

OLIGARCHIC FORESTS: WHEREEXTRACTION WORKS

The term 'oligarchic forest' was coined by Peterset al. (1989) for areas of forest within Amazoniathat, because of some environmental factor, arelargely dominated by a single species. If thedominant species in an oligarchic forest is ofeconomic use, then, because of its abundance,many of the constraints of extractive reserves inspecies-diverse rainforest are removed. Mostoligarchic forests in Amazonia are dominated bypalm species that are of considerable use, andsome cover extensive areas (cf Clement &Villachica, pp230-238).

Oligarchic forests of buriti or aguaje palm .(Mauritia llexuosa) and of agai palm (Euterpe

oleracea) occur in excessively swampy areaswhere little else will grow. There are vast areasof Mauritia, such as the area of over 10 000 hain extent around the Rio Jutai. This palm isexploited for its edible fruit in Amazonian Peru(Padoch 1988) and for both the fruit and aweaving fibre from the leaf epidermis forcraftwork in Maranhao, in eastern Amazonia,Brazil (Prance 1988). It could be a continuingand sustainable resource, but, unfortunately, inboth places local people often fell the trees tocollect the fruit. As the species is dioecious, itproduces stands of only male trees. The agaipalm is particularly abundant in the Amazondelta region in areas flooded daily by tidalback-up of the river. It is a truly multipurposepalm with many uses but especially for the fruit,the pulp of which commands a good price inlocal markets in Be lém. Its harvesting has beenstudied in detail by Anderson and associates(Anderson 1990; Anderson et al. 1985; Anderson& Ioris 1991; Anderson & Jardim 1989). Agai isalso an ideal species for the production of heartof palm, which is canned and exported, becauseit is a multistemmed palm that regenerates fromsuckers when cut. However, it is already beingseriously over-exploited on Maraj6 island inBrazil, and is another crop that could head for a'boom and bust' if not carefully controlled.

The babassu palm dominates large areas of thetransitional forests in eastern Amazonia betweenAmazon rainforest and the cerrado formation ofcentral Brazil. It forms oligarchic forests becauseof its resistance to fire, which can spread intothe drier forests of the transition region. The areaof babassu forest has increased dramatically inrecent years because individuals in deforestedareas survive and spread rapidly Babassu is alsoa multipurpose palm and has supported a largenumber of local people for many years. The fruitis divided into three parts, each of which isuseful. The outer mesocarp is rich incarbohydrate and is used for flour, the extremelyhard endocarp is used to make high-qualitycharcoal, and an oil that is both comestible andcombustible is extracted from the kernel. This oilis almost identical chemically with coconut oil.(For further details, see Anderson & Anderson1983; Anderson, May & Ba lick 1991; Balick 1987;May 1990a, b; May et al. 1985.)

The last example of an oligarchic forest is thatdominated by camu-camu (Myrciaria dubia) inthe riverside and lakeside areas of the upperAmazon. This member of the Myrtaceae familygrows in areas that are inundated for much ofthe year and where little else can survive. Thecherry-like fruit, which is produced as the riverlevel rises, can be easily harvested from canoes;it contains 30 times more vitamin C than citrus.It is already extremely popular in Iquitos, Peru,as a drink and as a flavouring for ice-cream,

and has much wider potential if a method fortransporting the fruit can be developed.Camu-camu is highly productive, easy to harvestand unlikeiy to be over-harvested. (For furtherdetails, see Peters et al. 1989; Peters &Hammond 1990; Peters & Vasquez 1987.)

EVALUATION OF EXTRACTION

Extractive reserves are not a panacea that willsave the whole Amazon region, but, if theirlimitations are recognised, they have an importantpart to play in the conservation of Amazonia.Extractive harvesting'works best in areas ofoligarchic forest, which by definition arespecies-poor, and not in the centres of diversitymost in need of conservation. However, improvedmanagement of oligarchic forests could deflectsome of the pressure from other areas. Extractivereserves in species-diverse rainforest have manysocial problems (Anderson 1992; Browder 1992;Salafsky Dugelby & Terborgh 1993), but theyshould not be under-rated. Nepstad et al. (1992)studied the Chico Mendes Reserve at Porongabain Acre State, Brazil, and showed some bioticimpoverishment of game, with one species, thetapir, eliminated. It was also noted that the Brazilnut was not regenerating, allegedly because ofover-harvesting. This study also showed that manymore rainforest species survived in the extractivereserve of Porongaba than in managementsystems of logging and ranching with which itwas compared. The absence of Brazil nutregeneration is normal because it is alight-demanding species that only develops whenthere are light gaps in the forest (Pires & Prance1977).

With human activity there is undoubtedly somebiotic impoverishment in extractive reserves.However, this impoverishment does not detractfrom their value because they have been createdin areas where the original forest is particularlythreatened by destruction. Most reserves wereformed to stop the progress of cattle ranchingand in this objective they have been successful.Apart from the conservation of biodiversityextractive reserves are most important becausethey maintain forest cover, protecting the soils,regulating the hydrological cycle (Salati & Vose,1984; Salati, Vose & Lovejoy 1986), andsequestering carbon, rather than releasing it intothe atmosphere. We should not under-estimatethese roles of extractive reserves.

If extractivism is to work well, markets must bedeveloped for products other than Brazil nutsand rubber, both of which have fluctuatingprices. This means that the livelihood of anextractivist is tenuous. The challenge before us isto search for and promote other products fromthe forest, and to develop markets forthem - locally regionally and internationally

9

EXTRACTION LEADS TO DOMESTICATION

Most of the Arnazonian trees or shrubs that arenow cultivated for their fruit, oils, nuts or otherproducts (eg cacao, rubber) were originallyextraction products. To support the concept ofextractivism is, therefore, to encourage futuredomestication. Extractivism needs to keep aheadof domestication, by constantly looldng for newproducts. One of the best-known examples ofextractivism is the Brazil nut. Until recently theentire production was from wild trees, as earlyefforts to grow plantations were a failure. Forexample, there is a large area of abandonedBrazil nut plantation near to Aleixo, Manaus.

Early attempts to cultivate Brazil nuts inplantations fared badly because the species doesnot thrive where the soil has lost its nutrients,and because of a lack of understanding about itsintricate pollination mechanism. We now laiow thepollinators (Prance 1985; Mori & Prance 1990;Nelson et al. 1986), and that Brazil nuts need tobe planted in clusters or strips near to forest sothat the pollinator bees work the plantation. Thefact that Brazil nuts naturally are often found inclusters of 50-100 trees (Balée 1989; Prance &Moh 1979) is probably because Indians usedthis method in the past. Today there are youngBrazil nut plantations in several parts of Brazil,some of which look very healthy Torres andMartine (1991) indicate that several thousandhectares of Brazil nut plantations have recentlybeen planted in Amazonas and Paled states. Brazilnuts are, therefore, in the process of becomingtruly domesticated, and much research is beingcarried out on their management at the HumidTropics Agricultural Research Station (CPATU) inBe lém (see bibliography in Correa, Pinto & dosSantos 1983).

Another crop that began as an extraction crop,was then grown in many private gardens, and isnow cultivated in plantations, is cupuagu(Theobroma grandiflora), one of the most popularfruits of the Brazilian Amazon. The plantations arestill low-yielding and are much attacked bywitch's broom (Marasmius perniciosus), also adisease of the related cacao. Domestication oftropical crops into plantations within their nativeareas is not always easy because of the endemicdiseases. Rubber plantations in Amazonia areentirely unsuccessful because of the leaf rustMicrocylus ulei and several other fungal diseases.There is a strong argument in favour ofdomestication through mixed-crop agroforestrysystems, such as those described in the nextsection, rather than in plantations, becauseagroforestry systems are less prone to diseasethan plantations, maintain a much greater degreeof biodiversity and retain more carbon, especiallyin the soil.

Domestication of some extraction products maydo a lot to avoid the 'boom and bust' economy

10

that has been the fate of so many Regular,controlled supplies of a product permit a muchbetter marketing strategy We should not beafraid of domesticating the most promisingextraction products, if this programme isaccompanied by adequate conservation measures.

AMAZONIAN AGROFORESTRY SYSTEMS

One of the most promising options for parts ofAmazonia is agroforestry It has been littleencouraged by developers or by governmentinstitutions, but has been widely practised formillennia by indigenous populations and morerecently by their successors, the caboclos orcampesinos, who have adopted many indigenoustechniques. Indigenous systems of agriculture areusually mixed-crop agroforestry involving a mixedplanting of herbaceous and tree crops. Aregeneration cycle is started which graduallyturns the field into an extraction forest. It is adynamic successional process rather than a staticone-crop or one-phase system. 'Pm Amazonianindigenous systems of agroforestry have beenparticularly well studied: those of the KayapóIndians (Hecht 1989; Posey 1982, 1983, 1984,1985), and the Bora Indians of Peru (Denevan etal. 1984, 1985; Denevan & Padoch 1988). Asimilar system to that of the Bora Indians hasbeen adopted by ribererio communities in Peru,such as the villages of Tamshiyacu and of SantaRosa (Padoch 1990; Padoch et al. 1985; Padoch& de Jong 1989). Agroforestry systems used bythe Huastec people of Mexico have also beenstudied in some depth (Alcorn 1984a, b).

With the Bora Indians, Denevan and Padoch(1988) demonstrated, through a series ofillustrations of fields of different ages, the gradualprogression from the initial planting to a19-year-old forest which was still visited by theIndians to harvest 22 useful species and whichcontained a small orchard of their crop tree,macarnbo (Theobroma bicolor). Similarly Posey(1982) described the Kayapó swiddens as'anything but abandoned fields'. Both thesesystems are intensive, can support a relativelydense population, and produce a wide range ofproducts, including food, oils for cooking, fibresand building materials.

In Tamshiyacu, Peru, the area used foragroforestry is now in the second or third phaseof a management cycle which the local peopleuse. The succession from the initial planting toforest and small patches of orchard is allowed todevelop for 35 years, and is then felled and thecycle repeated. The skilful use of variouscomponents of the system was noted. One of themost marketable fruits, the umari (Poraqueibasericea), is grown in small orchards, and anumber of different varieties are utilised. At theend of the cycle, the Poraqueiba trees are usedfor charcoal, another marketable product. Brazil

nut trees are also an important part of thissystem. At maturity a few of the trees are cut fortimber and others are left standing in the newcycle to continue the production of nuts, whichare not produced until a tree is at least 12 yearsold. In addition to this productive agroforestrysystem on upland non-flooded ground, theTarnshiyacenos cultivate rice and beans on thefloodplains of the river. These crops grow therewell in annual cycles between flood seasons.Alluvial matter enriches the soil so that it can bere-used each year without fertilizers. The moststriking result of this study was that the 1987income of a family in Tamshiyacu was just overUS$5,000, a remarkable amount anywhere in Peru.

Mother characteristic of both the indigenoussystems and that of l'amshiyacu is the number ofnamed varieties of each of their domesticatedcrops: at least six varieties of cassava (Manihot)and four of Poraqueiba were observed by thisauthor on a single visit. This point is furtheremphasised by the report that 130 varieties ofcassava are used by the Rikano Indians of theupper Rio Negro (Chernela 1986). This use ofgenetic biodiversity is the best protection theyhave for their crops against cli,sease. The fact thatvarieties exist of crops such as Poraqueibaindicates that they are already domesticated foruse in local agroforestry systems.

The forests of the Amazon delta region whichare dominated by the agai palm have alreadybeen mentioned (see earlier). These oligarchicforests are most effectively used when a form ofagroforestry rather than extraction is practised, anaspect that has been studied in detail on CombuIsland very close to the city of Belem (Anderson1988, 1990; Anderson et al, 1985; Anderson &Jardim 1989; Anderson & Ioris 1991). In the caseof Combu Island and other similar places in theAmazon delta, instead of initial clearcutting as inthe Peruvian systems described above, the areais managed by an initial selection thinning of theoriginal forest to encourage the growth ofselected species of economic importance, suchas cacao and agai palm. The structure of themanaged areas is remarkably similar to theoriginal forest (see profile diagram in Andersonet al. 1985). The combination of low annualinvestment into the system and access to anearby market made the Combu Islandagroforestry system the most profitable so fardescribed. The average landholding of a familyis 36 ha. Combu is an area particularly suited tothis type of farming system, which is onlyappropriate to the tidal delta region, andillustrates the need to adapt to local conditions,even within Amazonia, rather than design asingle formula for agroforestry throughout theregion.

Agroforestry is one of the most important areasfor further research and development in

Amazonia. It will only succeed if it is based ona diversity of clops, including species producingtimber and NTFPs. Agroforestry also offers anopportunity for further domestication of manynew plants (such as those listed in theAppendix). It is a relatively intensive land usethat can concentrate population density inwell-managed areas where agroforestry isappropriate, leaving sufficient room for anadequate number of biological reserves.

NON-TIMBER FOREST PRODUCTS FROMTIMBER TREES

There are many Amazonian timber treesproducing non-timber products that can beharvested while the timber is maturing. To maketimber trees productive in other ways wouldenable economic returns to be obtained from alonger growing cycle, and so encourage theharvesting of older trees. The agroforestry systemdescribed for Tamshiyacu includes Brazil nuttrees, some of which are eventually harvested fortheir excellent timber. The Brazil nut is one ofthe many possibilities for a mixed-harvestmanagement system. At present it is illegal inBrazil to use the wood of the Brazil nut, and thelaw which rightly protects the wild trees mightneed some modification to allow the use ofBrazil nut wood from a managed area.

Some examples of other possible crops fromdeveloping timber trees are: (i) ancliroba oil fromthe fruit of Carapa guianensis; pequi fruit andoil from Caryocar villosum; and (iii) virola oilfrom various species of Virola. Carapa is animportant wood in the Meliaceae or mahoganyfamily From quite a young age the trees producelarge seeds, that fall off and can be gatheredfrom the ground to extract the oil which alreadyhas many uses in Amazonia and which has hadan international market from time to time for usein soaps and medicines. It is now being testedfor cosmetics. A small plantation of Carapa at theDucke Forest Reserve near to Manaus isproducing abundant fruit. Virola oil, which hasbeen much used locally in soaps and candles, isobtained from species of Virola, some of themost important timber trees in Amazonia. Its usehas declined mainly because of timberharvesting, but it could instead be a by-productof the timber industry The wood of Caryocar ismuch used locally in boat-building rather than onthe international market. The fruits have anedible outer pulp and the kernel an oil that canbe used for cooking or lighting. Abundant fruitsare produced in young trees and could easily beharvested during the growth of timber (seePrance 1990).

Other crops that need to be further developedfor both agroforestry systems and timberplantations are the herbaceous and shrub cropswhich can be grown on the forest floor under

11

the shade of the canopy in a similar way tocacao, a long-established shade crop. Posey(1984) showed that the Kayapó Indians plantmany such crops along their trails through theforest to supply their needs as they move about.One possibility is the tuberous species ofCalathea (Marantaceae), already grown by sometribes.

CONCLUSIONS

Non-timber forest products must receive moreattention in future rainforest management systems.They are subject to the same dangers as thetimber trade of a 'boom and bust' use pattern(Vincent 1992), as has happened with Amazoniantimbers such as mahogany and cerejeira(Torresia acreana). A balanced mixture of use oftimber and NTFPs could lead to a more stableand sustainable land use pattern. The mostsuccessful models of land use developed so farin Amazonia are those based on traditionalmixed-crop agroforestry These systemsencourage a natural succession from herbaceouscrops through to an extraction forest from whichtimber can later be harvested.

Much further work is needed in Amazonia onthe domestication of further crops, including treecrops (see Appendix for species with particularpotential). However, even if further crops aredeveloped, it is vital to have an adequateconservation policy that conserves the geneticmaterial of the indigenous species of the region.If the wild relatives of either existing crops orones that are to be domesticated in the futureare lost, the genetic basis will be seriouslyweakened, and sustainability may becompromised. While some of the area can beused for extractive reserves and timberharvesting, a large proportion of the forest needsto be conserved in biological reserves, both forthe purpose of species conservation and forregulating local, regional and global climate.

I hope that this Conference will have encourageddialogue and collaboration between researchersinterested in timber trees and those interested innon-timber forest products. The systems that areworking sustainably are based largely onmultipurpose species which can yield bothtimber and non-timber products.

REFERENCESAlcorn, J.B. 1984a. Development policy forests and peasantfarms: reflections on Huastec-managed forests' contributions tocommemial production and resource conservation. EconomicBotany, 38, 389-406.

Alcorn, J.B. 1984b. Huastec Mayan ethnobotany. Austin, Texas:University of Texas Press.

Alcorn, J.B. 1989. An economic analysis of Huastec Mayanforest management. In: Fragile lands of Latin America:strategies for sustainable development, edited by J.0. Browder,182-206. Boulder, Colorado: Westview Press.

Allegretti, M.H. 1990. Extractive reserves: an alternative for

12

reconciling development and environmental conservation inAmazonia. In: Alternatives to deforestation: steps towardssustainable use of the Amazon rainforest, edited by A.B.Anderson, 252-264. New York: Columbia University Press.

Anderson, A.B. 1988. Use and management of native forestsdominated by acai palm (Euterpe oleracea Mart.) in theAmazon estuary Advances in Economic Botany, 6, 144-154.

Anderson, A.B. 1990. Extractivism and forest managementby inhabitants in the Amazon estuary In: Alternatives todeforestation: steps towards sustainable use of the Amazonrainforest, edited by A.B. Anderson, 65-85. New York:Columbia University Press.

Anderson, A.B. 1992. Land-use strategies for successfulextractive reserves in Amazonia. In: Non-timber products fromtropical forests: evaluation of a conservation and developmentstrategy, edited by D.C. Nepstad & S. Schwartzman, 67-77.(Advances in Economic Botany 9.) New York: New YorkBotanical Garden.

Anderson, A.B. & Anderson, S.E. 1983. People and thepalm forest: biology and utilization of Babacu palms inMaranhao, Brazil. (USMAB Project Report.) Gainesville, Florida:Department of Botany University of Florida.

•Anderson, A.B. & Ioris, E.M. 1991. The logic of extraction:resource management and income generation by exti-activeproducers in the Amazon estuary In: Traditional resource usein neotropical forests, edited by K.H. Redford & C. Padoch.New York: Columbia University Press.

Anderson, A.B. & Jardim, M.A.G. 1989. Costs andbenefits of floodplain forest management by rural inhabitantsin the Amazon estuary: a case study of acai palm production.In: Fragile lands of Latin America: strategies for sustainabledevelopment, edited by J.0. Browder, 114-126. Boulder,Colorado: Westview Press.

Anderson, A.B., May, P. & Balick, M.J. 1991. The subsidyfrom nature. New York: Columbia University Press.

Anderson, A.B., Gely, A., Strudwick, J., Sobel, G.L. &Pinto, M.G. 1985. Um sistema agroflorestal na varzea doestuario amazonico. Acta Amazonica, Suppl., 15 (1-2), 195-224.

Arkcoll, D.B. 1986. Couma utilis. Food and fruit-bearingforest species. Ronie, 44 (3), 109-111.

Balée, W. 1986. And lise preliminar de inventario florestal e aetnobotanica Ka'apor (Maranhao). Boletin do Museo ParenseEmilio Goekii, (Ser. Botanica), 2 (2), 14-167.

Balée, W. 1989. The culture of Amazon forests. Advances inEcdnomic Botany, 7, 1-21.

Balick, M.J. 1987. The economic utilization of the Babacupalm: a conservation strategy for sustaining tropical forestresources. Journal of the Washington Academy of Sciences,TT, 215-223.

Balick, M.J. & Mendelsohn, R. 1992. Assessing theeconomic value of traditional medicines from tropicalrainforests. Conservation Biology 6, 128-130.

Balslev, H., Luteyn, J., 1311gaard, B. & Holm-Nielsen,L.B. 1987. Composition and structure of adjacent unfloodedand floodplain forest in Amazonian Ecuador. Opera Botanica,92, 37-57.

Black, G.A., Dobzhansky, T. & Pavan, C. 1950. Someattempts to estimate species diversity and population densityof trees in Amazonian forests. Botanical Gazette, 111, 413-425.

Boom, B.M. 1985a. 'Advocacy botany' for the Neotropics.Garden, 9 (3), 24-32.

Boom, B.M. 1985b. Amazon Indians and the forestenvironment. Nature, London, 314, 324.

Boom, B.M. 1986. A forest inventory in Arnazonian Bolivia.Biotropica, 18, 287-294.

Browder, 1.0. 1990. Extractive reserves will not save thetropics. BioScience, 40, 626.

Browder, J.O. 1992. The limits of extractivism: tropical forest

sadtegies beyond extractive reserves. BioScience, 42, 174-182.

Campbell, D.G., Daly, D.C., Prance, G.T & Maciel, U.N.1986. Quantitative ecological inventory of terra finne andverzea tropical forest on the Rio Xingu, Brazilian Amazon.Brittonia, 38, 369-393.

Chavez, F.W.B. & Clement, C.R. 1983. Consideractiessobre o araca-boi (Eugenia stipitata McVaugh, Myrtaceae). In:

Amazonia Brasileira Comunicação 'Iecnica, 167-177.Florianopolis: SBF/EMPASC.

Chernela, J.M. 1986. Os cultivates de mandioca na Area do

Uaupés (flikano). In: Suma Etnologica Brasileira I, edited by

D. Ribeiro, 151-158. Petropolis: Etnobiologia, FINEP

Correa, M.P.F., Pinto, A. de A. & dos Santos, W.C. 1983.

Guaranå: resumos informativos. Brasilia: EMBRAPA.

Dean, W. 1987. Brazil and the struggle for rubberCambridge: Cambridge University Press.

Denevan, W.M. & Padoch, C. 1988. Swidden fallow

agroforestry in the Peruvian Amazon. (Advances in Economic

Botany 5.) New York: New York Botanical Garden.

Denevan, W.M., Treacy, J.M., Alcorn, J.B., Padoch, C.,Dens low, J. & Paitan, S.F. 1984. Indigenous agroforestry in

the Peruvian Amazon: Bora Indian management of swidden

fellows. Interciencia, 9, 346-356.

Denevan, WM., Treacy, J.M., Alcorn, J.B., Padoch, C.,Dens low, J. & Paitan, S.F. 1985. Indigenous agroforestry in

the Peruvian Amazon: Bora Indian management of swidden

fellows. In: Change in the Amazon Basin. 1. Man's impact onforests and rivers, edited by J. Hemming, 137-155.Manchester: Manchester University Press.

Fearnside, P.M. 1989. Extractive reserves in BrazilianAmazonia. BioScience, 39, 387-393.

Gentry, A.H. 1988. Tree species richness of upperAmazonian forests. Proceedings of the US National Academy of

Sciences, 85, 156-159.

Gentry, A.H. & Dodson, C. 1987. Contribution of nontrees

to species richness of a tropical min forest. Biotropica, 19,149-156.

Hecht, S.B. 1989. Indigenous soil management in the

Amazon Basin: some implications for development. In: Fragilelands of Latin America: strategies for sustainable development,

edited by J.0. Browder, 166-181. Boulder, Colorado: Westview

Press.

Lima, R.R., Trassato, L.C. & Coelho, V. 1986. 0 tucume

(Astrocaryum vulgare Mart.) principals caracteristicas epotencialidade agroindustrial. (Boletim de Pesquisa, 75.) Belem:

EMBRAPA-CPATU.

May, P.H. 1990a. Local product markets for babagu andagroindustrial change in Maranhao, Brazil. In: New directions

in the study of plants and people, edited by G.T. Prance &

M.J. Balick, 92-102. (Advances in Economic Botany 8.) NewYork: New York Botanical Garden.

May, P.H. 1990b. A tragedy of the non-commons: recentdevelopments in the babagu palm based industries inMaranhão, Brazil. Forest, Trees and People's Newsletter 11,23-27,

May, P.R., Anderson, A.B., Frazao, M. & Balick, M.J.1985. Babacu palm in the agroforestry systems in Brazil'smid-north region. Agroforestry Systems, 3, 275-295.

Mori, S.A. & Prance, G.T. 1990. Taxonomy, ecology and

economic botany of the Brazil nut (Bertholletia excelsa Humb.

& Bonpl.: Lecythidaceae). In: New directions in the study of

plants and people, edited by G.T. Prance & M.J. Balick,130-150. (Advances in Economic Botany 8.) New York: New

York Botanical Garden.

Nelson, B.W., Absy, M.L., Barbosa, E.M. & Prance,G.T. 1986. Observations on flower visitors to Bertholletia

excelsa H. & B. and Couratari tenuicarpa(Lecythidaceae). Acta Amazonica, Suppl., 15 (1-2), 225-234.

Nepstad, D.C., Foster Brown, I., Luz, L., Alechandre,A. & Viana, V. 1992. Biotic impoverishment of Amazonian

forests by rubber tappers, loggers and cattle ranchers. In:Non-timber products from tropical forests: evaluation of a

conservation and development strategy, edited by D.C. Nepstad& S. Schwartzman, 1-14. (Advances in Economic Botany 9.)

New York: New York Botanical Garden.

Padoch, C. 1988. Aguaje (Mauritia flexuosa) in the economy

of Iquitos, Peru. In: The palm - tree of life, edited by M.J.

Balick, 214-224. (Advances in Economic Botany 6.) New York:New York Botanical Garden.

Padoch, C. 1990. Santa Rosa: The impact of the forestproducts trade on an Amazonian place and population. In:New directions in the study of plants and people, edited by

G.T. Prance & M.J. Balick, 151-158. (Advances in EconomicBotany 8.) New York: New York Botanical Garden.

Padoch, C. & de Jong, W. 1989. Production and profit in

agroforestry: an example from the Peruvian Amazon. In:Fragile lands of Latin America: strategies for sustainabledevelopment, edited by J.0. Browder, 102-112. Boulder,Colorado: Westview Press.

Padoch, C., Chota Inuma, J., de Jong, W. & Unruh, J.1985. Amazonian agroforestry: a market-oriented system inPeru. Agroforestry Systems, 3, 47-58.

Peters, C.M. 1990. Population ecology and management of

forest fruit trees in Peruvian Arnazonia. In: Alternatives todeforestation: steps toward sustainable use of the Amazonian

rainforest, edited by A.B. Anderson, 86-98. New York:Columbia University'Press.

Peters, C.M. & Hammond, E.J. 1990. Fruits from theflooded forests of Peruvian Amazonia: yield estimates fornatural populations of three primary species. In: New

directions in the study of plants and people, edited by G.TPrance & M.J. Balick, 159-176. (Advances in Economic Botany

8.) New York: New York Botanical Garden.

Peters, C.M. & Vasquez, A. 1987. Estudios ecologicos deCamu-camu (Myrciaria dubia). I. Producción de frutos en

poblaciones naturales. Acta Amazonica, 16/17, 161-173.

Peters, C.M., Gentry, A.H. & Mendelsohn, R. 1989.Valuation of an Amazonian rainforest. Nature, London, 339,655-656.

Peters, C.M., Balick, M.J., Kahn, F. & Anderson, A.B.1989. Oligarchic forests of economic plants in Amazonia:utilization and conservation of an important tropical resource.

Conservation Biology, 3, 341-361.

Pinedo, P.M., Rarnirez, N.F. & Blasco, L.M. 1981. Notas

preliminares sobre el araza (Eugenia stipitata). Fruto nativo de

la Amazonia Peruana. (Publ. Disc., 229.) Lima: MAA-INTIA &

HCA.

Pinedo-Visquez, M., Zarin, D., jipp, P. & Chota-Inuna,J. 1990. Use values of tree species in a communal forestreserve in northeast Peru. Conservation Biology, 4, 405-416.

Pinheiro, C.V. & Balick, M.J. 1987. Brazilian palms.

Contributions from The New York Botanical Garden, 17, 25-26.

Plies, J.M. & Prance, G.T. 1977. The Amazon forest: anatural heritage to be preserved. In: Extinction is forever,

edited by G.T Prance & TS. Elias, 158-194. New York: New

York Botanical Garden.

13

Posey, D.A. 1982. The keepers of the forest. Garden, 6 (1),18-24.

Posey, D.A. 1983. Indigenous ecological knowledge anddevelopment of the Amazon. In: Dilemma of Amazondevelopment, edited by E. Moran, 225-257, Boulder, Colorado:Westview Press.

Posey, D.A. 1984. A preliminary report on secondary forestmanagement by the Kayapo Indians of the Brazilian Amazon.In: Ethnobotany in the neotropics, edited by G.T Prance & J.Kallunki, 112-126. (Advances in Economic Botany 1.) NewYork: New York Botanical Garden.

Posey, D.A. 1985. Native and indigenous guidelines for newAmazonian development strategies: understanding biologicaldiversity through ethnoecology In: Change in the AmazonBasin, 1. Man's impact on forests and rivers, edited by J.Hemming, 156-181. Manchester: Manchester University Press.

Prance, G.T. 1973. New and interesting Chrysobalanaceaefrom Amazonia. Acta Amazonica, 2 (1), 7-16.

Prance, G.T. 1985. The pollination of Amazonian plants. In:Amazonia, edited by G.T Prance & TE. Lovejoy 166-191. (Keyenvironments.) Oxford: Pergamon.

Prance, G.T. 1988. Manual de botânica econdmica doMaranh5o. São Luis: Grafica Universitaria, UF'MNCORSUP

Prance, G.T. 1989a. Economic prospects from tropicalrainforest ethnobotany In: Fragile lands of Latin America:strategies for sustainable development, edited by J.0. Browder,61-74. Boulder, Colorado: Westview Press.

Prance, G.T., ed. 1989b. Viihite gold: the diaries of anAmazon rubber tapper Oracle, Arizona: Synergetic Press.

Prance, G.T. 1990. The genus Caryocar L. (Caryocaraceae):an underexploited tropical resource. In: New directions in thestudy of plants and people, edited by G.T Prance & M.J. •Balick, 177-188. (Advances in Economic Botany 8.) New York:New York Botanical Garden.

Prance, G.T. & Mori, S.A. 1979. Theactinomorphic-flowered New World Lecythidaceae.Lecythidaceae Part I. Flora Neotropica, 21, 1-270.

Prance, G.T. & Schaller, G.B. 1982. Preliminary study ofsome vegetation types of the Pantanal, Mato GIOSSO, Brazil.Brittonia, 34, 228-251.

Prance, G.T., Rodrigues, W.A. & da Silva, M.F. 1976.Inventårio florestal de uma hectare de mata de terra firmekm 30 Estrada Manaus-Itacoatiara. Acta Amazomca, 6, 9-35.

Prance, G.T., Balée, W., Boom, B.M. & Carneiro, R.C.1987. Quantitative ethnobotany and the case for conservationin Amazonia. Conservation Biology, 1, 296-310.

Salafsky, N., Dugelby, B.L. & Terborgh, J.W. 1993. Canextractive reserves save the rain forest? Conservation Biology7, 39-52.

Salati, E. & Vose, P.B. 1984. Amazon Basin a system inequilibrium. Science, 225, 129-138.

Salati, E., Vose, P.B. & Lovejoy, T.E. 1986. Amazon rainfall,potential effects of deforestation and plans for future research.In: Tropical forests and world atmosphere, edited by G.T.Prance, 61-74. Boulder, Colorado: Westview Press.

Schwartzm.an, S. 1989. Extractive reserves: the nibbertappers' strategy for sustainable use of the Amazonianrainforest. In: Fragile lands of Latin America: strategies forsustainable development, edited by J.0. Browder, 150-165.Boulder, Colorado: Westview Press.

Torres, B. & Martine, G. 1991. Amazonian extractivism:prospects and pitfalls. (Documento de Trabalho, 5.) Brasilia:Instituto Sociedade, População e Natureza.

Vasquez, R. & Gentry, A.B. 1989. Use and misuse of forest

14

harvested fruits in the Iquitos area. Conservation Biology, 3,350-361.

Vincent, J.R. 1992. The tropical timber trade and sustainabledevelopment. Science, 256, 1651-1655.

Weinstein, B. 1983. The Amazon rubber boom 1850-1920.Stanford: Stanford University Press.

Whitmore, T.C., Peralta, R. & Brown, K. 1985. Totalspecies count in a Costa Rican tropical rain forest. Journal of

)Tropical Ecology 1, 375-378.

APPENDIX

Some non-timber forest products from Amazoniawhich should be considered for futuredomestication, and which are not alreadymentioned in text in detail, are described here.

Aragaiboi - Eugenia stipitata McVaugh(Myrtaceae)

A shrub that produces a most delicious juicyfruit with a tasty acidic juice that is mostly usedfor drinks and ice-creams. This species iscultivated in western Amazonia especially aroundIquitos, Peru, and has already undergone somedomestication. Field trials near Iquitos haveshown that it is highly productive (C Padoch,personal communication). (Details about araga-boiin Chavez & Clement 1983 and Pinedo, Ramirez& Blasco 1981.)

Castanha de cutia - Acioa edulis Prance(Chrysobalanaceae)

This large tree of central Amazonia producesabundant large ovoid fruits, the kernels of whichare used by local people to extract an oil usedfor both coolcing and soap-making. The kernelsare also nut-like to eat raw The mature fruits fallfrom the tree and consequently can be easilygathered. (Details in Prance 1973.)

Castanha de galinha - Couepia longipendula(Chrysobalanaceae)

This widespread tree has a flagelliflorousinflorescence and the fruits hang down from thecrown on long peduncles. When it is cultivatedout of the forest in open light, it forms a smallbushy tree from which the fruits can easily beharvested. The kernels contain a useful oil forcooldng or lighting. It was formerly cultivatedaround Manaus, but is no longer used and is anexcellent potential oil plant.

Curupira - Curupira tefeensis G.A.Black(Olacaceae)

This large tree from the same region as castanhade cutia, around Tefé in Amazonas, Brazil, isanother excellent source of oil that has beenmuch used by local people and has greatpotential for further domestication.

Sacha mangua - Grias peruviana Miers(Lecythidaceae)

This tree of the seasonally flooded forests ofAmazonian Peru has an avocado-like fruit that isborne on the trunk and branches. The fruits arerich in vitamin A and are much collected forsale in the market of Iquitos. Large monospecificstands occur in some places; the species isalready much cultivated in gardens. This is a fruitof great potential. (Details in Peters & Hammond1990.)

Sapota - Matisia cordata Hurnb. & Bonpl.(Bombacaceae)

A large tree of the western Amazonian rainforest,it is already semi-domesticated and produces alarge yellow fruit on the trunk and branches witha thick tasty pulp. This is one of the unsungfruits of the Amazon but deserves much greaterattention. Its local name causes confusion withmembers of the Sapotaceae, the sapodillas, withwhich it should not be confused because it isquite a different fruit. (Some details in Hodge1960.)

Sorva - Couma utilis Muell. Arg. - Coumamacrocarpa Barb. Rodr. (Apocynaceae)

These species produce both a delicious fruit anda latex that has been much used by thechewing gum industry Unfortunately until recentlythe trees have been felled to collect the latex, somuch genetic material has been lost and thespecies has become scarce in the forest. Sorvais a fast-growing tree that reaches 3 m in height,and begins to produce fruit three years afterplanting. It also grows well on poor soil. The treehas potential to be domesticated both for fruitand for latex. In 1985 Arnazonas exported overUS$1 million worth of sorva latex. (Details inArkcoll 1986.)

Tagua - Phytelephas macrocarpa Ruiz &Pavon (Arecaceae)

The fruit of this palm produces vegetable ivorywhich can be carved and used as a substitutefor ivory It was formerly an important crop inwestern Amazonia but was largely replaced byplastics. Recent efforts to revive the use of taguaby Conservation International are succeeding. If a

sufficient suppiy of the product is to be providedto the button industry further domestication isneeded.

Titara - Desmoncus spp. (Arecaceae)

This is the only genus of South American palmsto form vines like the rattans of Asia and Africa.These spinous vines can be used for cane as anexcellent substitute for rattan. The split cane isbeing used for chair bottoms in some places. Sofar, harvesting is entirely from the wild. Being avine which needs the support of trees, thisspecies lends itself to use in mixed croppingsystems. (Details in Pinheiro & Balick 1987.)

Tucuma - Astrocaryum vulgare Mart.(Arecaceae)

The various species of Astrocaryum are onegroup of the many palms of greater potential fordomestication in Amazonia as a source of fruitand of fibre from the leaf epidermis. The fruithas an edible mesocarp that is rich in vitamin Aand a kernel that produces an excellent oil forcooking and soap-making. The plants areabundant, especially in disturbed areas, and so itcould easily be domesticated. (Details in Lima,Trassato & Coelho 1986.)

Uvilla - Pourouma cecropiifolia Mart.(Cecropiaceae)

The fruits of the fast-growing short-lived treehave often been called 'grape of the Amazon'.They are grape-like, tasty and juicy This speciesis already an important crop of the Bora Indiansand in various agroforestry systems ofAmazonian Peru. This is a plant that is most

promising and should be one of the highestpriorities for future work on domestication.

Bacuri - Platonia insignis Mart. (Clusiaceae)

One of the most popular fruits of Para, Brazil,that is little laiown elsewhere. The fleshy pulpmakes a delicious sweet or ice-cream. It is atree of secondary forest that easily colonisescut-over areas and is therefore cultivatedextremely easily. The related bacupari (Rheediamacrophylla (Mart.) Pl. & Triam) is also a tastyfruit with considerable potential for domestication.

15

Designing commercially promising tropicaltimber species

J R PalmerDirector Tropical Forestry and Computing Ltd, 93 Gidley Way, Horspath, Oxford 0X9 1TQ, UK

ABSTRACTDomestication of tropical timber trees would be accelerated by establishing a more rational framework for research and

development. The 'ideotype' concept allows for information to be synthesised frOm a multilevel approach to research with five

major phases:

i. selection of end use characteristics and their ranking in order of priority;

ranking of the morphological characteristics of the tree which would give rise to the greater yield (quantity) or highest

priority end uses (quality), ie selection for a high 'harvest index';

identification of the physiological processes which affect the preferred morphological characteristics;

iv determination of the biochemical reactions which drive the physiological processes; and

v. identification of the gene sequences which control the preferred biochemical reactions.

When considering 'ideotype', thought has to be given to the size of the desirable harvested units. Some technologically promising

species of tropical timber trees have adequate growth rates and can provide adequate financial returns because of the high

value of pieces of small dimensions, which might be secondary products processed from the residues of the primary harvest.

INTRODUCTION

The idea of domestication is closely linked to theidea of selection, of fitness for purpose, ofpushing nature into a higher gear and in aparticular direction. Surely square trees in Swisscommunal selection forests would have given aneven higher yield of violin backs and doublebass fronts? Field foresters in every country arefascinated by the variety of breeds of Europeanlivestock, and the possibilities of introducing newstock, of cross-breeding, and the advantages anddisadvantages which result.

Although there is much justified concern at theloss of traditional crop varieties and ancientbreeds of domestic animals, we should not forgetthat production of new varieties and breeds isbeing undertaken in a more deliberate andinformed manner and at a greater pace thanperhaps ever before.

In this paper, the reasons for domesticating wildbreeds will be considered in relation to the designof trees for the specific product, timber. At present,our state of knowledge and development about thedomestication of promising tropical timber trees isprimitive. Is the domestication of tropical timbertrees worth the effort?

WHY DOMESTICATE?

It is not hard to explain the great number ofvarieties of the staple food crops which havebeen developed over the past 20 000 growing

16

seasons. Selection of desirable cultivars hasoccurred progressively from early cultivation ofpreferred batches of seed to modern breedingbacked by high technology

The number of palatable edible crop species is,however, limited. This limitation itself providessome pressure for selection and breeding of newvarieties, but the situation of tree species isdifferent. For those tree species whose productswe eat (leaves, flowers, fruits and nuts), theprocess of domestication has been quite rapid.There are national collections of the varieties oftemperate fruit trees: the International Board forPlant Genetic Resources (IBPGR) has produceddirectories to some of these collections, as wellas standardised descriptors for some fruit andnut species (Gulcan 1985; IBPGR 1986). Withmost horticultural crops, selection for precociousflowering may result in fertile seed in less thanfive years. The possibilities for breeding newvarieties of such crops are obviously morefavourable than for those forest tree specieswhich may take one or two decades to producefertile seed without selection for sexual precocity

Forest trees with non-edible or unpalatableproducts have been subject to much lessdomestication. Even the relatively weak senses oftaste and smell of urban people are morepowerful discriminating factors in driving thedomestication process than are preferences forparticular characteristics of wood or timber.

However, all long-established human communitieshave preferred particular forest species forparticular end uses, mostly on the basis ofgenerations of trial and enor. Why has thispreference not induced more domestication? Theexplanation may include the following reasons.

• Preferences are relatively weak comparedwith factors of availability and price, manytimbers being at least partly interchangeablefor many purposes.

• The end use characteristics of wood requiredin large volumes are tolerant of the variationbetween generally similar species.

• The market volumes are quite small whenend use characteristics are narrowly defined,and it may be more economic to search forparticular species in the natural forest, and topay the higher costs of harvesting, than toinvest in domestication.

Whether domestication is worthwhile, and howmuch domestication is desirable, is thus aneconomic question. The question may becomplicated by risk factors, which include therelatively long rotations of timber trees anduncertainty over the legal rights to products. Ifthe government of a nation/state claimssubstantial rights over all trees, whether wild ordomesticated, the incentives are small for anyone person or community to engage in eitherthe domestication of trees or the care of forests.

Forces which encourage domestication may begrouped into three classes:

i. diminishing availability of wild stocks, such as:

• physical loss through changes in land use,excessive harvesting, disease and pests, orfire;

• stocks increasingly remote or sparse,becoming more expensive to harvest.

Countering diminished availability involvesnot only the technical matters of conservingbiodiversity and multiplication of stocks;security of land and tree tenure must also beresolved. Loss of stocks may be associatedwith tenurial systems which have failed toadjust to changes in demand and supply offorest products.

unsuitability of wild types for current orforeseen needs;

appreciation of some elements of the rangeof natural diversity

The last two factors lead to a consideration ofthe design of trees.

DESIGN OF TREES FOR PARTICULARPURPOSES: IDEOTYPES

Major improvements in yields of staple cropshave been achieved in part by raising the totalbiomass, but also by increasing the harvest

index: the proportion of humanly useful materialin the plant. We could indeed define'domestication' in part as 'improvement of theharvest index' (Evans 1980).

The idea of a plant model (Donald 1968) wasrefined for trees in meetings of the InternationalUnion of Forestry Research Organizations atEdinburgh (Cannell 1979; Dickrnann 1985). Theconcept is helpful in the domestication processbecause it provides a goal to which the processcan be directed. An 'ideotype' is a plant modelwhich will yield a greater quantity or quality ofuseful products than will conventional cultivars orwild types. It is associated with predictableperformance in a defined environment, whichincludes the management regime as well as thebiophysical site where the plant grows.

The emphasis in the defuttion is on the word'useful'. An ideotype may push selection in adirection quite different from that which naturalprocesses might take. For example, the preferredmorphological characteristics of a forest treespecies planted into a farmer's fenceline mightbe quite different from those which nature wouldselect in the forest (Carmen 1979). The selectionpressures are for utility of product(s) versus 'thegene machine'.

There are perhaps five major phases in thedesign of tree ideotypes:

i. selection of end use characteristics and theirranking in order of priority;

ii ranking of the morphological characteristicsof the tree which would give rise to thegreatest yield (quantity) or highest priorityend uses (quality): ie selection for a highyield and high 'harvest index';

identification of the physiological processeswhich influence the preferred morphologicalcharacters;

iv. determination of the biochemical reactionswhich drive the physiological processes; and

v. identification of the gene sequences whichcontrol the preferred biochemical reactions.

In our present state of ignorance, we can bemuch more confident about the grosser aspectsof the ideotype than about the gene sequencesand biochemistry However, this multilayerednature of the ideotype allows progress to bemade in different fields at different speeds andat different times, all the while contributing to thedomestication of the target species. Moreover,because of the conceptual framework providedby the ideotype, interaction between theories ofimproved yield and practical tests should beimproved.

We know already that some preferredcharacteristics of trees are under strongergenetic control than others. Stem form andbranching characters are often strongly heritable,

17

the presence or absence of spines is usuallycontrolled by a single gene, blit growth vigour isonly weakly heritable. Furthermore, wood yield isa polygenic character and needs to be improvedthrough research on the components.

END USE CHARACTERISTICS

The end use characteristics of timber vary inrelation to the precise end use, so it is notappropriate to suggest a general ranking of the18 characteristics normally applicable to sawtimber. The needs of the end user also have tobe considered. Furniture for a rural school in thehigh Andes is less demanding in timberproperties than reproduction Chippendale chairs.Most users of saw timber, as opposed to usersof poles or pulpwood, would prefer the followingcharacteristics:

• long lengths, with straight, even grain;

• large width and depth, to improve theconversion percentage;

• uniformity throughout the piece;

• minimal differences between sapwood andheartwood, and between earlywood andlatewood in any growing season;

• rapid, easy seasoning, with low shrinkage,distortion and checking;

• naturally durable, or easy to impregnate;

• free of inclusions such as calcite, silica andresin pockets;

• free of damage by insect pests, fungalpathogens and forest fires;

• branch knots absent; or few, well scattered,small, tight, round and at right angles to thegrain;

• no bark inclusions;

• easy to work by machine or hand tools;

• easy to glue with modern adhesives, with noadveise reaction from wood chemicals;

• no adverse reaction (such as staining,swelling or corrosion) to metal fastenings orchemicals;

• no irritant dust or chemicals released onsawing or other working, and no bleeding ofnatural resins or gums over time;

• slow to change colour when exposed tolight, and resistant to photo-deterioration;

• slow rate of burning, with internal protectionby charring of the exterior of a piece;

• easy to match to other pieces of the sametimber species;

• strong for mass.

Most of these characteristics would need to bequalified with respect to any particular end use.A furniture manufacturer might value fine figureand high colour more than uniformity of surfaceappearance. A flooring manufacturer would prefer

18

hard-wearing properties more than freedom fromsilica, and resistance to chemicals spilled inwarehouses more than a constant colour.

For many uses, upper and lower densities andstrengths can be specified, together withacceptable limits of shrinkage. In some countries,these specifications are included in buildingcodes (such as Eurocode 5 for timber structuresin Europe and BS 5268 for the UK timber codeof practice). Developing countries can expectsales resistance against their exported timbers ifthe pieces do not conform to these and otherconsumer protection codes. However, somedeveloping countries or treaty areas can nowprovide at least qualitative indications formatching domestically common timbers to majorend uses.

TREE MORPHOLOGY AND GROSS WOODSTRUCTURE IN RELATION TO TIMBEREND USE

It is necessary to interpret the end usecharacteristics in terms of tree morphology andgross wood structure, as opposed to woodanatomy or fme structure. It is, however,recognised that for specific purposes, particularlythose which involve fine fmishes or appliedsurface films, anatomical characters may need tobe included in the ideotype.

• Long lengths of straight-grained large piecesobviously require long, straight, cylindricalboles developed on monopodial stems.(Incidentally trunks with square cross-sectionswould have much more of the undesirablereaction wood at the corners and thus alower yield or harvest index thanround-sectioned trunks.)

• Differences in properties between sapwoodand heartwood may be associated withprovenance as well as species. Impermeableheartwood is not necessarily naturallydurable, so there may be a preference forwide sapwood. A distinct pith is usuallyundesirable, being associated sometimes withweak fibres and a high content of extractives.

• Differences between earlywood and latewoodmay be amplified by strongly seasonalclimates and other environmental factors.Early flushing and late leaf abscission maycontribute to a cambium whose activity isrelatively constant rather than strongly pulsed.

• Freedom from extractives may aid timberprocessing and end use but may not bebeneficial to the tree. Extractives may help toprotect the tree from attack by insects andfungi.

• Well-scattered, small, tight, round knotstranslate to small branches inserted at rightangles to the stem. If knots are to be few,then branch abscission should be early and

efficient. Presence of few and small knotscould mean a small and thin crown,photosynthetically inefficient, and so opposedto the preference for rapid growth. On theother hand, a narrow, deep, persistent livecrown with branches inserted at an acuteangle to the main stem may be conducive torapid growth but would reduce the yield ofclear wood. If knotty timber is favoured foraesthetic reasons, the narrow and persistentcrown may be quite suitable.

• 'Ilee boles which are cylindrical, free frombuttresses, wings and flutes, and whosebranches are small and held horizontally, areunlikely to give rise to inclusions of bark inthe wood.

• Strength and density characteristics of woodmay be related to speed of growth and tothe degree of stress experienced at thegrowing site, and may be modified by themanagement regime. Wood showing closeand even growth rings has usually beengrown under strong inter-tree competition butalso equable conditions which change littlefrom year to year. However, dense wood isnot invariably associated with slow growth.

So, many components affect the growth rate oftrees, and, consequently many selection strategiesare possible. Uniform growth may be obtainedby combinations of some of the followingsilvicultural operations:

• close spacing of trees;

• frequent light thinning;

• modifying crown size through pruning;

• fertilization;

• irrigation.

The proper management of these operationsmight obviate the need for genetic selection.Alternatively if the components could becontrolled genetically the need for rigorousmanagement might be reduced. Tree architectureis described elsewhere (see Oldeman &Sieben-Binnekamp, pp25-33), so here it willsimply be noted that bud behaviour, modes ofbranching, leaf types, leaf sizes and leaforientations can be affected by these silviculturaloperations and, in turn, these may affect bolegrowth rates through alterations to the rate andamount of photosynthesis (see also Ladipo et al.,pp239-248).

Long-term studies on bud behaviour oftemperate fruit trees, coffee and oil palm at theCIRAD Modelling Laboratory in Montpellier,France, suggest that more attention should bepaid to improving our understanding of theplasticity of growth habits of tropical timber trees.Such studies might explain why some valuabletimber species seem to need a 'forest'environment in order to maintain a monopodial

habit. For example, numerous trials of some ofthe Brazilian furniture wood species, especiallylegumes, in open plantations have producedorchard-like trees instead of the hoped-for longclean boles ( j R Palmer, personal observation).

Studies on the relationships between treearchitecture and wood structure are beingextended to a few tropical tree species, butcould usefully include many others. It would beespecially interesting to examine the woodstructure of contrasted tree forms within a singletaxon. This information would help to determinewhether efforts to pursue greater volumetricyield were adversely affecting the quality of thewood, other than in terms of strength anddensity which have been broadly studied already

In tropical forest timber trees, it is common forthe usable bole to be only half the total treeheight. Could we delay the age or height ofcrown break until the main stem forms a greaterproportion of total tree height? Could theshoot/root ratio be shifted in favour of the shoot?How would this affect early competitiveness ofthe tree? How would it affect resistance to windlater in the tree's life? Many of these questionshave been considered for temperate trees.

PHYSIOLOGICAL PROCESSES WHICHAFFECT THE PREFERREDMORPHOLOGICAL CHARACTERS

Which physiological processes can be controlledor modified, either through the selection ofgenotypes or through silviculture to improve theharvest index? To date, the physiologicalprocesses themselves are rarely selected ormodified directly Control is through selection ofparticular morphological characters, or theirmodification. For example, leaf variation betweensome provenances of Eucalyptus camaldulensiscan be explained in terms of the wind climate inthe regions of 'origin (CSIRO, unpublished data).This is an adaptive response of leaf shape andorientation to minimise the effects of waterdeficits and temperature shock. Armed with thisknowledge, we can now choose provenanceswhich are better adapted to different climaticzones.

We still know little about the physiology oftropical trees beyond the seedling stage, whileinformation about temperate trees is increasingrapidly Much of what is known concerns waterrelations, and the effectiveness of morphologicaladaptations which allow trees to control waterflows. To construct an ideotype for a timber treeon any particular site, it is necessary to know:

• the probability of there being enough waterto sustain the tree at its age or size ofmaximum consumption;

• the nutrient requirements over the technicalrotation of the tree, the corresponding

19

nutrient availability and what the tree doeswith the nutrients;

• the size and variability of the various nutrientsinks in different species of tropical trees;

• how they change seasonally and as the treeages; and

• how they vary in response to site differencesand management regimes.

The role of root symbionts in water and nutrientrelations of tropical trees also needs to beexplored more thoroughly In some cases, theassociation is critical to the survival andcommercially adequate growth of a tree. Thisaspect and the matching of trees to their lightenvironments are covered elsewhere in thisvolume (see Lapeyrie & Hogberg, pp158-164;Mason & Wilson, pp165-175; Sprent, pp176-182;Fasehun & Grace, pp148-157).

BIOCHEMICAL REACTIONS AND GENESEQUENCES

Descriptive information about genetic variationbetween and within species is increasing rapidlythrough the use of biochemical, cytological andmolecular marker techniques. How thesevariations are functionally related to thephenotypic expression of traits is still little known.The combination of laboratory techniques withfield and herbarium studies has considerablyenlarged our knowledge of infraspecific variationin the few tropical species which have beenexamined so far. These new studies, plusconventional biosystematics and provenance/progeny trials of tropical trees, tend to confirmwhat we know in relation to staple crops and theprincipal breeds of animals: ie there is a wealthof infraspecific genetic variation which we havebarely begun to explore and use.

At present, to locate the ideotypes, we searchthrough the natural population for the mostsuitable phenotypes. In future, we can expectthat trees will be genetically engineered to meetthe specifications of ideotypes (or that geneticmarkers will be used to identify appropriategenotypes: selection based on the genotypedirectly not phenotype).

If this range and depth of studies for thedevelopment of ideotypes seem to bedepressingly formidable, one should rememberthat the framework can be entered at any level.Once the characteristics of the desired endproduct have been defmed, in a sense the restfollows. Because our current knowledge oftropical timber trees is so patchy even for thebetter-knoWn species, I conclude that it would notbe sensible to advocate a common approach tofurther studies on domestication. Priorities willvary according to how much is known alreadyabout a particular species, and what are themost promising lines to follow as indicated by

20

work on related species. The improvement ofcloning techniques for tropical timber trees will,however, allow the capture and use ofcharacterised ideotypes, in the same way as hasoccurred in horticulture.

DOMESTICATION OF TROPICAL TIMBERTREES

How does the sequence outlined above comparewith the way in which tropical timber trees havebeen domesticated so far? The answer is, verylittle. Some of the ideas may have been in theminds of the domesticators, but the procedureswhich have generally been used appear, withhindsight, to have been rather haphazard. Thenow quite numerous manuals on treeimprovement seem to have skipped over thestep of clarifying what the users would want andwhich kind of tree would satisfy those wants.

CASE STUDIES

Teak (Tectona grandis)

With few exceptions, the Tectona which is nowwidely distributed in the tropics and subtropicscomes from poorly documented sources, and itsperformance has been worse than was expected.Tectona was collected and distributed because ofthe good name it acquired from its observedgrowth in some of the natural stands, andbecause of the good performance of some ofthe early introductions as an exotic.Documentation is mostly too fragmentary todetermine whether the early collectors anddistributors of germplasm consciously sought tomatch source provenances to the intendeddestinations.

We now know that Tectona is quite responsive tosoil water and nutrient status and to climate(Kadambi 1972). Trees of the Verbenaceae aretenacious survivors, even when grown far fromtheir optimum sites, although stem form may beabysmal and growth may be poor. In someways, it has been a disadvantage that evenpoorly grown and off-site Tectona has usuallyfound a profitable market, thereby discouragingattempts to discover whether other provenancesor management methods might have been moreappropriate. Tectona has been well served bythe range-wide provenance collections andinternational trials organised by the Danish Forest

. Seed Centre, giving some quantitative informationabout genotype/environment interaction (GED inthis species (Keiding, Wellendorf & Lauridsen1986).

American mahogany (Swietenia znacrophylla)

Swietenia is still a much-favoured and prestigiousfurniture wood, being easy to process andpossessing many desirable features. Neotropicalplantations of Swietenia have failed repeatedly

mostly because of the persistent and cripplingattacks of the pyralid shoot borer Hypsipylagrandella (see Newton et al., pp256-266).

After many years of failures, it was noted thatwith the Hypsipyla in the palaeotropics, H.robusta, either it did not attack the Swietenia orthe attacks were much less damaging. As aconsequence of this natural resistance, there arenow extensive plantations in Java and lesserareas in Fiji and Sri Lanka.

There have been few attempts, perhaps only one,to make even a partial collection of Swietenia •provenances. The plots managed by the Instituteof Tropical Forestry in Puerto Rico probablyrepresent the only source of some germplasm,as the SMetenia-bearing forest at some originalsites in Central America has now beendestroyed. Because of the lack of genetic testing,only fragments of observations about GEI existfor Swietenia. This lack of modern research iscurious, in view of the value of the world tradein Swietenia over nearly five centuries and afterthe discovery that the shoot borer could becontrolled by systemic insecticides.

WHY HAVE FEW TROPICAL TREES BEENDOMESTICATED?

The early stages of the domestication of mostnon-food tropical trees seem to have beenstimulated either by a great shortage of wood ofany kind in a particular area, or by difficulties inregenerating the commercial species. Mostattempts to introduce woody species seem tohave started in a rather shotgun manner, usingseed and cuttings of any species which could beobtained. Some of the more advanced nationalforestry services then progressed to homoclinalclimatic matching and a phased succession ofelimination trials.

Where capital-intensive forest product industriesprovided a demand, domestication was morefocused, with a clear idea of the technicalproperties required in the bark (for tanning) orwood (mostly for pulp) from the trees. Incontrast, most timber-using primary processors offorest products in the tropics lack sufficientknowledge of the end use requirements of thetimber users, and of the pricing structures of theretail markets, to provide incentives for thedomestication of the species which they use.Outside the large-scale afforestation schemesdriven by commercial demand, it is rare to fmdany suggestion that the foresters had a particularproduct in mind, perhaps because of thedifficulty in obtaining adequate survival andgrowth of any species. Problems have arisenfrom a lack of knowledge about the degree ofspecies/symbiont plasticity in relation to siteconditions. Foresters have also tended tounderestimate the depletion of the sites which

they were planting, after several to many years ofinappropriate land use by others.

Tree species which have orthodox seed, areeconomical with nutrients and water, and grownaturally on old land surfaces in soils depletedover geological time and in harsh climates havegenerally been much more successful as speciesfor reforestation and land rehabilitation, beingamenable to seed storage and having lessstringent site requirements. However, now that wehave much improved knowledge about seedhandling and nursery techniques (includingvegetative propagation) and establishmentmethods, it can be argued that we can be moreselective about what we try to domesticate.Attention to ideotypes would provide a focus andenable us to concentrate the effectiveness of ourlimited resources for research.

One factor deterring foresters from plantingeither slower-growing quality hardwoods or thosewhich are not robust in the establishment phasehas been the small but vocal group of 'wholetree, any tree' advocates. This group takes theview that, on the one hand, wood fibre is notper se a scarce commodity in many countriesand, on the other hand, processing technology isnow capable of taking almost any fibre andreproducing or even improving upon manynatural products. For example, knowledge ofstructural engineering and wood strengths allowssubstantial houses to be built fromsmall-dimensioned sawn timber; tropicalexamples date back at least 30 years. Glulambeams can span larger spaces than natural woodbeams, with much less use of fibre and morepredictable strength properties. In Australia, the'scrimber' process, in principle, allows almost anysize of tree to be formed into a structuralmember with known engineering properties;unfortunately it is not yet economical to processnatural fibre in this way (A G Brown, CSIRO,personal communication).

Proponents of the 'whole tree, any tree'philosophy argue that processing technology isdeveloping much faster than trees grow. Theycan point to the shifts in demand which havetaken place in less than one timber tree rotationin peninsular Malaysia, from naturally durableheavy hardwoods to medium hardwoods to lighthardwoods, as national lifestyles have changedand industries have developed.

Because we can already simulate the appearanceof timber of many species, and improve uponnatural fmishes, the 'whole tree, any tree' grouponly advocates domestication of species whichare amenable to mass-production processing. Anindustrially bland species, such as Gmelinaarborea, is thus more useful as a substrate foroverlays, than timbers which are less amenableto industrial processing but more valuable forparticular purposes.

21

There are two main contrary arguments.

i. A market exists in industrialised countries forfmely fmished natural products, includingtimbers, for which high prices may be paid.

The 'whole tree, any tree' processingtechnology is usually far from cheap, and sois unlikely to substitute for less traditionalsawmills in developing countries.

WHERE ARE DOMESTICATED TREES MOSTNEEDED?

We can predict future needs for reforestation onlyin very general terms, and thus the dangers ofover-specifying the ideotypes must berecognised. We can hazard guesses at thealtered growing conditions which climate changesmight induce. We can be sure that insect pestsand fungal pathogens will challenge whatever wedomesticate, but we do not know what pest willattack which species, where or when, or whatdamage might be caused (Cannell 1979).

Demographic pressure alone will tend to restrictlarge-scale commercial/industrial bulk plantingsmore and more to marginal sites. These sitesmay be steep, rocky saline, sodic, droughtywindy poorly drained, or otherwise unsuitable forfanning. The sites may have been fertileoriginally but are now depleted through poorfarming practices. A variety of ideotypes may beneeded to suit these sites, each with particularphysiological adaptations.

However, as forest masses decline, trees areperhaps more likely to be grown deliberately onfarms, as single trees, in clumps and smallwoodlots. This perhaps is a better scenario fordomesticated trees, but timber species which aresite-demanding are candidates for domesticationonly if they cause minimal loss of arable croparea or yield. These considerations lead to threedistinct ideotypes.

i. The 'crop' ideotype is the model forcommercial plantations and smallholderwoodlots. Individual trees woilld be weakcompetitors with each other, but would beefficient exploiters of their own growingspace and better adapted to the range ofdegraded sites described above. Live crownswould be deep, dense and narrow. Stemgrowth would be strongly monopodial. Standswould be uniform in appearance, with anarrow range of diameters. Plantingspacement would need to be calculatedcarefully so that resources in the growingspace did not become limiting before thefirst commercial harvest. If sites weremarginal, probably with soils depleted byimproper use, initial spacing might need tobe wide in order to give the trees enoughexploitable nutrient-bearing soil to sustaingrowth. However, canopy closure should be

22

sufficiently rapid to keep weeding costsbearable.

The lenceline' ideotype is for the arablefarm. Individual trees would have deep roots,both to provide wind-firmness and fornutrient and water capture without competingwith the arable crop. Crowns would be thinand open, to avoid shading of the crop.Fastigiate branching would probably beundesirable because of the reduction intimber quality Strong monopodial growthwould carry the crown rapidly above thecrop. Ree species would not be allelopathicto crops. Water requirements would be low Aheavy leaf litter, which could be used asmulch, might be more beneficial thannitrogen-fixing ability especially if therequirement for deep rooting is contrary tothe conditions for abundant N fixation.

The 'shade-tree' ideotype can be envisaged,also for use on-farm. It would be mostsuitable for taungya-type agroforestry Farmersmight tolerate a moderate-to-heavy crownedtimber species standing over the arable crop,if the timber was of high value and if heightgrowth was sufficient to restrict crop-reducingshading to the first year. Paulovw2ia tomentosaover wheat in the People's Republic of Chinais a well-known example. Less well known isS. macrophylla over maize in the farms of theKetchi Maya in southern Belize. Moregenerally applicable might be agroforestrysystems of small numbers of timber treesper hectare, standing over perennial treecrops such as coffee and cacao and spices.Many indigenous polycultures of this kindhave been documented.

Just as proponents of clonal plantings recognisethe need for security by the use of severalclones in any one planting area, so forestersmight prefer to hedge their bets by usingphenotypically plastic genotypes. Such treescould modify their physiology and morphologyaccording to their spatially and temporallychanging environments. Detection of suchplasticity would require extensive testing ofgenotypes on a range of sites and under arange of management regimes. The chosenideotype could, of course, be aided byappropriate silvicultural methods, but for on-farmtrees it would be better to assume that farmercare would be minimal.

WHY HAS LITTLE EFFORT BEEN PUT INTODOMESTICATING TREES?

I suggest that one reason for the paucity ofeffort to domesticate tropical timber trees is therelative abundance, until recently of nearlyinterchangeable timbers. Some 120 species orspecies groups are used domestically in the

Amazon basin, perhaps 600 in peninsularMalaysia, almost all of them by harvesting fromthe wild state because it is perceived to beeconomical. As I said earlier, domestication is aneconomic decision, and it has not seemedworthwhile in most countries until very recently

Since colonial times, governments have laid claimto ultimate ownership of forests, whether or notthey were able to exert that authority effectivelyFor a variety of reasons, governments havegenerally treated forests as a residual use of theland and have made no serious attempts tovalue the goods and services of the forest in aholistic manner. Some governments continue tosubsidise harvesting of trees from the wild,perhaps because of a lack of understanding ofthe total costs which they are incurring for thenation.

Increasingly urbanised populations need more,not less, timber per head than rural populations,so domestic demand in developing countries isrising while natural forest resources are generallydeclining.

Legislation, national accounting and socialattitudes in many countries have failed to keeppace with the rapidly changing situation oversupply of and demand for timber (and manyother forest products). Thus, the law, economicsand society appear to conspire against thedomestication of tropical timbers. Wheregovernments have taken firm and soundly basedsteps to stimulate afforestation and reforestation(as in Brazil formerly and Costa Rica presently),the demand for suitable germplasm has greatlyexceeded the national capacity to supplydomesticated trees, at least in the short term.The absence of current incentives for thelarge-scale use of domesticated germplasmshould not lull national forestry research servicesinto believing that research is not needed.

IS DOMESTICATION OF TROPICAL TREESWORTH THE EFFORT?

There are perhaps three questions which have tobe asked to determine whether domestication isworthwhile.

Are there naturally occurring genotypeswhich correspond to the timber ideotypes?

Extrapolating from our laiowledge of geneticvariation in some temperate timber trees andtropical multipurpose trees whose geographicdistributions have been well explored, it seemslikely that satisfactory timber genotypes could belocated if the same effort were to be deployed.We can be fairly sure that simple silviculturaltechniques, such as mass selection of superiorseedlings and vegetative propagation of goodphenotypes, will boost both stem form andgrowth rate.

Are tree growth rates fmancially attractive?

Rapid growth is a necessary part of theideotype, because domestication is a processrequiring investment and the cost must berecovered. Although slow growth in thepost-establishment phase has sometimes beendue to neglect of silviculture (eg the stagnatedplantations of 7brminalia ivorensis in Ghanaduring the years of national economic crisis),problems have more usually occurred because ofunsuitable germplasm, poor nursery techniquesand off-site planting. There are enough examplesof excellent growth of trees in arboreta and trialplots to indicate that timber tree growth ratescan be quite adequate.

The idea of timber trees on farms is intrinsicallyattractive, as a biological bank deposit againstfuture needs. When silvicultural costs are low,harvesting costs become important. Single treeson farms may be more costly to harvest than ablock planting on a remote hillside. On the otherhand, transport costs may be lower because thefarm is likely to be closer to the timberprocessing plant.

Is the price adequate for the final product?Or, rather, will the price be right when thetrees are ready for harvest, and will theprices justify the cost of domestication?

Prices for standing trees in developing countriesare often conditioned in part by governmentfiscal measures which bear little or no relation tothe costs of forest or tree management or to thecosts of harvesting, transport and processing.Most countries have shelves of reports fromconsultancies over many years which recommendmore rational systems of valuing forest goodsand services. Timber growing, even withdomesticated germplasm, is a long-term businessand needs both incentives and assurances ofsecurity to encourage investors.

Even under the present, generally unfavourable,fiscal conditions, the continued willingness of aclientele to pay high prices for finely fmishedproducts from natural timber should provide itsown incentive (Johnson 1988). The high value ofmahogany products, including small-dimensioneditems (Past Times 1991), illustrates this point: acabinet priced at £10,220 (Prince George'sWoodcraft 1990), a George II mahoganycommode, valued at £400,000 (Duthy 1989), anda toilet seat and cover in solid mahogany pricedat £45 (Woodentops 1983).

Finally although this paper has consideredsingle-purpose timber trees, we all recognise thatresidues from timber trees can be used forcharcoal and fuel, at the very least. Demographicand other pressures on renewable naturalresources are rising so steeply in mostdeveloping countries that single-purpose land

23

uses are almost impossible to defend in socialterms. For multipurpose use (eg the productionof timber and fodder), it may be necessary todomesticate single-purpose lines. This aspect isconsidered later in this volume (see Felker,pp183-188; Owino, Oduol & Esegu, pp205-209;Simons, Mac Queen & Stewart, pp91-102).

ACKNOWLEDGEMENTS

The ideas in this paper were provoked byCannell (1979) and by drafts by P Wood andP Huxley in the mid-1980s for thenot-yet-completed ICRAF manual on research formultipurpose tree species in agroforestry Mythanks are due to the editors for their valiantconversion of a presentation to a paper.

REFERENCESCannell, M.G.R. 1979. Biological opportunities for geneticimprovement in forest productivity In: The ecology ofeven-aged forest plantations, edited by ED. Ford, D.C. Malcolm& J. Atterson, 119-144. Cambridge: Institute of TerrestrialEcology

Dickmann, DI. 1985. The ideotype concept applied toforest trees. In: Attributes of trees as crop plants, edited byM.GR. Connell & J.E. Jackson, 89-101. Abbots Ripton,Huntingdon: Institute of Terrestrial Ecology

Donald, C.M. 1968. The breeding of crop ideotypes.Euphytica, 17, 385-403.

24

Duthy, R. 1989. Furniture values climb. The Field, 271(7015), 39.

Evans, L.T. 1980. The natural history of crop yield. AmericanScientist, 68, 388-397.

Gulcan, R., ed. 1985. Descriptor list for almond (Prunusamygdalus). (Document no. AGPGIBPGR/85/36.) Rome:International Board for Plant Genetic Resources.

International Board for Plant Genetic Resources. 1986.Genetic resources of tropical and sub-tropical fruits and nuts(excluding Musa). Rome: IBPGR.

Johnson, N. 1988. Be one of the first to have a fitted officelike this. (Advertisement in many UK colour magazines/newspaper supplements.)

Kadarnbi, K. 1972. Silviculture and management of teak.(Bulletin 24.) Nagadoches, Texas: School of Forestry Stephen FAustin State University

Keiding, H., Wellendorf, H. & Lauridsen, E.B. 1986.Evaluation of an international series of teak provenance trials.Humlebaek: Danish Forest Seed Centre.

Past Times. 1991. A collection of fine and unusual gifts withan authentic feel of the past. Abingdon: Past Times. (Salescatalogue.)

Prince George's Woodcraft. 1990. This magnificentmahogany cabinetry is just $18,400. Washington: Museum ofArts. (Advertisement.)

Read, M. 1990. Lifting the lid on mahogany BBC Wildlife, 8(3), 160-163.

Woodentops Ltd. 1983. Bring back the warmth and naturalquality of wood. The Field, 28 March. (Advertisement.)

Timber trees: architecture and ecology

R A A Oldeman1 & A H M Sieben-Binnekamp 2

1Generaal Foulkesweg 76, 6703 BX Wageningen, The Netherlands2Samuel Mallerplein 11, 3023 SK Rotterdam, The Netherlands

ABSTRACTIn growing timber trees, life-long architectural adjustment of trees to their direct forest environment is topical. Hierarchical systems

analysis of trees examines their architecture in terms of architectural models, multiple sets of such models (metamorphosis and

reiteration), and the occurrence of specialised light-gathering shoots (phyllomorphism) and shoots which cease terminal activity

and thus branch (unstable monopodia). These different crown-forming processes can come together in a way that develops

branch complexes that are hierarchical, with one set dominating others, or polyarchical, where the complexes are repetitive and

co-dominate. Examination of these different levels of tree architecture allows one to zoom in on trunk formation, the basis of

silvicultural management for timber production. Tree architecture differs between tree groups (ie conifers, legumes, Fagaceae,

dipterocarps, Myrtaceae and 'miscellaneous'), determining the nature of forests in whole regions of the world. Hierarchical systems

analysis can also define the direct tree environment by zooming in from landscapes, to silvatic mosaics, to the conditions around

an individual tree (eco-unit level), determined by ecological interference. Domesticating and growing trees for wood are finally

examined in terms of architectural strategies. Silviculture should be overhauled and become more ecologically orientated, its

present combination of aims - to produce large timber volumes as straight stems of high-quality wood, in the shortest time -

being paradoxical.

INTRODUCTION

Trees perform a major feat of natural, biologicalengineering: their architecture continually adaptsto their changing environment. Every treeresponds to a temporal sequence ofenvironmental conditions as it grows larger (seeOldeman 1989a for an example of Acaciaspecies in the rather unpredictable East Africanclimate). This environmental sequence can beroughly predicted. It is not always true, however,that the seedling environment always differs fromthe surroundings of a fully grown tree crown,even for pioneer tree species. An alternativeview is that the tree crosses a spatial sequenceof environments from the soil to its adult canopyheight. However, this assumes the forestenvironment to be constant over time, which is avery rare condition (Oldeman 1989b).

This paper • concerns timber trees, which arethose that' form long and straight trunks, suitablefor extraction as sawlogs. One importantsilvicultural question, relevant to the process ofdomestication and the 'ideotype' concept (seePalmer, pp 16-24), is how to grow trees to thesespecifications. The answer is dependent on ourunderstanding of tree architecture, and theprocess of trunk formation in particular.

The long, straight stems required bysilviculturalists most often occur in natural forests.Only a few tree species maintain this form whengrown in the open. Indeed, timber plantations aregrown to mimic natural forests in this respect.

Both natural forest and plantation ecosystems arecomplex and alive, with the canopies of timbertrees being forced aloft by the living forestsurrounding them. Tree trunks may therefore, beseen as the trees' tools of ascent towards thelight available at canopy level.

The term 'environment' was used loosely above.In reality it covets several scales in the hierarchyof living systems. On the one hand, a 'forestenvironment' refers to the forest ecosystem, whichaccording to Oldeman (1990) exists on at leastthree nested scales, ie the eco-unit ('forest patch','gap'), the silvatic mosaic (the mosaic composedof interacting eco-units), and the site mosaic(composed of interacting silvatic mosaicsoccurring in a set of different sites). On theother hand, the 'environment' forcing a tree aloftis a subsystem of any 'forest environment', ie thatpart which directly surrounds an individual tree.Only this direct environment can fully explain theselection pressures on tree growth and theprocess of adaptation (Oldeman 1990, p513).

In the following sections, this paper will consider:

• an outline of tree architecture, its hierarchicalorganisation and the question of conformityversus flexibility;

• the trunk as an architectural component ofthe main forest-building trees, trunk size andstability and ways to favour the formation ofstraight trunks in timber trees;

• the direct environment of a tree versus theaverage stand environment, and architectural

25

strategy as a response to site dynamics; and

• some conclusions for timber tree form,growth rate, potential size and age obtainedin cultivation, in the context of thedomestication process.

TIMBER TREE ARCHITECTURE

Trunks, branches, roots

Every tree builds three interconnected vegetativesystems. These are the green, branchedabove-ground assimilation complex, the rootedbelow-ground assimilation complex, and the trunkas an intermediary transport and co-ordinationcomplex (Oldeman 1974; Ha 116, Oldeman &Tomlinson 1978; Ha 116 1991). Ha 116 and Oldeman(1970, p44) defined the trunk and branches asfollows: 'From the unique axis in unbranchedtrees,- the trunk mainly conserves the architecturalrole: it ensures the tree's foothold, its mechanicalresistance, its extension growth in height throughthe successive forest storeys, while producing atits apex new leaf-bearing branches which profitfrom the best light conditions; because itconstitutes a central communication and transportsystem, the trunk gives a tree its physiologicalcoherence; it determines the final stature of thetree on which depends, at least partially itscompetitivity and ecological niche . . . Thebranches in principle take over those functionswhich the trunk ceases to assume: photosynthesisand the formation of sexual organs. Thesefunctions can be fulfilled in many ways, andnumerous setups indeed exist both to captureincident light energy and to present the sexualorgans to the vectors of gametes'.

Some evidence suggests that the below-groundcomplex is organised not too differently from theabove-ground architecture (Atger 1992). However,no well-developed, large and persistenttrunk-equivalent (tap root) seems to be presentin most timber trees. This paper will refer toroots only in passing.

Above-ground tree architecture

Seedlings are usually unbranched (Figure 1, a).The first branches form a specialised interfacefor photosynthesis, the exchange of gases andwater, and temperature regulation. The peripheralbranches show such a set of properties lifelong(Edelin 1984). Sapling axes are of the first order(trunk) and of the last order (crown periphery)(Figure 1, b,e). The crown is later filled out byintercalation of branch orders (Edelin 1984; Atger1991), until the full inherited number of potentialbranch orders is formed (Figure 1, c,d).

If second- and higher-order axes are more orless equivalent, leaf properties determine thegreen interface between the tree and itsenvironment. Tree architecture then shows a cleardifference between skeletal functions (trunk and

26

main branch axes) and assimilation (leaves), andhence can be described (Figure 1, e-g) by onearchitectural tree model (Hallé & Oldeman 1970;Hallé et al. 1978), or intermediates. If, converselythe peripheral branches differ conspicuously fromthe axes between them and the trunk, botanistsperceive the formation of an intercalary axis asan abrupt change-over from a small crown to alarger crown built by smaller crowns. Edelin(1984) coined the term 'metamorphosis' for thisevent (Figure 1, b-d). The cauliflower-like crownsof SE Asian dipterocarps and neotropicalVochysiaceae are good examples (cf Oldeman &Fundter 1986).

Each tree species has inherited a limitednumber of differentiated branch orders, fivebeing common (C Edelin, personalcommunication). When axes of an order abovethat limit are needed for crown expansion,re-activated meristems yield a first-order axisagain, ie they start crown building over againwith an axis equivalent to the hypocotyl(Figure 1, h). Oldeman (1974) described thisprocess under the name of reiteration, but didnot give this explanation.

Phyllomorphs and unstable monopodia

'Rim concepts are important for understandingthe ways in which the basic architectural modeladjusts to its environment. These are (i)phyllomorphs (Hallé 1967) and (ii) unstablemonopodia (Oldeman 1974).

A phyllomorph is an axis with many foliarproperties, eg small size, limited lifespan, andlittle wood, while having many chlorophyllicappendages or tissues (leaves, phyllodes, etc).Originally phyllomorphism was assumed to beclosely linked to plagiotropy but architecturalanalysis of leguminous trees by Oldeman (1989a)showed that orthotropic short shoots may also bephyllomorphic. It can now be understood that thegreen interface functions of tree crowns areensured either by leaves, often on architecturalmodel-like crowns (Figure 1, c g), or byphyllomorphs, often on metamorphic crowns(Figure 1, b-d).

Figure 1. Principles of timber tree architecture

a-d Tree development with metamorphosis (Edelin 1984,1991); branch order 1 is the trunk, order 5 aperipheral branch (eg twig, phyllomorph), and orders2-4 are intercalated; sometimes orders 4 or 3 haveperipheral characteristics too

a c g Tree development without metamorphosis (Halle &Oldeman 1970), the periphery being made up byleaves and the sequence of branching orders soequivalent that intercalation is not proven

Reiteration tree with repetitive manifolding of themodel (cf 1g)

b-i Metamorphic tree with repetitive complication

Assemblage tree with both repetitive complication(numbers as for 1c) and repetitive manifolding(numbers as for If) in opportunistic sequences

5

3

2

••• doe

r

1

zf.--- 1.1\

r

d

1

.,,,••••• o• •••.••"'..• •• •

fe)f•

r

r

r

1

k l Si.—a—

e

.... .0'.

34

t..,

1

eirl°\4%

..,—.............

r

2

I

0c

3

..mommi,

4

gel•••••••,............. ..•••

In

/

s%

NS

RI

•

27

Selection pressure can make phyllomorphism'migrate' from the periphery to the trunk. Thismay happen in a very stereotyped way andcause modular growth, as in Petit's model (egthe rubiaceous sequence from the Rothmannia toTetrorchidium species originally studied by Ha 116in 1967, cf Ha 116 & Oldeman 1970), or fromRoux's to Aubréville's model (see the sequencefrom Trema to Terminalia described by Oldemanand Ha 116 1980). In the centre of the tree, thetrunk is eventually the only axis left with aconsiderable size, mass and lifespan (eg 'pagodatrees' such as Terminalia spp. - Corner 1952).Foresters speak of 'monopodial branching' insuch trees, in which the branches occur on abig, monopodial trunk.

On the other hand, migration of phyllomorphismto lower-branch orders can be opportunistic anddisorderly Leguminous trees in adverseenvironments (eg the dry tropics or heavy shadeunder forest) illustrate this behaviour; examplesare the Acacia species of East African treesavannahs (Oldeman 1989a), the Sahelianmimosoid Faidherbia albida (Sterck, van derZandt & Oldeman 1992) or Eperua falcata(Ede lin 1991), a caesalpinioid neotropicalrainforest tree.

Unstable monopodia, described by Oldeman(1974, eg in Guyanese, very hard-woodedmelastornataceous Mouriri species), are axes withan easily but unpredictably deactivated endmeriStem. In this instance, extension growth isoften taken over by another unstablemonopodium, known in temperate trees as shoottip abortion (cf Remphrey & Davidson 1991).This phenomenon is in the same strategy asphyllomorphism, for both the smaller size andshorter lifespan of phyllomorphs and earlymeristem abortion lead to small building blocksin tree architecture. Both phyllomorphism andmonopodial instability hence lead to theincreasing predominance of sympodial growthover monopodial growth in a tree.

Conmiand hierarchy or polyarchy

From a study on syrnpodial trees, Ede lin (1991)postulated two ways to build trees by branching.His first principle is hierarchical, the second ispolyarchical; both are determined by a commandhierarchy ('plan d'organisation' sensu Ede lin1991). In hierarchically built trees, an axiscommands or dominates the behaviour of itsbranches, or a branched complex commands ordominates the behaviour of other branched unitsit bears, such as reiterates (Figure 1).Polyarchical trees are colonies of equivalent unitsorganised by loose, ephemeral mutual interactions,

Systems hierarchy and tree architecture

With the above concepts, the architectural studyof timber trees can be reduced to a few clear

28

principles. The following systems hierarchy isdefmed especially to deal with timber treearchitecture.

Stems and architectural unit models

There is frequent confusion between the terms'axis', 'sympodium' and `monopodiurn', so treeswill be considered here to be built by 'stems'. Astem is defined as a growing cylindrical organbearing leaves, meristems or other appendages;it is perceived macroscopically; only itsgeometric origin, not its meristematic origin, isspecified; it can be macroscopically differentiatedaccording to simple criteria of apparentphyllotaxis and orientation (orthotropic,plagiotropic, mixed), size and longevity (trunk,branch, twig, phyllomorph), flowering (none,terminal, lateral), and timing (immediate, delayedor adventitious appearance). More refined criteriaare presented by Ha 116 et al. (1978) and Ede lin(1984).

All primary architectural unit tree models (egFigure 1, a-g) are defined as theoretical treesoriginating from seed, the architecture anddynamics of which are described in terms ofsimple branching patterns determined by woodyskeletal stems, ie trunk, branches and theirdifferentiation. This definition is close to, butdiffers from, those in Ha 116 and Oldeman (1970).

Repetitive architectural patterns

A metamorph (Figure 1, a-d) is an intercalaryregularly branched pattern caused by a spatialand temporal combination of stems being formedlater within the primary architectural unit treemodel (Figure 1, c,d). The metamorph is not aninitial part of this model (shown in Figure 1, b).A reiterate (Figure 1, h) is an apposed branchedcomplex, corresponding to the total or partialprimary architectural unit tree model. Thereiterate is not built within but apposed to thismodel. Reiteration is important for timber trees,because reiteration causes trees to have morethan one trunk. The tree has one trunk from theseed and one in each reiterate (word coined byC Ede lin, personal communication). In otherwords, a 'subtree' formed by reiteration alsopossessed a trunk. Metamorphs make morebranches within a tree; reiterates add additionalperipheral trunks and/or branches to a tree.

According to the above defmitions, all timbertrees, throughout their development fromgermination to senescence, belong to one of fourstrategic unit tree models: elementary trees(Figure 1, g), reiteration trees (Figure 1, h ),metamorphic trees (Figure 1, i) or assemblagetrees (Figure 1, j).

Architectural systems hierarchy for timbertrees

Timber trees can now be examined in theirecological context according to a simplified

systems hierarchy. As usual in such hierarchies, asystem is explained by the interactions betweenits subsystems, conceived to be situated onelevel lower than the system itself. A stem is thesum of its parts plus their interactions; a primaryor secondary architectural unit is the sum of itsstems plus their interactions, as apparent fromthe branching geometry and dynamics aroundone central stem; a strategic unit is the sum ofits architectural units plus their interactions asapparent from the geometry and dynamics oftrunks and major branches; a forest stand is thesum of its organisms plus their interactions asapparent from the distribution and dynamics ofits trees.

Conformity and flexibility

'Rees can either adjust or adapt (Ha 116 et al.1978): adjustment occurs as a change in growthand architecture of individual trees in response toenvironmental stimuli; adaptation occurs as theresult of inherited changes of behaviour andarchitectural strategy in a tree population inresponse to the selection of surviving andreproducing members. For the study of thearchitectural flexibility of timber trees, thebehaviour of big woody stems, trunks andbranches in primary and secondary architecturalunits is paramount. In general, the context hereis one of architectural conformity as in theoriginal architectural tree models of Ha 116 andOldeman (1970). Gigantism of the primaryarchitectural unit (Ha 116 1986) stands for maximalconformity regardless of what complexsubsystems its stems may conceal. It is byrepetition that big trees can adjust in a flexibleway The crux of the matter of predictingarchitecture and growth dynamics of timber treesis, therefore, the nature and number of thearchitectural units on the one hand, and theirsizes and timing of appearance on the other.

TIMBER TREE TRUNKS

In timber trees, a forester or tree farmer prefersthe development of one woody stem, the treetrunk, to be straightforward from the seedlingstage to maturity for harvest. This development,moreover, should be as fast, as voluminous andas stable as possible. Five tree groups are thesuperpowers in the forests of the world:

i. Coniferae, forming a major component inmany boreal, northern and montane forests;

Leguminosae, abundant in a majority oftropical African and American rainforests, andmany dry tropical forests worldwide;

Fagaceae, dominating the main cool andwarm temperate broadleaved forests andmany montane forests;

iv Dipterocarpaceae, moulding the basic featuresof the lowland rainforests of tropical Asia;

v. Myrtaceae, playing a decisive partparticularly in Australian forests and intropical forest plantations.

The trees in these groups represent a successstory because of only a few architectural andother responses to selection pressures (egOldeman & Fundter 1986). Another, more diffuse,group contains unrelated trees from variousfamilies (eg Meliaceae, Burseraceae, Lauraceae,Celastraceae or Vochysiaceae). Their successstories are unmistakable but rather moreindividual than collective, often parallelling one oranother feature of one of the above majorgroups. However, in this diffuse group, there alsoseems to occur one original, rather elementarystrategy enabling their success as tropicalrainforest trees. This group might be termed'miscellaneous giants', with the species buildingor rebuilding tropical rainforests sensu latofollowing major forest clearings or under latepioneer conditions.

Each of the above groups s' hows a particularformation, size and persistence of the trunk.Conifers are architecturally simple trees duringmuch of their life, their trunks being simplygrown-up seedling stems. Coniferous metamorphsseem to be rare, and reiterates occur late, ifever (also see Ede lin 1977). Both Leguminosae(Oldeman 1989a) and Fagaceae (Peters 1992;Roloff 1989, 1991) are widely spread over manyenvironments. They show an assemblage strategywith a pronounced tendency to form numerous,quite small metamorphs and reiterates, resultingin compound, unstable trunks and branches witha tendency towards crooked growth and forking.Dipterocarpaceae and Myrtaceae are most oftenmetamorphic trees with some, late, reiteration(Ede lin 1984). They conserve and enlarge astable trunk from the seedling stage onwards fora prolonged period, the complex crown beingbuilt mainly by metamorphs.miscellaneous giants are reiterative trees with alarge, stable and orthotropic primary architecturalunit and rather few late and massive reiterates,so that their woody mass is distributed overrather few massive and stable trunks andbranches.

This general survey of the main groups ofsuccessful forest trees, many of which areimportant as a source of timber, illustrates theirvery different responses to their forestenvironments. Therefore, both their domesticationand their silviculture have to be understood inthe light of such specific strategies.

The architectural strategies of conifers andmiscellaneous giants give a strong propensity forthe formation and persistence of one sturdytrunk. Domestication of such trees for timberincludes selection of genotypes in which thistrunk-building tendency is maintained or

29

enhanced. Silvicultural protection of these tnmksagainst breaking also has to be selected for,because trunk regeneration by early massivereiteration lowers timber quality Dipterocarps andMyrtaceae (including Eucalyptus spp.) should beselected in the same way but their susceptibilityto repetitive architectural complication needs tobe studied. Nutrients and energy have to beinvested in the trunks, not excessively in thecrown mass, and this feature also requiresappropriate silvicultural selection. FinallyFagaceae and Leguminosae are not programmedto form straight trunks, because they pile upmetamorphs and reiterates opportunistically theirenvironment dictating the compound trunk shape.The latter, therefore, depends on the silviculturalsystem used, as exemplified by two centuries ofQuercus (oak) and Fagus (beech) silviculture inEurope.

TIMBER TREE SITES

The folded forest model

Big trees exert a large influence on theirenvironment below, but big trees start life assmall trees. This statement defines the essentialpoints of 'inner' forest environment dynamics, towhich all forest organisms other than maturetrees are partly or totally subjected (Oldeman1983; Uribe 1991). Forest environment dynamicsare analysed at different scales below

The green mass of vegetation covers the land inirregular folds of forest, tree row, agricultural fieldor meadow (Figure 2, a; also cf Oldeman1992b). Within a forest, the canopy surface isitself folded by regeneration dynamics and theresulting patterns (cf GOrnez-Pompa, Whitrnore &Hadley 1991), such as large and small, open,aggrading and biostatic or mature patches(Figure 2, b) (see Oldeman 1990). Thispatchwork or silvatic mosaic acts as an irregularsieve on incoming climatic factors, such as light,wind and precipitation (Figure 2, c,d).

Ecological factors are usually dynamic, hence noclear ecological limits exist between eco-units.Geometrical limits (such as those proposed byBrokaw 1982, p102, later contested by Bongers &Popma 1988) are indeed irrelevant for timbertree ecology although valuable in forestdiagnosis (Koop 1989; Oldeman 1990).

Light, just like turbulent or rain-laden wind,penetrates into a forest through large gaps,through small openings between neighbouringcrowns, and through interstices between leafmasses borne by reiterates, metamorphs or largebranches (Figure 2, c). The distribution andrepetitive redistribution of light or other factorscreate a very mobile and heterogeneous •microclimate, manifest among others in themobile and elusive sunfleck pattern in the

30

undergrowth. These patterns permit thecalculation of the light available to plants in acertain spot in the forest. However, temporalresolution has to be kept low eg by usingmonthly figures of spotwise accumulatedavailable irradiation. A spot is defined in aspatial grid with quite a high spatial resolution,eg by fish-eye photograph scanning andsimulation. Horizontal and vertical light-and-shadepatterns depicted with high spatial and lowtemporal resolution are highly non-linear. Theyinvalidate the concepts of straight horizontalstrata or vertical gap patterns, even asapproximate working hypotheses. In man-madeforests, such simple patterns might be createdtemporarily However, a recent and realisticsimulation of simple, pure Pinus sylvestris (Scotspine) forest plantations in The Netherlands showsthat their heterogeneity as sieves for climaticfactors increases with time (Leersnijder 1992).

Small volumes of light or shade accumulatingover time in a non-linear model, as in theSILVI-STAR light climate simulation (Figure 2, c,d;Koop 1989), produce seemingly random points,accumulating over time like pixels producingfractal images on a computer screen.

Self-similarity of the climatic effects of stepwisedecreasing canopy openings also intimates fractalmathematics (Oldeman 1992a). Such forestmicroclimate patterns are generally repetitiveenough on a local scale that plant and animalspecies can live and reproduce in such forestsfrom generation to generation. This kind ofdynamic status quo brings to mind theclimatological 'Lorenz attractors' (Gleick 1989;also, F Lutz, personal communication). There areno data as yet to check these hypotheses. Thischecking would be worthwhile, however, becauseof the promise of new methods for the calculusof predictive models, based on ecosystemgeometry

Direct environments of timber trees

Using the cruder, hand-made approximation(Figure 2, c), however, forest environments nestedwithin the whole-forest mosaic, down to thedirect environment around one tree, can bediscussed. First, folding of the forest surface isthe consequence of different gap forms and sizesand eco-unit development (Oldeman 1990,1992b). Second, all forest folds have a canopyhigh or low, thin or thick. This canopy risesslowly reaches its highest level, and thencollapses, on an eco-unit scale.

In large eco-units (Figure 2, a, L), the gap phaseis all but devoid of a functional canopy Itgradually develops and diversifies into a climaticsieve, usually built by fast-growing, rathershort-lived pioneer trees. Generally these treesneither reach their highest level, nor die, at the

a

A

0

t.

•

clY)

/X

OD0

8 080g,Figure 2. Principles of nested tree environment dynamics

a Land covered by a folded green carpet (afterOldeman 1992b). r, row of trees; L, large surfacepioneer forest; f, forest; rS, rather small forest gap(young eco-unit); S, small gap (young eco-unit);a,agricultural canopy; p, park canopy

Folds are folded again by crowns, these are foldedagain by subcrowns (for formation see Figure 1);abbreviations as for Figure 2, a

Folded canopy acting as an ecological sieve (cfOldeman 199b), eg for light; open arrows, lightthrough gaps; black anows, light through chinksbetween crowns; stippled arrows, light betweensubcrowns, the latter two usually being included in'indirect lightmeasurements; note pattern not at all asin gaps defined as discrete 'holes', light not verticaland canopy neither homogeneous nor opaque;shading denser where darker

SILVI-STAR-like, hypothetical pattern (cf Koop 1989) oflighter and darker patches on the forest floor causedby the transmission patterns as given in Figure 2, c;shading denser when darker; thick lines approximateisoluxes, thin lines crown projections; upper half ofFigure forest as Figure 2, b, lower half large-surfacepioneer forest

Metamorphic timber trees (M) with later somereiterates (R), such as Asian dipterocarps, grow inrather small openings

Assemblage trees (A) develop in small, highlydynamic openings, such as legume trees in tropicalAfrica and America

g & h Development of elemehtary pioneer trees (E), such asMoraceae, above the pioneer forest canopy andassemblage trees (A), such as legumes, in thatcanopy; note repetitive phenomena coupled withdecreased size of architectural units

re

re

Pan

c00,Cw ) P, )n 8 c

o„

4 1%kV

8

_

h

o

A

same time. After the pioneers, the large eco-unitsplits up into a finer-meshed mosaic, with smallergaps as the origin of smaller eco-units. The latterrange in size from one to a few trees (Figure 2,a, rS and S). This development may take one ormore decades. Meanwhile, in and under thecanopy a first degree of non-linear complexity isborn, with shaded and illuminated spots andregions. This pattern does not conform to asimple mosaic of illuminated gaps versus shadedundergrowth. The interference of large and smalllight, wind and humidity gradients creates adiverse array of microclimatic conditions in whichthe seedlings become established (forinterference, see Oldeman 1989b, 1992b).

From the seedling populations come the saplingsthat shape the eco-units arising in small gaps.These trees modify the climatic sieve formed bythe canopy once they start to participate in it&formation. The more a forest develops beyondthe pioneer stage, the more differentiated itsinternal environment can become. There are twolimits to this complexity The first limit iswholesale forest elimination, such as by fire,clearcutting, or other major event. Pioneerdevelopment may then start all over again. Thesecond limit is over-diversification, whenenvironmental heterogeneity reaches the pointwhere a particular species population can nolonger survive because appropriate nichesbecome too few and far between.

31

Architectural strategies in the context of thedomestication of timber trees

The domestication process requires thatgenetically improved timber trees are able toexpress their potential productivity; to this end,there is a need for silviculture to create theappropriate environment for growth (see Fasehun& Grace, pp148-157). Timber tree domesticationis, therefore, inseparable from the choice ordesign of forest-like ecosystems in which suchtrees are to be cultivated. It is futile, for instance,to select Eperua genotypes (Leguminosae) forsimple plantations and clearcutting, because theywould never yield straight timber, but onlycrooked firewood. This is the intuitive base of theclassical choice of species related to their 'lightrequirements'. However, forest ecoclimates areless straightforward than is suggested bycontrasting 'light-demanding' against'shade-tolerant' species. Forest trees includearchitectural strategies opposing programmedconformity to near-total flexibility but also admitcombinations in order to survive in dynamicenvironments (Oldeman 1990; Oldeman & VanDijk in G6mez-Pompa et al. 1991).

Silvicultural systems have been designed tofavour the ecosystem's carrying capacity forchosen tree species at the expense of others.Selective silviculture usually eliminates pioneertrees to a large extent. Mixed-plantation timbersilviculture is usually considered difficult, beingintermediate between natural regeneration andartificial regeneration. With the above concepts, itcan be implemented. Amohg the best examplesare tropical agroforestry systems utilisingmultipurpose, non-timber species (see Simons,MacQueen & Stewart, pp91-102) and the mixeddipterocarp silviculture in East Kalimantan (seeOldeman & Van Dijk in Gómez-Pompa et al.1991).

The crux of the matter is timing. Foresters andtimber farmers want a high timber volume of thebest marketable tree species in as short a timeas possible, unless they work in a multiple-useforestry context. There are inbuilt paradoxes inthis set of aims. They can only be achievedeither if the desired species have pioneercharacteristics, such as short lifespan, highgrowth rates and straight trunk formation, or ifthe combination of rotation and volume isadapted to maximise tree size, rather than thevolume per year per hectare. Domestication of'new' timber tree species makes sense only ifthe concept of timber production itself isthoroughly overhauled (Oldeman 1991), adoptingprinciples of ecological forestry

ACKNOWLEDGEMENT

We thank Miss Dr J Ruinen for her thoroughreading of the text and illustrations.

32

REFERENCESAtger, C. 1991. L'architecture racinaire est-elle influence parle milieu? Naturalia Monspeliensia, special issue (L'arbre:biologie et développement), 71-84.

Atger, C. 1992. Essai sur l'architecture racinaire des arbres.Doctor's thesis, Institut Botanique USTL, Montpellier.

Bongers, F. & Popma, J. 1988. Trees and gaps in aMexican tropical rain forest. Utrecht: Rijksuniversiteit.

Brokaw, N.V.L. 1982. Treefalls: frequency timing andconsequences. In: The ecology of a tropical forest, edited byE.G. Leigh Jr, A.S. Rand & D.M. Windsor, 101-108. Oxford:Oxford University Press.

Corner, E.J.H. 1952. Wayside trees of Malaya. 2nd ed. 2vols. Singapore: The Government Printer.

Edelin, C. 1977. Images de l'architecture des coniferesMontpellier: Institut Botanique USTL.

Edelin, C. 1984. L'architecture monopodiale: l'exemple dequelques arbres dAsie tropicale. Montpellier: Institut BotaniqueUSTL.

Edelin, C. 1991. Nouvelles données sur l'architecture desarbres sympodiaux: le concept de Plan d'Organisation.Naturalia Monspeliensia, special issue (L'arbre, biologie etdéveloppement), 125-154.

Gleick, J. 1989. Chaos. London: Cardinal-Sphere Books,McDonald & Co.

Gómez-Pompa, A., Whitmore, T.C. & Hadley, M., eds.1991. Rain forest regeneration and management. (Man and thebiosphere series, no. 6.) Casterton: Parthenon.

Hallé, F. 1967. Etude biologique et morphologique de la tribudes Gardéniées (Rubiacées). (Mémoires ORSTOM, 22.) Paris:ORSTOM.

Hallé, F. 1986. Deux strategies pour l'arborescence:gigantism et repetition. Naturalia Monspeliensia, special issue(Colloque international sur Parbre), 159-170.

Hallé, F. 1991. Le bois constituant un tionc peut-il 'etre denature racinaire? Une hypothése. Naturalia Monspeliensia,special issue (L'arbre, biologie et développement), 97-111.

Hallé, F. & Oldeman, R.A.A. 1970. Essai sur l'architectureet la dynamique de cmissance des arbres tropicaux. Paris:Masson.

Hallé, E, Oldeman, R.A.A. & Tomlinson, P.B. 1978.Tropical trees and forests: an architectural analysis. Heidelberg:Springer.

Koop, H. 1989. Forest dynamics, SILVI-STAR: a comprehensivemonitoring system. Heidelberg: Springer.

Leersnijder, R.P. 1992. PINOGRAM, a pine growth areamodel. Wageningen: R.P Leersnijder.

Oldernan, R.A.A. 1974. L'architecture de la forét Guyanaise.(Mémoires ORSTOM 73.) Paris: ORSTOM.

Oldernan, R.A.A..1983. Tropical rain forest, architecture,silvigenesis and diversity In: Tropical rain forest: ecology andmanagement, edited by S.L. Sutton, TC. Whitmore & AC.Chadwick, 139-150. Oxford: Blackwell Scientific.

Oldeman, R.A.A. 1989a. Biological implications ofleguminous tree architecture. In: Advances in legume biology,edited by C.H. Stirton & J.L. Zarucchi, 17-34. (Monographs ofsystematic botany of the Missouri Botanical Garden, no. 29.)St. Louis, Missouri: Missouri Botanical Garden.

Oldeman, R.A.A. 1989b. Dynamics in tropical rain forests.In: Tropical forests, botanical dynamics, speciation and diversity,edited by LB. Holm-Nielsen, I.C. Nielsen & H. Balslev, 3-21.London: Academic Press.

Oldeman, R.A.A. 1990. Fbrests: elements of silvology.Heidelberg: Springer.

Oldeman, R.A.A. 1991. The paradox of forest management.(Pnaceedings 10th World Fbrestry Congress, Paris, 4, 153-182.)Revue Forestière Francaise, special issue. Nancy: ENGREF

Oldeman, R.A.A. 1992a. Architectural models, fractals andagroforestry design. Agriculture. Ecosystems and Environment,41, 179-188.

Oldentan, R.A.A. 1992b. Bosarchitectuur en diversiteit: wat iser gemengd in gemeng de bossen? (Werkdocurnent IKC-NBLE12, 1-12.) Wageningen: IKC-NBLE (In Dutch.)

Oldeman, R.A.A. & Fundter, J.M. 1986. OnDiptero-Vochysian tree strategy Naturalia Monspeliensia, specialissue (Colloque international stir l'arbre), 191-208.

Oldeman, R.A.A. & Ha 116, F. 1980. Sobre los ejes mixtosplagio-ortôtropos en algunos arboles tropicales.Landbouwhogeschool Wageningen Miscellaneous Papers, 19,281-287.

Peters, R. 1992. Ecology of beech forests in the northernhemisphere. Wageningen: Landbouwuniversiteit.

Remphrey, W.R. & Davidson, C.G. 1991. Crown shapevariation in Fmxinus pennsylvanica (Vahl) Fern.: its relation toarchitectural parameters including the effect of shoot-tipabortion. Naturalia Monspelien_qa, special issue (L'arbre:biologie et développement), 169-180.

Roloff, A. 1989. Entwicklung und Flexibilitat der Baumkroneund ihre Bedeutung als Vitalitasweiser. SchweizerischesZeitschrift ffir das Fbrstwesen, 140, 775-798; 943-963.

Roloff, A. 1991. Crown morphology as a tool to determinetree vitality Naturalia Monspeliensia, special issue (L'arbre,biologie et développemenf), 115-126.

Sterck, F.J., van der Zandt, G.J.A.W. & Oldeman, R.A.A.1992. Architectural development of Faiclherbia albida (Del.)A.Chev In: Physiologie des arbres et arbustes en zones arideset semi-arides, edited by Groupe d'Etude de l'Arbre. Paris:Groupe d'Etude de l'Arbre.

Uribe, V.II., ed. 1991. Bosques de niebla de Colombia.Bogot: Cristina Uribe Editores.

33

Fruit trees: diversity and conservation strategies

C Okaforl & A Lamb21Tree Crops and Tropical Ecology Consultants, 3 Kingsway Road, PO Box 3856, Enugu, Nigeria2Agricultural Research Station, PO Box 197, 7bnom, Sabah, Malaysia

ABSTRACTFruit-producing trees contribute significantly to the species diversity of tropical forests. Their .contributions to the nutrition and localdiet in developing countries are great, and their diversity is considerable. The availability and consumption of native fruits aredeclining in most regions, partly because of the greater attention given to research, development and marketing of exotic fruits.Population growth and severe forest degradation, arising from over-exploitation and infrastructural development, also adverselyaffect the availability of native fruits.

Strategies which can address the situation include conservation, domestication and enhanced utilisation of wild fruit trees, throughthe establishment of cottage industries and land use applications involving local participation. In this paper, priority species arerecommended for domestication in countries of Malaysia and West Africa, and include durians, rambutans, mata kucing, bushmango, African pear and African breadfruit. The domestication effort should be complemented by in situ conservation practices,such as establishment of forest reserves for the protection of genetic diversity Increased taxonomic, phenological and propagationstudies are necessary to enhance identification, conservation and sustained use of fruit trees.

INTRODUCTION

The term 'fruit' is broaclly applied here in bothits horticultural and botanical senses to includethe succulent and non-succulent portions ofripened ovary or ovaries which are eaten freshor used in various cooked food preparations.Apart from trees, other tropical woody plantssuch as erect shrubs and climbers also provideedible fruits. In addition, trees, shrubs andclimbers are also sources of edible leafyvegetables, edible flowers, mushrooms, and otherfoods and beverages:

Wild fruit trees form an important part of thespecies diversity of tropical forests. The diversityof tropical fruit trees is related to the range offamilies, genera and species involved, as well asthe extent of intraspecific variation. For example,in Borneo, considered to be the centre ofdiversity for many plant genera, the genusGarcinia (Guttiferae) consists of 50 species ormore, while Baccaurea (Euphorbiaceae)comprises at least 25 species (Wong & Lamb1990).

The wide range of edible products obtainedfrom wild fruit trees includes nuts and seedsused as staple foods or main dishes; those usedas minor food supplements; condiments;thickening agents and flavours; leafy vegetables;fresh fruits; fresh seeds; edible oil; spices; fruitdrinks and non-alcoholic beverages; andalcoholic drinks. These edible forest productsconstitute important and cheap sources ofvitamins, minerals, protein, carbohydrates and fats

34

and their contribution to the diet of local peopleis often great. Their dietary contribution isincreased because they are available duringmost seasons, including strategic periods in theyear when the conventional staples andvegetables are scarce (Okafor 1991). Thepotential of these wild plants for overcoming foodavailability problems in tropical countries isenormous (Getahun 1974; Jong, Stone &Soepadme 1973; Okafor 1975a, 1980a, b, 1981a;Oldgbo 1977; Roche 1975a; Wong & Lamb 1990).

Notwithstanding their diversity nutritional anddietary importance, the availability andconsumption of wild fruit trees appear to bedeclining in many regions, partly because of thegreater attention given to research, developmentand marketing of exotic fruits rather than toindigenous ones. Another important contributingfactor is the declining availability of edible forestproducts, owing to human population growth andsevere forest degradation. To redress the situationit is necessary to intensify programmes on theidentification and genetic improvement of specieswith high potential, and on the development ofstrategies for conservation, development -andproduction of both traditional and innovativeproducts (Okafor 1991). These objectives formthe focus of this paper, based largely on workand field experiences in Malaysia and countriesin West Africa.

TAXONOMY OF EDIBLE WILD PLANTS

There are few taxonomic studies on edible wild

plants of the tropics. Available information comesmostly from general notes and casual remarks inFloras and more general texts, and from oralreports as commonly documented inethnobotanical and socio-economic surveys(Okafor 1989, 1990a; Shepherd & Okafor 1991;Thomas et al. 1989). Okafor (1981a) presentedan account of 171 indigenous woody plants (53families and 119 genera) of nutritionalimportance within the forest zone of Nigeria.

Species diversity of wild fruit trees in the forestsof Sabah is illustrated by the genera Artocarpus(Moreaceae), Durio (Bombacaceae) andMangifera (Anacardiaceae) (Wong & Lamb 1990).There are at least 13 species of Artocarpusindigenous to Sabah, out of some 25 specieswhich are found in Borneo Garrett 1959a, b,1960). The durians (Durio spp.) are representedby about 28 species in SE Asia, while in Sabah14 species are recorded out of the 19 speciesfound in Borneo (Korsterman &Soegeng-Riksodihardjo 1958). Of these 14species, there are seven which have edible fruits.Similarly in the genus Mangifera, there are some19 species (with one introduced) in Sabah, outof a total of 24 found in the whole of Borneo(Bompard 1988). Other fruit species representing33 families and 65 genera, cited by Wong andLamb (1990), further illustrate the overall diversityof the wild and semi-wild fruits found in theforests of Sabah. Many of these families andgenera are also represented in West Africa(Table 1). For example, the family Sapindaceaeincludes many important edible fruits both inSabah and elsewhere. The akee apple (Blighia

sapida), which occurs in West Africa, is acommercially important fruit tree, while thebest-known fruits of this family in Sabah are therambutan, meritam or pulasan (genus Nephelium)

Table 1. A list of some genera producing edible fruits in Nigeria, West Africa, that are also reported to occur in Sabah,

Malaysia, by Wong and Lamb (1990). NB Several genera (and species) which occur in West Africa but were not cited in Wong

and Lamb (1990) have been excluded. For a more detailed list, see Okafor (1979)

Family

Annonaceae

Burseraceae

Ebenaceae

Euphorbiaceae

Guttiferae

Leguminosae

Moraceae

Palmae

Sterculiaceae

Species Local name

Uvaria chamae

Canarium schweinfurthii 1Jbe-okpoko

Dacryodes edulis African pear

D. Idaineana

Diospyros iturensis

fl elliotii

Antidesma venosum —

Garcinia kola Bitter kola

G smeathmannii

Dialium guineense Velvet tamarind

Par Ida biglobosa Locust bean

Ficus capensis —

F exasperata

vallis-choudae

Nypa fruticans Nipah

Sterculia oblonga

and mata kucing (genus Dimocarpus). The mostcommonly eaten rambutan is N lappaceum.

THE EXTENT OF INTRASPECIFICVARIATION

Examples of intraspecific variation abound intropical fruit tree species. For instance,Dimocarpus longan (mata kucing) is veryvariable morphologically especially in Borneo,where 30-40 local races can be distinguished(Van Welzen, Lamb & Wong 1988). All theseBornean races belong to the subspeciesmalesianus, which is divided into two varieties:var. malesianus and var. echinatus. The twospecies of Nephelium, N lappaceum andN cuspidatum, are also very variable inappearance. N lappaceum is divided into threevarieties, namely var. lappaceum, var. pallens andvar. xanthioides, all three occurring in Sabah.N cuspidatum has six varieties of which var.robustum (with large glabrous leaves) and var.eriopetalum (small and hairy leaves) are alsofound in Sabah.

The existence of natural variation within fruittrees is a crucial factor in efforts aimed at thedomestication of edible forest species (Whitmore1976; Okafor 1980a). Examples of varietaldelimitation in West African fruit trees, such asIrvingia gabonensis (Okafor 1975b), Treculiaafricana subsp. africana (Okafor 1981b) andDacryodes edulis (Okafor 1983), have indicatedhigh prospects for extending the period of fruitavailability increasing the yield, and choosing thedesired pattern as well as season of yield(Okafor 1978, 1981a; Okigbo 1977). If the twovarieties of I. gabonensis, which fruit in the rainy(var. gabonensis) and dry (var. excelsa) seasonsrespectively were each to be developed, the

Product

Fruit

Fruit

Fruit

Fruit

Fruit

Fruit

Fruit

Pulp and seeds

Pulp and seeds

Seed aril

Fruit pulp, seed

Fruit, leaves

Fruit

Fruit

Fruit

Seed

Potential for development

Low

Medium

High

Low

Low

Low

Low

High

Medium

High; very rich invitamin C

High; rich in vitamin C

Low

Low

Low

Low

Low

35

combined period of availability of the productswould be significantly extended (Okafor 1991).

PROPAGATION TECHNIQUES

Large-scale production of fruit trees requires thedevelopment of nursery procedures involving theuse of seeds or vegetative propagation, either forthe production of stockplants to be used forbudding (budgrafting), or for seedling trees tobe outplanted in seed orchards or plantations.For example, in Nigeria, standard nurserypractices have been developed for severalspecies including Dacryodes edulis, Irvingiagabonensis and Treculia africana (Okafor 1981a,1990b).

Seed propagation

Traditionally tropical trees are propagated andconserved by seed, either by establishingplantations or by encouraging naturalregeneration. There are three main advantages ofseeds.

• Large numbers can usually be obtainedeasily

• Spare seeds can be stored for later use.

• Variation is retained in succeedinggenerations, although this may not strictly beregarded as an advantage if a uniformproduct is desired.

Vegetative propagation

Vegetative multiplication is an alternative to theuse of seeds as a method of raising plantingstock of young trees. The advantages of thismethod are as follows.

• Genetically uniform planting stock may beproduced.

• When mature adult budwood is used, earlyfruiting and fruit set at low height can beachieved.

• Superior genotypes can be selected andmultiplied, leading to improvements in qualityand yield.

• Selection and conservation of clones that areresistant to pests and diseases may bepossible, and clones may be developed thatare suitable for a particular region, site andend use.

• Conservation of genetic variation (such ashybrids produced by breeding) can beachieved through vegetative multiplication, inthe form of clone banks of both 'adult' and'juvenile' material.

The main objection to the development of clonaltree crops is the increased risk of biotic orphysical catastrophe which is associated with thenarrowing of the genetic base (Barnes & Burley1987). However, frequent introduction . of newclones and the maintenance of genetic diversity

36

should reduce risk (see Foster & Bertolucci,pp103-111). The advantages of vegetativepropagation techniques are greatest when thereare problems with seed availability or storage,when only the male or female plant of dioeciousspecies is required, or when there is aparticularly high value attached to certainindividual trees (see Barnes & Burley 1987).

Van Welzen, Lamb and Wong (1988) successfullyproduced cultivars of rambutan with constantquality by using budding techniques (Forkertmethod). In a study in West Africa, some 26species of trees producing edible products werefound to be buddable in experiments usingscions from mature trees (Okafor 1978, 1980a,1981a) (Table 2). Budded trees of some of thesespecies (Chrysophyllurn albidum, Dacryodesedulis, Dialium guineense, Irvingia gabonensis,Pentaclethra macrophylla, Spondias mombin andTreculia.africana) produced viable fruits close toground level after only two to four years (insteadof five or more years in trees raised from seed).

7bble 2. Trees with edible parts, successfully propagated bybudgrafting in West Africa (source: Okafor 1981a)

Species name Family

Afzelia africana CaesalpiniaceaeA. belle var. belle CaesalpiniaceaeBosqueia angolensis* Moraceae

Canarium schweinfurthii BurseraceaeCeiba pentandra BombacaceaeChrysophyllum albidum Sapotaceae

Cola acuminate SterculiaceaeC. gigantea SterculiaceaeC hispida StemuliaceaeDacryodes edulis BurseraceaeDetarium microcarpum CaesalpiniaceaeDia lium guineense CaesalpiniaceaeHildegarrlia barteri StemuliaceaeIrvingia gabonensis IrvingiaceaeMonodora myristica AnnonaceaeMyrianthus arboreus Moraceae

Parka biglobosa MimosaceaePentaclethra macrophylla MimosaceaePterocarpus mildbreedii PapilionaceaeP santalinoides PapilionaceaeP soyauxil PapilionaceaeSpondias mombin Anacandiaceae7btrapleura tetraptere MirnosaceaeTreculia africana Moraceae

Vitex doniana VerbanaceaeXylopia sp. Annonaceae

* Now Trilepisiurn madagascariense

Propagation by stem cuttings (without applicationof auxin) using adult shoots was also successfulfor 21 West African tree species (Table 3).Fruiting was subsequently recorded in trees ofT africana subsp. africana var. inversa andD. guineense derived from stem cuttings. Withjuvenile cuttings, R R B Leakey (personal

communication) recorded 69% rooting ofT africana under intermittent mist, using a 0.2%solution of IBA in 95% ethyl alcohol.

Table 3. Woody plants with edible parts, successfully

propagated by stem cuttings (source: Okafor 1981a)

Species

Ceiba pentandra

Co/a gigantea

Detarium microcarpum

Dialiurn guineense

Dioscoreophyllum cumminsii*

Ficus capensis

Gneturn spp.*

Gongronema latifoliurn*

Heinsia crinata

Hildegazdia barteri

Monodora myristica

Lasianthera africana

Myrianthus arboreus

Piper guineense*

Pterocarpus mildbraedii

P santalinoides

P soyauxii

7'riplochiton scleroxylon

7btracarpidium conophorum

Treculia africana

Vernonia amygdalina+

* Woody climber; + Shrub

Family

Bombacaceae

Sterculiaceae

Caesalpiniaceae

Caesalpiniaceae

Menispermaceae

Moraceae

Gnetaceae

Asclepiadaceae

Rubiaceae

Sterculiaceae

Annonaceae

Icacinaceae

Moraceae

Piperaceae

Papilionaceae

Papilionaceae

PaPilionaceae

Sterculiaceae

Euphorbiaceae

Moraceae

Compositae

Research studies on vegetative propagation havealso been carried out on a number of tropicaltimber species (see Leakey Newton & Dick,pp72-83). For timber production, juvenile shootsare usually preferred for ease of rooting and forvigorous shoot growth, while, for fruit production,adult scions are used in order to obtain earlyfruiting at low heights. Furthermore, aplagiogeotropic growth habit, which may be atopophytic or 'adult' character, is regarded asundesirable for timber species, except for seedproduction in clonal seed orchards. Such traitsmay however, be desirable for fruit treeproduction. It is, therefore, necessary to developpropagation methodologies and selection criteria

specifically for fruit trees (see Okafor 1990b),even when the same species is important forboth fruit and wood products.

CONSERVATION OF THE GENETICRESOURCES OF FRUIT TREES

Conservation of genetic resources can beimplemented by adopting both in situ and ex situ

methods (Roche 1975b). Forest reserves, strictnatural reserves (SNRS) and protected fetishgroves contain fruit trees and serve as potentialreservoirs "of the various species. However, owingto high rates of forest clearance, natural reservesare no longer dependable sources of the genetic

material of fruit trees. As a result of bushburning, farming, grazing and infrastructural

development, over 26 000 ha of high forest arelost annually in the forest zone of Nigeria(Ola-Adams & Iyamabo 1977), and more than90% of natural forest vegetation has already beencleared (World Wide Fund for Nature 1989).These losses adversely affect the availability offruit trees in Nigeria and elsewhere in the tropics.

The density of fruit trees in rainforests is usuallylow, sometimes around three trees per hectare(Jong et al. 1973; Okafor 1981a), indicating thatlarge areas are needed for sustainable in situ

conservation (Roche 1979; Okafor 1981a). Thisfact also emphasises the need for ex situconservation, which may involve seed storage,establishment of clone banks, tissue andmeristem culture techniques, and theestablishment of living collections (see Tompsett,pp61-71; Engelmann, pp49-52; Khuspe et al.,pp53-60; Ladipo et al., pp239-248). Varietalselection must form a key part of suchconservation programmes for fruit trees.

The adoption of agroforestry practices holdsgood prospects for ex situ conservation,especially in the support zone of national parks,such as Cross River National Park in Nigeria andthe Korup Project in Cameroon (Okafor 1989,1990a). Agroforestry can contribute to theconservation of biodiversity if people areencouraged to cultivate local forest species,especially edible varieties, in their farmingsystems (Okafor 1991). Species which can beused in agroforestry systems such as alleyfarming and intercropping include Treculiaafricana, Ricinodendron heudelotii, Dacuodesedulis, Irvingia gabonensis, Cola acuminata, etc(Okafor & Fernandes 1987).

OPPORTUNITIES FOR COMMERCIALDEVELOPMENT OF INDIGENOUSTROPICAL FRUIT TREES

The large number of species of fruit trees invarious parts of the tropics and the wide rangeof their food products have been welldocumented (eg Van Welzen et al. 1988; Abbiw1990; Falconer 1990; Getahun 1974). In additionto the direct nutritional contribution of these localfruit trees, there is significant potential fordeveloping cottage industries based upon them.These industries, in turn, could enhånce theimproved use of their edible products, as well aspromote the development and conservation ofthe species for the large-scale supply of rawmaterials (see Prance, pp7-15; Clement &Villachica, pp230-238; Maghembe et al.,pp220-229). A number of food products thatcould be produced commercially fromlesser-known tropical fruits are described below

jams and jellies

Suitable species include Irvingia gabonensis var.

37

gabonensis, Chrysophyllum albidum, Dialiumguineense, Ficus sycomorus, Sclerocarya birrea andNephelium lappaceum (rambutan) (Okafor 1973;Okafor & Okolo 1974; Van Welzen et al. 1988).

Fruit juice

The following species are suitable: Irvingiagabonensis, Parihari curatellifolia, Tamarindusindica, Parka biglobosa (Okafor 1980a).

Confectionery

Breadfruit flour, processed from Treculia africana,can be used to produce a variety of bakedfoods (Anazonwu-Bello 1981).

Beverages

Recently Ejiofor, Obiajulu and Okafor (1988) haveprepared a non-alcoholic beverage frompowdered T africana seeds. Various fruits arealso suitable for brewed alcoholic drinks,including Spondias mombin, Uapaca kirkiana andDiospyros mespiliformis.

Composite seasoning

Several indigenous seeds can be blended intoseasoning. Suitable species include Afrostyraxlepidophyllus (country onion), Monodora brevipes,Piper guiheense (bush pepper), Tetrapleuratetraptera and Xylopia aethiopica (Ajayi 1986).

Fats and oils

The high fat and oil content of the fruits ofDacryodes edulis, Elae.is guineensis,Butyrospermum spp., Baillonella toxisperma,Irvingia gabonensis, Ricinodendron heudelotii andNephelium lappaceum (rambutan) indicates theirsuitability for commercial production of cookingoil and margarine, manufacture of soaps andpharmaceutical preparations, etc (Okafor & Okolo1974; Udeala, Onyechi & Agu 1980; Van Welzenet al. 1988).

Livestock feeds

Kernels of Dacryodes edulis, which contain 3.2%protein (Okafor & Okolo 1974), and fruit headpulp and bran of Treculia africana, which contain9.4% and 5.7% protein respectively (Ejiofor et al.1988), can be used in livestock feedformulations.

Medicinal uses

Many fruit trees are used in traditional medicines.For example, seeds of Garcinia kola (bitter kola)are useful in the treatment of coughs andhepatitis (Iwu et al. 1987). The roots ofNephelium lappaceum (rambutan) are used indecoctions for treating fever, the fruit fordigestive problems, and the leaves in poulticesfor headaches (Van Welzen et al. 1988).

PRIORITIES FOR DOMESTICATION

Many tropical wild fruit, nut and spice trees havebeen selected by generations of local people in

38

various tropical countries, and have become partof their everyday life in food supply and farmingsystems. For example, in Borneo, the commonlycultivated fruit trees include the tampois(Baccaurea macrocarpa), the rambais(B. motleyana), the rambutan group (Nepheliumspp.), mangosteen (Garcinia mangostana) anddurians (Durio spp.). In West African countriessuch as Nigeria and Cameroon, commonlycultivated fruit trees include bush mango (Irvingiagabonensis), African pear (Dacryodes edulis),monkey kolas (Cola lepidota, C. pachycarpa),kola (Cola acuminata, C. nitida), and Africanbreadfruit (Treculia africana), among others.

The collection of fruits and other products fromthe species listed above represents a start to thedomestication process, but much greaterprogress could be made with some well-targetedresearch. Domestication of such species shouldhelp overcome the problems enountered whenthe only source of the fruit is from natural forest,such as the lack of information on periods offruit maturity irregular fruit supply variableproduct quality and the sparse density of fruittrees. Economically important species whichprovide a range of useful products and whichare easily propagated should be accordedhighest priority for domestication. Similarlyspecies which have restricted distribution, sparsedensity or are threatened with extinction shouldalso be the focus of research attention (seeTable 4).

CONCLUSIONS

Despite the importance of indigenous fruit treesas sources of food products and cash income tolocal people in tropical countries, the future ofsuch trees appears bleak, because of theever-increasing pressures on the tropical forestresources base, caused by increasing population,intensification of various land use practices,infrastructural development, and associatedenvironmental problems. Many of the tropicalwild fruit trees are fast disappearing, and manymay soon face extinction if not conserved orprotected. In addition, many tropical wild fruitspecies are markedly under-exploited. Toillustrate this point, according to Zakri, Saw andRajanaidu (1989), there are approximately 12 000species of seed plants recorded in Malaysia, outof which only about 300 species (or 2.5%) arecurrently being used. The rest are still relativelyunknown (VVong & Lamb 1990).

Efforts are currently being made by theDepartment of Agriculture in Sabah, Malaysia(Wong & Lamb 1990), and the ForestryCommission, Enugu, Nigeria (Okafor 1981a),towards the conservation and domestication ofthe wild fruit trees in these countries, bydeveloping suitable propagation techniques andby collecting wild species and planting them in

Table 4. Fruit tree species with apparent high potential for selection and genetic improvement

Species

Malaysia (eg Borneo, Sabah)

Artocarpus spp.

Baccaurea spp.

Canarium spp.

Dacryodes spp.

Dimocarpus spp.

Durio spp.

Garcinia spp.

Mangifera spp.

Nephelium slop.

West Africa (eg Nigeria, Camemon)

Afrostyrax lepidophyllus

Baillonella toxisperma

Chrysophyllum albidum

Cola acuminate, C. nitida

C. lepidota, C. pachycarpa

Dacryodes edulis*

Dia liurn guineense

Garcinia kola

Irvingia gabonensis*

Monodora spp.

Parkia biglobosa

Pentaclethra macrophylla

Piper guineense*

Ricinodendron heudelotii

.Spondias mombin

Tbtracarpidium conophorum*

Treculia africana*

Local name

Breadfruits

Tarnpois

Dabai, kembayau, kedondong

Kembayau

Mata kucing

Durians

Bebata, kandis, mangosteen

Mangoes

Rambutans

Country onion

Shellnut

African star apple

Kola

Monkey kola

African pear

Velvet tamarind

Bitter kola

Bush mango

Nutmeg

Locust bean, dorewa

Oil bean

Guinea pepper, uziza

Gmundnut tree

African hog plum

Conophor, ukpa, kasu

African breadfruit

*Species among the priority list of 20 under-utilised species considered as needing immediate attention by a symposium onconserving biodiversity held at the International Institute of Tropical Agriculture, lbadan, Nigeria, in 1990

orchards. To complement these efforts, the in situconservation of wild fruit trees in strict naturalreserves (VVong & Lamb 1990) needs to beproperly maintained and expanded to covermore areas, for the protection of geneticresources of the species. This work should besupplemented by further ex situ conservation.

Vegetative propagation techniques and nurseryprocedures developed for several fruit tree specieshave great potential as tools for domestication andex situ conservation, including incorporation of fruittree species in agroforestry systems. Theinvolvement of local communities is necessary forthe identification, design and implementation ofsuch agioforestry programmes. Development andutilisation of food and industrial raw materialsbased on indigenous tropical fruit trees should beundertaken as a purposeful conservation strategyfor the species. Such tree species should also beused in land use, management and developmentoptions, to enhance the sustainable supply ofedible products and their role in environmentalprotection.

Ethnobotanical surveys, including taxonomic andphenological studies involving local fruit trees,

Family

Moraceae

Euphorbiaceae

Burseraceae

Burseraceae

Sapindaceae

Bombacaeae

Guttiferae

Anacardiaceae

Sapindaceae

Styracaceae

Sapotaceae

Sapotaceae

Sterculiaceae

Stemuliaceae

Burseraceae

Caesalpiniaceae

Guttiferae

Irvingiaceae

Annonaceae

Mimosaceae

Mimosaceae

Piperaceae

Euphorbiaceae

Anacardiaceae

Euphorbiaceae

Moraceae

are required in tropical countries, in order toidentify species and variants with high potential,and their traditional uses. Improved knowledgeabout tropical fruit trees will facilitate theirconservation and efficient use.

ACKNOWLEDGEMENTS

The authors thank Dr R R B Leakey and allthose who assisted him in organising theConference and for inviting this contribution. Wealso thank the editors for improving themanuscript. The financial support provided toJ C Okafor which enabled him to attend theConference is also gratefully acknowledged.

REFERENCES

Abbiw, D.K. 1990. Useful plants of Ghana. IntermediateTechnology Publications and the Royal Botanic Gardens Kew

Ajayi, G.A. 1986. Production of composite seasoning fromlocal spices. HND research project, Department of FoodTechnology Federal Polytechnic, Idah, Nigeria.

Anazonwu-Bello, J.N. 1981. Indigenous foods and nutritionaladequacy (Symposium on development of indigenoustechnology.) Enugu: Ministry of Science and Technology(Unpublished.)

39

Barnes, R.D. & Burley, J. 1987. Vegetative propagation forimproved topical forest trees. In: Improving vegetativelypropagated crops, edited by A.J. Abbott & R.K. Atkin, 211-277.London: Academic Press,

Bompard, J.M. 1988. Wild Mangifera species in Kalimantan(Indonesia) and in Malaysia. (Unpublished final report on the1986-1988 collecting Missions.) IBPGR-WWE

Ejiofor, M.A.N., Obiajulu, aR. & Okafor, J.C. 1988.

Diversifying utilities of African breadfruit (Treculia africanaDecne. subsp. africana) as food and feed. International TreeCrops Journal, 5, 125-134.

Falconer, J. 1990. The major significance of 'minor' forestproducts. The local use and value of forests in the WestAfrican humid forest zone. (Community Forestry Note, no. 6.)Rome: Food and Agriculture Organisation.

Getahun, A. 1974. The role of wild plants in the native dietin Ethiopia. Agroecosystems, 1, 45-56.

Iwu, M.M., Igboko, O.A., Onwuchekwa, U.A. & Okonji,C.O. 1987. Evaluation of the antihepatotoxic activity of thebiflavonoids of Garcinia kola seeds. Journal ofEthnopharmacology 21, 127-138.

Jarrett, F.M. 1959a. Studies in Artocarpus and allied genera,I. General considerations. Journal of the Arnold Arboretum, 40,1-29.

Jarrett, F.M. 1959b. Studies in Artocarpus and allied genera,III. A revision of subgenus Artocarpus. Journal of the ArnoldArboretum, 40, 113-115, 298-368.

Jarrett, F.M. 1960. Studies in Artocarpus and allied genera,IV A revision of Artocarpus subgenus pseudojaca. Journal ofthe Arnold Arboretum, 41, 73-88.

Jong, K., Stone, B.C. & Soepadme, E. 1973. Malaysiantopical forest: an under-exploited genetic reservoir of ediblefruit tree species. In: Proceedings of the Symposium forBiological Resources and Natural Development, edited byE. Soepadmo & K.G. Singh, 113-121. Kuala Lumpur: MalaysianNature Society

Kio, P.R.O., Gbile, Z.O., Ola-Adams, B.A. & Ladipo,ao. 1985. Strategies for conservation of natural forests andgrazing lands in Nigeria. (Technical note. Technical Report(Savanna Series).) lbadan: Forestry Research Institute ofNigeria.

Korsterman, A.J.G.H. & Soegeng-Riksodihardjo. 1958. Amonograph of the genus Durio Adans (Bombacaceae). Part I.Bornean species. Commur2ications of the Forest ResearchInstitute (Bogor), 61, 1-88.

Okafor, J.C. 1973. Prospects for large-scale production ofIrvingia gabonensis (Irvingiaceae). Proceedings 4th AnnualConference Forestry Association of Nigeria, Enugu, edited byE.E. Enabor, 172-186. Lagos: Federal Department of Forestry

Okafor, J.C. 1975a. The place of wild (uncultivated) fruitsand vegetables in the Nigerian diet. In: Proceedings of theNational Seminar on Fruits and Vegetables, edited by0.0. Ojehomon, PA. Town & K. Mclean, 153-171. Ibadan:National Horticultural Research Institute.

Okafor, J. C. 1975b. Varietal delimitation in Irvingiagabonensis (Irvingiaceae). Bulletin du Jardin Nationale Belgique,45, 211-221.

Okafor, J.C. 1978. Development of forest tree crops for foodsupplies in Nigeria. Forest Ecology and Management, 1,235-247.

Okafor, J.C. 1979. Edible indigenous woody plants in therural economy of the Nigerian forests zone. In: The Nigerianrainforest ecosystem, edited by D.U.0 Okal. (Proceedings ofthe MAB Workshop on the Nigerian Rainforest Ecosystem.)llpadan: University of lbadan.

40

Okafor, J.C. 1980a. Eclible indigenous woody plants in therural economy of the Nigerian forest zone. Forest Ecology andManagement, 3, 45-55.

Okafor, J.C. 1980b. Trees for food and fodder in thesavanna areas of Nigeria. International Tree Crops Journal, 1,131-141.

Okafor, J.C. 1981a. Woody plants of nutritional importance intraditional farming systems of the Nigerian humid tropics. PhDthesis, University of Ibadan, Nigeria.

Okafor, J.C. 1981b. Delimitation of a new variety of Treculiaafricana Decne. Subsp. africana (Moraceae). Bulletin du JardinNationale Belgique, 51, 191-199.

Okafor, J.C. 1983. Varietal delimitation in Dacryodes edulis(G. Don) H.J. Lam. (Burseraceae). International Tree CropsJournal, 2, 255-265.

Okafor, J.C. 1989. Agroforestry aspects. In: Cross RiverNational Park, Oban Division: plan for developing the Park andits support zone, Appendix no. 2. Godalming: World WideFund for Nature.

Okafor, J.C. 1990a. Agroforestry development in the BoshiOkwangwo Division of the Cross River National Park.Godalming: World Wide Fund for Nature.

Okafor, J.C. 1990b. Development and selection ofcommemially viable cultivars from forest species for fruit. In:Proceedings of the 12th Plenary Meeting of AETFAT Hamburg,edited by H.-D. Ihlenfeldt, 81-97. Gottingen: Hubert.

Okafor, J.C. 1991. Improving edible species of forestproducts. Unasylva, 165 , (42), 17-23.

Okafor, J.C. & Fernandes, E.C.M. 1987. Compound farmsof southeastern Nigeria. A predominant agroforestry homegarden system with crops and small livestock. AgroforestrySystems, 5, 153-168.

Okafor, J.C. & Okolo, H.C. 1974. Potentialities of someindigenous fruit trees of Nigeria. In: Proceedings of the 5thAnnual Conference. Forestry Association of Nigeria.(Unpublished.)

Okigbo, B.N. 1977. Neglected plants of horticulturalimportance in traditional farming systems of tropical Africa.Acta Horticulturae, 53, 131-150.

Ola-Adams, B.A. & Iyamabo, D.E. 1977. Conservation ofnatural vegetation in Nigeria. Environmental Conservation, 4,217-226.

Roche, L. 1975a. Major trends and issues in forestryeducation in Africa: a view from lbadan. CommonwealthForestry Review, 54, 166-175.

Roche, L. 1975b. The methodology of conservation of forestgenetic resources. (Miscellaneous publication, no. 75/8.) Rome:Food and Agriculture Organisation.

Roche, L. 1979. Forestry and the conservation of plants andanimals in the tropics. Forest Ecology and Management, 2,103-122.

Shepherd, G. & Okafor, J.C. 1991. Cameroon forestmanagement and regeneration project. socioeconomic surveyreport. A socioeconomic survey work carried out duringNovember 1991 in the villages around Mbalmayo ForestReserve. Proposals for the future, based on team findings.London: Overseas Development Institute.

Thomas, D.W., Thomas, J.M.C., Bromley, F.T. &Mbenkum, F.T. 1989. Korup ethnobotany survey Godalming:World Wide Fund for Nature.

Udeala, O.K., Onyechi 1.0. & Agu, S.I. 1980. Preliminaryevaluation of dika fat, a new tablet lubricant. Journal ofPharmaceuticals and Pharmacology, 32, 6-9.

Van Welzen, P.C., Lamb, A. & Wong, W.W.W. 1988. EdibleSapindaceae in Sabah. Nature Malaysiana, 13 (1), 10-25.

Whitmore, T.C. 1976. Natural variation and its taxonomictreatment within tropical tee species as seen in the Far EastIn: Tropical trees: variation, breeding and conservation, editedby J. Burley & B.T. Styles, 25-34. (Linnean Society Symposiumseries, no. 2.) London: Academic Press.

Wong, W.W.W. & Lamb, A. 1990. Species diversity of wildfruit trees in the forests of Sabah as illustrated by the generaArtocarpus, Durio and Mangifera. (Paper presented at the

International Conference on Forest Biology and Conservationin Borneo, Kota Kmabalu, Sabah, Malaysia.) (Unpublished.)

World Wide Fund for Nature. 1989. Crass River NationalPart Oban Division: plan for developing the Park and itssupport zone. Godatining: WWF

Zakri, A.H., Saw, L.G. & Rajanaidu, N. 1989. Nationalconservation strategies of plant genetic resources in Malaysia.In: Genetic resources of under-utilized plants in Malaysia.Malaysian National Committee on Plant Genetic Resources.

41

Trees for drugs and other extractives: futureprospects

P G WatermanPhytochemistry Research Laboratories, Department of Pharmaceutical Sciences, University of Strathclyde,204 George Street, Glasgow G I I XW UK

ABSTRACTHistorically plants and their extracts have been exploited as medicines and for a range of other uses, such as dyestuffs, perfumesand industrial adjuncts. In the majority of cases, supplies have been obtained by collection from wild plants. Relatively fewspecies have been cultivated to any great extent. Hevea and Cinchona are two exceptions. In the past, the absence of reliable

(Strophanthus spp.) or sustainable (Rauvo)tia serpentina) sources significantly harmed the success of potentially valuable medicinal

plants.

Currently in the search for new lead molecules for drug development, there is a major increase in the screening of plantextracts. The major developments taking place in screening technology (high throughput and very specific screening targets) willensure that re-assessment of the bioactivity of an extractive can take place over and over again in the coming years. Evidencefrom our own participationiin such screening exercises suggests that there will be a significant 'hit rate', demanding further, moredetailed, investigation of the active extracts.

For the Western pharmaceutical industry the follow-up to these hits will be directed toward identification of the activecomponent(s) and subsequent development of a 'manageable' product. 'Management' will, by preference, be by synthesis, but, ifa non-synthesisable natural product is required, it will be produced by biotechnology Only where both synthesis andbiotechnology fail would domestication be contemplated. There may be greater opportunities for domestication in relation toherbal medicines and non-medical products, such as perfumes.

INTRODUCTION

Historically plants and extracts or purecompounds isolated from them have been animportant source of medicinal substances, as wellas dyestuffs, perfumes, tannin substances andvarious industrial adjuncts and foodstuffs. In thispaper, the current situation is reviewed aridconsideration is given to-what might happen inthe future, restricting the discussion more or lessto tropical tree species and to their potentialvalue in the pharmaceutical and allied sectors.First, however, a few generalisations are madeabout attitudes and philosophies that currentlyexist with regard to plant-derived pharmaceuticals.

CURRENT ATTITUDES

As pointed out previously (Waterman 1989), therehas not, in the pharmaceutical context, been anydecrease in the use of drugs derived fromnatural sources during the period from 1930 to1988 because of the advent of antibiotics and •developments in the use of animal products andsemi-synthetic derivatives of natural products(notably antibiotics and steroidal drugs). Therehas, however, been a very considerable changein emphasis away from the use of powderedplant materials and extracts toward the use of

42

isolated pure compounds, because of the greatercontrol this allows in maintaining product qualityand dosage.

Today in many pharmaceutical companies, thereis a re-awakening of interest in the living worldas a source of leadcompounds for drugdevelopment programmes. This interest must notbe interpreted as a change in emphasis awayfrom synthetic chemistry When a drug discoveryteam identifies a novel compound withtherapeutic potential, the most likely responsewill be, simultaneously to find out how it worksand to find a way of synthesising it. If the resultsare of sufficient interest, this stage will then befollowed by a major synthetic programme aimedat producing analogues which maximise therequired activity while minimising side-effectsand toxicity The final marketed product willgenerally be wholly synthetic. Only on thoseoccasions where an isolated metabolite is thecompound of choice, or where the marketableproduct is a semi-synthetic derivative that defiestotal synthesis, would the original natural sourcebe likely d.to retain lasting importance.

One recent example is the new anti-cancer drugtaxol (Figure 1), which was originally obtainedby extraction from the bark of Taxus brevifolia(Taxaceae). The structural complexity of taxol

•••

Figure I. Chemical structure of taxol

0 0

o w..

OH

means that synthesis of the compound is notcurrently a viable option, but there is scope forproducing a wide range of semi-syntheticderivatives (Kingston, Samaranayake & Ivey1990). Unfortunately as it is estimated (Kelsey &Vance 1992) that 2000-3000 trees are needed toproduce 1 kg of taxol, there is currently asevere supply problem which is limitingavailability of the drug. As a result, other speciesof Taxus have been studied and taxol and/orallied compounds have now been obtained fromthe leaf as well as the woody parts, giving hopethat a supply can be maintained from a resourcethat is renewed annually (Kelsey & Vance 1992).However, at present, the only source for taxolrecognised by the US Food and DrugAdministration is the bark of T brevifolia.Harringtonine (Figure 2) from Cephalotaxusharringtonia (Cephalotaxaceae) is anotherpromising anti-tumour compound that seems topresent similar problems.

Undoubtedly a major effort will be made toproduce taxol under controlled laboratoryconditions. If this cannot be achieved bysynthesis, then tissue culture and otherbiotechnological methods will certainly beattempted. To date there has been much effort inthis area but limited success. Shikonin (Figure 3),a dye and antiseptic from Lithospermumerythrorhizon (Boraginaceae), berberine fromCoptis japoriica (Ranunculaceae), and thealkaloids of ergot (Claviceps purpurea) arecurrently produced commercially by tissueculture (Samuelsson 1992), but it is fair to saythat, despite all the efforts that have gone intothis technique, there have been relatively fewsuccesses. Given all the problems confrontingproduction, there has been some suggestion ofcultivating T brevifolia, but as yet there is noserious programme to bring this about.

AcO0

= HO

0

A c 6

These arguments reflect the attitudes of Westernpharmaceutical companies producing productsthat are aimed primarily at a developed worldmarket. The option of a developing countrybeing able to supply its own pharmaceuticalrequirements by growing medicinal plants is avery different matter, as evidenced by theimportance placed on growing medicinal plantsin the People's Republic of China (Hildno 1989)and in India (Aslam 1989). The Chinese, inparticular, are also great importers of medicinal

0

<0

0

OH

0

0OMe

Me0OH

Figure 2 Chemical structure of harringtonine

OH 0 HO H

I

OH 0

Figure 3. Chemical structure of shikonin

43

plants from other parts of the world, and there isobviously considerable scope for improving thequality of materials to meet this demand.

If the preferred strategy of the Westernpharmaceutical industry seems largelyunsympathetic to tropical forestry andagroforestry the opposite is true for otherproducts which can be labelled as 'green' orherbal medicines and health foods. Whatever thescientific basis for the advocacy of thesematerials, the fact they they are 'natural' is amajor attraction to a subset of the population ofthe developed world. The problems with manysuch products is that their popularity is oftentransient, which is not conducive to long-termdevelopment. A 'fad' is all too often accompaniedby the rapid exploitation of limited naturalresources by haphazard collecting from the wildpopulation. One example, seen in Kenya in the1980s, resulted in the harvesting of Aloe species,such as A. marsabitensis, to a state ofnear-extinction. However, where interest issustained, as in the case of ginseng (Panaxspp.), then there is very considerable scope forthe development of cultivated material (Evans1989).

There is currently a more balanced situation inallied areas, such as perfumery and in spice andgum production, where stable, long-term marketsexist. Among the most important spice trees arecinnamon (Cinnamomum verum) which is nowextensively grown in Indonesia, Sri Lanka and theSeychelles, nutmeg (Myristica fragrans) fromIndonesia, Malaysia, Sri Lanka and Grenada, andclove (Syzygium aromaticum) for which MalagasyBrazil and Malaysia (but not Zanzibar) are nowthe most important sources. Otherwell-established crops are those producing gums,particularly for confectionery and tanningmaterials.

THE PRESENT SITUATION

Pharmaceuticals, resins and major spices

In the past, there have been some notablesuccesses in the domestication of trees

44

MOD

Figure 4. Chemical structure of quinine (RI=OH, R2=H) andquinidine (R1=H. R2=0H)

containing pharmaceuticals, the most widespreadbeing a number of species of Cinchona(Rubiaceae), the source of quinine (the firstanti-malarial), and quinidine (useful against someforms of heart disease) (see Figure 4). It isarguable whether the protection that Cinchonabark extracts provided against the ravages ofmalarial fever contributed very significantly to thecolonisation of the tropics by European peoples.Originating in South America, Cinchona wasinitially collected from wild plants and soonbecame locally extinct. Seeds obtained by theDutch were taken to Java where through anextensivebreeding programme hybrids wereproduced with a much-elevated alkaloid content(pharmaceutical-grade bark must have aminimum of 6% alkaloids, and some strains arerecorded as reaching the astonishing level of15%). Because of the other use of quinine, asthe bitter principle in tonic water, its cultivationand production continue to be a major industryIn addition to Indonesia, there is nowappreciable production in Bolivia, Guatemala andthe Congo, and plantations exist in many othercountries.

The continued buoyancy of the Cinchona industryis not, unfortunately matched by many of theother drug plants that have been importantcomponents of the materia medica. A list ofsome of the tree species of the tropics whichhave been or are still cultivated for medicinalproducts is given in Table 1. Their use isgenerally in decline in the developed world astheir products become less important inmedicine (senna and the spices are exceptions),or occasionally as in the cases of tragacanth andacacia gum, demand outstrips supply

Some materials, such as the burseraceous resinsfrom Commiophora and Boswellia species(franldncense, myrrh, bissabol, opopanax), mayhave only small Western markets but are inheavy demand in the Far East, both as incenseand as ingredients in Chinese medicine.According to Thu lin and others (Thu lin & Warfa1987; Thu lin & Claeson 1991), they are amongthe major exports of Somalia and are alsocollected extensively in north-east Kenya andsouthern Ethiopia. Throughout their range thecollection of these resins appears to be basedmore or less entirely on wild plants, and there iseither a total lack of, or inadequate, qualitycontrol of what is collected and sold, except inEthiopia. As a consequence, material for exportfrom Somalia and Kenya is almost always ofvariable quality a fact that buyers are able toexploit to keep the price low. This seems to bea classic example of an industry that wouldbenefit from rationalisation, selection anddevelopment of appropriate varieties andcultivation, and from the production of standards

Table 1. Pharmaceutical products (with uses, where appropriate) obtained from culfivated tree species in tropical or subtropicalcountries (source: Evans 1989)

Species (family)

Styrax benzoin (Styracaceae)

Styrax paralleloneurus

Myroxylon balsamum (Leguminosae)

Myroxylon pereirae

Liquidambar orientalis (Hamamelidaceae)

Astragalus sp. (Leguminosae)

Acacia senegal (Leguminosae)

Sterculia urens (Sterculiaceae)

Cassia senna (Legurninosae)

Rhamnus purshianus (Rhamnaceae)

Myristica fragrans (Myristicaceae)

Cinnamomum verum (Lauraceae)

Cinnamomum caqsia

Cinnamomum camphora

Syzygium aromaticum (Myrtaceae)

Eucalyptus spp. (Myrtaceae)

CommiPhora myrrha (Burseraceae)

Boswellia carteri (Burseraceae)

Garcinia hanburii (Guttiferae)

Pistacia lentiscus (Anacardiaceae)

Quillaja saponaria (Rosaceae)

Picrasma excelsa (Simaroubaceae)

Prunus africana (Rosaceae)

Erythroxylum coca (Erythroxylaceae)

Cephaelis ipecacuanha (Rubiaceae)

Physostigma venenosum (Leguminosae)

Strychnos nux-vornica (Loganiaceae)

Rauvolfia serpentina (Apocynaceae)

Rauvolfia vomitoria

Voacanga spp. (Apocynaceae)

Pausinystalia johimbe (Rubiaceae)

Cinchona sp. (Rubiaceae)

Pilocarpus sp. (Rutaceae)

*Probably collected entirely from wild populations

Product

Balsamic gum

Balsamic gum

Balsamic gum

Balsamic gum

Balsam

Tragacanth gum

Acacia gum

Stemulia gum

Purgative

Purgative

Carminative, flavour

Carminative, flavour

Carminative, flavour

Rubefacient

Antiseptic, stimulant

Decongestant

Antiseptic

Incense, fumigant

Veterinary purgative

Medical paints

Emulsifier

Bitter tonic

For prostate treatment

For ENT surgery

Expectorant, emeticfor amoebic dysentery

Anti-mydriatic

Respiratory stimulant

Anti-hypertensive

Anti-hypertensive

Anti-hypertensive

For sexual inadequacy

Anti-malarial, bittertonic, for carcliac disease

For glaucoma

for identifying and quantifying the monoterpene,sesquiterpene and diterpene components of theresins.

As already observed, there is currently anupsurge of interest in medicinal plants and theirproducts; one of the most tangible results is thepresent interest in taxol. Unfortunately the taxolsituation points out the problems for plantbreeders when faced with the sudden rise toprominence of a previously obscure species. Thepharmaceutical industry needs taxol now, not in10-15 years' time. The situation with taxol isappreciably different from that which followed thediscovery of the last major anti-cancer drugs tobe derived from a plant source, vinblastine andvincristine (Figure 5), which were isolated fromthe Madagascan periwinkle (Catharanthus roseus,Apocynaceae). These have proven, as taxolcurrently appears to be, intractable to synthesis.However, the Madagascan periwinkle is a weedthat will grow prolifically in more or less anymoist tropical or subtropical climate. Yet, despite

Me0

Source

Indonesia

Thailand

Cuba, Colombia*, Venezuela*

Guatemala, Honduras, San Salvador

Turkey

'Ilirkey*, Iran*, Iraq*

Sudan

India, Pakistan

Sudan, India

N America*

Indonesia, Malaysia, Sri Lanka, Grenada

Sri Lanka (widespread)

S China

S China, Japan, India, Georgia

Madagascar, Indonesia, Brazil, Tanzania

Widespread

Somalia*, Ethiopia*

Ethiopia*, Yemen*

SE Asia*

Greece

Chile*, Peru*, Bolivia*, USA, India

W Indies*

Africa*

Peru, Bolivia, Colombia, Indonesia

Brazil*, India

W Africa*

India

India, Pakistan

Africa*

Madagascar*, Indonesia*

W Africa*

Indonesia, Tanzania, Kenya, Guatemala,Bolivia

S America

OH,,,,,,

iHI OAc

I"COOMe

N H OH

lit

Figure 5. Chemical structure of vinblastine (R=Me) andvincristine (R=CHO)

45

the ease of cultivation, there have still beenenormous efforts to produce these alkaloidsunder controlled conditions: in this case, throughtissue culture. It has proved to be a difficult andfrustrating process, and is still not a viableproposition after some 20 years of research.

The demand for an immediate supply of a plantspecies and the pressures this demand placeson natural populations are currently being seenwith the harvesting of seeds and bark ofVoacanga species (Apocynaceae) and the bark ofPrunus (Pygeum) africana (Rosaceae) in tropicalAfrica. Both provide extracts that are being usedin medicines popular in continental Europe; bothare collected entirely from wild plants. Accordingto Cunningham (1990), during the 1980s anaverage of about 800 tonnes of Prunus africanabark were exported annually from Cameroon toFrance, a level of output that can no longer beSustained from easily accessible material. Thesetwo taxa, and Prunus in particular, while notimmediately endangered, will be lost over partof their ranges if the problem of indiscriminatecollection is not overcome. This is the challengethat tropical forestry faces: what can it do toanswer the demand for species like Prunus?

Gums

There are major problems with the traditionalsources of natural gums. Tragacanth, fromAstragalus species (Leguminosae), has neverbeen widely cultivated; over-collection andpolitical problems (native to Iran and Iraq) haveled to a major reduction in its availabilityProduction of acacia gum, from Acacia senegal(Legurninosae), is centred in Sudan, but currentsupply is unable to keep pace with demand, aproblem which is magnified by political andworsening climatic conditions. In response to theproblems that have affected tragacanth andacacia production, there has been a considerableupsurge in the use of the gum of Sterculia urens(Sterculiaceae), a species presently cultivatedprimarily in India. Where supplies of gums fail,semi-synthetic materials (methoxycellulose) orby-products of crop plants (legume seed gums)have been taking an increasing market share, butthe traditional products are still sold at apremium price. The traditional gums are allproduced as exudates from injured bark. Twofurther species of Leguminosae also producegums of commercial value: Ceratonia siliqua(carob) and Cyamopsis tetragonolobus (guar).Both guar and carob gums are obtained fromthe seed.

At first sight, the market seems ripe for theintroduction of further natural gums. However,there is another set of problems which must beconfronted by someone wishing to produce anew product to replace and compete with thetraditional gums, which are at present approved

46

by regulatory authorities. Acacia gum, forexample, if it were to appear on the market asa new product today would have to run thegauntlet of examination by bodies such as theFood and Drug Administration in the USA. Theremust be severe doubts that it would proveacceptable in the raw state on the grounds ofvariable quality and microbial contamination. Itescapes censure primarily on the basis ofprolonged use without strong evidence ofdeleterious effects. To overcome thesepresentation difficulties, spray-dried acacia,produced from imported crude material, isbecoming common in Western countries; onesuspects that any new large-scale introductionswould need to be presented in somecomparable, modified, form.

Because of the pitfalls that surround theintroduction of a natural replacement for theedible gums, the answer seems to lie with thedevelopment of better management and controlof those currently in use. Of these the mostimportant is acacia. There certainly seems to beplenty of scope for such developments.Astonishingly according to Evans (1989), 'apartfrom some "acacia gardens", wild, self-sownplants are the main source of the gum', withSudan and, to a much lesser degree, Senegaland Nigeria providing the best product. Giventhe shortfall in supply and the apparently safelong-term need for such gums (for the foodindustry as well as pharmaceuticals), acacia, andperhaps some of the others as well, seem to bestrong candidates for development as crops forarid and semi-arid areas in the Sahel.

Tannins

Historically the use of plant-derived tanninextracts to make leather has been a major itemof commerce throughout the world. Numerousspecies have been used to produce tannin, fromvarious plant parts (barks, woods, leaves, fruitsand galls). Most of the major sources have beenlisted by Has lam (1989).- Among the mostimportant are Acacia species (wattle, babul,cutch), Eucalyptus species (wandoo), Cassiaspecies (avaram), Caesalpinia species (divi-divi,algarobilla, tara), Terminalia species (myrobalans),Quercus (valonea, aleppo, garoilille), variousAnacardiaceae (Schinopsis — quebracho; Rhus,Anogeissus—sumach), Castanea species (chestnut),Uncaria• species (gambier), and variousmonocotyledonous tree species (Picea, Tsuga,Pinus, Larix).

In the UK, vegetable tanning has now beensupplanted to a large degree by mineral tanning.Tannin from vegetable sources (most commonlyfrom oak, Quercus spp.) is used largely forspeciality fashion wear and for orthopaedicequipment (Haslam 1989). More vegetable tanninis now probably imported into the UK for

commercial uses other than for use in thetanning industry However, elsewhere in theworld, vegetable tannins remain important. In theUSA, about 15% of all leather tanning still usesvegetable materials (Seigler et al. 1986). In 1976,approximately 18.5 million lb (8400 t) of wattle(Acacia meanisit) derivatives, 30 million lb(13 600 t) of quebracho and a small amount ofEuropean chestnut were imported into the USAto meet the need for vegetable tannins. India isestimated to import some US$124M worth ofvegetable tannin, which is made up primarily ofwaffle, quebracho, oak galls, gambir andmyrobalans (Devi & Prasad 1991).

One of the major sources of wattle for theAmerican market has been Kenya, which in 1978had 14.3 kha under plantation with Acaciameanisii. However, despite the fact that the priceof waffle increased five-fold between 1977 and1986, the area under waffle plantationsdecreased in the same period to 9.8 kha(Mugedo & Waterman 1992) because of theneed to convert to food-producing farmland. Ifcurrent trends continue, then Kenya will soonneed to import vegetable tannin to support itsown tanning industry rather than export it.

As a result of the realisation that there will soonbe a serious shortfall in vegetable tannage,attempts have been made to identify newsources. The focus has been Acacia, with surveysbeing carried out of species indigenous to theUSA (Seigler et al. 1986), India (Devi & Prasad1991) and Kenya (Mugedo & Waterman 1992). Ineach case, species have been located withpromising yields of tannins and which will growin situations unsuitable for the growth of foodcrops (usually semi-arid environments). There isclearly considerable scope, as well as need, forthe development of new sources of tannins.However, this development must be donecarefully and with close co-operation betweentannin chemist and forester. It is not just thequantity but also the quality of tannin producedthat is important in selecting species, andassessing quality in a tannin is a very difficultprocedure.

THE FUTURE

At the present time, the potential for drugdiscovery based on natural products is higherthan it has been at any point in the past 30years. The development of high-throughputscreens and the identification of target receptorsand enzymes on which to base highly selectivescreens, are the subject of major efforts by drugcompanies. Many of these screens are capableof assaying crude extracts, as well as isolatednatural products or synthetic chemicals. Eachtime a new screen target is developed, theneach extract needs to be tested again,irrespective of past record of activity or inactivity

To help feed these screens, many drugcompanies are accessing natural biodiversity Acurrent topical example is the agreementbetween the American company Merck and theInstitut Biodiversidad of Costa Rica (INBio),whereby Merck have agreed to pay INBioUS$1M for exclusive access (for a limitedperiod) to an agreed number of species (plantand animal). The payment will be used to helpconserve and protect Costa Rican biodiversityThe desired outcome is the identification ofuseful bioactivity and the isolation, synthesis andpatenting of an active compound. The key isaccess to maximum biodiversity and only whenthere is a coincidence of:

• optimal bioactivity in an extract or naturalproduct, and

• intractability to synthesis

will there be any chance of cultivation for a newproduct. Even when this coincidence occurs, itseems increasingly likely that the solution will befound in the realms of biotechnology and not intraditional methods of production.

In looking to the future and trying to identifynew opportunities for forestry it seems best todisregard the Western pharmaceutical industryHowever, there are plenty of opportunities inother areas. These include:

• improving the production of alreadyestablished crops for food andpharmaceutical industries, notably amonggums and resins for both Western and FarEastern markets;

• identifying regional needs in the developingworld and ensuring continuity of supply ofdrugs for local markets; and

• identifying and establishing new sources oftannins.

REFERENCESAs lain, M. 1989. Asian medicine and its practice in Britain.In: Pharmacognosy, edited by W C. Evans, 704-725. London:Bail liere Tindall.

Cunningham, A.B. 1990. African medicinal plants: settingpriorities at the interface of conservation and primary healthcare. (WWF Report, no. 3331, Plants Conservation Section.)Gland: World Wide Fund for Nature.

Devi, S.R. & Prasad, M.N.V. 1991. Tannins and relatedpolyphenols from ten common Acacia species of India.Bioresource Technology, 36, 189-192.

Evans, W.C., ed. 1989. Pharmacognosy London: BailliereTindall.

Has lam, E. 1989. Plant polyphenols: vegetable tanninsrevisited. Cambridge: Cambridge University Press.

Hikino, H. 1989. Oriental drugs. In: Pharmacognosy, editedby W C. Evans, 726-738. London: Bailliere Tindall.

Kelsey, R.G. & Vance, N.C. 1992. Taxol andcephalpmannine concentrations in the foliage and bark ofshade-grown and sun-exposed Taxus brevifolia trees. Journal ofNatural Products, 55, 912-917.

47

Kingston, D.G.I., Samaranayake, G. & Ivey, C.A. 1990.The chemistry of taxol, a clinically useful anticancer agent.

Journal of Natural Products, 53, 1-12.

Mugedo, & Waterman, P.c. 1992. Sources of tannin:alternatives to wattle (Acacia mearnsh) among indigenousKenyan species. Economic Botany 46, 55-63.

Sarnuelsson, G. 1992. Drugs of natural origin. Stockholm:Swedish Pharmaceutical Press,

Seigler, D.S., Seilheimer, S., Keesy, J. & Huang, H.F.1986. Tannins from four common Acacia species of Texas and

Northeastern Mexico. Economic Botany 40, 220-232.

48

Thu lin, M. & Claeson, P. 1991. The botanical origin ofscented myrrh (bissabol or habak hadi). Economic Botany 45,487-494.

Thu lin, M. & Warfa, A.M. 1987. The frankincense trees(Boswellia spp., Burseraceae) of northern Somalia andsouthern Arabia. Kew Bulletin, 42, 487-500.

Waterman, P.G. 1989. Bioactive phytochemicals - the search

for new sources. In: New crops for food and industry, editedby G.E. Wickens, N. Hag & P Day 378-390. London:Chapman & Hall.

Techniques to domesticate trees

In vitro systems for the conservation andutilisation of germplasm

F EngelmannOffice de la Recherche Scientifique et Tbchnique d'Outre-Mer, Centre de Montpellier, 2501 Avenue du Valde Montferrand, BP 5045, 34031 Montpellier Cedex 1, France

ABSTRACTThe use of in vitro techniques is essential for the safe conservation of numerous plant species which have recalcitrant seeds orare vegetatively propagated. For medium-term conservation, growth reduction is achieved by modifying various environmentalvariables (temperature, culture medium, gaseous environment). These techniques are now routinely employed in many laboratories.For long-term conservation, cryopreservation (in liquid nitrogen at -196°C) is the only method available. Cryopreservationprotocols have been set up for more than 70 plant species under various forms (cell suspensions, calluses, meristems, somaticand zygotic embryos). However, the large-scale application of these techniques is exceptional at present. Conventionalcryopreservation protocols are generally complex and comprise numerous successive steps which have to be determined foreach new material. In the past two years, simpler techniques (encapsulation/dehydration, desiccation, vitrification) have beendeveloped which may facilitate the long-term conservation of the germplasm of numerous plant species.

INTRODUCTION

For seed storage purposes, plant species havebeen divided into two categories (Roberts 1973):those with orthodox seeds and those withrecalcitrant seeds. Orthodox seeds can withstandextensive dehydration to low water contents andcan be stored in the dry state at lowtemperatures for extended periods of time.Recalcitrant seeds, which have a high moisturecontent, are sensitive to dehydration and remainviable only for a short time (weeks or months),even if stored optimally (high humidity andtemperature). Species with recalcitrant seeds arepredominantly of tropical and subtropical originand include different groups of plants (fruit trees,beverage crops, oleaginous crops, spices,horticultural and medicinal species) which areoften of great economic value. Long-term seedstorage is also difficult for many forest treesbecause they have a very long juvenile phaseand do not produce seeds for several years. Theconservation of species which reproduceasexually also poses considerable problems.

PRESERVATION OPTIONS

In situ conservation is often problematic becauseof the widespread destruction of naturalvegetation and the disappearance of large wildareas. Ex situ conservation is limited by variousparameters, notably the size of the samplesnecessary to retain genetic diversity (from 20plants for a single population to 5000-20 000plants to maintain the heterozygosity of aspecies). Consequently the land requirementsand the cost of maintaining the ex situ field sites

can become very extensive, particularly in thecase of large plants such as forest trees.Furthermore, plants preserved in naturalconditions remain exposed to pests, pathogens,natural and man-made disasters such ashurricanes, fire, etc, which can be especiallyserious for ex situ conservation areas. Analternative, therefore, is to developlaboratory-based storage systems, such as seedstores (see Tompsett, pp61-71) or in vitro culture.

IN VITRO CULTURE FOR GERMPLASMPRESERVATION

The use of in vitro techniques, which have nowbeen applied to more than 1000 plant species, isof great interest for the collection, storage,exchange and multiplication of plant germplasm.Such techniques are particularly useful forrecalcitrant seeds or for species whichreproduce asexually Tissue culture techniqueshave the additional advantage that they providethe opportunity for very high multiplication ratesin an aseptic environment. Moreover, the spacerequirements and labour costs are veryconsiderably less than those of in vivo systems.Finally tissue culture techniques greatly facilitatethe international exchange of plant germplasm,reducing the volume to be shipped andsatisfying phytosanitary regulations by providingsterile conditions.

Depending on the duration of storage required,the conservation methods employed vary Forshort- or medium-term storage, the aim is toreduce the growth rate of the plant material,usually by lowering the culture temperature. For

49

long-term conservation, cryopreservation in liquidnitrogen (-196°C) arrests all metabolic eventsand cell divisions. The plant material can thus bestored without any alteration for unlimitedperiods of time. In this instance, the cultures arealso stored in a small volume and protectedfrom contamination.

There is an extensive literature on in vitroconservation techniques for plant germplasm,covering general aspects, the use of differentculture systems (Kartha 1985; Dereucklre &Engelmann 1987) and their application todifferent groups of plants, such as woody andhorticultural species (Aitken-Christie & Singh1987; Engelmann 1991a) and tropical crops(Zalai et al. 1991; Engelmann 1991b). This paperbriefly presents the classical short- andlong-term storage techniques, and then discussesmore extensively the new in vitro conservationtechnologies and their potential for germplasmconservation, focusing on tropical woody species.

USE OF IN VITRO TECHNIQUES FORCOLLECTING GERMPLASM

The use of in vitro methods when collectinggermplasm has many advantages andapplications (Withers 1987). For example, thesetechnologies are ideal when only vegetativeexplants are available, when clonal genotypesare required, or when collecting missions fail tocoincide with the time of seed production.Similarly the collection of mature embryos of -recalcitrant seeds can help to prolong theirviability and, at the same time, reduce their bulk.In vitro techniques can be carried out underfield conditions using minimal equipment. Todate, in vitro techniques have been used tocollect germplasm of Gossypium sp., Manihot sp.and Cocos nucifera. They are also being testedfor other species such as Citrus, Theobromacacao, Musa spp., forage grasses and temperatefruits; techniques are under development forPersea americana and Vitis spp., and are plannedfor Coffea spp. (Withers 1991).

SHORT- AND MEDIUM-TERM STORAGEUSING SLOW GROWTH

Technical approaches

Various conditions can be modified to reducethe growth of plant cell cultures and to extendthe intervals between successive transfers. Themost successful storage conditions for slowgrowth are those which result in as fewmodifications as possible, in comparison withstandard culture conditions. In most cases, theculture temperature is reduced. For example,with temperate woody species, the temperatureis decreased to between 2°C and 20°C, allowinga conservation duration ranging between tenyears and four months, depending on the

50

species (Mascarenhas & Agarwal 1991). Fortropical species which are generallycold-sensitive, other modifications have to besought. Slow-growing species such as Coffea canbe stored in standard conditions and transferredonly every six months to one year,(Bertrand-Desbrunais & Charrier 1990).Alternatively the composition of the culturemedium can be modified by reducing thecontent of mineral elements and/or sugar, oradding compounds such as osmoticums orgrowth retardants (Engelmann 1991b). Anotheroption is to reduce the quantity of oxygenavailable to the cultures. This reduction can beachieved by covering the material with mineraloil, as with calluses (Augereau, Courtois &Pétiard 1986) or shoot cultures of Coffea (Jouve,Engelrnann & Charrier 1991) and Costus(Dekkers, Rao & Goh 1991). The use ofcontrolled gaseous atmospheres has beensuccessfully tested by Bridgen and Staby (1981)with Nicotiana tabacum plantlets and byEngelmann (1990) with somatic embryos ofElaeis guiheensis.

Desiccation represents another promisingapproach. It was first investigated by Nitzsche(1980) with carrot calluses. Recently McKersie etal. (1990) and Lecouteux et al. (1992)successfully stored desiccated somatic embryosof Medicago sativa and Daucus carota,respectively for eight months.

Present development of slow-growth storagetechniques

Slow-growth techniques are presently usedroutinely in many commercial laboratories, aswell as in conservation centres for the storage ofgermplasm as shoot cultures. The assumption isthat they are considered the best methods.However, there are no well-documented dataavailable concerning the stability of shootcultures or conformity to type during storage.Similarly nothing is known of the possibleselection and adaptation processes which maytake place during the prolonged conservation ofshoot cultures (Withers 1991). Other types ofmaterial, such as callus, can be interesting forconservation purposes, but are known to be veryunstable, even under slow-growth conditions(Seitz 1987). There is, therefore, a need todevelop alternative conservation techniques.

LONG-TERM STORAGE USINGCRYOPRESERVATION

Technical approaches

Up until now most plant material has beencryopreserved using classical techniques,involving a cryoprotective treatment andcontrolled freezing (Dereuddre & Engelmann1987). In the past three years, new techniques

such as vitrification, encapsulation/dehydrationand desiccation have been developed. Thesefacilitate the cryopreservation of numerous planttissues (Dereuddre 1992), such as cellsuspensions, calluses, meristems and embryos,which are considered separately below. Thedifficulty of defining an efficient cryopreservationprocess increases with the structural complexityand cellular heterogeneity of the material.

Cell suspensions and calluses

Cryopreservation of cell suspensions is stillroutinely performed in many laboratories, usingor adapting the classical protocol (includingpre-growth and slow cooling) proposed byWit-hers and King (1980), although simplerfreezing techniques can be developed inparticular cases (Tessereau et al. 1990). Similartechniques, derived from those used with cellsuspensions, can be applied to calluses.

Meristems

Meristems are very complex structures:consequently the application of standardtechniques generally leads to poor andnon-reproducible results. Dramatic improvementshave now been made using vitrification andencapsulation/dehydration (Dereuddre 1992).During the vitrification procedure, theexperimental conditions induce the formation ofan amorphous glassy structure which replacesthe intracellular water, thus avoiding thedetrimental formation of ice crystals which candisrupt cellular structure. Vitrification has beennotably applied recently to different cultivars ofMalus and Pyrus meristems (Niino et al. 1992).However, the high toxicity of the cryoprotectivesubstances employed may reduce the routineapplication of this technique.

Encapsulation/dehydration uses synthetic seedtechnology encapsulating the structure to befrozen in an alginate .bead which protects itduring the successive steps of thecryopreservation procedure. This approach seemsthe most promising in the case of meristems,and has enabled high survival rates with shoottips to be obtained repeatedly . notably with Pyrus(Dereuddre et al. 1990), Vitis (Plessis, Leddet &Dereuddre 1991) and Eucalyptus (Monod et al.1992). Finally partly desiccating stem segmentsof Asparagus allowed them to withstand freezingin liquid nitrogen (Uragarni, Sakai & Nagai 1990)

Embryos

Zygotic and somatic embryos also representcomplex structures for cryopreservation. Forthese tissues, especially somatic embryos, thefreezing procedure consists of the classicaltechnique. A simpler process, involvingdesiccation followed by rapid freezing, is mainlyused for zygotic embryos. A combination of bothmethods is used in particular cases. The

cryopreservation of woody plant embryos hasbeen applied to a wide range of species (seereview by Engelmann 1992) and most recently tozygotic embryos of Coffea (Normah &Vengadasalam 1992; Abdelnour-Esquivel,Villalobos & Engelmann 1992) and Citrus(Normah & Hamidah 1992).

In summary cryopreservation techniques havenow been developed for a wide range ofspecies, including many tropical crops(Engelmann 1991b). However, these techniqueshave only rarely been applied to these crops.One successful example concerns thecryopreservation of somatic embryos of around150 oil palm clones (Engelmann 1991c).

CONCLUSIONS

In vitro conservation techniques for the safestorage of plant germplasm have becometechnically more feasible, particularly as a resultof recent progress in cryopreservation. However,for many tropical tree species, there are severalprerequisites which have to be fulfilled beforethe application of cryopreservation techniques ispossible. A minimal knowledge of the biologyand physiology of the species to be conservedis needed in order to define the necessary invitro culture conditions. Moreover, the germplasmmust be evaluated, in order to store arepresentative sample of the genetic variability ofthe species. Finally practical problems such asthe provision of minimal tissue culture facilities,trained personnel, etc, have to be overcome inorder to apply these techniques in many tropicalcountries. In the near future, efficient techniquesfor the reliable storage of tropical treegermplasm should be developed with thecontinuing efforts of various internationalinstitutes.

REFERENCES

Abdelnour-Esquivel, A., Villalobos, V. & Engelmann, F.1992. Cryopreservation of zygotic embryos of Coffea spp.Cryo-Letters,. 13, 297-302.

Aitken-Christie, J. & Singh, A.P. 1987. Cold storage oftissue culture. In: Cell and tissue culture in forestry vol. 2.Specific principles and methods, growth and development,edited by J.M. Bonga & D.J. Durzan, 285-304. Dordrecht:Nijhoff.

Augereau, J.M., Courtois, D. & Pétiard, V. 1986. Longterm storage of callus cultures at low temperatures or undermineral oil layer. Plant Cell Reports, 5, 372-376.

Bertrand-Desbrunais, A. & Charrier, A. 1990.Conservation des ressources génetiques cafeières envitrothéque. In: Proceedings of the 13th International ScientificColloquium on Coffee, Paipa, Colombia, 1989, 438-447. Paris:ASIC.

Bridgen, M.P. & Staby, G.L. 1981. Low pressure and lowoxygen storage of Nicotiana tabacum and Chrysanthemum xMorifolium tissue cultures. Plant Science Letters, 22, 177-186.

51

Dekkers, A.J., Rao, A.N. & Goh, C.J. 1991. In vitrostorage of multiple shoot cultures of gingers at ambienttemperatures of 24-29°C. Scientia Horticulturae, 47, 157-167.

Dereuddre, J. 1992. Cryopreservation of in vitro cultures ofplant cells and organs by vitrification and dehydration. In:Reproductive biology and plant breeding, edited by Y Dattée,C. Dumas & A. Gallais, 291-300. Berlin: Springer.

Dereuddre, J. & Engelmann, F. 1987. The use ofcryopreservation for setting up banks of plant germplasm. In:Cell culture techniques applied to plant production and plantbreeding, edited by J. Boccond-Gibod, A. Benbadis & K.C.Short, 48-78. Angers: ENITPH.

Dereuddre, J., Scottez, C., Arnaud, Y. & Duron, M.1990. Resistance d'apex axillaires de poirier (Pyrus communisL.) a une déshydratation puis a une congélation dans l'azoteliquide: effet d'un tTaitement au froid des vitroplants. ComptesRendus de l'Académie des Sciences, Paris, 310 (Série III),317-323.

Engelmann, F. 1990. Utilisation d'atmosphéres a teneur enoxygéne réduite pour la conservation de cultures d'embryonssomatiques de palmiers a huile (Elaeis guineensis Jacq.).Comptes Rendus de l'Académie des Sciences, Paris, 310 (SerieHI), 679-684.

Engelmann, F. 1991a. In vitro conservation of horticulturalspecies. Acta Horticulturae, 298, 327-334.

Engelmann, F. 1991b. In vitro conservation of tropical plantgermplasm - a review Euphytica, 57, 227-243.

Engelmann, F. 1991c. Le sdéveloppement actuel de lacryoconservation des embryons' somatiques de palmier ahuile. In: Proceedings of the 18th International RefrigerationCongress, Vol. IV New Challenges in Refrigeration, Montréal,1991, edited by International Institute of Refrigeration,1676-1680. Paris: International Institute of Refrigeration.

Engelmann, F. 1992. Cryopreservation of embryos. In:Reproductive biology and plant breeding, edited by Y Dattée,C. Dumas & A. Gallais, 281-290. Berlin: Springer.

Jouve, L., Engelntann, F. & Charrier, A. 1991. Effets del'hypoxie et de la temperature sur la conservation in vitro depousses feuillées de Coffea arabica L. Café Cacao Thé, 35,205-210.

Kartha, K.K., ed. 1985. Cryopreservation of plant cells andorgans. Boca Raton: CRC Press.

Lecouteux, C., Tessereau, H., Florin, B., Courtois, D. &Pétiard, V. 1992. Conservation d'embryons somatiques decarotte (Daucus carota L.) par déshydratation apres .prétraitement avec du saccharose. Comptes Rendus del'Académie des Sciences, Paris, 314 (Serie III), 423-428.

Mascarenhas, A.F. & Agarwal, D.C. 1991. Prospects for invitro conservation of tree germplasm. In: Conservation of plantgenetic resources through in vitro methods, edited by A.H.Zakri, M.N. Normah, M.T Senawi & AG. Abdul Karim,231-252. Kuala Lumpur: Forest Research InstituteMalaysia/Malaysian National Committee on Plant GeneticResources.

McKersie, B.D., Senaratna, T., Bowley, S.R., Brown,D.C.W, Kielly A., Krochko, J.E. & Bewley, J.D. 1990.Artificial seeds: application in the production of hybrid alfalfa(Medicago sativa L.). In: Abstracts from the VIlth InternationalCongress on Plant Tissue and Cell Culture, Amsterdam, editedby H.J.J. Nijkamp, L.H.W Van der Plas & J. Van Aartrijk, 259.Dordrecht: Kluwer.

52

Monod, V., Poissonier, M., Paques, M. & Dereuddre, J.1992. Cryopreservation of shoot-tips of in vitro plantlets ofeucalyptus after encapsulation and air dehydration. In:Abstracts from CRY0'92, Society for Cryobiology 29th AnnualMeeting, Ithaca, 1992. Cryobiology, 29, 737-738.

Niino, T., Sakai, A., Yakuwa, H. & Nojiri, K. 1992.Cryopreservation of in vitro grown shoot tips of apple andpear by vitrification. Plant Cell, Tissue and Organ Culture, 28,261-266.

Nitzsche, E.H. 1980. One year storage of dried carrotcallus. Zeitschrifi fiir Pflanzenphysiologie, 100, 269-271.

Normah, M.N. & Hamidah, S. 1992. Cryopreservation ofGarcinia mangostana L. and Citrus halimii embryos. In: Bookof poster abstracts. Proceedings and Abstracts from the XIIIthEUCARPIA Congress on Reproductive Biology and PlantBreeding, edited by Y Daltée, C. Dumas & A. Gallais, 423-424.

Norrnah, M.N. & Vengadasalarn, M. 1992. Effect ofmoisture content on cryopreservation of Coffea and Vignaseeds and embryos. Cryo-Letters, 13, 199-208.

Plessis, P., Leddet, C. & Dereuddre, J. 1991. Resistance ala déshydratation et a la congélation dans l'azote liquided'apex enrobés de vigne (Vitis vinifera L. cv Chardonnay).Comptes Rendus de l'Académie des Sciences, Paris, 313 (SérieIII), 373-378.

Roberts, E.H. 1973. Predicting the viability of seeds. SeedScience and 7bchnology, 1, 499-514.

Seitz, U. 1987. Cryopreservarion of plant cell cultures. Plantamedica, 4, 311-314.

Tessereau, H., Lecouteux, C., Florin, B., Schlienger, C.& Pétiard, V. 1990. Use of a simplified freezing process anddehydration for the storage of embryogenic cell lines andsomatic embryos. In: Seeds: genesis of natural and artificialforms, edited by Le Biopole Végétal, 123-133.

Uragami, A., Sakai, A. & Nagai, M. 1990.Cryopreservation of dried axillary buds from plantlets ofAsparagus officinalis L. grown in vitro. Plant Cell Reports, 9,328-331.

Withers, L.A. 1987. In vitro methods for collectinggermplasm in the field. FAO/1BPGR Plant Genetic ResourcesNewsletter, 69, 2-6.

Withers, L.A. 1991. Tissue culture in the conservation ofplant genetic resources. In: Conservation of plant geneticresources through in vitro methods, edited by A.H. Zakri, M.N.Normah, M.T Senawi & AG. Abdul Karim, 1-18. KualaLumpur: Forest Research Institute Malaysia/Malaysian NationalCommittee on Plant Genetic Resources.

Withers, L.A. & King, P.J. 1980. A simple freezing unit androutine cryopreservation method 'for plant cell cultures.Cryo-Letters, 1, 213-220.

Zakri, A.H., Normah, M.N., Senawi, M.T. & AbdulKarim, A.G., eds. 1991. Conservation of plant geneticresources through in vitro methods. Kuala Lumpur: ForestResearch Institute Malaysia/Malaysian National Committee onPlant Genetic Resources.

Capture of genetic resources by invitro culture: field evaluation andprecocious flowering

S S Khuspe, R S Nadgauda, S M jagtap, C K John &A F MascarenhasPlant Tissue Culture Division, National Chemical Laboratory, Pune 411008, India

ABSTRACTClonal propagation using either plant tissue culture or rooted cuttings can now be achieved with several species. This paper

deals with results obtained under two tissue culture programmes initiated in our laboratory on propagation and field evaluation of

different tree species. The objectives under these programmes were to capture the maximum gains in biomass production and

timber quality utilising explanted material from proven mature trees of Eucalyptus spp. In vitro precocious flowering in bamboos

was an additional method employed to reduce conventional regeneration cycles, and increase the rate of genetic improvement.

Field trials have been conducted at several locations and growth rates evaluated over periods ranging fr0111 WO to ten years. In

general, plantlets derived from tissue culture showed a high degree of uniformity with an increased biomass capacity compared

to seedlings. The advantages of in vitro culture methods are discussed in the light of these results.

INTRODUCTION

The world consumption of wood today is around3 billion tonnes, nearly half of which is used asfuel. The total wood consumption is more thanthat of steel and plastics combined, and inanother decade the world's population will beusing over 4 billion tonnes of wood annually Thetotal forest loss on the earth is over 12 Mhaevery year, which is roughly the area of England.To meet the increasing demands for forest rawmaterials in India, emphasis has been laid onincreasing the productivity of our forests from thepresent low average yield of 0.5 m3 ha-1 yr-1.This increase is being planned through rationalland policies, technological innovation andimproved management. Tree improvementprogrammes for the genetic improvement ofplanting stock are also being introduced byseveral forestry-based companies to increaseproductivity These programmes involve:

• raising seedlings only from selected mother

trees;

• producing clonal planting stock of selected

plus-trees, through the conventional rooting ofcuttings or by micropropagation;

• creating seed orchards for the production ofplants which combine the selected traits fromthe best parents and show high vigour atthe nursery stage; and

• applying and integrating biotechnologicalmethods (Cheliak & Rogers 1990).

Considerable progress has been made over thepast two decades in developing tissue culturemethodologies from explants of juvenile andmature trees for industrially important species,

and applying them to forestry (Mascarenhas etal. 1988; Dunstan 1988; Mascarenhas &Muralidharan 1989). An integrated approach toreforestation involving the application of clonaltechniques offers a realistic hope for enhancingforest production. However, the success of suchan approach is limited by the efficiency withwhich selected trees can be reproducedvegetatively and by the high production costs ofsuch plants, relative to likely gains. Woodyspecies in the juvenile phase are generally easyto clone from embryos and young seedlings viaconventional vegetative methods andmicropropagation (Whitehead & Giles 1977).However, the ease of propagation tends todecline as trees mature to a stage where amore reliable evaluation of their crop potentialcan be made. This problem is especially seriousin plants whose cuttings are difficult to root byconventional means. In such cases, tissue culturecan be extremely useful for the production ofclonal material for testing and/or theestablishment of clonal germplasm banks.

A large number of tropical tree species havenow been propagated in vitro using juveniletissues (eg Khaya ivorensis and Naucleadiderrichii; see Newton et al., pp256-266). Morerecently a few species have been culturedfollowing rejuvenation of the trees by specialtreatments such as in vitro subculturing (VonArnold, Clapham & Ekberg 1990), and sprayingwith benzyl amino purine (BAP), eg Eucalyptusficifolia (Magalewski & Hackett 1979). Similareffects can also be . obtained by serial rooting orgrafting on juvenile rootstocks (Franc let et al.1987) and by repeated hedging (Campinhos &

53

Ikemori 1983). These latter methods aretime-consuming and may be more expensivethan rejuvenation induced by in vitro subculture.

In addition to developing the procedures formicropropagation, there is now an increasingawareness of the importance of the fieldevaluation of plants raised ih vitro. However, thenumber of reports on such studies is limitedand, with a few exceptions (Table 1), examplesare confined to a few temperate coniferous(Frampton & Amerson 1989; Horgan & Holland1990; Ritchie & Long 1986) and broadleaved treespecies (McCown & Amos 1979),

7'able I. Observations from field evaluation ofmicropropagated tropical trees. The list is not comprehensive

ilee speciesSource of explant

Dendrocalamusstrictus (bamboo)Seedlings

Eucalyptus spp.Mature trees

Hevea brasiliensisSeedlings andmature trees

Tamarindus indica(tamarind)Seedlings

Observations

Early culrnformationNumber of culmshigher

Increased earlygrowth rates insome clonesUniformity

Higher growth rates

Early floweringUniformity

Tbctona grandis(teak) Increased earlyMature trees growth rates

54

Reference

Mascarenhas et a/.(1988, 1989)Ramanuja-Rao &Rao (1990)Mishra et al. (1990)

Khuspe et al. (1987)Mascarenhas et al.(1988, 1989)V Patil & T NBhandari (personalcommunication)

M P Asokan & MSethuraj (personalcommunication,1992)M P Carron(personalcommunication1992)

Mascarenhas et al.(1988, 1989)

Mascarenhas et al(1988, 1989)

Populus deltoides Higher growth rates H C ChaturvediSeedlings (personal

communication,1992)

This paper is restricted to the results of fieldevaluation studies conducted on tissue culturepropagules obtained from: (i) mature trees ofthree Eucalyptus species which are important forrapid production of woody biomass and also, insome species, for their oil, and (ii) seedlings ofbamboo (Dendrocalamus strictus). Studies carriedout on the in vitro precocious flowering andseeding of bamboos as an alternative method ofpropagation are also described.

CULTURE PROCEDURES

The experimental culture procedures followed toinduce, grow and multiply roots and shoots fromtrees of Eucalyptus tereticornis, E. torelliana andE. camaldulensis and from Dendrocalamus strictus(bamboo) seedlings have been described inearlier publications (Gupta, Mascarenhas &Jagannathan 1981; Gupta, Mehta & Mascarenhas1983; Gupta & Mascarenhas 1983, 1987; Nadgiret al. 1984). The ages of the Eucalyptus treesfrom which vegetative buds were collected forinitiation of cultures were between ten and 20years. Both apical and axillary buds werecollected from the crown regions of the trees.The rooted plantlets or microcuttings of bothEucalyptus and Dendrocalamus were transferredto polyethylene bags (2.5 cm diameter x 10 cmheight), containing autoclaved sand, and placedin a greenhouse maintained at 30°C with arelative humidity of 80-90%. On attaining aheight of 10 cm, they were transferred to largerbags (10 cm x 20 cm) containing river sediment,and further incubated in the greenhouse untiltheir transfer to different locations for theestablishment of field trials.

FIELD EVALUATION OF EUCALYPTUS

Seedlings versus clones: field trials ofE. tereticornis and E. torelliana

Individual trees were selected on the basis oftheir phenotypic traits (an apparent rate ofbiomass production ten times higher than theaverage) in forest plantations. Clones weresubsequently produced ,by tissue culturemethods, and then grown on a pilot scale (30trees) in Pune, Maharashtra, India, in 1983. Theclones were compared with plants raised fromseed according to a non-replicated experimentaldesign described previously (Ktiuspe et al. 1987).Plants of both sp.ecies grown from tissue cultureshowed height, diameter and biomass valueshigher than control plants over the first fewyears. With E. tereticornis, the biomass yields oftissue-cultured plants at 12, 34 and 52 monthswere 220%, 60% and 16%, respectively higherthan controls (Table 2). After the same timeintervals, the biomass yields of E. torelliana were900%, 115% and 105% higher than controls,respectively (Table 3). After 84 months' growth ofE. torelliana, biomass yield was reduced belowthat of the seedling controls, while in E.tereticornis yields remained higher than controlseven after 108 months. Density of the wood at52 months was similar to that of six-year-oldseedlings (Mascarenhas et al. 1988).

These clones did not achieve growth rates tomatch their estimated (ten-fold) superiorityexcept perhaps during the first few months,when, in addition to culture conditions, manyother factors, such as acclimation (Boulay 1987),

Table 2. Growth and biomass evaluation of Eucalyptus planting stock, size at planting, etc, would have atereticornis plants pioduced by tissue culture (TC) (average major effect. It could be argued that thesebased on 30 plants) plus-tree tissue-cultured plants were genetically

superior to the control seedlings, but that thisLocation NCL, Pune advantage was lost owing to the onset of earlierPlanting date Juty 1982 maturation. In this study, the tissue-cultured plants

Flowering date TC - July 1985 started flowering after 24 months, when 3 m tall,Contml - January 1987 whereas the seedlings did not flower until they

Observation date Juty 1992 were 42 months old. However, the generalmorphology of the plantlets resembled that of

Diameter at seedlings with respect to colour and shape ofHeight breast Biomass plant-1 leaves, stem growth, branching habitand canopy

Age (m) height (cm) O(g) This study demonstrates the difficulties associated(months) TC Control TC Control TC Control with phenotypic plus-tree selection for

fast-growing clones, and emphasises the need forproper genetic testing in clonal trials beforeselecting superior material for multiplication bytissue culture.

Table 3. Growth and biomass evaluation of Eucalyptustorelliana plants produced by tissue culture (TC) (averagebased on 30 plants)

Location NCL, Pune

Planting date July 1982

Flowering date TC - July 1985Control - January 1987

Observation date July 1992

Diameter atHeight breast Biomass plant-1

Age (m) height (cm) (kg)(months) TC Control TC Control TC Control

Table 4. Comparative field trial of four clones of Eucalyptuscamaldulensis produced by tissue culture

Location: Kundewadi, Nasik, MS

Nos of replications 4

Nos of plants/replication 16

Planting date April 1991

Observation date May 1992

Treatments Height (m) Girth (cm)

R-6 2.6 8.8

ERK-4 3.7 11.7

NSPR 1.2

APFDC(R-5) 2.5 7.8

Control 2.1 7.3

SE

CD (5% level)

±0.2 ±0.1

0.6 0.4

Clone/site interaction in field trials ofE. camaldulensis

Clones from four candidate trees of E.camaldulensis (R-6, ERK-4, APFDC(R-5) andNSPR, collected from Andhra Pradesh) weregrown in a replicated trial to compare thegrowth performance of different clones. Thecandidate trees ranged from ten to 20 years old,and were .13-22 m tall with girths of 30-65 cm.The trial was conducted at Nasik, India, in April1991 and replicated at six other sites (Table 4).Seed-raised plants were used as controls. Rootedplantlets were potted and repotted in December1990 and February 1991, respectively thenplanted out in April 1991. Maximum increases inheight and girth measurements were observed inclone ERK-4 followed by R-6 and APFDC (R-5)(Table 4). The uniformity in height and girthmeasurement was also striking. Clone NSPRdisplayed very poor growth. These results clearlydemonstrate the importance of testing a range ofdifferent clones. At other locations (Pune andTirupati), ERK-4 was also the clone whichdisplayed the highest growth rate. Assessment ofthe growth performance of these plants over alonger period will be necessary to arrive atmore defmite conclusions about their relativeperformance.

Effects of subculturing tissue-culturedclones from mature Eucalyptus trees

In this study plant propagules were producedfrom one elite tree of both E. tereticornis (CloneT-8) and E. camaldulensis (Clone R-3) afterdifferent numbers of subcultures. Ten plantletswere field-grown at each subculture to testwhether any significant differences occurred ingrowth performance. In E. camaldulensis and E.tereticornis, plants were tested from the fourth toseventh and ninth to eleventh subcultures,respectively A striking uniformity was seenbetween subcultures (Table 5) in both species

55

Table 5. Effect of number of subcultures on giowth ofmicropropagated Eucalyptus camaldulensis and E. tereticornisplants (average based on ten plants)

56

after 36 months' field growth. This experimenthighlights the possibility of utilising plantlets fromrepeated subculture. In Eucalyptus this is animportant factor because initiation of freshcultures from mature trees is beset by two majorproblems: (i) the very low percentage of budssprouting even in the best seasons, and (ii) thelong initiation period with primary explants.These results indicate that multiplication in thelaboratory can be carried out throughout theyear without any seasonal influences on cultureinitiation or field performance. No evidence wasfound in this study for rejuvenation by serialsubculturing.

Mature clones of E. citriodora for essentialoil production

Eucalyptus citriodora is important for its foliar oil,which is used in the cosmetic andpharmaceutical industries. The importantingredients of the oil are citronellal (90-95%) andcitronellol (5-10%), The oil content varies from0.5% to 4% in the green leaves in different trees,suggesting the immediate application ofvegetative methods of propagation and clonalselection (Gupta & Mascarenhas 1987) to capturethis variation. Tissue-cultured plants were raisedfrom a 20-year-old parent tree containing highconcentrations of oil and citronellal, and growingat an altitude of over 2000 m inOotacamangalam, Tamil Nadu, in the south ofIndia. The regenerated plants were grown atPune, Maharashtra, India, and the oil and itsingredients were analysed when the plants wereone and three years old, as described by Guptaand Mascarenhas (1983). The concentrations ofoil (3.5%), citronellal (90.0%) and citronellol(5.0%) in one- and three-year-oldmicropropagated plants were more or lesssimilar to those of the parent tree. These results

indicate the advantages of tissue-culturing maturetrees to obtain early returns in one-year-oldplants. Seedling plants take three to four yearsbefore the oil contents stabilise. Themicropropagation of elite trees, therefore, hasgreat potential for high value chemicalproduction (Mascarenhas et al. 1988).

Early flowering in micropropagatedE. tereticornis

Plant lets were produced from an elite tree,K(1)-37 of E. tereticornis, collected from Shimoga(Karnataka), India, and planted at two locations,Neyveli and Pune, India. The heights oftissue-cultured plants after 13 months at Punewere significantly greater than those of controlseedlings (121 versus 80 cm; P). Similar growthrates were observed at Neyveli. Interestinglythese plants flowered within six months afterplanting, at both Pune and Neyveli. Althoughearly flowering has been recorded intissue-cultured plants, the reason why thisphenomenon occurred only in clone K(1)-37, 18months after transfer to polythene bags, isunclear, and is currently being investigated. Inorder to determine whether soil factorsinfluenced flowering, soil samples from Neyveliwere compared with samples collected at Puneusing the analytical methods of Perur et al.(1973). The main difference observed was thehigher potassium content of the Pune soils.

The time required for forest trees to reachsexual maturity has implications for the breeding,selection and propagation stages. In the former,the major influence is the time of first floweringand, consequently the time required to providethe next generation in the breeding programme.In the latter, if propagation is to be via thesexual process, then not only is the initiation offlowering important but so are the consistencyand scale of flowering (Cheliak & Rogers 1990).

The treatments to promote flowering in foresttrees include drought, root pruning, girdling, hightemperature, and the application of nitrogenfertilizers and appropriate external growthregulators, such as gibberellins. There is someevidence that flowering can be initiated earlierthrough in vitro culture. For example, Hackett(1983) reported that somatic embryos derivedfrom callus ft-om roots of mature Panax ginseng,and from the callus of zygotic embryos of datepalm flower in the embryoid stage, whereassexual embryos do not flower for several years.McKeand (1985) reported that two-year-oldtissue-cultured plantlets of Pinus taeda grown inthe field exhibited high growth rates andmorphological characters typical of mature trees.This result was not expected as the plantletsoriginated from juvenile embryonic tissues. Asmentioned earlier, the causes of the earlyflowering response in Eucalyptus in the present

studies could be attributed to the origin of theexplants from mature trees, but it is also possiblethat flowering was promoted by the ingredientspresent in the culture medium.

FIELD EVALUATION OFMICROPROPAGATED DENDROCALAMUSSTRICTUS (BAMBOO)

A number of bamboo species flowergregariously and unpredictably. Most bamboospecies are monocarpic, ie they flower oncebefore culm death (McClure 1966). Theproduction of seed in bamboo is irregularbecause, in the majority of species, floweringcycles range from four to 120 years (Janzen1976). Most bamboo species have considerableeconomic and social importance. Some of thebest qualities of paper are produced frombamboo. Throughout Asia and other tropicalregions, bamboo is a vital source of lightweightrot-resistant wood used in the construction ofhouses, furniture, tools, baskets and otherhousehold items. Bamboo hay which has fourtimes the protein of other fodder grasses, is astaple for livestock.

Bamboos are propagated by seed or vegetativelyConventionally offsets, rhizomes, cuttings orlayering have been used for this purpose(Seethalakshmi, Venkatesh & Surendran 1983;Banik 1985) and have given successful results.Micropropagation methods through organogenesis,from embryonic tissue and seedlings, have alsobeen developed recently (Ramanuja-Rao & Rao1990; Mascarenhas et al. 1990; Nadgir et al. 1984;Yeh & Chang 1986a, b, 1987).

In our laboratory bamboo plantlets were raisedby tissue culture using the procedures of Nadgiret al. (1984) and Mascarenhas et al. (1990). Thecultures were isolated from germinated seedlings.The plantlets were grown in a greenhouse, asdescribed for Eucalyptus above, and planted outat Pune, India, in August 1990, using a pairedplot design with 16 replications. Significantlyincreased heights and culm numbers wereobserved after 16 months. Mishra, Nair andVidya-Sagar (1990) reported the results of asimilar trial conducted at Kumarapatnam,Karnataka, India, with plants collected from ourlaboratory The increase in height of cultures andrate of culm production was 20% and 75%respectively compared with plants raised fromseed as controls.

The results of these two experiments indicate theimportance of micropropagation in bamboo forobtaining earlier and higher growth rates both inculm formation and in the number of culms. Thegestation period of about two years for plants ofseed origin can be reduced to six or sevenmonths using tissue-cultured plants (Mishra et al.1990). Moreover, multiplication of plants can becarried out all year round within the confines of

a laboratory and need not be determined by thelong and unpredictable flowering cycles in thisspecies. Micropropagation can be used toproduce sufficient planting material between theflowering cycles when seed is scarce.

Natural stands and plantations of bamboosderived from seedlings show considerablevariation, unlike clonally propagated plantations.The totally solid culms of some D. strictusclumps fetch much higher prices than hollowones. It is not laiown whether this unusual trait isunder genetic or environmental control, or both.Even if it proves to have low heritability byselection and clonal propagation it should bepossible to increase the frequency of solid culmsin a clump.

There are several examples of spontaneousmutants produced in different bamboo species.Thornless mutants of normally thorny species areknown to occur in Bambusa anmdinacea of Indiaand Guadua angustifolia of tropical America. In B.vulgaris, culm colour variations are also known.Clonal multiplication by tissue culture, if possiblewith these species, has great potential formultiplying such selections (Venkatesh 1992).

IN VITRO FLOWERING AND SEEDING INBAMBOOS

In vitro flowering experiments

In India, the mast flowering cycle for Ochlandratravancorica (Venkatesh 1984) is seven years, forBambusa arundinacea 30-40 years, and forDendrocalamus strictus 20-65 years. Thisinfrequent flowering at long intervals has been amajor obstacle for the genetic improvement ofbamboo by conventional breeding. The ForestResearch Institute of Guangdong province in thePeople's Republic of China has recentlyproduced several artificial interspecific andintergeneric bamboo sexual hybrids by makingopportunistic crossings with sporadically floweringclumps of the different species of bamboo(Zhang & Chen 1985). This chance crossing,however, is not as useful as a controlledbreeding programme would be.

In our laboratory using tissue culture techniques,we have recently succeeded in inducingprecocious flowering and normal seed setting ofthree Indian bamboos: Bambusa arundinacea andDendrocalamus brandish. (Nadgauda, Parasharami& Mascarenhas 1990), and, more recently D.strictus. In brief, the steps followed for obtainingflowering and seeding in these species are asfollows.

i. Seedlings of the species are raised onWhite's medium.

Excised seedling segments are transferred toMurashige and Skoog's liquid medium with

57

After three passages on the above medium,vegetative shoots develop an inflorescencecontaining panicles of normal spikelets in70% of B. arundinacea and 40% of D.brandisii shoots.

iv On separation of inflorescence segments andsubculture to MS-1 medium, they multiplygiving rise to an inflorescence culture forrepeated seed production.

v. In 30-50% of the cultures, seed set wasobserved with viability of 3-5%.

Recently the in vitro flowering of D. hamiltonlihas also been reported (Chambers, Heuch &Pirrie 1991). These preliminary results suggestan alternative approach for obtaining acontinuous year-round supply of seed, providedthe extent of seed production and seed settingcan be increased.

The causes of low seed set in bambooraised in vitro

In order to determine the cause for the lowseed setting in the flowers of bamboo speciesraised in vitro, experiments were carried outwith Bambusa arundinacea. This species hadflowered extensively at Pune and so allowed acomparison of morphology and pollen fertility offlowers raised in vivo and in vitro. Pollen fertilitywas determined by following Alexander's stainingprocedure (Alexander 1969). Floral parts offlowers raised both in vivo and in vitro werecomplete. However, the opening of tissue-culturedflowers was only partial; in 60-70% of theflowers, the lemma and palea were closed.Anthesis and pollen fertility in flowers raised invivo were almost 100%, compared to the verylow values for flowers raised in vitro. Thesepreliminary data suggest an explanation for thelow fertility of flowers produced in vitro.Moreover, it is also reported that in vivo thestigmas are receptive before pollen production insome bamboo species. This characteristic couldexplain the outcrossing that is assumed to occur.

The main goal of flower induction technology isto provide flowering on demand, with consistencyand predictability In the bamboos, this isextremely important because of their longvegetative phase before flowering and seeding,and the difficulties associated with theirvegetative propagation. The discovery of ameans to stimulate precocious flowering andseeding in four bamboo species opens up theprospect of:

• controlled breeding by a systematicintercrossing between different species to

58

BAP and coconut milk (MS-1) for initiation ofshoot cultures and for the first threesubcultures.

produce intraspecific or interspecific hybridswith desirable characteristics;

• continuous supplies of bamboo seeds forplanting during the long inter-mast periodswhere seed is scarce (this method can becombined with micropropagation procedures);

• the continuous availability of anthers (pollen)for production of haploids;

• studies of the mechanisms for theunpredictable flowering of bamboos, and oftheir reproductive development.

By inducing synchronous flowering and seeding,it will be possible to apply conventionalbreeding and hybridisation procedures to thegenetic improvement of bamboos.

CONCLUSIONS

Modern agricultural techniques require cropuniformity Plants of the same height make forthe easier placement of agricultural chemicalsand for weeding by machines. Crops whichmature at the same time with uniform fruit orcob size facilitate machine harvesting orplucking. In the limited number of field testsreported to date, the morphological appearancearid the performance of micropropagated plantsof forest tree species have varied considerablyfrom species to species. Results with Eucalyptusspecies and Dendrocalamus strictus are similar tosome of the results obtained with other species,and are marked by a higher degree ofuniformity with a rapid early growth rate. As hasbeen observed in our studies, it seems feasibleto increase the productivity of forest lands beforethe trees reach their biological limits. Althoughmicropropagation may not be justifiedimmediately for producing millions of plants byorganogenesis, at a low unit cost it still has thefollowing genetic and economic advantages:

• the exploitation of the early floweringbehaviour of superior mature trees tocapture their general combining ability formating at random in a seed orchard (in thisway superior individuals could be bredwithin a shorter period);

• the production of plants in very highnumbers and within shorter timespans thanis possible by conventional vegetativepropagation methods;

• increased gains in biomass per tree(however small, these can result in highlysignificant absolute gains when the annualplanting numbers are considered).

In India, the forestry plantation programmeinvolves several million Eucalyptus plants annuallyWith a regeneration programme of thismagnitude, even small gains per tree representsignificant absolute gains. To capitalise fully on

clonal forestry there must also be a means ofconserving germplasm, ensuring that it ismaintained in a state capable of subsequentpropagation for a length of time commensuratewith its field testing regime (see Engelmann,pp49-52).

Bridging the widening gap between supply anddemand of firewood, timber and raw materialsfor pulp and fibre is possible only throughwell-conceived, highly efficient productiontechnologies, involving high-yielding certifiedplanting stock, under the aegis of agroforestryand industrial forestry programmes. Although ourresults already indicate the possibility ofobtaining higher gains in biomass in Eucalyptusand bamboo, there are several other factorswhich have to be studied in far greater detail.The factors which influence field performance ofmicropropagated plants can be summarised as:

• culture conditions involving origin of explants,period in culture, etc;

• greenhouse and ultimately agroclimaticconditions in which the propagules areplanted; and

• the agricultural practices adopted.

These factors, individually or in combination, canraise the biological limits of productivity ingenetically superior planting stock, and ensuresustainability

ACKNOWLEDGEMENTS

Financial assistance received from the NationalBank for Agriculture and Rural Development,Bombay India, and the Department ofBiotechnology New Delhi, is gratefullyacknowledged. The assistance and co-operationof the Forest Departments, Forest Corporationsand other forest agencies is acknowledged,particularly for help in collecting budwoodmaterial and making the observations on fieldtrials. Thanks are also due to Mrs S Bhondve, SV Shirke, D Chauk and P R Akkadkar for thetyping of the manuscript and computer graphics.

REFERENCES

Alexander, M.P. 1969. Differential staining of aborted andnon-aborted pollen. Staining Tbchnology, 4, 117-122.

Banik, R.L. 1985. Techniques of bamboo propagation withspecial reference to prerooted and pre rhizomed branchcuttings and tissue culture. In: Recent research on bamboos,edited by A.N. Rao, G. Dhanarajan & C.B. Sastry 160-169.CAE China/International Development Research Centre, Canada.

Bou lay, M. 1987. In vitro propagation of tree species. In:Plant tissue and cell culture edited by C.E. Green, DA.Somers & WP Hackett, 367-382. New York: Liss.

Campinhos, E. & Ikemori, Y.M. 1983. Production ofvegetative propagules of Eucalyptus spp. by rooting ofcuttings. In: Proceedings of the 2nd Symposium on PlantationForests in the Neo-tmpics - its Role as a Source of Energy,1-8. Vienna: International Union of Forestry ResearchOrganizations.

Chambers, S.M., Bench, J.H.R. & Pirrie, A. 1991.Micropropagation and in vitro flowering of the bambooDendrocamalus hamiltonii Munro. Plant Cell, Time and OrganCulture, 27, 45-48.

Cheliak, WM. & Rogers, D.L. 1990. Integratingbiotechnology into tree improvement programmes. CanadianJournal of Forest Research, 20, 452-463.

Dunstan, D.I. 1988. Prospects and progress in coniferbiotechnology Canadian Journal of Forest Research, 18,1497-1506.

Frampton, L.J. Jr & Amerson, H.V. 1989. Influence of

tissue culture method on field growth of loblolly pine. In VitroCellular and Developmental Biology, 25 (3), 13A.

Franc let, A., Boulay, M., Bekkaoui, F., Fourest, Y.,Verschoore-Martouzet, B. & Walker, N. 1987. Rejuvenation.In: Cell and tissue culture in forestry, vol. 1, edited by J.M.Bonga & D.J. Durzan, 232-247. Dordrecht: Nijhoff.

Gupta, RIC. & Mascarenhas, A.F. 1983. Essential oilproduction in relation to organogenesis in tissue culture ofEucalyptus citriodora Hook. In: Plant cell culture in cropimprovement, edited by S.K. Sen & K.L. Giles, 299-308. NewYork: Plenum.

Gupta, P.K. & Mascarenhas, A.F. 1987. Eucalyptus. In: Celland tissue culture in forestry, vol. 3, edited by "M. Bonga &D.J. Durzan, 385-399. Dordrecht: Nijhoff.

Gupta, P.K., Mascarenhas, A.F. & Jagannathan, V. 1981.Tissue culture of forest trees - clonal propagation of maturetrees of Eucalyptus citriodora Hook by tissue culture. PlantScience Letters, 220, 195-201.

Gupta, P.K., Mehta, U. & Mascarenhas, A.F. 1983. Atissue culture method for rapid multiplication of mature treesof Eucalyptus torelliana and E. camaldulensis Plant CellReports, 2, 296-299.

Hackett, W.P. 1983. Phase change and intraclonal variabilityHorticulture Science, 18 (6), 12-16.

Horgan, K. & Holland, L. 1990. Micropropagation ofRadiata pine. Abstract. In: Proceedings of Conifer BiotechnologyWorking Group's 5th Meeting, 15. Sittingbourne: Shell ResearchLtd.

Janzen, D.H. 1976. Why bamboos Wait so long to flower.Annual Review of Ecology and Systematics, 7, 347-391.

Khuspe, S.S., Gupta, P.K., Kulkarni, D.K., Mehta, U. &Mascarenhas A.F. 1987. Increased biomass production bytissue culture of Eucalyptus. Canadian Journal of ForestResearch, 17, 1361-1363.

Magalewski, R.L. & Hackett, W.P. 1979. Cuttingpropagation of Eucalyptus ficifolia using cytokinin induCed

basal trunk sprouts. Combined Proceedings of the InternationalPlant Propagators Society, 29, 118-124.

Mascarenhas, -A.F. & Muralidharan, E.M. 1989. Tissueculture of forest trees in India. Current Science, 58, 606-613,

Mascarenhas, A.F., Khuspe, S.S., Nadgauda, R.S.,Gupta, P.K. & Khan, B.M. 1988. Potential of cell culture inplantation forestry programmes. In: Genetic manipulation of

woody plants, edited by J.W Hanover & DE. Keathley391-412. New York: Plenum.

59

Mascarenhas, A.F., Khuspe S.S., Nadgauda, R.S.,Gupta, P.K., Muralidhaxan, E.M. & Khan, B.M. 1989.Biotechnological application of plant tissue culture to forestryin Inc lia. In: Application of biotechnology in forestry andagriculture, edited by V Dhawan, 73-86. New York: Plenum.

Mascarenhas, A.F., Nadgir, A.L., Thengane, S.R.,Phadke, C.H., Khuspe, S.S., Shirgurkar, M.V.,Parashrami, V.A. & Nadgauda, R.S. 1990. Potentialapplication of tissue culture for propagation of Dendrocalamusstrictus In: Bamboos - current research, edited by I.VRamanuja Rao, R. Gnanaharan & C.B. Sastry 159-166. KeralaForest Research Institute, India/International DevelopmentResearch Centre, Canada.

McClure, F.A. 1966. The bamboos - a fresh perspective.Cambridge, Massachusetts: Harvard University Press.

McCown, B. & Amos, R. 1979. Initial trial with commercialmicropropagation of birch selections. Combined Proceedings ofthe International Plant Propagators Society, 29, 387-393.

McKeand, S.E. 1985. Expression of mature characteristics bytissue culture plantlets derived from embryos of loblolly pine.Journal of the American Society of Horticulture Science, 110,619-623.

Mishra, R.P., Nair, P.M. & Vidya-Sagar. 1990. Bamboopropagation by tissue culture. Abstracts. In: Proceedings of theNational Seminar on Bamboo, Bangalore, 1990, 59. KarnatakaForest Department and the Bamboo Society of India.

Nadgauda, R.S., Parasharami, V.A. & Mascarenhas, A.F.1990. Precocious flowering and seeding behaviour in tissueculture bamboos. Nature, London, 344, 335-336.

Nadgir, A.L., Phadke, C.H., Gupta, P.K., Parasharami,V.A., Nair, S. & Mascarenhas, A.F. 1984. Rapidmultiplication of bamboo by tissue culture. Silvae Genetica, 33,219-223.

Perur, N.G., Subramanium, C.K., Muhr, G.R. & Ray,H.E. 1973. Soil fertility evaluation to serve Indian farmers.Mysore: Department of Agriculture, Mysore University ofAgricultural Sciences, Bangalore, India.

Ramanuja-Rao, I.V. & Rao, ILI. 1990. Tissue cultureapproaches to the mass propagation and genetic improvementof bamboos. In: Bamboos - current research, edited by I.V

60

Ramanuja Rao, R. Gnanaharan & C.B. Sastry 151-158. KeralaForest Research Institute, India/International DevelopmentResearch Centre, Canada.

Ritchie, G.A. & Long, A.J. 1986. Field performance ofmicropropagated Douglas fir. New Zealand Journal of ForestScience, 16, 343-356.

Seethalakshmi, K.K., Venkatesh, C.S. & Surendran, T.1983. Vegetative propagation of bamboos using growthpromoting substances. I. Bambusa balcooa Roxb. Indian Journalof Forestry, 6, 98-103.

Venkatesh, C.S. 1984. Dichogamy and breeding system in atropical bamboo Ochlandra travancorica. Biotropica, 16,309-312.

Venkatesh, C.S. 1992. Biotechnology in tree and bambooimprovement. Indian Forester, 118, 249-255.

Von Arnold, S., Clapham, D. & Ekberg, I. 1990. Hasbiotechnology a future in forest tree breecling? Forest TreeImprovement, 23, 31-47.

Whitehead, H.C.M. & Giles, K.L. 1977. Rapid propagationof poplars by tissue culture methods. New Zealand Journal ofForest Science, 7, 40-43.

Yeh, M.L. & Chang, W.C. 1986a. Somatic embryogenesisand subsequent plant regeneration from inflorescence callusof Bambusa beechyana Munro var. beechyana. Plant CellReports, 5, 409-411.

Yeh, M.L. & Chang, W.C. 1986b. Plant regeneration throughsomatic embryogenesis in callus culture of green bamboo(Bambusa oldhamii Munro). Theoretical and Applied Genetics,73, 161-163.

Yeh, M.L. & Chang, W.C. 1987. Plant regeneration viasomatic embryogenesis in mature embryo derived callusculture of Sinocalamus latifolia (Munro) McClure. Plant Science,51, 93-96.

Zhang, Guang-zhu & Chen, Fu-qui. 1985. Studies onbamboo hybridization In: Recent research on bamboos, editedby AN. Rao, G. Dhanarajan & C.B. Sastry CAF,China/International Development Research Centre, Canada.

Capture of genetic resources by collection andstorage of seed: a physiological approach

P B TompsettRoyal Botanic Gardens Kew, Wakehurst Place, Ardingly, Nr Haywards Heath, West Sussex RH17 6TN, UK

ABSTRACTSeed provides the most natural vehicle for the collection, transport and storage of genetic variation, often avoiding the technical

difficulties and expense inherent in other methods. Consideration is given to the advantages and limitations of using seed, with

examples drawn from the Meliaceae, Dipterocarpaceae and Araucariaceae families. The importance of placing seed into

desiccation-intolerant ('recalcitrant') or desiccation-tolerant ('orthodox') storage physiological categories is emphasised, and

progress in this regard is reported. Handling procedures to ensure maximum viability for the two types are discussed in relation

to moisture content, temperature and gaseous factors. It is concluded that long-term conservation of some tree species as seed is

a feasible strategy but that other species may require further research to ensure maximum viability at harvest and greatest

possible longevity in storage. Achievement of high initial viability will compensate for the shorter natural longevity of tree seed

compared with seed of herbaceous crop species.

INTRODUCTION

Relevant general articles on seed storage includethose by Bonner (1990), Stubsgaard (1992),Wil lan (1985) and De Muckadell and Pilim(1983). The present paper is based mainly onpersonal experiences of the author. The term'seed' is used herein to encompass fruit.Availability collection and storage topics arecovered, with the emphasis on physiologicalrather than genetic aspects; sampling strategieshave been considered by Tirnbull (1978) and byWil lan, Hughes and Lauridsen (1990).

'Orthodox' and 'recalcitrant' seed types

A consideration of supreme importance indeciding practical handling methods is thestorage physiological category of the seed;'orthodox' seed can be stored dry and cold,whereas 'recalcitrant' seed cannot be dried tolow moisture contents without loss of viabilityand so must be stored moist (Roberts 1973).Orthodox seeds tend to be small and areproduced on both woody and herbaceousspecies; recalcitrant seeds are generally largeand are found mainly if not exclusively onwoody species. Assessment of storage type isaccomplished by determining the relationshipbetween germination and moisture content,preferably using several desiccation methods;misclassification can occur if detailed studies arenot carried out. Recalcitrant seed cannot bestored for long periods so cannot be consideredas a means for genetic conservation. Seed oforthodox species can be stored for much longer,and its use in conservation is discussed below

Percentages of species possessing recalcitrantseeds are 35%, 94% and 33% in the Meliaceae(Table 1), Dipterocarpaceae (Tompsett 1992) andAraucariaceae (Table 2), respectively for thosemembers examined so far; the percentage withrecalcitrant seed for all three families combined,calculated by pooling all data in the abovesources, is 72%. The extent to which this smallsample represents woody species as a whole ishard to assess. However, the probability thatthere is a large percentage of recalcitrant-seededspecies in rainforests, combined with the knownhigh species density in these forests, indicatesthat this figure may not be entirelyunrepresentative.

Recently a third 'intermediate' type of seedstorage physiology has been proposed, whichhas desiccation characteristics intermediatebetween those of the orthodox and recalcitranttypes (Ellis, Hong & Roberts 1990, 1991). Inaddition to the partial desiccation toleranceexhibited by this category, three furtherdifferences relating to dry seed were reported todifferentiate it from the orthodox type: cooliemperatures near 0°C were more damagingthan warm temperatures; freezing temperatureswere likewise more damaging than warmtemperatures; and, within a range extendingapproximately from 5% to 10% moisture content,the usual improvement in longevity observed ondrying orthodox seed was reversed. Thequestion of whether this new category isphysiologically distinct from the orthodox type isconsidered later in relation to earlier research onAraucaria columnaris (Tompsett 1984b).

61

Table 1. Lowest-safe moisture contents (LSMC), mass values and optimum recorded storage conditions for Meliaceae seed; storagetemperatures, and moisture contents are given with the best fmal germinations (germ.) achieved. Seeds were supposed mature.Mass is that at or near LSMC. Authorities are those of Pennington and Styles (1975) for subfamilies, tribes and genera (data arethose of P B Tompsett (unpublished) unless otherwise stated)

Orthodoxor

recalcitrant

MELIOIDEAE

62

Lowest-safe

moisturecontent

(%)

*NA, not applicable; PB, ventilated polythene bags; MC,B Tompsett, K Manger & R D Smith (unpublished)

2Arlls removed before testing or storageChomski (1990)

The need for tree seed conservation

Mass

(g) Period

The seed stores currently existing in forestry andallied institutions provide material for breedingand plantation programmes in the short tomedium term. Long-term tree seed storagefacilities, as employed for the conservation ofherbaceous crop species, are almost non-existent.Physiological studies have revealed that, onaverage, tree seed appears to possess aninherently shorter storage life than crop seed;long-term storage is nonetheless feasible for at

moisture content

Optimum storage recorded

MoistureTemperature Germination content

(°C) (%) (%) Comment

least some species. This approach is urgentlyneeded where a species is under threat ofextinction in the whole or in a part of its range,as is the case for Entandrophragma and Khayaspp. Although in situ conservation is the methodof first choice, and ex situ conservation standshave been established for many species, it isrecognised that other ex situ conservationmethods should be more closely examinedNational Research Council 1991). The presentpaper considers the role seed can play in the exsitu conservation and utilisation of tropical trees.

Table 2. Lowest-safe moisture contents, mass and optimum recorded storage conditions for Araucariaceae. All detailg except those

concerning taxonomy are as in Table 1. Genera and species sections are given for Araucaria. Germinations are not adjusted for

percentage empty seed, which averages over 50Vo in orthodox species (data are of P B Tompsett (tuipublished) except where

otherwise stated)

COMPARATIVE MERITS OF SEED AS ACONSERVATION MEDIUM

Objectives of seed storage studies

The objectives of seed storage studies may beviewed as two-foid:

• to improve short-term storage formaintenance of viability during transport andtemporary holding for use in species trialsand plantation programmes; and

• to enable long-term genetic conservation inseed form.

Advantages

First, seed is the natural vehicle for genedispersal and has advantages in ex situ geneconservation, being immediately available forseedling production. Alternative methods, such asthose involving tissue culture and pollen, requireintermediate processes before seedlings can beproduced. Second, for cross-pollinated species,each propagule may have a wider range of

lin the case of orthodox seed, some damage may occur on drying if seed is either immature or over-mature2Storage data from Shea and Armstrong (1978)3D Doley (personal communication)4R D Smith and J B Dickie (unpublished)5Preest (1980)

genes than vegetative material from the sametree. Third, maintaining material as seed is lessexpensive than other methods such as tissueculture slow growth; in addition, seed storage is

genetically more stable. Fourth, conservation bypollen storage entails the uncertainties ofpollination and seed production (or equivalenttissue culture procedures) before seedlings canbe fmally produced.

Disadvantages

The problems associated with moist storage (forexample, premature germination) imply that anylong-term ex situ conservation ofrecalcitrant-seeded species must probably be informs other than whole seed. Second, for somespecies seed is produced only at intervals ofbetween two and five years in any particularlocation, as for example in the case of rainforestdipterocarps, creating difficulties in acquiringmaterial. Third, some orthodox-seeded specieshave propagules requiring particular harvestingand handling procedures to ensure successful

63

desiccation for safe storage; such seed mayinclude the so-called intermediate category(discussed below). With adequate research,however, the latter problems can be alleviated.Fourth, the male parent may be unknown forseed of cross-pollinating species, adding anelement of uncertainty not present whenvegetative materials are collected. Fifth,large-seeded species may be relativelyexpensive to store, because of the volume theyoccupy although storage of the de-coated dryembryo may provide a cheaper alternative. Sixth,seedlots of tree species sometimes contain ahigh percentage of empty (ernbryoless) seeds; insome cases, separation from sound seed isimpracticable (eg Araucaria cunninghamii).many tree species have special processingrequirements, such as the de-pulping ofdrupaceous fruits (see Maghembe et al.,pp220-229); further research is needed to assessoptimum procedures.

SEED AVAILABILITY

Periodicity of seeding

As referred to above, availability can be limitedby infrequent seeding, particularly in relativelymoist environments; examples are provided byDipterocarpus grandiflorus and D retusus. Inseasonal areas, however, some seed may beproduced every year, as for example with D.intricatus and D. tuberculatus in dry dipterocarpforest.

Amount of material available

When fruiting occurs it may be prolific, as withAzadirachta indica or Swietenia species, or it maybe more restricted, particularly for species withlarger propagules such as those of Shoreamacrophylla (the illipe nut) or Dipterocarpusgrandiflorus. Obtaining the required spread ofgenetic material can be more difficult when onlya restricted quantity is available.

Crown accessibility

The large crowns of many species providedifficulties in attaining the necessary fresh, viableseeds. Araucaria hunsteinii, which can exceed80 m in height, is an example of a speciesneeding skilled collectors, leading to high cost ofthe seed. Seed shaken from tall trees possessingwind-dispersed seed, such as Hopea foxworthyi,can also provide problems, in that abscised seedmay be hard to locate.

Parasite infestation

Pest problems can restrict availability Speciessuch as Shorea amplexicaulis and Dipterocarpustuberculatus can have seed which approaches100% infestation with insect larvae.

64

Premature abscission

Failure of the flowers or young fruit to developmay be caused by dry conditions. In addition,localised high winds associated with rainfall nearthe end of the hot season in Thailand havestripped Shorea roxburghii trees of seed; similarlyyoung Hopea foxworthyi seed has been lostduring a typhoon in the Philippines in 1992.

Geographical and topographical isolation

Separation of seed sources from populationcentres by great distances or by physicalfeatures such as mountains or rivers can reduceavailability Collection of dipterocarps in PapuaNew Guinea is difficult because rugged terrainhinders the location of suitable seed-bearing trees.

SEED COLLECTION

It is not proposed to cover collection techniquesin detail; information on this topic may be foundelsewhere. Broadly seed can be collected fromthe ground after abscission, from the crown usingground access, and from the crown by climbing.Many dipterocarps have large fruits which falldirectly below the mother tree and can be easilycollected from the ground. Such Material, even ifselected for apparent freshness, can, however, beof poor quality; crown collections by climbingare preferable.. Collecting from the tree by handpicldng is feasible for low branches of smallindividuals of Azadirachta indica and Meliaazedarach. Rifles may be used to shoot downbranches (Green & Williams 1969), a techniqueapplicable for harvesting seed of trees in openhabitats. flexible saws are suitable for severingfruit-bearing branches on medium-sized trees(Robbins 1984). A recent relevant review is thatof L Thomson and colleagues at the Food andAgriculture Organisation (Guarino, Rao & Reid1993).

SEED STORAGE

Viability constants for orthodox seeds

The most important storage characteristics oforthodox seeds relate to the mathematicalrelationships between longevity temperature andmoisture content; these relations were originallyworked out for barley (Ellis & Roberts 1980a, b).A detailed exposition of the reasoning is given inEllis (1984). Within a range of moisture contents(lower limit 2-6% and upper limit 15-28%according to species), longevity increases in apredictable way as moisture content is reduced.In addition, for temperature within the range fromabout -13°C to 90°C, longevity increases astemperature is reduced, again in a predictablemanner. The relationship found to fit theobserved data best takes the form:

10g100 = KE - Cw10g10M - CHt - CQt2

where KE, Cw, CH and C0 are species constants,t is temperature in °C and m is percentagemoisture content. Longevity is represented by a,which is the reciprocal of the slope of therelationship between the probit of germinationpercentage and time (d) during storage.Following the work of Ellis and Roberts, thisequation was shown to apply to tree seed usingUlmus carpinifolia and 7brmina1ia brassii(Tompsett 1986a).

The most important concept on which thisequation rests is the statistically normaldistribution of seed deaths over time for aseedlot held under constant storage conditions.The seedlot must be homogeneous, having beencollected at the same time and place and atabout the same stage of maturity A consequenceof normal distribution is the existence of asigmoid shape to the relationship betweenpercentage germination and time; accordinglywhen this relation is plotted on a scale ofnormal equivalent deviates (probits), a straightline is produced. The slopes of many of theseprobit lines must be assessed for a largenumber of combinations of temperature andmoisture content, the results being subjected tomultiple regression analysis in order to produceaccurate species viability constants.

A quicker method for determining the viabilityconstants has, however, been published (Dickieet al. 1990). The method uses the results from a

Table 3. Viability constants for seeds of ten tree species. Values in parentheses are those obtained when KE and Cw arecalculated from an experiment including very few temperatures and combined with the best available temperature constants(Diclde et al. 1990 describe the method)

set of probit lines determined at several moisturecontents using a single high temperature forstorage. The slope of log loo against logiomoisture content for this temperature is thentaken as Cw, from which KE can be derived.Tlie values of the temperature constants, CH andCQ, have been found not to differ significantlyamong a wide range of species (Dickie et al.1990), and may thus justifiably be used inconjunction with the values for KE and Cwobtained. Most of the values for the constants inTable 3 were obtained in this way.

A fifth viability constant, IC1, represents the probitof percentage germination at the start of storage,and is a characteristic of a particular seedlot. Itis important in determining the future storage lifeof the seedlot.

A further notable consideration is the relationshipbetween longevity and moisture content ofultra-dry seed. It has been found that, below amoisture content of around 5%, there is a lowermoisture content limit below which no furtherincrease in longevity occurs, but no decrease isfound either (Ellis et al. 1990).

Experimental approaches

There is not enough space to describe allmaterials and methods used by the author; mostdetails can be found in the publications cited.On receipt, a new species is assessed for itsseed storage physiology; desiccation for thispurpose is generally achieved in a drying room

65

at 15°C and 15% relative humidity sampling atmany moisture contents. If the seed is orthodox,viability constants are then determined. Theviability constants for prediction of longevity inTable 3 are based on at least four accuratelydetermined survival curves (ie the relationshipsbetween viability and period of storage forcombinations of moisture content andtemperature). The value of CW is assessed usingonly the linear part of the relationship betweenlogloo and logia moisture content. The author'sresults for the longest storage periods achieved(Tables 1 & 2) are based on survival curvesdetermined in hermetic dry storage for orthodoxseed.

For recalcitrant seed, a range of moist storageconditions is tested. Results for seed which wassubjected to weekly ventilation are presented inTables 1 and 2; in the experiments in Table 1, atemperature of 18°C was employed, the seedsbeing retained in small rib-channel type plasticbags with a high volume ratio of air to seed, thebags in turn being kept in a Conviron G30germinator at approximately 99% relativehumidity Lowest safe moisture content provides aguide for use in practice, and is defined as themoisture content below which a reduction ingerminability fast appears (Tompsett 1986b). Theeffect of gaseous environment was assessed bycontinuously passing the relevant moisturised gasthrough the seedlot to maintain a constant gascomposition in the environment and constantmoisture in the seed.

Assessments of relationships between relativehumidity and seed moisture content atequilibrium were made with a Michell S4020dewpoint hygrometer.

Theoretical considerations

In the following sections, consideration is given tothe possibility of adapting the present short-termseed storage practice to achieve long-termstorage for conservation purposes. Suitable seedmoisture contents will need to be determined forthis to be undertaken. The present section givessome background to theoretical considerations.

The original research contributing to thepredictive viability equation above was conductedusing conventional moisture content assessmentto represent the water status of seed. However, amore accurate measure of seed water status inrelation to physiological activity is seed waterpotential, which is in turn related to the relativehumidity producing, at equilibrium, the moisturecontent under consideration. These relationshipshave been considered in connection with storagelife by Roberts and Ellis (1989). The reason whyrelative humidity is of importance may beillustrated by considering the influence onlongevity of the reserves in an oily seed. For aspecies with an oil content of 50%,

66

ageing-associated physiological responses wouldbe predicted at a moisture content which isabout half the moisture content for the sameresponses in a non-oily seed, provided all otherfactors are identical; this is because of thehydrophobic nature of the oily reserve. Therelative humidity value at equilibrium, however,would be expected to be similar for the seed ofboth species for the same predictedphysiological responses. Other chemical factorsbesides oil content may affect the relationbetween moisture content and relative humidityat equilibrium. Appreciation of the involvement ofoil, however, may help understanding of whydifferent moisture contents were selected forinclusion in the calculations contributing to Table4; Swietenia humilis and Araucaria cunninghamiihave oily embryos, whereas those ofDipterocarpus alatus are starchy (P B Tompsett,unpublished), so it is not surprising that, optimumlongevity tends to be at relatively low moisturecontents for the former, compared with the latter.The moisture content values used in Table 4 forS. humilis, A. cunninghamil and Ulmus carpinifoliaare based on the author's unpublished work andrepresent the lowest moisture contents at whichthe linear relation between logma and logimoisture content applies; equilibrium relativehumidity of U carpinifolia at this moisture contentis about 6%. The 7% moisture content value forDipterocarpus alatus was chosen because it hasproved difficult to dry the seed further at normaltemperatures.

Both orthodox and recalcitrant seeds ofMeliaceae and Dipterocarpaceae are susceptibleto chilling damage below about 16°C when theyare moist (P B Tompsett, unpublished), so caremust be taken to avoid subjecting seed of eithertype to low temperatures at high moisturecontents.

The genetic stability of material stored as seedhas not been found to be a severe problem forlettuce (Rao et al. 1987), but research to assessgenetic changes in stored tree seed would beadvantageous. Storage at subzero temperatures ofdry orthodox seed should enable seed tosurvive over long periods. The question whethertemperatures much below —13°C conveysignificant benefit requires further attention;Tompsett (1986a) found similar longevity forUlmus seed stored at —13°C and —75°C.

ARAUCARIACEAE

The Araucariaceae family contains the twogenera Araucaria and Agathis.

Desiccation tolerance and mass

Desiccation experiments summarised in Table 2have indicated that there are at least sixorthodox- and four recalcitrant-seeded Araucaria

species (Tompsett 1982, 1984a). The small seedof A cunninghamii (mean 0.15 g dry mass)could be dried to 2% moisture content withoutloss of germinability The medium-sized seed offive New Caledonian species (mean 0.35 g drymass) could be dried to about 12% moisturecontent, and those of four recalcitrant species(mean 4.7 g dry mass) to a mean 35% moisturecontent. One species, A. heterophylla, remains ofuncertain type.

At first, the intermediate desiccation tolerance ofthe New Caledonian species casts doubt on thestorage physiology category of this seed.However, subsequent studies on A. columnarisclearly showed the usual relations amonglongevity moisture content, and temperature; forthis reason, a decision was taken not to create anew storage category (Tompsett 1984b). Asudden drop in viability on placing dry materialin freezing temperatures was indeed observed,but this drop may have been due to thepresence of small proportions of seed in theseedlots which were at an above-averagemoisture content and which were unable tosurvive freezing. This explanation is supported,first, by the absence of any further loss over a20-month period of subsequent storage and,second, by the fact that the driest seedsustained no loss of viability on freezing. Seed ofAgathis australis has similarly been found tosuffer a loss of some viability on drying to 5%moisture content (R D Smith & J B Dickie,unpublished).

Seed longevity

Orthodox Araucaria seeds are relativelylong-lived (Table 2) (Tompsett 1984a). Seed of A.cunninghamii has been stored for 12 years withlittle viability loss (P B Tompsett, unpublished).The seed is predicted to remain at above 85%viability (on a full-seed basis) for over 3000years, provided initial viability is 99.4% (Table 4)and ideal conditions are maintained. Relativelygreat longevity is suggested by the KE and Cwconstants in Table 3; these figures are relativelyhigh for A. cunninghamii with the consequencethat, if all other factors in the above equation areconstant, a relatively greater longevity is achievedfor this species than for all others in the list,except Acer platanoides. In general, either KE orCw can indicate relative longevity among speciesfor two reasons: first, KE appears to be linearlyrelated to Cw for the eight best-determinedspecies (P B Tompsett, unpublished); and,second, CH and CQ probably have the samevalues for all species (Dickie et al. 1990).However, this method of longevity comparisondoes not take into account such factors as seedoil composition, as explained above.

The low viability of many Eutacta species in Table2 after six months' storage is a consequence of

low initial gemination rather than of loss duringstorage. The related species, Agathis australis, canbe stored for over 11 years (Table 2).

Recalcitrant seeds of Araucaria have been storedat Wakehurst Place, Surrey, UK, for longer thanthose of recalcitrant dipterocarps, the periodamounting to over two years for A. araucana.Such seeds are not subject to chilling damage at2°C, unlike the moist seeds of Meliaceae andDipterocarpaceae. Cool storage is a means ofreducing the germination rate of recalcitrantseeds, which maintain metabolism for a longperiod provided chilling damage does not occur.

Effect of the gas environment

Detailed studies concerning the effect of the gasenvironment on the recalcitrant seed of A.hunsteinii have revealed a decrease in longevityas oxygen concentration is reduced from 21% tozero, and a reduction in longevity in anatmosphere with 0.01% ethylene (Tompsett 1983).Ventilation of recalcitrant seed appears essentialto remove damaging gases and to replenishoxygen lost in respiration.

MELIACEAE

Desiccation tolerance, habitat, mass andshape of seed

The seeds in both of the subfamilies which werestudied contain both orthodox and recalcitranttypes (Table 1). Various associations have beenobserved between storage physiology type andother factors. The lighter-weight seeds of thespecies which were studied in the Swietenieae,the mother trees of which tend to come frommore open habitats and from seasonal areas, areorthodox, whereas the often larger seeds of thegenera Guarea and Trichilia, deriving broadlyfrom moister areas, are recalcitrant. Seed ofCarapa procera, which is found in marshylocations among others, is both large andrecalcitrant-seeded. Seed of Azadirachta indica, adrought-tolerant tree, is orthodox and small; seedof Ekebergia senegalensis, similar in size andappearance to that of A. indica, is probably alsoorthodox.

Seed shape may be correlated with physiologytype. The thin seeds of the orthodox-seedednaya and Entandrophragma species are capableof drying quicldy because of their high surfaceto volume ratio. The recalcitrant seeds of Trichiliaspecies and of Carapa procera, however, aremore globular and thus resist desiccation. Fororthodox-seeded species in the forest, thesecharacteristics lead to dry 'storage' on theground, ensuring survival until rain stimulatesgermination. With recalcitrant seeds, however,desiccation after abscission must obviously beavoided.

67

Azadirachta indica seed - an intermediatetype?

Seed of A. ihdica (neem) has been found by theauthor to have partial desiccation tolerance(unpublished), as was the case for Araucariacolumnaris (Tompsett 1984b). Furthermore, neemseed is contained within a drupe, as is coffee,which was the first intermediate seed describedby Ellis, Hong and Roberts. (1990, 1991). Thereare other similarities with coffee, such as theimportance of the developmental stage of thefruit in determining desiccation tolerance andhence potential storage life of the extractedseed. In addition, chilling and freezing damagehas been found to occur in dry A. indica seed(P B Tompsett, unpublished), as was describedfor coffee. Furthermore, considerable variation inthe extent of desiccation tolerance amongseedlots has been observed for both neem andcoffee seed. However, despite all thesephysiological parallels, Azadirachta indica seedhas been stored for 12 years at -20°C and 4%moisture content in the Kew Seed Bank (Table1). The existence of the intermediate categorythus needs further confirmation before it can beaccepted as distinct from the orthodox type. Thevariation in desiccation tolerance observedduring development for A. indica and Coffeaarabica suggests a possible explanation forseedlot variation; on different collection occasions,different proportions of the seedlot may havebeen at desiccation-sensitive stages.

Seed longevity

Orthodox-seeded species in the Swietenieae tribehave been stored for several years (Table 1) andSwietenia humilis is predicted to survive well for266 years under the best storage conditions(Table 4). Although the KE and Cw values forMeliaceae in Table 3 are somewhat low, the

Table 4. Predicted period (in years) for loss of viability from99.4% to 85% germination (1.5 probits) for four tree speciesstored at -13°C and 2°C. Moisture contents selected werethose below which further desiccation causes little furtherincrease in longevity except in D. alatus where 7% waschosen to illustrate potential only Viability constants used arethose in Table 3. Equilibrium relative humidity at 21°C isgiven where known

Temperature

-13°C

2°C

Moisturecontent (%)

Equilibriumrelativehumidity

68

Swietenia Araucaria Ulmus Dipterocarpushurnilis cunninghamii carpinifolia alatus

266 3295

99 1264

2.0 3.3

145 41

52 8

3.3 7.0

c6 0

ability of the seeds illustrated to be dried tovery low moisture contents ensures a lengthystorage life, provided initial viability is highenough. Even the seed of recalcitrant Trichiliaspecies has been kept at very high viability forcfuite long periods (seven months in Tmegalantha), if kept in inflated polythene bags at18°C and carefully ventilated (Table 1). AllMeliaceae seeds so far studied suffer chillingdamage if kept moist and below about 11-16°C.

DIPTEROCARPACEAE

Desiccation tolerance, habitat, size andshape of seed

The relationship between moisture content andgermination was studied for five Dipterocarpusspecies; two desiccation tolerance groups wereidentified (Tompsett 1987). In addition, associationsbetween desiccation tolerance and species habitat,seed mass and seed shape were observed(Tompsett 1987). Seed of D alatus, D. intricatus andD. tuberculatus could be dried to 10-17% moisturecontent without damage. Seed of this fiist gmupincreased in longevity as moisture content wasreduced and as temperature was reduced,confirming their orthodox nature. Despite partialdesiccation tolerance, there was no evidence thatthis type should be categorised as intermediate,which would imply fundamental differences fromthe orthodox type of seed. The species D.turbinatus and D obtusifolius could not be driedbelow 45% moisture content, and wereconsequently classified as recalcitrant.

In agreement with seed physiology type, theorthodox group is found partly or. wholly inseasonal areas, the seed falling on dry ground atthe end , of the dry season, whereas recalcitrantD. turbinatus is found in moister areas. AlthoughD. obtusifolius is found in the dry dipterocarpforest, it also possesses recalcitrant seed; thisspecies has an alternative strategy to survive dryconditions, which is to develop a thick andrelatively impervious calyx around the embryo.Other habitat relationships for dipterocarps aregiven in Tompsett (1986b).

The heaviest seed, D. obtusifolius (mean mass6.79 g), was recalcitrant, whereas the lightestseed, D. intricatus (mean mass 0.48 g), wasorthodox. Seeds of all orthodox species haveprojections from the calyx (tubercles, lamellae orwings), whereas seeds of both of the recalcitrantspecies are globular, giving the smallest possibleratio of surface area to volume. These featuresrespectively maximise and minimise desiccationrates, resulting in optimum survival in nature forboth quickly dried orthodox seed anddesiccation-resistant recalcitrant seed during theperiod on the ground between abscission andgermination. The relationship between size anddesiccation tolerance for other dipterocarps is

given in Tompsett (1986b). Another associationwhich was observed concerned species habitatand seed longevity; those from the drier habitatshad the greatest longevity (Tompsett 1986b).Some of the associations described above werealso observed in the genus Shorea, although inthis case the three species examined wererecalcitrant (Tompsett 1985).

All species of dipterocarp examined sO far, otherthan the three described above, have been foundto be recalcitrant; many authors have publisheddata, and their results are referred to in a recentreview (Tompsett 1992).

Seed longevity

Recalcitrant dipterocarp species vary in longevitysurviving for periods of up to nine months forShorea roxburghii with a final germination of about50%, when moisture content was 40% and storagetemperature was 16°C (Tompsett 1985). Up to oneyear of storage has been reported for Hopeahainanensis with 80% final germination, storageconditions being 37% moisture- content and 18°C(Song et a). 1984). Among the orthodox-seededspecies of Dipterocarpus, D. alatus and D. intricatushave survived without observed loss of viability for4.5 years at near 0°C with 12% moisture content,whilst recalcitrant D. zeylanicus has survived threemonths with 40% moisture content at 21°C, givinga final 53% germination (P B Tompsett,unpublished). Seed of D. alatus is predicted to last41 years before ageing to the regenerationstandard of 85% germination, provided initialviability is 99.4% (Table 4).

FINAL CONSIDERATIONS

Prediction of storage physiology type

The studies described suggest that it may bepossible to provide a provisional classification ofseed of unlaiown storage physiology which mayaid collection and transport until its true type isknown/Characteristics indicating orthodox seedstorage physiology are as follows: small size.(means of 0.3 and 0.5 g in Tables 1 & 2respectively); projections from the seed, such aslamellae; and a relatively dry habitat. Conversely,features suggesting a recalcitrant seed type areas follows: large size (means of 2.4 and 4.7 g inTable 1 & 2 respectively); globular shape; and arelatively moist habitat. Dry soil surfaces areassociated with the occurrence oforthodox-seeded species, even if water ispresent below the soil (eg D. alatus, which growsby waterways but sheds its seed on to dryground before the rainy season).

Prediction of longevity from viabilityconstants

The storage life of orthodox seed can bepredicted if the four species constants in the above

equation are known and if the initial germination(referred to as K,) is accurately aP,sessed asprobability units (see Ellis 1984). 'lb obtainaccurate results, however, the seedlot must behomogeneous, having been collected at the sametime and at approximately the same maturity Otherfactors are important; there is a need to remove allempty seeds, all black-embryo 'seeds in the caseof Dipterocarpus species, and all parasitised seeds.The inclusion of such material would distort thenormal distribution of seed deaths aver time, andso lead to false predictions. If removal is notpossible, account must be taken of these factorsby adjusting results using estimates of thepercentage of empty black-embryo or parasitisedseeds.

Relative longevity of crop and tree seed

A comparison of the mean KE and Cw valuesfor herbaceous and tree species is presented inTable 5. Although the crop values are based ononly five species, they agree well with those fora wider range of less well-determined values forcrop species. The values presented show that, ifall other factors remain constant, longevity isgreater for crop than for tree species. Futureresearch will require consideration of equilibriumrelative humidity values .(or water potentials,which are related to relative humidity atequilibrium), rather than moisture contents.

7'able 5. Mean values for KE and Cw for the five

best-determined herbaceous species compared with meanvalues for ten tee species

Herbaceous species

'free species

KE

8.39(SE 0.47)

5.77(SE 0.30)

Cw Source

4.72(SE 0.24)

2.80(SE 0.22)

Dickie et al.(1990)

Table 3

Strategies for conservation

Despite the shorter storage life of tree seedcompared with that of crop seed, it is evidentfrom data in the present paper (Table 4) thatcertain orthodox-seeded species, such asAraucaria cunninghamli, could be safelyconserved in seed form, as currently practisedfor crop species. Provided seed is harvested withhigh initial viability (K,), survival to theregeneration standard of 85% germination maybe over 3000 years for seed kept at -13°C and3.3% moisture content. For the other speciesconsidered, a shorter storage life is predicted(Table 4). For Dipterocarpus alatus, the predictedseed survival at -13°C and 7% moisture contentis 41 years. This period is adequate to justifyconservation as seed, as it is longer than thejuvenile stage of development, a criterionsuggested by Bonner (1990). However, for certain

69

species, 2°C would appear unacceptable as aconservation temperature. If there is doubtconcerning the ability to attain a high initialviability (Ki) for this species, a dual strategy ofseed conservation combined with ex situ treestand establishment may be advisable. Orthodoxseed exhibiting partial desiccation tolerance(when' present harvesting and processingtechniques are employed) may require furtherresearch, but should eventually be capable ofstorage for very long periods, as evidenced bythe storage of Azadirachta indica seed for 12years (see above). .

In the case of recalcitrant-seeded species, wholeseed is probably not an option for long-termconservation; ex situ tree stands and in situprotection are the only possibilities currentlypractised. Current studies, such as those usingabscisic acid and lowered water potential to slowgermination, have not yet extended storage lifeby substantial amounts. Further research isneeded to develop new ex situ techniques, suchas cryopreservation, in order to conserverecalcitrant-seeded species. Whole seeds ofrecalcitrant species can, however, be used fortransport and short-term storage if they areproperly handled.

Regeneration requirements of material stored asseed will need attention, but the problemsinvolved should not be insuperable.

Practical transport and storage conditions

If seed is harvested at the highest possible initialgerminability (Ki), the greatest possible longevityin storage is ensured. For long-termconservation, orthodox seed should be kept atthe moisture content below which desiccationproduces no further increase in longevity (about2-7%, according to species) and at -13°C orlower. Hermetic storage containers are essentialand will need to be larger than those employedfor crop seed in many cases.

Storage of orthodox seed of the partlydesiccation-tolerant type should at present be atslightly higher moisture contents (12% is oftensuitable) in order to avoid the loss of aproportion of the seedlot by desiccation damage.When further research has resolved theparticular harvesting and processing techniquesrequired for each species, it will probably bebetter to dry to 2-7% moisture content.

Recalcitrant seed can be transported and storedin the short to medium term at above the lowestsafe moisture contents; seed of Araucariaceaecan be kept at 2°C, but seed of Meliaceae andDipterocarpaceae must be retained at about18°C to avoid chilling damage. Ventilation isessential. Seed should be quickly transported inboxes with small holes and packed intocontainers with a high ratio of air to seed in

70

order to ensure maximum viability on arrival; thisprocedure minirnises the dual dangers of anoxiaon the one hand and desiccation due toover-ventilation on the other. For long-termstorage, the possibility of storing excised seedparts must be further examined, building oninitial fmdings for dried A. hunsteinii embryos,which can retain the ability to produce callusafter storage at -20°C for over four years (H WPritchard, P B Tompsett & K Manger,unpublished).

ACKNOWLEDGEMENTSI thank my colleagues at Kew for helpfulcomments on this manuscript. Financial supportwas provided by the UK Overseas DevelopmentAdministration.

REFERENCESBonner, F.T. 1990. Storage of seeds: potential and limitationsfor germplasm conservation. Forest Ecology and Management,35, 35-43.

Choinski, J.S. 1990. Aspects of viability and post-germinativegrowth in seeds of the tropical tree, Trichina dregeanaSonder. Annals of Botany, 66, 437-442.

De Muckadell, J.S. & Pilim, M. 1983. Preliminaryobservations on harvesting, handling and storage of seeds fromsome diPterocarps. (FAO/UNDP-MAL/78/009 Working Paper, no.18.) Sandakan, Sabah: Forest Research Centre.

Dickie, J.B., Ellis, R.H., Kraak, H.L., Ryder, K. &Tompsett, P.B. 1990. Temperature and seed storagelongevity Annals of Botany, 65, 197-204

Dickie, J.B., May, K., Morris, S.V.A. & Tit ley, S.E. 1991.The effects of desiccation on seed survival in Acerplatanoides L. and Acer pseudoplatanus L. Seed ScienceResearch, 1, 149-162.

Ellis, R.H. 1984. The meaning of viability In: Seedmanagement techniques for genebanks, edited by J.B. Dickie,S.H. Linington & J.T Williams, 146-181. Rome: InternationalBoard for Plant Genetic Resources.

Ellis, R.H. & Roberts, E.H. 1980a. Improved equations forthe prediction of seed longevity Annals of Botany, 45, 13-30.

Ellis, R.H. & Roberts, E.H. 1980b. The influence oftemperature and mofsture on seed viability period in barley(Hordeum distichum L.). Annals of Botany, 45, 31-37.

Ellis, R.H., Hong, T.D. & Roberts, E.H. 1986. Logarithmicrelationship between moisture content and longevity insesame seeds. Annals of Botany, 57, 499-503.

Ellis, R.H., Hong, T.D. & Roberts, E.H. 1990. Anintermediate category of seed storage behaviour? I. Coffee.Journal of Experimental Botany, 41, 1167-1174.

Ellis, R.H., Hong, T.D. & Roberts, E.H. 1991. Anintermediate category of seed storage behaviour? II. Effectsof provenance, immaturity and imbibition ondesiccation-tolerance in coffee. Journal of Experimental Botany,42, 653-657.

Ellis, R.H., Hong, T.D., Roberts, E.H. & Tao, K.-L. 1990.Low moisture content limits to relations between seedlongevity and moisture. Annals of Botany, 65, 493-504.

Green, J.W. & Williams, A.V. 1969. Collections ofEucalyptus branch specimens with the aid of a rifle. AustralianJournal of Forest Research, 4, 19-30.

Guarino, L., Rao, V.R. & Reid, R. 1993 A manual of plant

gennplasm collection. Rome. International Board for PlantGenetic Resources. In press.

National Research Council. 1991. Managing global geneticresources. Forest trees. Washington DC: National AcademyPress.

Pennington, T.D. & Styles, B.T. 1975. A genericmonograph of the Meliaceae. Blumea, 22, 419-540.

Preest, D. 1980. Seed storage of several New Zealandindigenous trees. I. Kauri (Agathis australis). New ZealandJournal of Forestry Science, 9, 337-343.

Rao, N.H., Roberts, E.H. & Ellis, R.H. 1987. Loss ofviability in lettuce seeds and the accumulation of geneticdamage under different storage conditions. Annals of Botany,60, 85-96.

Robbins, A.M.J. 1984. 7bols for harvesting tree fruits.(Technical Note, no. 13.) Humlebaek: Danida Forest SeedCentre.

Roberts, E.H. 1973. Predicting the storage life of seeds.Seed Science and lbchnology, 1, 499-514.

Roberts, E.H. & Ellis, R.H. 1989. Water and seed survival.Annals of Botany, 63, 39-52.

Shea, G.M. & Armstrong, P.A. 1978. The effect ofpost-harvest factors on the longevity of Hoop pine seed.(Research Note of the Queensland Forestry Department, no.24.) Queensland: Queensland Forestry Department.

Song, X., Chen, Q., Wang, D. & Yang, J. 1984. A study onthe principal storage conditions of Hopea hainanensis seeds.Scientia Silvae Sinicae, 20, 225-236.

Stubsgaard, F. 1992. Seed storage. (Lecture Note, no. C-9.)Humlebaek: Danida Forest Seed Centre.

Tompsett, P.B. 1982. The effect of desiccation on thelongevity of seed of Araucaria hunsteinii and A. cunninghamii.Annals of Botany, 50, 693-704.

Tompsett, P.B. 1983. The influence of gaseous environment

on the storage life of Araucaria hunsteinii seed. Annals ofBotany, 52, 229-237.

Tompsett, P.B. 1984a. Desiccation studies in relation to thestorage of Araucaria seed. Annals of Applied Biology, 105,581-586.

Tompsett, P.B. 1984b. The effect of moisture content on theseed storage life of Araucaria columnaris. Seed Science and7bchnology, 12, 801-816.

Tompsett, P.B. 1985. The influence of moisture content andstorage temperature on the viability of Shorea almon, S.robusta and S. roxburghii seed. Canadian Journal of ForestResearch, 15, 1074-1079.

Tompseft, P.B. 1986a. The effect of temperature andmoisture content on the longevity of seed of Ulmuscarpinifolia and 7brminalia brassii. Annals of Botany, 57,875-883.

Tompsett, P.B. 1986b. The effect of desiccation on theviability of dipterocarp seed. In: Proceedings of theInternational Symposium on Seed Problems under StressfulConditions, edited by J. Nather, 181-202. (Report, no. 12.)Vienna: Federal Research Institute.

Tompsett, PB. 1987. Desiccation and storage studies onDipterocarpus seeds. Annals of Applied Biology, 110, 371-379.

Tompsett, P.B. 1992. A review of the literature on storage ofdipterocarp seeds. Seed Science and Tbchnology, 20, 251-267.

Turnbull, J.W. 1978. Seed collection and certification. In:International training course in forest tree breeding, 81-94.Canberra: Australian Development Assistance Bureau.

Willan, R.L. 1985. A guide to forest seed handling withspecial reference to the tropics. (FAO Forestry Paper, no.20/2.) Rome: Food and Agriculture Organisation.

Willan, R.L., Hughes, C.E. & Lauridsen, E.B. 1990. Seedcollections for tree improvement. In: Tree improvement ofmultipurpose species, edited by N. Glover & N. Adams, 1137.Arlington, Virginia: Winrock International Institute forAgricultural Development.

71

Capture of genetic variation by vegetativepropagation: processes determining success

R R B Leakey, A C Newton & J McP DickInstitute of Terrestrial Ecology*, Bush Estate, Penicuik, Midlothian EH26 OQB, UK

*A component of the Edinburgh Centre for Tropical Forests

ABSTRACTVegetative propagation offers the opportunity to rapidly overcome the limitations to domestication imposed by long generationtimes, irregular fruiting/flowering and outbreeding. Vegetative propagation techniques are increasingly being applied to a widerange of tree species, of both moist and dry tropics. Identification of the critical factors determining adventitious root developmentis crucial to sustained, cost-effective propagation, even in 'species in which these factors are not currently limiting.

Over the past ten years, research at ITE has identified a wide range of factors which influence rooting, including the stockplantgrowth environment, cutting origin, post-severance treatments applied to cuttings, and the propagation environment. In order tounderstand the influence of these different factors and their interactions, an appreciation of the physiological, biochemical andcytological processes involved in rooting is required. The primary processes occurring in the leaf are net photosynthesis andtranspiration, while those in the stem are starch hydrolysis, translocation of sugars, water and nutrients, respiration, mitosis and celldifferentiation. Each of these processes is influenced by a number of environmental, morphological and physiological variables,such as leaf area and thickness, internode length, chlorophyll content, stomatal density stem Uglification, etc.

Suggestions are made for experimental approaches to improve further the understanding of the mechanisms of rooting of leafystem cuttings. These approaches include the development of a mechanistic model of adventitious root development. The practicalimplications arising from a process-based approach to propagation are also discussed.

INTRODUCTION

In agricultural crops, substantial increases inproductivity and quality of produce have beenobtained by breeding and selection over manygenerations. In contrast, less progress has beenmade in the domestication of trees because oftheir (i) relatively long generation times, (ii)irregularity in flowering and fruiting, and (iii)high incidence of outbreeding, with consequentloss of genetic gain in subsequent generations.Vegetative propagation offers the opportunityrapidly to overcome these limitations, bycircumventing the need for sexual reproductionand facilitating the capture of individualgenotypes.

Vegetative propagation techniques areincreasingly being applied to the domesticationof tropical tree species (Leakey et al. 1990;Mesén, Boshier & Cornelius, pp249-255; Ladipoet al., pp239-248). A range of approaches canbe utilised (see Hartmann & Kester 1983; Leakey1985), including grafting, stem cuttings (Table 1),hardwood cuttings, marcotting (air-layering),suckering, and in vitro techniques such asmeristem proliferation, organogenesis andernbryogenesis. Several of these techniques havethe disadvantages of a low rate of multiplication,a high requirement for skilled labour, or theneed for high capital investment. These problemshave been overcome by recent advances in the

72

development of a low-technology propagationsystem (Leakey et al. 1990; Newton et al. 1992;Newton & Jones 1993a), which has enabled thesuccessful propagation of a wide range ofspecies by leafy stem cuttings (see Table 1).This system utilises non-mist propagatorsconstructed out of cheap and readily availablematerials, with no requirement for a piped waterand electrical supply (Leakey 1991; see alsoLadipo et al., pp239-248).

Although many species can be propagated easilyby leafy stem cuttings, some species arerecalcitrant. For example, early attempts at thepropagation of Albizia falcataria resulted in leafletshedding and death of all the cuttings within tendays. Pre-conditioning the cuttings by intermittentdroughting of the stockplants resulted in leafretention and high rooting percentages, withmore than 100 roots per cutting in some cases(R R B Leakey unpublished data). In suchsituations, identification of the factors limitingrooting is critical to successful propagation.Enhanced rooting is important even in specieswhich are propagated easily as small gains inrooting percentage may be of considerableeconomic value when the species is masspropagated on a commercial scale. In somecircumstances, the rooting ability of cuttings maydecline with successive harvests of cuttings froma set of stockplants. This phenomenon can arise

when over-frequent harvesting of cuttings causesthe depletion of the endogenous reserves of thestump, the death of the stockplant's fine rootsystem, or a reduction in soil fertility For allthese reasons, an understanding of the factorswhich influence rooting is fundamental tosuccessful and sustained vegetative propagation.

Another key reason for a better understanding ofthe rooting process is the accumulation of abody of apparently contradictory data in thescientific literature. For example, the role ofcarbohydrates in the rooting process is still thematter of great debate (see review by Veirskov1988). Highly contrasting results have beenobtained from the same species in differentsituations. Most propagation experiments usuallyconsider only one or two of the many factorsknown to influence rooting, and fail to recognisetheir interactions. Many investigations also fail tomeasure and record sources of variation inrooting ability which can differ betweenexperiments, such as propagation environmentand stockplant growth conditions. Consequentlythe mechanisms of root formation remain unclear.

In order to understand the interactions betweenthe different factors influencing rooting, an

West Africa

Afzelia africana

Afrostyrax lepidophyllus

Bailonella toxisperma

Carapa procera

Ceiba pentanclra

Milicia (syn.

Chlorophora) excelsa

Entandrophragma

angolense

Khaya ivorensis

naya senegalensis

Lovoa trichilioides

Nauclea diderrichil

Ricinodendron heudelottii

7bclea verdoorniana

7brminalia ivorensis

lbrinir2alia superba

7btrapleura tetraptera

neculia africana

Triplochiton scleroxylon

Central and SouthAmerica

Albizia caribaea

Albizia guachapele

Alnus acuminata

Annona muricata

Caesalpina spinosa

Casimiroa edulis

Cedrela odorata

Chrysophyllum cainito

Coidia alliodora

Eugenia jambos

Ochroma pyramidale

Prosopis cineraria

Prosopis juliflora

Psidium guajava

Swietenia macrophylla

Swietenia mahagoni

Tipuana tipu

Vochysia hondurensis

East and Central Africa

Acacia nilotica

Acacia senegal

Acacia tortilis

Dalbergia melanoxylon

Faidherbia albida

Maesopsis

Melia volkensii

Sesbania sesban var.

sesban

Sesbania sesban var.

nubica

7brminalia brovvnii

7brminalia prunioides

7brminalia spinosa

Vateria seychellarum

Vitex keniensis

appreciation of the physiological processesinvolved is required. In recent years, there havebeen a number of literature reviews on this topic(eg Leakey 1985; Andersen 1986; Haissig 1986;Davis, Haissig & Sankhla 1988; Thompson 1992).Consequently this paper attempts to pulltogether data on these processes in leafysoftwood cuttings obtained from a limitednumber of experiments with tropical treespecies. In particular, it suggests how theseprocesses may be influenced by the followingfactors: (i) the stockplant's growing environment;(ii) the cutting's position of origin within a shootand the shoot's position within the canopy of thestockplant; (iii) the post-severance treatmentsapplied to cuttings; and (iv) the environment ofthe propagating system. The implications of theseconclusions for both research and the practicalapplication of vegetative propagation techniquesare discussed.

PROCESSES INVOLVED IN ADVENTITIOUSROOT DEVELOPMENT

The key physiological, biochemical andcytological processes influencing adventitious rootdevelopment in a leafy cutting are

l'able 1. Tropical tree species vegetatively propagated at the Institute of Terrestrial Ecology Edinburgh

South Africa

Acacia karoo

Colophospermum

mopane

Triplochiton zambesiacus

Vangueria infausta

SE Asia and Australasia

Acacia mangium

Agathis australis

Agathis dammara -

Agathis macrophylla

Agathis obtusa

Agathis robusta

Agathis vitiensis

Albizia falcataria

Anthocephalus chinensis

Araucaria huristeinii

Camposperma

brevipetiolata

Citrus halimi

Dipterocarpus turbinatus

Durio zibe thinus

Gmelina arborea

Shorea

Shorea

Shorea

Shorea

Shorea

albida

contorta

curtisii

leprosula

macrophylla

Tamarindus indica

7brininalia calamansanii

7brminalia brassii

?bona ciliata

73

photosynthesis, transpiration, respiration, starchhydrolysis, translocation of sugars, water andnutrients, mitosis, cell differentiation andelongation. While all of these processes mayoperate in different parts of leafy softwoodcuthngs, photosynthesis and transpirationprimarily occur in the leaf, and mitosis and celldifferentiation are generally of greatestimportance in the cutting base. The stem is theprimary organ of translocation for nutrients,carbohydrates and water between the leaf andthe cutting base.

The interactions between these processes andthe factors affecting rooting, mentioned earlier,are considered separately below although, inreality they are interlinked.

FACTORS INFLUENCING THE PROCESSESTHAT DETERMINE ROOTING

Stockplant growth environment

The physiology and morphology of stockplantsare influenced by the growth environment, whichmay include nutrients, water, temperature andlight (both irradiance and spectral composition).In addition, the individual cuttings are influencedby the variation in microclimate within thecanopy of the stockplant. The characteristics ofthe cutting (Figure 1) that are influenced bythese environmental factors include:

• the area and thickness of the leaf;

• leaf photosynthetic capacity as affected bystomatal density chlorophyll content, etc;

• internode length and stem diameter;

• leaf and stem starch, soluble carbohydrateand nutrient contents;

• pre-severance leaf and stem water potentials;and, probably

• concentrations of endogenous growthsubstances.

Some influences of stockplant growth environmentare illustrated by results from experiments withTriplochiton scleroxylon, in which rooting abilitywas found to be correlated with pre-severancenet photosynthetic rate, which in turn wasdetermined by irradiance and nutrient applicationtreatments during growth. In stockplants grownunder artificial lighting (6.3 red/far-red (RFR)ratio) with an irradiance of 250 wnol n1-2 S-1

and nutrient addition, photosynthetic rate variedbetween 0.003 and 0.009 mg 002 s-1 1eaF1, anda mean rooting percentage of 33.3±5% wasobtained. These figures compared withphotosynthetic rates of 0.0005-0.002 mg 002 s-1leaF1 and an associated mean rootingpercentage of 9.3±3.1% when stockplants weregrown under an irradiance treatment of 650 grriol

-2. - 1m s and without added nutrients. The lowrooting percentages obtained under this high

74

Pre-severance stockplant environment

1. Light* 2. Water 3. Nutrients 4. Temperature

Leaf areaLeaf structure

Stomata! densityChlorophyll contentLeaf water potentialSugar/starch content

Nutrient content

Leaf

Net photosynthesis

Transpiration

Number of shoots

Internode lengthLignification

Secondary thickeningStem diameter

Stem water potentialSugar/starch content

Nutrient content

Mitosis

Stem

Cell differentiationand elongation

Figure 1. Schematic diagram of the effects of thepre-severance stockplant environment on the morphologicaland physiological factors affecting rooting ability insingle-node, leafy cuttings via their influences on thebiochemical processes of their leaf and stem (*irradiance and

qualitY)

irradiance/low nutrient treatment were attributedto the accumulation of starch in the cuttingspre-severance (Leakey & Storeton-West 1992).

The importance of stockplant illumination isfurther illustrated by other experiments withconstant irradiance but varying light quality InT scleroxylon under artificial lighting, higherrooting percentages (92%) were obtained at a

ratio of 1.6, while lower rooting percentages(53%) were obtained at a R:FR ratio of 6.3(Leakey & Storeton-West 1992). These results,together with those on the effects of irradiance,concur with results examining the varying rootingability of cuttings from different shoots within thestockplant canopy (Leakey 1983). However, incommon with many other vegetative propagationexperiments in the scientific literature, no datawere collected on the physiological activity ofthese various shoots. To hypothesise how thesetreatments affected the physiological processesand morphological condition of these shoots, it isnecessary to examine a more extensive data set

from similar experiments with Eucalyptus grandis(Hoad & Leakey 1992) testing a wider range ofR:FR ratios (including more natural ratios of RFR,eg 0.4-1.3). In this more recent study, rootingability was related to treatment-induced changesin leaf and stem morphology and to pre- andpost-severance gas exchange and carbohydratedynamics. In addition, pre-severancephotosynthetic rate per unit leaf area, chlorophyllconcentration, stomatal conductance and wateruse efficiency increased with an increase in theR:FR ratio. The photosynthetic rate per unit ofchlorophyll was, however, greater at low R:FRratios, and there was a positive linear relationshipbetween the percentage of cuttings rooted andthe pre-severance photosynthetic rate per leaf.

As mentioned above, there are importantinteractions between the light environment of thestockplant and its nutrient regime. Theseinteractions have been demonstrated inT scleroxylon as stockplant environment effects,using controlled conditions and shade treesabove rows of stockplants in the field (Leakey &Storeton-West 1992). Similar interactions havealso been demonstrated between the shoots ofstockplants with and without fertilizers (Leakey1983). In both instances, the highest rooting

Leaf size and structureLeaf water potential

Nutrient contentSugar/starch content

Leaf

Net photosynthesis

Transpiration

Cutting origin within shoot

Internode lengthStem diameter

Secondary thickeningLignification

Nutrient contentSugar/starch contentStem water potential

Mitosis

Stem

Cell differentiationand elongation

0

rT_I

.(acc

Figure 2. Schematic diagram of the effects of cutting originwithin shoot on the factors affecting looting ability insingle-node, leafy cuttings via their influences on thebiochemical processes of leaf and stem

percentages occurred in shaded shoots fromplants with added fertilizers. The unrest" ictedapplication of fertilizers is, however, notrecommended as, at very high fertilizerapplications, plants of Khaya ivorensis yieldedcuttings which suffered high mortalities. In thiscase, mortalities due to rotting were associatedwith increasing reducing sugar and decliningstarch contents of the leaves, during the periodof propagation (Tchoundjeu 1989).

Cutting origin

The physiology morphology and subsequentrooting potential of a cutting are also influencedby:

• the position within a stem from which thecutting originates (Leakey & Coutts 1989;Figure 2); and

• the position of the shoot within the stockplantcanopy (Leakey 1983, 1985; Tchoundjeu1989; Figure 3).

Cutting mortality can also be related to nodeposition in the shoot (Leakey 1983), suggestingthat the factors determining rooting and mortalityare inversely related and form some kind ofcontinuum.

Leaf areaLeaf structure

Stomata! densityChlorophyll contentLeaf water potential

Sugar/starch contentNutrient content

Leaf

Net photosynthesis

Transpiration

9

Shoot position within stockplant

Internode lengthLignification

Secondary thickeningStem diameter

Stem water potentialSugar/starch content

Nutrient content

Mitosis

Stem

Cell differentiationand elongation

Figure 3. Schematic diagram of the effects of the shootposition within stockplant on the morphological andphysiological factors affecting rooting ability in single-node,leafy cuttings via their influences on the biochemicalprocesses of leaf and stem

0

773.

a)

75

Within a stem, there are gradients in:

• succulence,

• leaf size, age and morphology

• leaf water potential (Leakey 1983), and

• mutual leaf shading

which affect both the amount and quality of lightreceived by the leaves (Leakey & Storeton-West1992), and consequently net photosynthetic rateand stomatal conductance. In addition, there aregradients in:

• stem morphology such as the extent ofsecondary thickening and lignification, andinternode length (Leakey & Mohammed1985), stem diameter and, consequently stemvolume (Leakey Dick & Newton 1992);

• starch and sugar content (Leakey & Coutts1989);

• nutrient content (Tchoundjeu & Leakey 1993);and, almost certainly

• the content of plant growth regulators androoting co-factors.

The most important of all these factors inT scleroxylon seems to be the effect of cuttingposition on the internode length and diameter.Together, these two variables result in variationin cutting volume. From these results and thoseof other species (see, for example, Hoad &Leakey 1992; Dick, East & Leakey 1991), wehave developed the hypothesis that cuttingvolume determines the capacity of a cutting tostore assimilates produced both pre- andpost-severance (cf Leakey & Storeton-West 1992).Cuttings with a small stem volume may becomesaturated with starch, which can inhibitphotosynthesis and consequently rooting.However, an often overlooked aspect of thecarbon budget is the stem respiration rate, whichis influenced by the cutting's size and itsenvironment (pre- and post-severance). Toexamine the mean respiration rate of internodetissue, a one cm section was taken from thebase of Prosopis juliflora cuttings with a range ofbasal diameters (Dick, Blackburn & Mc Beath1993). Using an oxygen electrode, the respirationrate per gram dry mass of tissue was found todecrease within sequentially more basal cuttings,possibly as a result of the increased lignificationof older tissues. However, because of the largermass of these basal cuttings, the total respirationfor the whole 1 cm section of tissue (Figure 4)was greatest in the cuttings with the largestbasal diameter. Thus, the larger mass of tissue inthe larger cuttings more than compensated fortheir smaller respiration rate per unit mass. Inaddition, the larger diameter cuttings had alarger area of wound tissue which may also havecontributed to their increased respiration. Theseresults show that, compared with smallerdiameter cuttings, larger diameter cuttings have

76

0.014

0.012

0.010

0.008

0.006 •

•

•• •

• •

•

•

• 1.0 1.5 2.0 2.5 3.0

Stem diameter (mm)

Figure 4. Total respiration rate of the basal 1 cm section ofProsopis julillora cuttings measured at the time of severance,using an oxygen electrode (soume: Dick et al. 1993)

greater total respiratory/metabolic activity at theirbasal (wounded) end. This feature may contributeto the increased rooting potential of largediameter cuttings (Dick, Dewar & Leakey 1992).

Between shoots there are variables that affectrooting which are similar to those environmentalfactors affecting the whole stockplant, except thatthey are associated with the microclimaticenvironment of the individual shoot. As in thewhole plant, the interactions between light(quality and quantity) and nutrient availabilityseem to be particularly important (Leakey 1983),presumably mediated by effects on shootmorphology and gas exchange. Cuttings frombasal shaded shoots of T scleroxylon stockplantsgiven added nutrients have a high rooting abilityIn addition to these microclimatic effects, there isalso inter-shoot competition, which is related tothe number of shoots on the stockplant and totheir position within the dominance hierarchybetween shoots (Leakey 1983). In smallT scleroxylon stockplants (height 0.5 m), higherrooting percentages were associated with cuttingsfrom the top shoot, and progressively declinedwith each shoot down the plant. In larger plants(ie 1.5 m), rooting percentages of cuttings fromthe upper shoots were low and those from basalshaded shoots were relatively high. In Lovoatrichilioides, another West African hardwoodspecies, rooting percentages of cuttings frombasal shoots were highest, even in smallstockplants (Tchoundjeu 1989). The rooting abilityof these cuttings from different shoots alsodepended on how many shoots there were perstockplant.

From this examination of the effects of cuttingorigin and stockplant environment, it is clear thatthe physiological and morphological condition of

the cuttings at the time of severance from thestockplant is very important, and that, even insmall, 'juvenile' plants, rooting ability can rangefrom 0-20% in unsuitable material to 80-100% insuitable material. It is also clear that, asstociciplants get bigger and more complex, itbecomes more difficult to obtain shoots in aphysiological/morphological condition that confersgood rooting ability This loss of rooting ability isfrequently attributed to the loss of juvenility (ieattainment of sexual maturity or ontogeneticageing). Numerous experiments havedemonstrated this loss of rooting ability but theyhave all been confounded by the many otherchanges that occur as plants grow bigger andmore complex (Leakey 1985). There is, therefore,no good evidence from in vivo studies thatsexual -maturation per se affects- rooting abilityalthough there are currently some studies from invitro culture that suggest that there are events invitro which apparently 'rejuvenate' plant tissues(Jones & Webster 1989). As yet, there is nounderstanding of what these events are or howthey are mediated. For the moment, therefore, itseems most appropriate to indicate (seeFigure 3) that the effects of ontogenetic ageingon the rooting process are not lmown. What isclear is that physiological ageing, as manifestedby the physiological and morphological conditionof the cutting, does have major effects on thesuccess of vegetative propagation programmes(Leakey et al. 1992).

Trimming leaf, ie leaf area Cutting lengthLeaf removal, ie leaf number Auxin application

Physical treatment of cut end

Leaf

Net photosynthesis

Transpiration

Post-severance treatment

Mitosis

Stem

Cell differentiationand elongation

Figure 5. Schematic diagram of the effects of post-severance

treatment on the morphological and physiological factors

affecting moting ability in single-node, leafy cuttings via theirinfluences on the biochemical processes of leaf and stem

Post-severance treatments applied to thecuttings

The rooting of cuttings is influenced by theapplication of a number of post-severancetreatments (Figure 5). The most commontreatment is application of an auxin 'rootinghormone'. Auxins often hasten rooting, increasethe percentage of cuttings rooted, and increasethe number of roots formed. However, the resultsof even this most commonly applied treatmentare subject to considerable variation, because ofthe interactions with the factors listed previously(Newton et al. 1992). Other post-severancefactors which influence rooting include leaf area,which is often reduced to minimise waterdeficits. Evidence from a number of speciessuggests that there is an optimum leaf area, as acompromise between having a large enough leafto produce assimilates and a small enough leafto minimise water loss (Leakey & Coutts 1989).In T scleroxylon and several other species, theoptimal leaf area is about 50 cm2 (Leakey 1985).However, in L. trichilioides and Khaya ivorensis,about 200 cm2 and 20 cm2 respectively havebeen shown to be optimal, under the conditionstested (Tchoundjeu 1989). In contrast, otherspecies which root easily have displayed notendency towards an optimum leaf area (Leakey1990; Newton, Muthoka & Dick 1992). Insmall-leaved species, the optimal leaf area canperhaps only be achieved by retaining morethan one leaf per cutting and thus more thanone node, but there is some evidence that thereare interactions between cutting length and thenumbers of leaves and nodes.

Cutting length is another variable that can bemanipulated at the time of severance. The effectof stockplant/shoot environment on cutting lengthand volume has already been discussed.However, it is not often realised how trimming acutting to a standard length rather than acceptingnatural variation in internode length affects thecutting volume (Leakey et al. 1992). Standardisingcutting length has the advantage that all cuttingscan be inserted into the medium to the samedepth and also have their leaf held at the sameheight above the cutting medium, which can beimportant for cutting survival (S P Hoad &R R B Leakey unpublished). As cutting diameterusually increases down a shoot, takingconstant-length cuttings usually confers anadvantage on more basal cuttings as their stemvolume is greater. In contrast, taking cuttings of aconstant number of nodes may result in anadvantage to apical cuttings, if their greaterlength exceeds the effect of cutting diameter oncutting volume (Leakey et al. 1992).

It may sometimes be necessary to store cuttingsprior to insertion in the propagator, particularly ifthey are collected from the field at somedistance from the propagation unit. Rooting ability

77

declines with increased storage time, and mayalso be influenced by storage temperature. InT scleroxylon, storage for 24 hours in aninsulated box (18-23°C) had no effect on rooting,but, after 72 hours' storage, rooting percentageswere significantly decreased (Njoya 1991).

The propagation environment

Successful propagation of leafy cuttings isdependent on the maintenance of suitable airand leaf temperatures, irradiances and leaf/airvapour pressure deficits (VPDs) duringpropagation (see review by Loach 1988b). Leafycuttings are vulnerable to desiccation duringpropagation, particularly prior to rootdevelopment and when VPD is high. The primaryeffects of water deficits are lower leaf and stemwater potentials and the consequent inhibition ofphysiological activity and of photosynthetic ratein particular. The association of low leaf waterpotentials with low rooting percentages has beenillustrated by Loach (1988a) and Newton andJones (1993b).

Although many authors have suggested thatcurrent assimilates are important for rooting,there have been relatively few studies that haveactually measured photosynthesis andtranspiration during the rooting process (seereview by Davis 1988). Recent studies haveshown that many tropical tree species activelyphotosynthesise and transpire prior to rootformation, supporting the suggestion that gasexchange during propagation is one of thecritical factors influencing rooting (Leakey &Storeton-West 1992; Newton, Muthoka & Dick1992; Newton et al. 1992; Newton & Jones1993b; Hoad & Leakey 1992; Mesén, Leakey &Newton 1992; Mesén 1993). In Terminalia spihosa,mean photosynthetic rates of 2 gmol m 2 S-1 anda maximum rate of over 6 gmol ITC2 s-1 wererecorded in a non-mist propagator prior to rootdevelopment. These photosynthetic rates wereassociated with rooting percentages of over 80%;the rate at any given time was found to bestrongly dependent on stomatal conductance andirradiance (Newton et al. 1992b). Transpirationrates of 0.002-0.007 g H20 crn-2 1-1.-1 wererecorded in cuttings of Cordia alliodora,depending on the rooting medium (Newton,Muthoka & Dick 1992; Newton et al. 1992), andstomatal conductances in the range 0.1-1.4 cms were recorded for four species in both mistand non-mist propagation systems (Newton &Jones 1993b). The contrasting effects ofirradiance on VPD, photosynthetic activity andcutting water relations will determine theoptimum leaf area of the cutting, and the shadingregime to be employed during propagation(Newton & Jones 1993a, b).

Apart from influencing VPD, air and bedtemperatures in the propagator also influence the

78

rates of all the processes directly (Figure 6),particularly in metabolic processes such asstarch mobilisation, respiration, mitosis and celldifferentiation. Some species display cleartemperature optima in terms of rooting. Forexample, from a range of 20-38°C, an optimumbed temperature of 29-31°C was identified inT scleroxylon, at an air temperature of 25-30°C(Leakey Chapman & Longman 1982). Therelationship between air, leaf and bedtemperatures is generally considered to influencethe relative activity of the different parts of thecutting, and may account for shoot growth priorto root formation during propagation.

Application of nutrients during propagation mayalso facilitate root development (see review byBlazich 1988), by influencing photosynthetic rateand other metabolic processes. Nutrientapplication may be particularly beneficial inslow-rooting species under mist propagationsystems, Where nutrient leaching from the cuttingcan be a problem. Foliar nitrogen concentrationis particularly influential in determiningphotosynthetic rate, and nitrogen is alsoimportant for starch mobilisation and othermetabolic processes.

Variation in ambient CO2 concentration of the airsurrounding the cuttings may also influencerooting, primarily by its effect on photosyntheticrate and dry mass accumulation (Davis & Potter1983). In non-mist propagators, CO2concentrations have been found to vary diurnally

Propagation environment

1. Light' 2. Water' 3. Nutrients 4. Temperature 5. CO,

Leaf

Net photosynthesis

Transpiration ;15

-6

tTs-

Pis

Mitosis

Stem

Cell differentiationand elongation

Figure 6. Schematic diagram of the effects of propagationenvironment on the rooting ability of single-node, leafycuttings via their direct influences on the biochemicalprocesses of leaf and stem (Iirradiance and quality; 2VPD,humidity, moisture cOntentof medium, air/water ratio ofmedium)

between 150 and 550 limol mor-1 (Leakey et al.1990; Matin 1989; Newton & Jones 1993a), whichmay be expected to have a major effect onphotosynthetic activity

The rooting medium determines the amount ofwater and air available to the base of thecutting, which will influence the rates oftranspiration and respiration respectively as wellas the other metabolic processes in the cuttingbase. The effect of different concentrations ofindole-3-butyric acid (IBA) on rooting ofBombacopsis quinata and Vochysia hondurensishas been found to interact with the type ofrooting medium utilised (Newton et al. 1992).

Genetic variation

Different species and clones vary in theirphysiological and morphological characteristics. Itis likely therefore, that much of the geneticvariation in rooting success can be attributed tothe interactions between the above processesand factors influencing rooting. In this connection,there is increasing evidence for the role ofgenes which encode enzymes involved in thebiosynthesis and metabolism of the auxinindole-3-acetic acid (Blakesley & Chaldecott1993).

SUGGESTIONS FOR EXPERIMENTALAPPROACHES TO UNDERSTAND THEMECHANISMS OF ROOTING

For research projects aimed at optimising rootingfor a given species, it is necessary to conductan experimental programme. Ideally thisprogramme should involve the followingstrategies, although it is recognised that the idealsituation is rarely achieved in practice.

Experimental analysis and design

All experiments should obviously follow standardexperimental design practices and be wellreplicated. The minimum number of cuttings pertreatment or clone/treatment interaction should be40-60. The major difficulty with analysis is thateither a cutting roots or it does not, and thusthere are only two possible outcomes, giving abinomial distribution. With moderate numbers ofreplicates, it may be possible to use standardANOVA or regression procedures on either theraw data or a transformed data set, but care is'needed and the validity of various assumptionsmust be tested. Alternative approaches involveeither the contingency table approach using X2

tests, although these results can be difficult tointerpret, or use of a generalised linear modelwith its associated analysis of deviance, whichrequires more specialised computing software.

Depending on the chosen technique of analysis,an appropriate experimental design should beadopted. Ideally cuttings should be blocked in

the propagator according to the spatial variationin the propagation environment, which may bepronounced (Newton & Jones 1993a). Treatmentsshould be replicated across all blocks andrandomised within a block, following standardpractice. Other sources of variation should ideallyalso be randomised within each block, butfrequently this is impracticable. For example,cuttings from a single shoot may be most easilylaid out in node order rather than fullyrandomised with respect to node position.

A number of recent studies (Leakey &Storeton-West 1992; Hoad & Leakey 1992) havedemonstrated that, if the experimental analysis islimited to the overall effect of treatments, a greatdeal of the variation in rooting may remainunexplained. By accounting for the variationattributable to node position, for example, moreinformation on different processes is oftenobtained (Dick et al. 1991; Dick & East 1992). Ifthe experiment also includes destructive harveststaken at different times during propagation, /additional information on the dynamics ofdifferent processes may be obtained. Forexample, studies of the carbohydrate dynamicsin T scleroxylon have demonstrated that there isno relationship between soluble carbohydratecontent at insertion (day 0) and subsequentrooting, but a strong relationship between solublecarbohydrate content at day 28 and rooting atday 42 (Leakey & Coutts 1989). Destructiveharvests of this sort obviously increase thenumbers of cuttings required per treatment.These harvests can and should ideally beassociated with non-destructive assesSments ofphysiological variables (gas exchange, respiration,chlorophyll fluorescence, etc), as anunderstanding of the dynamics of theseprocesses is fundamental to understanding theachievement of successful rooting.

Control of variation

In any given experiment, some variables will bemanipulated, whereas others will either becontrolled or simply measured. Any variableswhich cannot be controlled should be measuredand reported. The variables which should benoted include all the factors which may influencerooting ability as outlined above, eg leaf area,stem length and diameter, node position, numberof nodes, number of leaves, leaf shedding,changes in leaf colour, genotype, propagationenvironment, pre-severance environment,physiological age, chronological age, ontogeneticage, shoot number and position, nutritional status,etc. Although it is recognised that practicallimitations will restrict which variables can bemeasured, experimenters should be aware of allthe potential sources of variation in rootingability and should seek to control or recordthem as far as possible.

79

Modelling adventitious root development

As rooting is influenced by so many interactingfactors, there is a need for process-basedmodels as a tool to understanding. To date, amodel has been constructed which describes thedevelopment of root structural dry matter frompre-formed initials, as well as the dynamics ofsugar and starch pools within the cutting duringthe rooting process. This model (Dick & Dewar1992) was constructed on the basis that thefollowing previously mentioned factors werepositively correlated with rooting: (i) respirationrate at the base of the cutting; (ii) labile sugarcontent, rather than total carbohydrates; and (iii)post-severance photosynthetic rate of the cutting.

The leafy softwood cutting is considered to havethree structural compartments: leaf, internode androot, but can easily be adapted to include agrowing shoot or to be a leafless hardwoodcutting. A leafy softwood cutting is represented(see Figure 7) by boxes (i) and (ii) and definedby six dynamic variables: the soluble sugarpools (mg sugar) in the leaf (Lsg), internode(Isg), and root (Rsg); the starch pools (mg starch)in the leaf (Lst) and internode (Ist); and thestructural dry matter pool (mg structural dm)-1in the root (Rx). If, as in Cordia alliodora (Mesén1993) or Prosopis julillora (Dick et al. 1991), thecutting produces a new axillary shoot duringpropagation, this can be represented by theaddition of an additional box In this case,the bud has a dry matter pool (13x) whichcompetes for sugars with the new root (Rx). Inboth these cases, the leaf and internodestructural dry matter pools (Lx and Ixrespectively) are held constant in the model. Inthe case of hardwood cuttings (which can berepresented by box ii, with or without box iii;see Figure 7), the dynamic variables are thesame as for the softwood cutting with a growingshoot, except that the leaf sugar and starchpools are omitted. Given the initial values of thedynamic variables at severance (t=0), the modelsimulates the development of root initials, androoting is said to have occurred when the rootstructural dry mass reaches a threshold (x mg).

As explained earlier in this paper, many factorsinfluence the processes that determine rooting.As an example of the situation when structuralroot growth is limited by the amount of sugarreaching the cutting base, the output of themodel has been parameterised with publisheddata for single-node cuttings of Triplochitonscleroxylon (Leakey & Coutts 1989), with leafareas of 10, 50 or 100 cm2.

To calibrate the model in the absence ofmeasured rates of photosynthesis, respiration andsugar transport, physiologically reasonableparameter values were assumed (Table 2). Thesevalues were taken to be the same for cuttings

80

Figure 7. Schematic diagram of the rooting model. Box (i)represents the leaf, (ii) the internode and rooting zone, and(iii) a growing bud or shoot. Solid boxes represent thedynamic starch (st), sugar (sg) and structural dry matter (x)

pools. Arrows represent fluxes of sugar assimilation (A),starch/sugar interconversion (I), sugar transport (T), mot sugarutilisation (U), root structural growth (G), internodemaintenance respiration (Rm) and root growth respiration (Rg)

Table 2. Parameter values and initial values of state variablesused to simulate the root development of single-node leafycuttings of T scleroxylon, for three different leaf areas(source: Dick & Dewar 1992)

Leaf area (cm2)

10 50 100

Parameters

gij Leaf/internode conductance(mg structural dm d-1)

gir Internode-root conductance

150

150

150

150

150

150

ix Internode structural dm (mg) 300 300 300

ki Starch/sugar conversioncoefficients (d-1)

k2

0.0001

0.03

0.0001

0.03

0.0001

0.03

0.0001 0.0001 0.0001

Ica 0.05 0.05 0.05

Root growth efficiency 0.5 0.5 0.5

li Utilisation constant (d-1) 10.0 10.0 10.0

cJi Specific internode respiration rate 0.0 0.0 0.0[10-3 mg sugar d-1(mg structural dm)-1]

Specific leaf photosynthetic activity 8.0 9.4 4.1[10-3mg sugar d-1(mg structuialdm)i]

lx Leaf structural dry matter (mg) 300 500 700

Initial values of state variables (mg)

iss Internode sugar 9.9 9.9 9.9

ist Internode starch 10.8 0.8 10.8

Iss Leaf sugar 3.9 6.5 9.1

1st Leaf starch 0.3 0.5 0.7

rsg Root sugar 0.033 0.033 0.033

rx Root structural dry matter 1.0 1.0 1.0

with 10, 50 and 100 cm2, except for the specificleaf photosynthetic activity (al) which wasderived, such that 50% rooting occurred after 24,19 and 28 days for cuttings with 10, 50 and100 cm2 leaf areas respectively as in Leakeyand Coutts (1989). The results of this simulationsuggest that the 10 cm2 leaves werepredominantly area-limited, while the 100 cm2leaves were predominantly limited byphotosynthetic rate per unit area. This finding isconsistent with the observation by Leakey andCoutts (1989) that 50 cm2 was optimal forrooting and that assimilation rates were limitedby low leaf water potential in cuttings with leafareas of 100 cm2.

This example serves to show that the modelprovides a framework for examining the role offactors known to affect rooting. As a furtherillustration, if an increasing range of specificinternode respiration rates is used in the model,rooting in a cutting with a leaf area of 50 cm2 isprogressively delayed until the root sugar pooldeclines to zero, at which point further increasesin respiration rate result in cutting mortality(Table 3).

Table 3. Predicted time of rooting or death of T scleroxyloncuttings with leaf area 50 cm2, for various values of theinternode respiration constant, i. All other parameters andinitial values of state variables are as given in Table 2(soume: Dick & Dewar 1992)

Internode respiration constant[10-3 mg sugar c1-1(mg structural dm)-I]

0

5

10

13

14

14.3

14.4

14.5

15

20

Days torooting

19

26

46

137

576

Days todeath

35

19

13

5

2

As already mentioned, the model, like allmechanistic models, makes various assumptions.So far, the only assumption tested is that the rateof respiration at the base of the cutting (rootingzone) is higher than that of the stem above it(Dick et al. 1993) This study confirmed thisassumption and suggests that a concentrationgradient is created which drives sugar transportbasipetally. The measurements also supportedone of the predictions of the model: during theformation of callus and new roots, the respirationrate at the base of the cutting increases withtime (Dick et al. 1993).

In summary the model provides a potentiallyuseful framework for: (i) the design and

interpretation of rooting experiments on awhole-cutting basis, and (ii) understanding themechanistic control of adventitious rooting. Withfurther development and a better understandingof the interactions between carbon, nutrients andwater, we hope that it will be possible to usethe model to improve stockplant managementand achieve more predictable and sustainedrooting during mass propagation. However, it isalready clear, from the experience of developingthe model, that there may be several contrastingsets of conditions that will give acceptableprotocols for commercial vegetative propagation,depending on whether the stockplants andcuttings are managed in a way that makes themdependent on stored reserves or currentassimilates.

PRACTICAL IMPLICATIONS

In situations where a full-scale researchprogramme is inappropriate, the propagation ofan undomesticated species may be approachedby considering a limited number of key factors.These include leaf area, cutting length, auxin andpropagating conditions (eg Tchoundjeu & Leakey1993). As a starting point, cuttings should be5-10 cm in length, with a leaf area reduced by50% and an applied auxin concentration of0.2-0.4% LBA in a solution of alcohol, or in theform of a commercial rooting powder. Cuttingsshould be inserted to a depth of 15-25 mm inthe rooting medium, ensuring that the leaf doesnot touch the medium surface, and that themedium at the cutting base is not waterlogged.The rooting medium should consist of an inert,well-aerated moisture-retaining substrate, such asmixtures of fine gravel, sand or rotted sawdust.The propagator should be shaded to about 25%of full sunlight in order to keep the propagatorsas cool and as humid as possible and yetprovide enough light for the promotion ofphysiological activity

In any propagation programme, the reasons forlack of success may be difficult to ascertain, asso many factors influence rooting. From theabove, it should be clear that the provision of agood rooting environment and the application ofstandard post-severance treatments are relativelyeasy and likely to result in a good measure ofsuccess. However, the commonest problemsassociated with vegetative propagation are thosecaused by inappropriate stockplant managementor the unsuitable physiological state of thematerial to be propagated. To sustain goodrooting over many cutting harvests and severalyears will require (i) much greater understandingof stockplant management in order to maintainstockplants in a good physiological condition,and (ii) a knowledge of how to manipulaterooting ability by pre-severance conditioning. Wehope that, in due course, the mechanistic model

81

will be of practical use for this task. This needto sustain physiological youth, which is oftenconfused with the retention of juvenility isperhaps the most difficult aspect of vegetativepropagation. It is important to improve ourunderstanding of these ageing phenomena(Leakey et al. 1992).

ACKNOWLEDGEMENTS

We gratefully acknowledge the work of anumber of students and the provision of fundsby the UK Overseas Development Administration,the British Council and Shell Research Ltd forcontract research and studentships, and NERC forcore funding. We also wish to thank the 11Enursery glasshouse and technical staff, especiallyMr F Harvey and Mr C Mc Beath, for theirattention to the plants and propagation facilities.

REFERENCESAndersen, A.S. 1986. Environmental influences onadventitious rooting in cuttings of non-woody species. In: Newroot formation in plants and cuttings, edited by M.B. Jackson,223-253. Dordrecht: Nijhoff.

Blakesley, D. & Chaldecott, M.A. 1993. The role ofendogenous auxin in root initiation: part II. Sensitivity andevidence from studies on transgenic plant tissues. PlantGrowth Regulation, 13, 77-84.

Blazich, F.A. 1988. Mineral nutrition and adventitious rooting.In: Adventitious root formation in cuttings, edited by T.D. Davis,B.E. Haissig & N. Sankhla, 61-69. Portland, Oregon:Dioscorides Press.

Davis, T.D. 1988. Photosynthesis during adventitious rooting.In: Adventitious root formation in cuttings, edited by TD. Davis,B.E. Haissig & N. Sankhla, 79-87. Portland, Oregon:Dioscorides Press.

Davis, T.D. & Potter, J.R. 1983. High CO2 applied tocuffings: effects on rooting and subsequent growth inornamental species. Horticultural Science, 18, 194-196.

Davis, T.D., Haissig, B.E. & Sankhla, N., eds. 1988.Adventitious root formation in cuttings. Portland, Oregon:Dioscorides Press.

Dick, J.McP. & Dewar, R. 1992. A mechanistic model ofcarbohydrate dynamics during adventitious root developmentin leafy cuttings. Annals of Botany, 70, 371-377.

Dick, J.McP. & East, K. 1992. The influence of light qualityon growth and rooting of Acacia tortzlis a semi-arid treespecies of East Africa. Nitrogen-fixing Tree Research Reports,10, 97-101.

Dick, J.McP., Blackburn, D.G. & Mc Beath, C. 1993.Stem respiration in leafy cuttings of Prosopis juliflora duringthe propagation process. New Forests. In press.

Dick, J.McR, Dewar, R. & Leakey, R.R.B. 1992. Amechanistic model of rooting for sustained mass propagationof stem cuttings. In: Mass production technology for geneticallyimproved fast growing forest tree species, vol. 1, 59-66.Nangis: AFOCEL.

Dick, J.McP., East, K. & Leakey, R.R.B. 1991. Influenceof propagation environment and cutting length on the rootingof Prosopis juliflora (Swartz) D.C. Nitrogen Fixing TreeResearch Reports, 9, 114-116.

Haissig, B.E. 1986. Metabolic processes in adventitiousrooting of cuttings. In: New root formation in plants and

. cuttings, edited by M.B. Jackson, 141-189. Dordrecht: Nijhoff.

82

Hartmann, H.T. & Kester, D.E. 983. Plant propagation:principles and practices. Englewood Cliffs, New Jersey:Prentice Hall.

Hoad, S.P. & Leakey, R.R.B. 1992. Morphological andphysiological factors induced by light quality and affectingrooting in Eucalyptus grandis. In: Mass production technologyfor genetically improved fast growing forest tree species, vol. 1,51-58. Nangis: AFOCEL.

Jones, art &Webster, C.A. 1989. Improved rooting fromconventional cuttings taken from micropropagated plants ofPyrus communis rootstocks. journal of Horticultural Science, 64,429-434.

Leakey, R.R.B. 1983. Stockplant factors affecting rootinitiation in cuttings of Triplochiton scleroxylon K. Schum., anindigenous hardwood of West Africa. Journal of HorticulturalScience, 58, 277-290.

Leakey, R.R.B. 1985. The capacity for vegetativepropagation in trees. In: Attributes of trees as crop plants,edited by M.G.R. Cannell & J.E. Jackson, 110-133. AbbotsRipton, Huntingdon: Institute of Terrestrial Ecology

Leakey, R.R.B. 1990. Nauclea didemPhii: rooting of stemcuttings, clonal variation in shoot dominance and branchplagiotropism. Trees, 4, 164-169.

Leakey, R.R.B. 1991. Low-tech cloning of tropical trees.Botanical Gardens Conservation News, 1 (9), 39-40.

Leakey, R.R.B. & Coutts, M.P. 1989. The dynamics ofrooting in Triplochiton scleroxylon cuttings: their relation toleaf area, node position, dry weight accumulation, leaf waterpotential and carbohydrate composition. Tree Physiology, 5,135-146.

Leakey, R.R.B. & Mohammed, H.R.S. 1985. Effects ofstem length on root initiation in sequential single-nodecuttings of Triplochiton scleroxylon K. Schum. Journal ofHorticultural Science, 60, 431-437.

Leakey, R.R.B. & Storeton-West, R. 1992. The rootingability of Tnplochiton scleroxylon cuttings: the interactionsbetween stockplant irradiance, light quality and nutrients.Forest Ecology and Management, 49, 133-150.

Leakey, R.R.B., Chapman, V.R. & Longman, K.A. 1982.Physiological studies for tree improvement and conservation.Some factors affecting root initiation in cuttings of Triplochitonscleroxylon K. Schum. Tbrest Ecology and Management, 4, 53-66.

Leakey, R.R.B., Dick, J.McP. & Newton, A.C. 1992.Stockplant-derived variation in rooting ability: the source ofphysiological youth. In: Mass production technology forgenetically improved fast growing forest tree species, vol. 1,171-178. Nangis: AFOCEL.

Leakey, R.R.B., Mesén, J.F., Tchoundjeu, Z., Longman,K.A., Dick, J.McP., Newton, A., Matin, A., Grace, J.,Munro, R.C. & Muthoka, P.N. 1990. Low-technologytechniques for the vegetative propagation of tropical trees.Commonwealth Forestry Review, 69, 247-257.

Loach, K. 1988a. Water relations and adventitious rooting. In:Adventitious root formation in cuttings, edited by T.D. Davis,B.E. Haissig & N. Sankhla, 102-116. Portland, Oregon:Dioscorides Press.

Loach, K. 1988b. Controlling environmental conditions toimprove adventitious rooting. In: Adventitious root formation incuttings, edited by T D. Davis, B. E. Haissig & N. Sankhla,248-273. Portland, Oregon: Dioscorides Press.

Matin, M.A. 1989. Carbon economy during rooting ofcuttings of Nauclea diderrichii (de Wild. and Th. Dur) Merill.MPhil thesis, University of Edinburgh.

Mesén, J.F. 1993. Vegetative propagation of Central Americanhardwoods. PhD thesis, University of Edinburgh.

Mesén, J.E, Leakey R.R.B & Newton, A.C. 1992. Haciael desarollo de tecnicas de silvicultura clonal para elpequeriofinguero. El Chasqui, 28, 6-18.

Newton, A.C. & Jones, A.C. 1993a. Characterization ofmicroclimate in mist and non-mist propagation systems.Journal of Horticultural Science, 68, 421-430.

Newton, A.C. & Jones, A.C. 1993b. The water status ofleafy cuttings of four tropical tree species in mist andnon-mist propagation systems. Journal of Horticultural Science,68, 653-663.

Newton, A.C., Muthoka, P. & Dick, J.McP. 1992. Theinfluence of leaf area on the rooting physiology of leafy stemcuttings of gerrninalia spinosa Engl. Trees - Structure andFunction, 6, 210-215.

Newton, A.C., Mesén, J.F., Dick, J.McP. & Leakey,R.R.B. 1992. Low technology propagation of tropical trees:rooting physiology and its practical implications. In: Massproduction technology for genetically improved fast growingforest tree species, vol. 2, 417-424. Nangis: AFOCEL

Njoya, C.Q. 1991. The effects of storage on the optimumrooting potential of leafy stem cuttings in hardwoods: r./sr,study of Thplochiton scleroxylon K Schum. MSc thesis,University of Aberdeen.

Tchoundjen, Z. 1989. Vegetative propagation of the tropicalhardwoods Khaya ivorensis AChev and Lovoa trichilioidesHarms. PhD thesis, University of Edinburgh.

Tchoundjen, Z. & Leakey R.R.B. 1993. Vegetativepropagation of African mahogany: effects of auxin, nodeposition, leaf area and cutting length. New Fbrests. In press.

Thompson, D.G. 1992. Current state-of-the-art of rootingcuttings and a view to the future. In: Mass productiontechnology for genetically improved fast growing forest treespecies, vol. 2, 333-349. Nangis: AFOGEL.

Veirskov, B. 1988. Relations between carbohydratesandadventitious root formation. In: Adventitious root formation incuttings, edited by T.D. Davis, RE.Haissig & N. Sankhla,70-78. Fbrtland, Oregon: Dioscorides Press.

83

Selection and breeding to conserve and utilisetropical tree germplasm

R D Barnes & A J SimonsOxford Forestry Institute, Department of Plant Sciences, University of Oxford, South Parks Road,Oxford OX1 3RB, UK

ABSTRACTThe breeding of non-industrial trees is a relatively new science involving many species and multiple traits, sites and managementpractices. Most tree improvement programrnes, especially in the tropics, will therefore have to deal with large arrays ofgermplasm which cannot be managed efficiently with traditional breeding approaches. To conserve and utilise genetic resources,strategies are needed to ensure efficient sampling, avoid inbreeding depression, exploit heterosis, develop the potential of cloning,manage the large arrays of taxa and accessions, and use genotype/environment interaction in a framework that does not restrictfuture options.

Efficient sampling, proficient breeding and cumulative gains can only be achieved if there is a good understanding of naturalbreeding systems and reproductive biology yet these remain undetermined for many multipurpose trees. Estimates of inbreedingdepression have also been lacking, and this has prevented breeders florri risking high selection intensities to capture maximumgenetic gain in advanced generations. The consequences of severe inbreeding depression for industrial and non-industrial speciesare disparate, and considerable discernment is required to balance risks versus gains.

Spectacular improvements in growth rate have come from the heterosis expressed by species hybrids. These advances have beentranslated into increases in productivity of plantations through the development of operational cloning techniques. Breeding tocapitalise on these often unpredictable advances must be accomplished by adopting strategies to manage the large arrays oftaxa and accessions that hold further potential and, at the same time, avoid reliance on too few clones.

For species where recurrent selection . is operative, multiple population breeding strategies are described which assist in theutilisation and conservation of tropical tree germplasm. Analogous concepts apply to improved material of non-industrial speciesthat is released to small-scale farmers, given the likelihood that they will collect their own seed in the future, thus developinglocal landraces.

INTRODUCTION

Breeding of industrial forest trees began inearnest in the 1950s with Pinus species. Pinesare analogous to maize in that they are bothoutcrossing, wind-pollinated species in which thetraits of greatest economic importance arequantitative and under multigenic control.Therefore, it was a logical progression fortraditional strategies developed forcross-pollinating agricultural crops to be used fortheir genetic improvement, albeit for species withlargely inflated generation intervals andcontrasting population structures. Through astrategy of recurrent selection for generalcombining ability (gca) and seed production fromprogeny-tested clonal orchards, average gains ofaround 20% in volume production alone wereaccomplished in the first generation of selectionin many programmes.

The potential to increase productivity ofplantation trees by selection among naturalpopulations was also firmly realised in the 1950s.A profusion of provenance trials has subsequentlybeen established which demonstrated for anumber of species that improved adaptabilityand even greater increases in productivity may

84

be available from reselection in the natural range(Mullin, Barnes & Prevost 1978). Thisdevelopment has resulted in a crucial need forstrategies to manage a plethora of potentiallyvaluable material and to retain the option toinclude that material in the initial breedingpopulations.

The improvement of non-industrial trees is anewer science involving many more species (seeSimons, Mac Queen & Stewart, pp91-102). Thetaxonomy distribution, genetic variation andreproductive biology are unknown for many ofthem. The demands of beneficiaries ofimprovement programmes of non-industrial treeswill be far more diverse than those of industrialspecies in terms of the traits of interest, the sitesfor planting and the flexibility of management.

Most tropical tree improvement programmes, bothfor industrial and for non-industrial tree species,will therefore have to deal with large arrays ofgermplasm of several species. This is quite adifferent situation from agricultural crops, whereone breeder or a single organisation usuallydeals with better characterised material of farfewer species. It has become clear that, in manycases, forest tree germplasm cannot be managed

efficiently with traditional breeding approaches toproduce the genetically improved materialrequired by a wide range of clients. To conserveand utilise this resource, strategies are needed toensure efficient sampling, avoid inbreedingdepression, exploit heterosis, develop thepotential of cloning and use genotype/environment interaction in a framework that doesnot res ict future options.

SAMPLING

Experience with industrial trees has shown thatthere can be economically significant intraspecificpopulation differences in the natural range of aspecies. In some cases, this knowledge hascome from provenance trials planted long afterthe breeding programme has started. Forexample, Pinus taeda was introduced intosouthern Africa at the beginning of the centuryprobably from a Georgia Piedmont source, andbreeding of the resulting landrace has producedsubstantial increases in productivity Rangewideprovenance trials, however, have subsequentlyshown that some of the southernmostprovenances from Rorida have the ability tooutproduce the bred material in volume, althoughstem form is unacceptable (Mullin et al. 1978).International provenance trials of P caribaea var.hondurensis have revealed the Guanaja Islandpopulation not only to be among the highestvolume producers, but also to have a wooddensity approximately 10% higher than themainland provenances (Barnes, Gibson & Bardey1983). In addition, these trials have demonstratedthe superior wind-firmness of the coastal sources(Birks & Barnes 1990).

International provenance trials also identified asubspecies of P patula, previously unrecognisedand untested, that will have a profound effect onthe productivity and exotic range of that speciesin the tropics and subtropics (Barnes & Styles1983). Moreover, in P patula, provenance trials insouthern and eastern Africa have shown theZimbabwean landrace to have superiority overnative provenances in its tolerance of the pinewoolly aphid (Pineus pint) after the latter'sinadvertent introduction from Australia (Barnes etal. 1976). Such provenance differences, however,are not restricted to the pines. For example,Eucalyptus camaldulensis was used for nearly 100years in southern Africa before provenance trials'showed the great superiority in productivity stemform and drought tolerance of some of thehitherto untested natural populations fromnorthern Australia, such as Petford and Katherine(Barrett & Carter 1976).

These examples highlight the need to ensurethat provenances over the full natural and exoticrange of a species are effectively sampled beforethe breeding populations are closed by the gapcreated by a number of generations of selection.

They show that a shategy is required to maintainand develop new material to a point where itcan contribute its potential to the propagules foroperational use, and that variation should beconserved in the breeding population to meetunpredictable events.

Trees for non-industrial use will often includespecies that are not naturally gregarious. Theyare likely to have much more variable populationstructures than the traditional plantation species,which normally occur naturally as dense stands,and this variability will affect sampling,domestication, breeding and conservationstiategies profoundly Logically, therefore, classicbreeding should not be started before thenatural breeding systems are properlyunderstood. For instance, several legumesdisperse their pollen in aggregates of 8, 16 oreven 32 monads; thus, seed from individual podsare likely to be full-sibs (eg Calliandracalothyrsus).

If it is accepted that for most non-industrial treespecies there will be little advanced generationbreeding and that farmers will ultimately derivetheir seed requirements from their own orneighbours' fields (see Simons et al., pp91-102),then the correct choice of base populationbecomes even more critical. Consider thescenario where population A produces a yield of130 units relative to the mean of all otherpopulations (100), and is chosen for multiplicationand release. When it is grown in a seedproduction area with moderate selection, theseed produced will feasibly have the potential of145 units of production. When released, thismaterial would spread throughout ruralcommunities, largely by fanner-to-farmer diffusionmechanisms (Cromwell 1990). Meanwhile, furtherexploration of the native range reveals a newpopulation (B) with a 170 unit production. Even ifpopulation A underwent recurrent selection, itwould take three or four generations to attain theproduction of population B at the expense of anarrowing of the genetic base. Depending uponthe rate of uptake of population A (in terms ofnumbers of fanners and proportion of their landcultivated), and the saturation level for thespecies in an area, the release of population B(even immediately in an unimproved state) mayhave little effect. This situation would occur evenif farmers later collected seed from trees ofpopulation B only because of contamination bypollen of population A. Furthermore, there mayeven be reduced yields, owing to the negativeeffects of creating population hybrids.

INBREEDING

Inbreeding in cross-pollinated species almostinvariably leads to a loss of vigour and fecundityAs most of the industrial tree species undergenetic improvement are outcrossing and single

85

undifferentiated breeding populations have beenthe norm, breeders have been concerned tokeep co-ancestry to low levels, not only in theseed orchards but also in the breedingpopulations themselves. This fact may haveprevented the full potential genetic gain beingpassed on to operational clients; the gainachieved through higher selection differentials inthe early generations is unlikely to have beennullified by inbreeding depression. The materialsand mating designs to estimate inbreedingeffects independently from gca are readilyavailable in many programmes, but, surprisinglythe experimental work has rarely been done.

Selfmg is the most extreme form of inbreeding.There are advantages in using selling both totest and to recombine genotypes. It conservesthe effective population size, does not increaseco-ancestry does not dilute valuable genecomplexes, can be used for recombination, and,theoretically can rank parents precisely (Lindgren1991). The problems are that it is common forself-fertility seed yield and seedling survival tobe low when outcrossing industrial tree speciesare inbred. For these reasons, selfmg has rarelybeen used by tree breeders, although there areinstances of self-fertility and outstandingperformance of selfs and of the latter beingpositively correlated with gca, eg with Pinuselliottii in Zimbabwe (Mullin et al. 1978).

There is clearly a need for tree breeders ofindustrial species both to be released from thefear of the consequences of inbreedingdepression, so that they can be moreadventurous in using high selection differentials,and to explore the possibility of developinginbred lines through selling and establishinghomozygous balance in subpopulations, so thatthey can enhance the prospects of achievinglarge heterotic effects.

HETEROSIS

Heterosis at the species hybrid level has beenresponsible for the greatest genetic increases in theproductivity of plantations. They have been mostspectacular among the Eucalyptus. Most hybridshave initially been produced by chance when plotsof two species whose natural distributions do notoverlap have been grown close to each other asexotics, eg E. grandis and E. urophylla in Brazil(Campinhos & aemori 1989). An added condition isalmost invariably that the hybrid exhibits heterosisonly in a hybrid habitat' (ie in an environmentwhere neither of the parent species is generally welladapted), eg E. grandis x E. tereticomis in theCongo (Delwaulle, Laplace & Quillet 1980), orwhere the hybrid tolerates a single factor in theenvironment that is limiting for both parent species,eg cyclone in the case of P caribaea var.hondurensis x P patula ssp. tecunumanil (Nikles1989; Nikles & Robinson 1989).

86

The potential to increase productivity with otherspecies hybrids is enormous in Eucalyptus (eg tointroduce cold tolerance into some of thefast-growing tropical species), Pinus (eg withinthe closed-cone pine group to extend the rangeat present covered by P patula), and in manyother genera, particularly those of somenon-industrial trees such as Leucaena and Acacia.There is evidence that those individuals withhighest gca within species combine to give thebest hybrid families. There is, therefore, anadvantage in selecting to improve the breedingpopulation genotypes of potential hybrid parentspecies. One current constraint to hybridbreeding is the labour-intensive controlledcrossing required, although, with recent advancesin identifying male sterility genes which functionin heterologous species, there exists the potentialto render one of the parental species malesterile, and thus form an ideal seed parent fornatural hybrid crosses.

The potential of interspecific hybrids fornon-industrial species is compromised by thelikely collection of seed by farmers for their ownuse, and the need for broadly based plantingstock. Hybrids are being promoted in the genusLeucaena to combine growth vigour of L.leucocephala with the acid tolerance or insectresistance of L. diversifolia (see Brewbaker &Sorensson, pp195-204). To fix gene frequenciesfor desirable traits and to prevent segregation inlater generations, breeding populations arehaving to be taken to the F4 and beyond. Thehigh selection intensities being employed,however, are narrowing genetic diversity in thishybrid, thus making the material more vulnerableto environmental disasters and perhaps restrictingits management and environmental flexibilityWithout complementary mating tests todetermine genetic correlations, there alsoremains the possibility that other characters arebeing selected against. The occurrence of steriletriploid interspecific hybrids in Leucaenacircumvents problems of segregation or fixation,but would tie the resource-poor farmer toreturning to the releasing agency each time fornew seed.

Although the value of interspecific hybrids hasbeen confirmed for many species, there areconflicting results on the performance ofinter-provenance hybrids'. In Pinus radiata, apopulation composited from two provenancesoutperformed the parental provenances by 20%in volume at seven years of age (Shelbourne etal. 1979). Several authors conclude that hybridprovenances are usually intermediate betweenparents, as found in Picea abies (Huhn & Muhs1981), Pinus sylvestris (Parks & Gerhold 1986)and Pinus banksiana (Magnussen & Yeatman1988). Such results have led workers to considerhybrid provenances as a way of combining

desirable traits uniquely present in only one ofthe parental provenances. Ericsson (1975) relatedthe character-specific nature of heterosis in Piceaabies, where positive effects were seen forvolume, yet negative effects were seen for insectresistance.

Whether or not it is possible to predict aheterotic response when combining two or moreprovenances, the compositing of provenances ofnon-industrial species may be inevitable, giventhe multiplicity of organisations involved in thetesting, promotion and release of these species.Compositing of provenances could arise becausefarmers will collect future seed from trees intheir own or their neighbours' fields. Thequestion arises as to how much provenancecompositing should be actively discouraged, or,alternatively, how much it should be promoted.Despite the label inultipurpose', mostnon-industrial trees are grown for a singlepurpose, and the additional products or benefitsare viewed as bonuses. Therefore, awood-producing provenance may not becompatible to composite with a leaf-producingprovenance, as the benefits of either may becompromised. Alternatively it may be possible tocombine characteristics such as tolerance of soilacidity with high growth rates by mixingprovenances with these attributes.

When improved germplasm is to be releasedwhere the material already occurs naturallyfurther thought must be given as to whether theimprovements will survive many generationsbefore the mixing with wild germplasm makesthe improvements redundant. One such case iswith Sesbania sesban, where the InternationalCentre for Research in Agroforestry (ICRAF') isplanning to test and release new provenances inAfrica (D Boland, ICRAF, Kenya, personalcommunication; see also Owino, Oduol & Esegu,pp205-209). The success of these efforts will relyupon being able to monitor the dilution ofimproved material. For instance, it may be thatfarmers can carry the improved germplasmthrough three or four generations before needingto resort back to pure stocks. A current study toinvestigate the breeding system and floralphenology is underway to resolve these issues.For species where vegetative propagation is apossibility such concerns do not arise.

CLONING

The huge increases in plantation productivityoffered by species hybrids have been madeoperationally possible through the developmentof new techniques of vegetative propagation bycuttings both in pines and in eucalypts. In fact, ithas been the development of this facility forspecies that were hitherto considered not to beclonable that has been the significanttechnological breakthrough, not breeding. Many

hybrid clones in use, particularly among theeuca/ypts, remain those that have beenproduced, often by chance, between unselectedparents. Breeding for yield within the parent -species populations should bring with it thepotential for cumulative increases in yield fromthe hybrids. There are, however, indications thatthe correlations between yield of the ortet andthat of its ramets are poor (eg in Eucalyptusgrandis, P A Clegg, SAPPI Research, South Africa,personal communication). Is it possible that yieldin the ramets may be more dependent on thegenetic capacity of the individual to produce anefficient root architecture in its clone? Better•correlations may be found between yield andsOme rooting trait,or between juvenile andmature characteristics (see Ladipo et al.,pp239-248). If such correlations are found, thensome form of reciprocal recurrent selection willoffer the prospect of cumulative genetic gains inyields from hybrids in the future. li suchcorrelations are not found, identifying a sufficientnumber of good clones will rely upon thecontinuous screening of large numbers, gains willnot be cumulative, and a point may be reachedwhere it is no longer profitable to look forfurther increases in yield.

Foresters continue to debate the wisdom ofreleasing few versus many clones to captureoptimal gains for industrial tree species. Thearguments about susceptibility to environmental(biotic or abiotic) disasters with few clones applymore rigorously to non-industrial trees, given therisk averseness of small-scale farmers and theirdesire to maximise stability of production ratherthan production per se. The consequences ofclonal effects on number of cuttings producedper year may be a climax clonal populationbased on a few clones. In Gliricidia sepium,number of stems is under strong genetic controland, after coppicing, there is a high correlationwith number of resprouts. Similarly the numberof resprouts in each successive coppice is highlycorrelated with original stem numbers. Therefore,after ten generations of vegetative propagation ofclones which produced two, three, four and fiveresprouts per generation, the population wouldconsist of 1024, 59049, 1 048 576 and 9 765 625cuttings per clone, respectively. These figurescorrespond to 0.01%, 0.5%, 10% and 89%representation of the population, respectively Tomitigate against these effects, it would benecessary to release hundreds or thousands ofclones rather than the tens being released forindustrial trees.

Cloning is and will continue to be used fornon-industrial trees for a great variety of reasons.Foremost is the ability to capture and fixdesirable traits. Sexual reproduction may not befavoured on genetic grounds because:

87

• the traits are difficult to breed for (lowheritability);

• there is a negative association betweendesirable traits (negative genetic correlation);

• the traits are not controlled by additivegenes (non-additive).

In addition, seed propagation may bedisadvantageous for biological reasons, such aspoor seed set (eg Calliandra calothyrsus inCameroon), recalcitrant seed (eg Simaroubaglauca) or difficulties with seedling establishment.

There may also be other consequences of usingvegetative propagation. For instance, cuttings mayhave a different root architecture and competewith crops differently (Liyanage & Jayasundera1988). For Gliricidia sepium in West Africa,hedgerows derived from cuttings lastapproximately six years, in contrast to ten yearsfor those derived from seedlings (J Cobbinah,International Institute of Tropical Agriculture,Ibadan, Nigeria, personal communication).

GENOTYPE/ENVIRONMENT INTERACTION

Genotype/environment interaction (gel) can occurat the species, subspecies, variety provenance,family and clonal levels. It can be caused byclimatic, edaphic, biotic and managementvariables in the growing environment. Some can,and some cannot, be controlled, and some maynot even be predictable. Those that arepredictable may interact with those that areunpredictable, thus rendering inconclusive theresults of experiments established to test theeffects of predictable factors. Climatic factorstend to be responsible for most interactionshigher up in the species-to-clone hierarchyphysical and biotic factors in the middle, andedaphic factors at the lowest levels. Theseinteractions have been clearly illustrated byplantation species (Matheson & Cotterill 1990).

Among the tropical and subtropical pine species,mean, minimum and maximum temperatures arethe most significant factors in the environmentgoverning a species' performance. For example,Pinus patula is at its optimum at a mean annualtemperature of about 14°C, P taeda at about16°C, P elliottii at about 18°C, and P caribaea atabout 22°C. As temperatures get too high for Ppatula, productivity falls off progressively untilmale and female flowering becomesasynchronous and no seed is produced. Incontrast, as it gets too cold for P elliottii and Pcaribaea, wood density becomes unacceptablylow, and no pollen or no seed are produced(Barnes & Mullin 1976). Within climatic zones,there may or may not be cyclonic winds. In Pcaribaea, there are very distinct provenancedifferences in resistance to wind damage; thecoastal populations are much more wind-firmthan the inland populations (Eisemann, Nikles &

88

Newton 1984). At the family level, heritability ofresistance to aphid (eg Pineus pint) and rust (egCronartium fusiformis) are high in P elliottii; if thedisease is present in one site and the insect inanother, a complete reversal in the ranldng offamilies can occur. At the individual tree level, itis laiown that certain clones of P radiata rank atthe top where there are high phosphorus levelsin the soil, but rank lowest where there is adeficiency of this element (Wilcox 1982).

Clear evidence of gei at all levels has led treeresearchers and breeders to expect it to haveimportant consequences for both conservationand genetic improvement. Many provenance,progeny and, more recently clonal trials havebeen replicated over a range of sites in theexpectation of observing gei and being able toexplain and predict it through statistical analysisto establish the causal factors in the environment.This has not happened, and there seems to belittle prospect even now of being able to gainmore than empirical indications of interaction,except where there is one over-riding effect inthe environment, eg frost, cyclone, disease,predators or elemental deficiency The failure todo more is largely attributable to the difficulty ofmeasuring the variables required to characterisean environment and the inherent micrositevariability Climatic data are the easiest to record,yet at many trial sites it is necessary tointerpolate statistics from distant meteorologicalstations with attendant inaccuracies. Soils arenotoriously difficult and expensive to characterise,especially in the case of forestry experimentswhich tend to cover large and variable areas.Most biotic effects are by nature unpredictable;however, strildng examples of the effects ofmicrosymbionts on performance have beenrecorded (see papers by Mason & Wilson,pp165-175; Lapeyrie & Hogberg, pp158-164;Sprent, pp176-182). The interaction of all thesefactors with management, which inevitably variesfrom site to site, further confounds the effects.

Replicated progeny tests within a species' rangehave regularly shown, for many industrial trees,that gei for productivity at the family level ispresent. The interaction variance is usuallycaused by a few families that performoutstandingly well at a specific site. There areusually however, also some families that aregood overall performers. Because of theempirical nature of these results, the expense oftesting over a large number of sites, and theimpracticality of running a large number ofbreeding programmes for different site types,breeders have tended to select the goodall-rounders and confine themselves to a singlebreeding programme to cover all sites. There isno doubt that potential to increase productivityand maintain variation is forfeited with thisstrategy both because adding Stability to the list

of selection criteria will reduce gain for othertraits, and because the specific high-performersare lost. Further, wide testing to identify thegood all-rounders is costiy, if it is to be effective.

The situation of gei with respect to non-industrialtrees is important and relates large/y toperformance. Following testing of provenances orclones at experimental sites, material will bereleased to resource-poor farmers who willmanage it in a wide range of ways and physicalenvironments. The crucial issue is one ofreliability of performance, as farmers seek tomaximise stability of production rather thanproduction per se. Therefore, provenances orclones should be sought that show stability ofperformance across differing site types ormanagement regimes, rather than tailoringspecific provenances or clones to restrictedgeographic areas.

A STRATEGY

A breeding strategy has been developed at theOxford Forestry Institute which accommodatesmost of the above requirements (Namkoong,Barnes & Burley 1980; Barnes 1986). It has beencalled the Multiple Population Breeding Strategy(IV1PBS). In it, the breeding population for asingle species consists of a variable number ofsubpopulations, which are kept separate so as toproduce trees with different gene complexes. Theobjective is to maintain or create diversitybetween them, while practising various intensitiesof selection within them. Diversity will occurbetween the subpopulations even if the selectioncriteria, the selection pressure and the breedingenvironment are the same, because most traits ofinterest are under gca control and different setswill be brought together in different populations.Currently neutral alleles will not be lost in thesame frequencies. Variation can be reinstated,inbreeding depression overcome, and newpopulations with specific attributes created bycrossing between the subpopulations. Undesirableor superfluous populations can be dropped atany time.

There are three types of multiple population inthis breeding strategy:

i. replicate populations, where independentsubpopulations are bred for the same trait -and elements from each are broughttogether for operational seed production toensure outcrossing;

diversified populations, where existingdifferences are used or new differencescreated by selecting for different traits and/orenvironments, and commercial seedproduction occurs within subpopulations forspecial products and/or environments;

heterotic populations, where both additive andheterotic gene effects are used through

hybrid or reciprocal recurrent selection inthe different subpopulations, and heterosis isexploited in operational seed production bycrossing between specific subpopulations.

Experience with industrial plantation species hasshown that there are subspecific differences atthe provenance (natural and exotic) level thathave-a lasting operational significance.Knowledge of these differences and the materialsto represent them rarely become accessibletogether; in fact, both characteristically becomeavailable over an extended period of time. Withnon-industrial species, the situation iscomplicated further because domestication maystart before there is sufficient knowledge onreproductive biology to permit an informedsampling of the genetic range of the species.The IVIPBS provides pedigree structure at thepopulation level that permits the managementand conservation of this variation.

By providing structure at the population level, theMPBS ensures a backstop of pedigree controland escape routes from inbreeding depression,which gives the breeder confidence to be moreadventurous and pursue potentially fertileavenues of research, including those into theproduction of inbred lines or high-performanceclones, in order to increase genetic gain.

The MPBS is also a suitable framework in whichto breed and conserve parent populations forcreating species hybrids and high-performanceclones. The ways in which traits can becombined and heterosis used are notpredictable. Selection within carefully defmedprovenance and species breeding populationswill preserve the potential to make successfulhybrids for possible subsequent cloning, asneeds unfold.

There is an argument for using the MPBS toretain the potential to use gei and, at the sametime, to benefit from it by default. Rather thantesting the same population over a large numberof sites, replicate populations can be establishedin very different environments with different setsof families in each. When the best families areselected, both the best all-rounders and the bestperformers in specific environments can besaved. The average operational performance ofprogeny of the combined top performers is notlikely to be below that of the good all-roundersalone. If routine seed source tests showed thatone particular breeding population were doingbetter than seed from the combined seedproduction orchard, this would be an indicationthat a separate seed orchard should beestablished for that area, thus taking advantageof any substantial gei. An additional advantage ofthis strategy over the conventional singlebreeding population is that the selectiondifferential for the seed production populationcould be much higher because of the larger

89

number of families constantly kept under test. Abreeding population with a much broadergenetic base could be maintained for theequivalent cost of a conventional programme,and would serve as a buffer againstunpredictable environmental events.

Finally the MPBS is seen as an effectiveframework in which to conserve geneticvariation, not only as a safeguard againstinbreeding and as a source of variation forbreeding, but also to conserve genes for anuncertain future in which environmental changeand material demands are unkriown.

REFERENCESBarnes, R.D. 1986. Multiple population tree breeding inZimbabwe. In: IUFRO Conference - Joint Meeting of WorkingPartiés on Breeding Theory, Progeny 7bsting, Seed Orchards,edited by AV Hatcher & R.J. Weir, 285-297. Raleigh, NorthCarolina: Tree Improvement Cooperative.

Barnes, R.D. & Mullin, L.J. 1976. Selection of forest treespecies in Rhodesia. South African Forestry Journal, 98, 16-20.

Barnes, R.D. & Styles, B.T. 1983. The closed-cone pines ofMexico and Central America. Commonwealth Forestry Review62, 81-84.

Barnes, R.D., Gibson, G.L. & Bardey, M.A. 1983.Variation and genotype-environment interaction in internationalprovenance trials of Pinus caribaea and implications forpopulation improvement strategies. Silvicultura, 32, 761-764.

Barnes, R.D., Jarvis, R.E, Schweppehauser, M.A. &Mullin, L.J. 1976. Introduction, spread and control of thepine woolly aphid, Pineus pini (L.), in Rhodesia. South AfricanForestry Journal, 96, 1-11.

Barrett, R.L. & Carter, D.T. 1976. Eucalyptuscamaldulensis provenance trials in Rhodesia. (The RhodesiaBulletin of Forestry Research, no. 2, part 2.) Rhodesia ForestryCommission.

Birks, J.S. & Barnes, R.D. 1990. Provenance variation inPinus caribaea, P oocarpa and P patula ssp. tecunumanii.(Tropical Forestry Papers, no. 21.) Oxford: Oxford ForestryInstitute.

Campinhos, E. & Ikemori, Y.K. 1989. Selection andmanagement of the basic population Eucalyptus grandis andE. urophylla established at Aracruz. In: Breeding tropical trees,edited by G.L. Gibson, A.R. Griffen & AC. Matheson,169-175. Oxford: Oxford Forestry Institute/Arlington, Virginia:Winrock International.

Cromwell, E. 1990. Seed diffirsion mechanisms in smallfarmer communities. (ODI Network Paper, no. 21.) London:Overseas Development Institute.

Delwaulle, J.C., Laplace, Y. & Quillet, G. 1980.Production massive de boutures d'Eucalyptus en RepubliquePopulaire de Congo. Silvicultura, 8, 779-781.

90

Eiseirtann, R.L., Nikles, D.G. & Newton, R.S. 1984. Theprovenance question in Queensland planting of Pinus caribaeavar. hondurensis. In: Provenance and genetic improvement intropical forest trees, edited by RD. Barnes & G.L. Gibson,211-213. Oxford: Commonwealth Forestry Institute/Harare:Forest Research Centre.

Ericsson, T. 1975. Results from some trials in northernSweden with provenance hybrids of Picea abies. Arsbok, 1974,114-159.

Huhn, M. & Muhs, H.J. 1981. Height growth of someinter-racial hybrids of Norway spruce using selected trees ofa northern and central European provenance with specialregard to between-plot competition. Silvae Genetica, 30, 25-29.

Lindgren, D. 1991. Progeny testing. In: Genetics of Scotspine, edited by M. Giertych, 193. Oxford: Elsevier.

Liyanage, L.V.I{. & Jayasundera, H.P.S. 1988. GlinCidia asa multipurpose tree for coconut plantations. Coconut Bulletin,5, 1-4.

Magnussen, S. & Yeatman, C.W. 1988. Provenance hybridsin jack pine: 15-year results in eastern Canada. SilvaeGenetica, 37, 206-218.

Matheson, A.C. & Cotterill, P.P. 1990. Utility of genotypex environment interactions. Forest Ecology and Management,30, 159-174.

Mullin, L.J., Barnes, R.D. & Prevost, M.J. 1978. A reviewof the southern pines in Rhodesia. (The Rhodesia Bulletin ofForestry Research, no. 7.) Rhodesia Forestry Commission

Namkoong, G., Barnes, R.D. & Burley, J. 1980. Aphilosophy of breeding strategy for tropical forest trees.(Tropical Forestry Paper, no. 16.) Oxford: CommonwealthForestry Institute.

Nikles, D.G. 1989. Early growth and potential value of Pcaribaea var. hondurensis x P tecunumanii x P oocarpa Flhybrids in Australia. In: Breeding tropical trees, edited by G.L.Gibson, AR. Griffen & AC. Matheson, 412-414. Oxford:Oxford Forestry Institute/Arlington, Virginia: WiruockInternational.

Nikles, D.G. & Robinson, M.A. 1989. The development ofPinus hybrids for operational use in Queensland. In: Breedingtropical trees, edited by G.L. Gibson, AR. Griffen & AC.Matheson, 272-282. Oxford: Oxford Forestry Institute/Arlington,Vhginia. Winrock International.

Parks, Y.S. & Gerhold, H.D. 1986. Population hybridizationin Scots pine. Silvae Genetica, 35, 159-165.

Shelbourne, C.J.A., Burdon, R.D., Bannister, M.H. &Thulin, I.J. 1979. Choosing the best provenances of radiatapine for different sites in New Zealand. New Zealand Journalof Forestry 24, 288-300.

Simons, A.J. 1992. Genetic improvement of non-industrialtrees. Agroforestry ,Systems, 18, 197-212.

Wilcox, M.D. 1982. Genotype x fertilizer interaction study(Paper presented at Meeting of IUFRO Working Party onBreeding Radiata Pine, Rotorua, New Zealand, 1982.)(Unpublished.)

Strategic concepts in the domestication ofnon-industrial trees

A J Simons, D J Mac Queen & J L StewartOxford Forestry Institute, Department of Plant Sciences, University of Oxford, South Parks Road,

Oxford OXI 3RB, UK

ABSTRACT

The planting of non-industrial trees has been widely promoted in the tropics to counter the effects of fuelwood shortages, to

increase sustainability of agriculture, and to prevent land degradation. 'Ilvo key components to the success of these endeavours

have been identified as (i) the particular speciesused, and (ii) the source of seed, as these may enhance productivity

independently of other inputs. Given that most non-industrial tree species have not undergone organised breeding, there exists a

potential for genetic improvement through selection and recombination of desirable individuals from selected seed sources. There

is likely to be some disparity however, between what is technically feasible and what is practically appropriate. In the absence of

theory and strategies for breeding non-industrial trees, an adoption of agricultural and commercial forest methodology seems

probable, although the case for advanced generation breeding and cumulative gain through recurrent selection in non-industrial

tres is equivocal, and in many cases illusory In addition, little consideration has been given to the demand-side variables and

the distribution mechanisms for improved seed of non-industrial trees. This paper examines the potential and scope for

improvement of non-industrial trees, and the likely demand for genetically improved trees.

INTRODUCTION

The importance of non-industrial trees in thefarming systems of rural communities in thetropics has recently been elevated by humanpopulation increases, accelerated deforestationand the lack of land for agricultural expansion.One reason for the popularity of such trees isthe diverse spectrum of products (eg fuelwood,fodder, green manure, food) and benefits (egshade, erosion control, soil amelioration) that theyprovide. Although the growing of non-industrial(or 'multipurpose') trees as part of an integratedagricultural system is not a new innovation, itsgreater frequency of occurrence is a recentphenomenon. Palmberg (1989) reported that 2.6million hectares (Mha) in the tropics wereplanted with non-industrial tree species duringthe period 1981-85, almost matching the areaplanted with industrial species (2.9 Mha) duringthe same period. Unlike commercial forestrywhere a few species dominate plantings, morethan 2000 tree species have been planted fornon-industrial use (Burley Hughes & Styles 1986).Consequently individual non-industrial treespecies may not warrant such intensive breedingeffort as their industrial counterparts.

With a few exceptions (see Wil lan, Hughes &Lauridsen 1990), there has been no organisedbreeding of non-industrial trees, suggesting anuntapped potential for genetic selection andimprovement (Simons 1992). In the absence oftheory and strategies specifically for the

breeding of non-industrial trees, an adoption ofagricultural and commercial forestry methodologyappears likely (eg Ventakesh 1988). Thefeasibility and appropriateness of suchmethodology however, remain untested, and inmany cases it may be unsuitable.

STATUS OF NON-INDUSTRIAL TREES

The economic status of non-industrial trees isinherently low Nonetheless, they may contributeto the income of small- and medium-scalefarmers through enhanced crop/livestockproduction, or may occasionally be solddirectly (eg for firewood or construction),although markets for the latter are often notwell established. The greatest benefits fromcultivating non-industrial trees are likely toaccrue outside the cash economy, so efforts toincrease their production should be consideredprimarily as rural development rather than aneconomic return on investment. Given thissituation, questions arise as to who will fundthe breeding of non-industrial trees andwhether this will be a sustained activity Therecent emergence of two new ConsultativeGroups on International Agricultural Research(CG1AR) — the Centre for International ForestryResearch (CIFOR) and the International Centrefor Research in Agroforestry (ICRAF) — to focuson tropical trees suggests that a long-termcommitment to non-industrial tree improvementmay develop.

91

An often quoted output from the GreenRevolution, in which the CGIAR agriculturalcentres were prominent, was the breeding ofhigh-yielding crop varieties, albeit with relianceon chemical fertilizers and irrigation. Thisadvance, however, was allied to major outbreaksof pests and diseases leading to disastrous cropfailures (Cooper, Vellve & Hobbelink 1992).Agricultural history includes many examples ofthe imprudence of over-narrowing the geneticbase of planting stock (eg the susceptibility tosouthern leaf blight of Texas male-sterilecytoplasm in maize) (Simmonds 1979). Morerecently examples of attacks by pathogens ongenetically uniform trees have emerged, such asrust diseases of Populus and Citrus. Pestepidemics are of equal importance, and thedevastation of large areas of Leucaenaleucocephala by a defoliating psyllid highlightsthe need to ensure the genetic diversity ofplanting material of non-industrial trees (Hughes1989). It may be that outbreaks of insects onearly introductions of other non-industrial trees(eg mealybug attack on Gliricidia sepium inUganda in 1914 - Tothill 1940; stem-borer onCalliandra calothyrsus in Philippines - Luego1989) were a result of narrowly based germplasm.

A broad genetic base in non-industrial trees isrequired not only for biological insurance, butalso because of the vagaries of the physicalenvironment where they are grown. Furthermore,farmers who grow these trees operate with lowinputs, because of insignificant cash reserves andminimal capacity to withstand high losses(Haugerud & Collinson 1990). Such farmers seekto realise stability of production, rather than tomaximise production per se, which incursunacceptably greater risk. Consequentlyimproved material of non-industrial trees shouldretain a broad genetic base to cope with currentand unforeseen production and environmentalvariables.

DEMAND FOR NON-INDUSTRIAL TREES

It is evident that natural tropical forests aredisappearing and that land for agriculturalexpansion is often restricted, indicating in generalterms a need to plant trees. The demand fortrees from people in rural communities, however,remains largely unquantified. Information on thedemand for individual species, many of whichare largely unknown, is also scant. Yet it is thisdemand for planting stock of individual speciesthat should drive tree improvement.

'raking an example of demand for an agriculturalcrop, Griliches (1957) fitted a logistic model tothe origin (a), slope (b) and ceiling level (c) forcultivation of hybrid maize when it wasintroduced in the USA. Despite the cleartechnical advantage of hybrids, there were

92

marked differences between regions in theirrates of acceptance (b=0.35-1.02) and ceilinglevels (c=53-100%) of this new technology Thesedifferences were explained partly bydissimilarities in farm size, isolation of farms andfarmer wealth. The point to draw is that even ina relatively resource-rich environment, such asthe USA farming community demand andacceptability varied. For non-industrial trees, it isreasonable to expect even greater differencesbetween different rural communities in theiracceptance and uptake of germplasm for whatmay not only be a new species but also awhole new technology (eg growing trees inalleys between crops).

Clearly without economic models of the processof adapting and distributing non-industrial treegermplasm, it is difficult to know how much seedto produce or, indeed, by which strategy itshould be produced. Figure 1 represents ageneralised, hypothetical case for the introductionof a new tree species into an exotic location.Not all farmers would take up the new species(Figure li), and farmers who did would not plantall of their available land with it (Figure lii).There may in fact, be a peak acceptance, with astabilised ceiling level below this peak owing toalternatives becoming available or todissatisfaction with over-promoted material.Knowledge of the initial uptake, the rate of

co

a

a

Time -

Time -

c

c

FVure 1. Generalised hypothetical case for uptake of treeplanting in terms of (i) number of farmers and (ii) area ofland (see text for further explanation)

uptake and the ceiling level would greatly assistin the formulation of breeding strategies andmechanisms for release. The way in which thetimescale along the x axis in Figure 1 relates tothe generation interval for that species is alsoimportant. If most farmers have taken up plantingthe tree (on whatever proportion of land theydeem fitting) before one generation interval haselapsed, then the scope for introducing the nextgeneration of material is limited to replacements.The significance of replacements will be high forshort-lived species such as Sesbania sesban, butfor most non-industrial trees, which arelonger-lived, it will be minimal.

To convince smallholders of the merits ofplanting material from an advanced generation,the more genetically advanced material wouldhave to be demonstrably superior. Forquantitative characters such as wood yield, asmall increase (say 10%) would be difficult forsmall farmers to perceive, particularly formultistemmed species. This does not mean that amodest gain is not important, but rather that itmay, be difficult to implement. Conversely forqualitative characters such as tree form or insectresistance, the superiority of improved materialwould be readily observable, thus increasing itslikely acceptance.

The qualities of a tree species are likely to beas important as the amount of materialdemanded. For crops, it has been found thatfarmers have different cultivar preferences,depending upon farm size, family structure,gender, wealth and market opportunities(Haugerud & Collinson 1990). Regionaldifferences will also prevail, such as with thespecies Moringa oleifera, which is used for oil inCentral America and leaf in Indonesia (Jahn1989). Differences can also be seen in traitpreferences in one region, such as the case withCalliandra calothyrsus in Flores, Indonesia, wherecertain farmers want single-stemmed aggressivetrees for posts, whereas neighbours desire highleaf biomass trees, which display minimal rootcompetition with crops. Therefore, if preferencesare ignored, trees may be bred which farmersdo not want.

SEED ORIGIN

Exotic species are being used in many caseswhere non-industrial trees are being promoted.For species where no naturalised populations orlandraCes already exist, smallholders must derivetheir material from farmer co-operatives,government extensionists, non-governmentalorganisations (NG0s) or donor agencies(collectively referred to here as 'seeddistributors'). In these cases, however, after theinitial planting material has fruited, farmers havethe option of collecting seed from their owntrees or from those of a neighbour. For

Calhandra calothyrsus in Java, farmers leave onetree every 10-20 m of hedgerow to act as seedtrees. Therefore, as time progresses, theproportion of seed derived from seed distributorswill decrease as that from local sourcesincreases (see Figure 2). The relative proportionsof the two sources will vary according to region,and perhaps species. The way in which seedorigin relates to fanner uptake is also important(cf Figures 1 & 2). One consequence of themajority of seed being derived from a farmer'sown fields is that other farmers who do plantimproved seed from the seed distributors will

as

600.

100

80

•

40a)

20

0

Generationinterval

•• •

• - - -Govt/NGO/donor—Self/neighbour

•

Time —•

Figure 2. Hypothetical relationship showing change in origin ofseed with time

obtain increased performance for the life ofthose trees, but that any seed collected fromthem will be diluted by pollen from unimprovedsources.

The relative proportions of the two seed sourcesover time can only be guessed; though, bydrawing an analogy to agriculture it could beexpected that most seed will ultimately comefrom farmers or their neighbours' fields(Cromwell 1990). In this way farmer seeddiffusion mechanisms will have more impact thanpublic sector seed organisations. For cropspecies, Cromwell (1990) described fivecharacteristics for the success of the farmer seeddiffusion mechanism, namely:

• its traditional nature;

• the informality of the mechanism;

• the operation at the community level;

• the evidence of a variety of exchangemechanisms; and

• small quantities of seed involved pertransaction.

The significance of fanner seed diffusionmechanisms can be seen for many agriculturalcrops. In the Punjab Region of Paldstan, it wasfound that 86% of seed replacement of anexisting wheat variety came from farmers' ownfields or those of neighbours. New varieties

93

entered the system through a few large farmerswith later farmer-to-farmer exchange at thevillage level (Cromwell 1990). For maize, it wasfound that farmers in Kenya wanted toexperiment with new varieties, and wouldevaluate risidness, taste, marketability andsusceptibility to pests. Interestingly yield was notmeasured. They did not feel the need for the,formal sector to provide improved seed in largequantities on a regular basis for replacement,and had the capability to multiply tested varietiesthemselves (Cromwell 1990). With a new ricevariety in Nepal, six years after introduction 94%of farmers obtained seed of it from otherfarmers and not from the research centre(Cromwell 1990). The majority of farmers whoreceived seed did not pass it on; rather, a fewkey individuals were responsible for passingseed on to other farmers.

A critical point with respect to farmermultiplication of cereal crops is that most cerealsare self-pollinating and therefore breed true(Simmonds 1979). Consequently the quality ofnewly released material would not be diluted byexisting material present. Most tree species, onthe other hand, are strongly outcrossing, so suchdilution may occur. Another important distinctionis that crops are generally annuals so the scopefor new introductions is greater because newseed is required each year. Conversely trees arelonger-lived and their rate of replacement is low,so new releases would have less impact.

From the studies conducted on agricultural crops,it appears that farmers procure seed from localand readily available sources. A key element inthe success of the release of non-industrial treeswill be close liaison of technical scientists withsociologists and extension workers. One key areawould be the planning and functioning ofcommunity-based nurseries (Scherr 1992);another would be the identification of the seeddiffusers in rural communities. It may be thatmost gain through selection will come aboutfrom participatory breeding schemes, assuggested by MacDicken and Bhurnibhamon(1990). Certainly participatory plant breedingapproaches have had great success with legumecrops in India at the International Crop ResearchInstitute for the Semi-Arid 11-opics and with beanbreeding in Colombia at the Centro InternationalAgricultura Tropical (CIAT) (Cooper et al. 1992).

Greater awareness of the potential risks (egweediness) of introducing exotic pioneer species(Hughes & Styles 1987) has, in part, led toinitiatives to concentrate on native or naturalisedspecies. Concerns over conservation issues havealso led to the increasing sensitivity of donor

_agencies to the promotion of exotic pioneerspecies, as reflected by the emergence ofregional biodiversity programmes (eg theSouthern African Development Coordination

94

Conference (SADCC) regional gene bank). Withthese species, trees may be planted or simplyallowed to regenerate naturally. Where trees areplanted, local farmers may possess knowledge ofseed collection, ripening and treatment, yet maybe unaware of the consequences of propagationfrom a restricted genetic base. In this case,forestry extension efforts could be redoubled topromote more optimal sampling and selection.Naturally regenerated material may also sufferfrom reduced outcrossing; supplemental plantingschemes or recommendations on minimalnumbers of trees could be proposed tocounteract this effect.

CONTINUITY OF RELEASE

Before breeding of non-industrial treescommences, long-term decisions need to betaken regarding the continuity of release ofimproved germplasm. If the assumption thatfarmers will largely collect their own seed infuture is correct, then this factor should guidethe breeding strategy

Once particular species have become known orsuperior provenances have been identified, therewill be some demand by innovative farmers orstrong supply initiatives by seed distributors toplant that material. The question arises as towhether or not unimproved seed of superiorprovenances should be released immediately orwhether release should be postponed until afterone generation of improvement. Following theformer course would jeopardise gain fromimproved material released at a later date,whereas the latter situation requires farmers towait one to several years. The relativeadvantages of these options could be evaluatedif the information presented in Figure 1 wereknown. In the absence of such models and inthe face of strong demand, it seems prudent torelease unimproved material as many farmersare likely to plant what is available rather thanwait, which is a luxury available only tonon-subsistence farmers. Research workers alsosometimes succumb to demand by releasingwhatever is available even if this meanscollecting seed from Provenance trials (J Timyan,SECID, Petion-Ville, Haiti, personal communication).

In the absence of strong demand from farmers,restraint in the wide promotion of identifiedsuperior sources by research organisationsshould be practised until sufficient seed isavailable. Some of the failures in popularisationof superior non-industrial trees in southern Indiahave followed cases of over-promotion ofinsufficient seed supplies of these species (RJambulingam, Tamil Nadu Agricultural UniversityCoimbatore, India, personal communication).

Some control over the amount of the dilution ofgain may be possible with the release ofunimproved material comprising families of low

fecundity This situation could be further rectifiedby selecting for high fecundity in the improvedmaterial to be released at a later date (providingthis is not negatively correlated to desirable traitsand there is no weediness potential).

It can be angued that there is a one-off chance forthe release of improved material because futureseed will largely come from farmers' own fields,which means that this material must be geneticallydiverse to avoid inbreeding and large foundereffects. It does not rule out, however, future smallerreleases to ensure a maintenance of geneticdiversity Future releases would ideally comprisedifferent families from the initial releases, inessence being a sublining across time. In fact, thegreater the opportunity for secondary releases (notnecessarily second generations), the greater thescope for more intense selection, and hence gain,in the first generation. It is important for wide/yplanted species that the genetic diversity oflandraces is monitored to permit ameliorativeaction, if required.

POTENTIAL FOR IMPROVEMENT

Most tropical non-industrial tree species have notbeen improved beyond their wild state (Simons1992). From reviews of isoenzymic variation, it isapparent that tropical trees contain pronouncedintraspecific variation and show high levels ofpopulation differentiation (Hamrick & Godt 1989).Within each species there exists a breadth ofvariation that has been created and partitionedby evolutionary forces (mutation, migration,selection and drift), and which may also residewithin naturalised or landrace populations.

Population differentiation is laiown to differ inmagnitude between different species (Hamrick &Godt 1989), but for many of the tree speciesinvestigated there is significant variation at theprovenance level, with four-fold differences inproductivity not uncommon. For instance, thedifference between best and worst provenancesof Eucalyptus spp. grown in Africa was as muchas 800% (Palmberg 1989). In attempting to breedfrom even an average-performing provenance,such increases would probably not beachievable. Clearly then, significant improvementcan be anticipated from the correct choice ofpopulation, suggesting that comprehensiveprovenance testing should be carried out beforebreeding begins.

From the large expense associated withrangewide collections (Wil lan et al. 1990), for aparticular species to warrant intensivegenecological investigation it must have shownsome superiority or desirability For lesser-laiownspecies, however, there exists the paradox that,unless they are evidently superior, no rangewidecollection will be carried out, yet theirsuperiority may only become manifest once

rangewide samples are available for testing.Therefore, particular species may be discountedon the basis of the performance of one or a fewprovenances. Indeed, Stewart and Dunsdon(1991) found this to be the case where a poorpopulation of Prosopis juliflora was included in aninternational trial series of Central American dryzone species.

The need to test more than one provenance perspecies, however, raises some difficulties withrespect to the size of species elimination trials. Ifprovenance identity is retained, then the largenumber of treatments may cause largeblock/treatment interactions.An alternativeapproach would be to bulk as many seedsources as possible for each species andexamine not only mean performance, but alsothe variation around the mean. In this way,promising species could then be investigated indetail at the provenance level, an activity thatwould be required even if provenance identityhad been maintained, unless a large number ofprovenances were included.

In commercial forestry there has been a strongtendency to maintain the fidelity of provenanceswith an avoidance of provenance mixing,following the assumption that such mixing wouldcause breakdown of co-adapted gene complexes.Deliberate attempts at provenance hybridisationhave had mixed success (see Barnes & Simons,pp84-90). Independent of whether or not there isa heterotic response to combining two or moreprovenances, the hybridisation (compositing) ofprovenances of non-industrial species may beinevitable as many organisations are involved inthe testing, promotion and release of thesespecies. Provenance hybrids would arise wheretwo organisations released different provenancesto the same area and farmers subsequentlycollected seed from trees in their own fields. Inaddition, where a landrace or naturalisedpopulation already existed (eg Calliandracalothyrsus in Indonesia), any new materialreleased would hybridise with the existingpopulation. It is unclear how much provenancecompositing should be actively discouraged, or,alternatively how much it should be promoted.For instance, while it may seem intuitivelyobvious not to permit compositing of awood-producing provenance with a leaf-producingprovenance because the benefits of each mightbe compromised, it may be possible to blendcharacteristics not found in combination inseparate provenances (eg acid tolerance, insectresistance, high growth rate). Theseconsiderations are currently being examined for

sepium in collaboration with theconservation and genetic improvement ofHonduras forest resources (CONSEFORH) project,where a three-provenance hybrid is beingevaluated against each of the parent provenances.

95

Whilst variation is likely to exist for themultiplicity of traits that farmers seek from treesin general, it may not be present for all traits inone species. Moreover, despite the label'multipurpose' that is attached to mostnon-industrial trees, in many cases small-scalefarmers cultivate trees for one or a few reasonsand accept (but do not demand) the associatedbenefits. There may even exist negative geneticcorrelations between characters, making theirsimultaneous improvement difficult or impossible.Therefore, the concept of the ideotype that canfulfil all functions (ie shade, green manure, wood,fodder) is often illusory Furthermore, inattempting to breed trees, it is important to beaware of the selection criteria of beneficiaries, asthese may not coincide with those of scientistsblinkered by mechanised, replicated andstraight-line ideologies. Indeed, what will berequired are trees which interact in a positiveway with farming practices, rather than the mostvigorously growing provenances, as these couldoverly reduce crop production in agroforestry

A strategic concept in the breeding ofnon-industrial trees is not whether desirablevariation exists for a particular species, butrather what chance there is of incorporating itinto the fields of farmers. Figure 3 presentssome generalised options for improvement,together with estimated gains. The way in whichthese relate to the generation interval and tofarmer uptake and seed origin (Figures 1 & 2)should contribute to decisions on animprovement strategy

Recurrent selection is expected to yield greaterultimate gains. However, if most farmers haveplanted the tree and seed is currently beingderived from local sources, then suchimprovements are redundant. Another problemwith recurrent selection is that, in order toachieve cumulative gains, there is a need to

> 80 Generation.1T1 interval3 60

40

20

96

0

First clonal selections- — - - Recurrent selection- - - - • Broad-base landrace

- In-bred landrace

______________________________

Time

Figure 3. Hypothetical relationship showing genetic gain withtime obtained from different breeding strategies

modify gene frequencies and thus narrow thegenetic base. For some species in certain areas,there may be an opportunity to practiserecurrent selection on different base populationsof the same provenance, and to releasesecond-generation material comprising differentfamilies from earlier releases (unimproved or firstgeneration).

Selection of vegetative propagules has theadvantage that it captures non-additive as well asadditive gene effects (Simmonds 1979). Somespecies may not be amenable to vegetativepropagation (usually by cuttings); however, forthose which can easily be propagated, thismethod may offer the greatest scope forimprovement. In cases where farmers cultivate.only a few trees, then they are interested in allof their own trees doing well and not interestedin the fact that the mean performance of alltrees in their district is high. Accordingly cuttingscould be used to achieve this aim. ForArtocarpus heterophyllus in Sri Lanka, twopredominant varieties exist (with hard and softflesh) which are used for different purposes,although they do not breed true. The use ofcuttings allows hard-fruit trees (most desired) tobe guaranteed.

Stakes as vegetative propagules are commonlyused to create 'instant' fences or boundaries. ForGhricidia sepium, stakes 2 m in length are oftenused, which show excellent survival. It is unclearin some areas whether stakes are used becauseof the problems of setting seed (eg northernHonduras, upland Sri Lanka) or because they areinherently preferred. Their advantages includethat they do not take time to grow up, theycompete well with other vegetation, and theyrequire little tending. Unfortunately this option isnot available for all species (eg Calliandracalothyrsus).

Physiological differences between seedling- andcutting-derived plants may restrict the use. ofcuttings in some cases, such as with Gliricidiasepium where cuttings can be coppiced for fourto eight years before declining whereasseedling-derived trees can be coppiced for overten yeats (Liyanage & Jayasundera 1988). Inaddition, the more lateral root architecture ofcuttings may not be suitable where below-groundcompetition with crops is important (eg in alleycropping). Another difficulty which can be seenwith vegetative propagation is that, if few clonesare released or come to dominate the clonalpopulation, then seed collected from them mayhave a narrow genetic base, resulting in inbreeding.

If little organised breeding is operative, then it ishoped that landrace populations will develop thatare more productive and relatively well adapted.The performance of the resultant landrace will

depend upon the quality and quantity of differentseed releases. To ensure sufficient variation tominimise inbreeding, to provide scope forfarmers to select, and to exploit genotype/environment interactions, broadly based materialmust be released (see below, section on geneticgain and diversity).

METHODS FOR PROVISION OF SEED

Assuming that a provenance has been selectedor that little population differentiation exists for aspecies, a range of methods are available todeliver seed to farmers: it may be:

• collected from the native range;

• produced from base populations; or

• bred in seed orchards.

Native range

Seed for distribution to small-scale farmers maybe collected from selected or unselected trees innative stands. Some populations, however, maycontain too few individuals for phenotypicselection to be practised (eg Calliandracalothyrsus, Acacia torah's). Collection of seedfrom selected trees is not possible for someother species, because desirable phenotypes aremasked as a result of disturbance (eg grazingor lopping for firewood). Even where numbers ordisturbance are not restrictions, the efficiency ofplus-tree selection from natural stands isdebatable (Ledig 1974) so seed should ideallybe collected from a representative sample oftrees.

Willan et al, (1990) comprehensively reviewed themethodology for sampling germplasm frompopulations of multipurpose trees. These authorsstressed the importance of adequate numbers oftrees, minimum distance between them, and equalrepresentation. For some species, however, thismethod may not satisfy the demand. For example,with Gliricidia sepium, every seed was collectedfrom native trees of the most desired provenance(Retalhuleu, Guatemala) in 1992; the 10 kg thuscollected, however, constituted less than 1% ofwhat was currently requested. Similarly withCallianclra calothyrsus, because of the small amountof seed produced per tree, combined withprolonged ripening and explosive pod dehiscence,the combined yield of a rangewide collection of31 provenances in the native range in 1991-92 bystaff from the Oxford Forestry Institute (OR) wasonly 45 kg. During the same period, demand fromjust two countries (Haiti and Sri Lanka) exceededthis amount three times.

Some logic is contained in the statement thebest way to conserveis to utilise', and thiscertainly applies for genetic resources underheavy pressure. Measures to prevent the loss ofkey genetic resources are fundamental toimprovement capability

Base populations

Seed may also be obtained from basepopulations for release to farmers. This systemcan be used when collection from nativepopulations is insufficient or difficult, and is asimple, low-cost and quick option. Although nodeliberate selection would be practised, it wouldbe prudent to raise these populations in thetarget environments in case natural or inadvertentselection were prominent. This is essentially theconcept behind the multiple population breedingsU tegy developed in Zimbabwe, which isdesigned to exploit genotype/environmentinteraction (Barnes & Mullin 1989).

A broad genetic base should be ensuredthrough the planting of seed from as many treesas possible (ideally more than 50). The largenumber of families should mitigate against theeffects of non-random mating and the build-up ofco-ancestry in future generatiohs.

Seed orchards

Lastly seed may be produced from seedorchards where deliberate selection is made toenhance desirable traits. Within this format thereis a range of options (described below insection on breeding options). The relevance ofeach of these methods will hinge upon thedemand, the mechanism for release, and theavailable resources. There may even be amixture of the three mechanisms, although thesuccess of the third will depend on how muchprevious material has been released (if theassumption that farmers will collect their ownseed in future is correct). If one organisationcame into an area releasing unimprovedgermplasm and satisfied most of the demandbefore another organisation had bred its firstgeneration of material, the overall effect of thelatter on mean population performance would bediminished. Some co-ordination of effort betweendifferent organisations releasing germplasm isclearly required; the new CGIAR centres couldhave a role to play in this area. Understandingtherole of improved seed and the mechanismsby which seed reaches small farmers has beencritical to agricultural development, with thetimely availability and physical accessibility ofseed being identified as major determinants forplanting of individual varieties (Cromwell 1990).For each of the methods used for the release ofnon-industrial tree germplasm, cognisance shouldbe taken of these points.

For species that are already planted in an areabefore new (improved?) germplasm is released,the situation will be different. Where vegetativepropagation is the mode of reproduction, then itmay be easy to introduce new material andcapitalise on its benefits (eg with Gliricidiasepium in Sri Lanka). It may be harder, however,if material is sexually reproduced, because there

97

would be a mixing of unimproved and improvedsources through the self-collection of seed. Insome locations, particularly with good extensionservices and a high appreciation of improvedgermplasm, it is conceivable that multiplereleases could satisfy planting requirements for aperiod of time until most previous material wasreplaced or newer material became available.Such situations, however, are likely to be theexception rather than the norm.

BREEDING OPTIONS

Where seed is to be produced in managedseed orchards, there are a number of importantconsiderations which will be largely species- andregion-specific. They include:

• seed orchard design;

• intensity of selection; and

• orchard management.

Seed orchard design

Seed orchards can be clone- or seedling-based.The classic clonal seed orchards of commercialtree species contain cloned genotypes selectedat high intensities (eg 1/10 000; Zobel & Talbert1984) following half-sib progeny tests or complexmating designs (eg diallels). This level of activitywould be difficult to justify for non-industrial treespecies because:

• intense selection- reduces the genetic baseand incurs risk;

• germplasm is required immediately thusruling out sequential testing as gain per yearis more important than gain per generation;

• it may be difficult to select plus-treesbecause of disturbance;

• there are high and recurrent costs involvedin this method; and

• the vast array of germplasm with whichprogrammes have to deal makes intensework on any one species difficult to justify

Seedling seed orchards (SS0s) or breedingseedling orchards (BS05) are already being usedfor commercial trees for some of these reasons(eg Pinus pinaster — Alazard 1986; Eucalyptusspp. — Barnes & Mullin 1989). Reddy Rockwoodand Meskimen (1986) and Franklin (1989)examined the gain after one generation in anSSO of Eucalyptus grandis in Florida. Throughretention of 117 families from 500 and onaverage 14 individuals per family from 60, therewas a realised gain of 95% in wood volume inthe first generation of selection. It is likelytherefore, that SSOs will prevail for non-industrialtrees. Design of SSOs is important and includeselements of composition (ie number of families,number of individuals) and layout (spacing, plottype, number of replicates). Each of them shouldreflect the objectives of the seed orchard, the

98

management of the orchard, and the selectionintensity that can be practised.

Composition

Ideally the larger the number of familiesincluded the better because it will ensure abroad genetic base and also allow for somebetween-family selection. For many species,though, family number will be limited by smallpopulation size (eg 25 in some Acacia tortilispopulations) (Olong'otie 1992). If an objective ofthe SSO is to provide genetic parameters, thenfamily number should be at least 20-25;otherwise, the standard errors are likely toswamp the parameter estimates (Crockford, Birks& Barnes 1989; A J Dunsdon, OFI, personalcommunication). To obtain accurate estimates offamily means, numbers in the order of 20 per .family are required (Cotterill & James 1984).With only 20 trees per family however, thisnarrows the scope for within-family selection to5% at most.

Layout

Assuming that there are fixed numbers offamilies and individuals per family (half-sibs), itis possible to arrange them differently within andbetween replicates. If a key objective is toestimate family means accurately then smallerplots (down to single trees) should be used,particularly if family size is small. Apart fromconsiderations of the statistical validity of singletree plots (Shuie & Pauley 1961; Franklin 1971),they are not always the ultimate arrangement.For example, work with Picea sitchensis (Johnstone & Samuel 1974) and Pinus radiata(Correll & Cellier 1987) found that four plantsper plot were more efficient. If, however,within-farnily selection is more important, thenlarger family plots should be used to increasethe efficiency of comparisons between individuals(Cotterill & James 1984). There will be atrade-off between these two considerations,depending on the method of selection(individual, family or combined), the relativevalues of family and individual heritabilities, thesignificance of final spacing, and the intensity ofselection. Ideally plot size should be equal to orless than 1/p (where p=proportion selected);otherwise, relatives will occur in the same plot,thus increasing the chances of inbreeding.

Intensity of selection

Intensity of selection comprises two elements:between- and within-family selection.Between-family selection will be most beneficialwhen number of families is high and whenfamily heritabilities are high or individualheritabilities are low Conversely within-familyselection is better when number of families issmall and individual heritabilities are high. The

largest family differences are like/y to be evidentat good growth sites, but these may not berepresentative of target environments, and thereis the chance of family/environment effects.

The prime objective of a managed seed orchardmay be to capture genetic gain (G) throughselection. It is possible to calculate thisalgebraically using the standard equation:

G = i oph2

where i = intensity of selection; op = phenotypicstandard deviation; and h2= narrow senseheritability

It is easy to examine the effect of intensity ofselection (i) on gain and to express this in termsof increased production. Increasing selectionintensity however, will reduce the geneticdiversity of planting stock. If the proportion ofselected individuals (p) is plotted against theintensity of selection (see Figure 4), it is clearthat, with an increase in the proportion beyond10%, there is a near-linear relation. A more rapidincrease in (i) comes about if proportions below10% are selected, when the relation is highlynon-linear. Therefore, to obtain the greatest gainrequires reduction of the selected population to avery small number of individuals. However, thisincurs an exponential decline in diversity

Lindgren (1986) plotted heritability h2, againstoptimal proportion of parents selected fordifferent numbers of parents, and demonstrated

•-

3.0

2.5

g to

0.5

0.00 10 20 30 40

Selected,p (%)

FVure 4. Relationship between percentage of populationselected and the intensity of selection (see text for furtherexplanation)

that for h2 between 0.1 and 0.3 the optimum isbetween 8% and 4%, respectively for n=50. Suchintense selection, however, would beinappropriate for non-industrial trees where thelikelihood of few releases would cause geneticerosion.

Orchard management

A conflict of purposes may arise in an SSO withrespect to thinning. If the orchard is a progeny

test, and this is a prime function of it, thenselective thinning will bias genetic parameters.Matheson and Raymond (1984) found that h2 ofdiameter at 19 years in Pinus radiata was inflatedfrom 0.24 (SE 0.059) to 0.56 (SE 0.132) usingsystematic versus selective thinning, even at themild selection rate of 50%. Other problems mayarise with early selective thinning ifjuvenile/mature correlations are not good.Selective thinning will also cause uneven spacing,with resultant problems in comparing individualor family performance at later dates.

Early systematic thinning overcomes problems ofheritability biases, spacing and juvenile/maturecorrelations, but there is the chance ofeliminating useful individuals. If there are largenumbers of trees in an SSO beyond thoseneeded to calculate genetic parameters, then acompromise could be struck by having a centralcore to the trial which was systematicallymanaged with surrounding blocks, whereselective thinning was operative. The trees in thecentral core would provide information ongenetic parameters yet could still contribute asseed or pollen parents after parameters wereobtained. Meanwhile selection, and possibly earlyseed collection, could proceed in thesurrounding block.

Caution will need to be exercised with selectivethinning of dioecious species (eg Simaroubaglauca). If thinning were carried out beforeflowering occurred, and morphological characterswere sex-linked (eg as in Fraxinus spp.), then itis possible that the outcome could be anall-male orchard.

One advantage of the family-based seed orchardover a base population, even if no selection ispractised, is that it provides control of pedigree.In this way inbreeding can be minimised bymaintaining sufficient individuals of selectedfamilies (or all farnilies if no between-familyselection is practised).

50 60 GENETIC GAIN AND DIVERSITYAlthough calculation of the relative merit ofgreater genetic gain through intense selection ispossible, the value of maintaining geneticdiversity remains obscure because it is difficult toestimate. It includes a component of immediateeconomic value and a component of futurepotential to adapt to changing conditions(Narnkoong 1984). Given that small-scale farmerswho will use the trees need genetically diversematerial, the numbers of parents retained shouldbe high. Various published estimates of minimumpopulation size are available, although they differgreatly in magnitude (n=25-1000) (Gregorius1980; Namkoong 1984). There is no reason tosuggest, however, that inferior trees within .families are more likely than superior trees to

99

contain useful variation for some unforeseenfuture purpose. Within-family selection, therefore,does not need to be compromised if largenumbers of families are available.

With recent advances in molecular biology (egRFLPs, RAPDs) (Waugh & Powell 1992), there isthe opportunity to measure the total genomicdiversity of individuals, families or populationsfree from the restrictions of sampling limitednumbers of loci of largely structural genes (ieisoenzymes) (Hamrick & Godt 1989).Consequently relative measures of diversitybetween families or individuals could be used toguide selection for seed parents in an orchard.This prospect is currently being studied incollaboration between the Oxford ForestryInstitute and the Scottish Crops Research Institute,Dundee.

Associated with the estimation of geneticparameters, and hence gain, are a number ofassumptions (eg random mating, no inbreeding,no selfing, additive gene action, etc) which maynot necessarily be fulfilled. Although adjustmentscan be made for inbreeding or selling (egReddy et al. 1986), it is important to verify thepredicted gains by examining realised gains ingenetic check trials. For example, Franklin (1986)found realised gain to be far below expectedgain for growth traits in Eucalyptus grandis.

Maintaining a large genetic base allows aminimisation of inbreeding but conflicts with theobjective of maximising gain. Studying inbreedingin Pinus radiata, Griffin, Raymond and Lindgren(1986) found that for each increase of 0,1 in F(inbreeding coefficient) there was a 5% declinein height growth rate after three years. Given thestatus of non-industrial trees, the avoidance ofinbreeding cannot be overstated. When Gliricidiasepium was first introduced into Sri Lanka duringthe last century it is believed that seed camefrom only a few trees; this restricted geneticbase led to severe inbreeding in the resultantlandrace (Hughes 1987). In support of this claimis the inferiority of this landrace when testedagainst unselected provenances from throughoutthe native range (Liyanage, Jayasundera &Liyanage 1991).

An often neglected aspect of tree improvement isan understanding of the reproductive biology ofthe species. This can affect genetic diversity andgain in several ways. For instance, with Calliandracalothyrsus, which has been shown to be largelybat-pollinated (D J Mac Queen, unpublished data),there may be problems with fruit set in exoticlocations (eg parts of Africa) because of a lackof the primary pollinator. In addition, floweringphenology may be significant: El-KassabyDavidson and Webber (1986) found moreinbreeding in early and late flowering classes ina seed orchard of Pseudotsuga menziesii.

100

Tremendous differences often exist betweenindividual trees in their precocity of flowering so,although equal amounts of seed may becollected from trees, the intensity of sampling ofindividuals as male parents may differ greatlyOne approach to this problem would be tocollect seed from trees in amounts inverselyproportional to the number of fruits theycontained (assuming there were no sex-linkedcharacters or that functional dioecy was notoperating).

CONCLUSION

Although it is difficult to generalise about theimprovement of all non-industrial tree species,similar features and decisions for differentspecies are evident. In the formulation ofdecision models to guide tree breeding strategyof non-industrial trees, a number of salientconcepts emerge, including status, demand, seedorigin, continuity of release, potential forimprovement, and genetic gain versus diversity

Although no specific methodology for breedingnon-industrial trees exists, workers in this fieldshould avoid attempting to emulate work onagricultural crops and commercial forestry as thestatus of non-industrial trees is more varied. Arequirement to ensure sufficient genetic diversityfor current and unforeseen needs demands thatbroadly based material is released and selectionintensities are relaxed. Recurrent selection andadvanced generation breeding are generallyirrelevant for non-industrial trees, except in areaswhere established extension services exist 'orappreciation of bred material is high. Gainsshould largely be seen as improvements to ruraldevelopment, rather than as economic returns oninvestmént.

Demand for planting materials of non-industrialtrees will arise owing to farmer requests and thepromotional activities of external organisations.Without knowledge of these demands, thedirection that breeding should take is not welldefined. Accordingly some flexibility in approachshould be retained, but not at the expense ofmaldng bred trees redundant because demandhas already been satisfied from unimprovedsources. Material from advanced generations willhave to be demonstrably superior to make itpreferable to previously planted stock, and thismaterial will differ for quantitative and qualitativecharacters.

It is reasonable to assume that farmers willcollect their own seed of non-industrial treesonce this option is available. Consequently therewill largely be a one-off chance of release, whichhas implications for co-ordination of release:separate organisations should not release differentpopulations which will subsequently hybridiseand may reduce performance. Liaison between

breeders and extension workers in the setting upof community-based nurseries and participatorybreeding schemes is essential to ensure thatplanting material is appropriate and optimallyproductive.

Despite the undomesticated state of manynon-industrial trees, the potential for improvementis tempered by the risk-averse and diversegroup of beneficiaries who require stable asopposed to maximal production. In the majorityof cases, selection of superior seed sources willprovide the largest benefits, and it is of littlepoint to begin breeding inferior populations.Multiplication of limited seed in seedling seedorchards provides some scope for moderateselection and gain.

ACKNOWLEDGEMENTS

The authors gratefully recognise the financialsupport of the UK Overseas DevelopmentAdministration Forest Research Programme(Research Projects R4525, R4585 and R4726).Comments and suggestions on the draftmanuscript were gratefully received from DBoshier, J Burley A Dunsdon, C Hughes and PKanowski.

REFERENCESAlazard, P. 1986. Management of progeny test/seedling seed

orchards for Pinus pinaster In: Proceedings of IUFRO

Conference, edited by A.V Hatcher & R.J. Weir, 528-539.Raleigh, North Carolina: North Carolina State University

Barnes, R.D. & Mullin, L.J. 1989. The multiple population

breeding strategy in Zimbabwe - five year results. In:Breeding tropical trees, edited by G.L. Gibson, AR. Griffin &

AC. Matheson, 148-158. Oxford: Oxford ForestryInstitute/Arlington, Virginia: Winrock International.

Burley, J., Hughes, C.E. & Styles, B.T. 1986. Geneticsystems of tree species for arid and semi-arid lands. Forest

Ecology and Management, 16, 367-390.

Cooper, D., Vellve, R. & Hobbelink, H. 1992. Growing

diversity: genetic resources and local food security London:Intermediate Technology Publications.

Correll, R.L. & Cellier, K.M. 1987. Effects of plot size,

block size and buffer rows on the precision of forestry trials.

Australian Forest Research, 17, 11-18.

Cotterill, P.P. & James, J.W. 1984. Number of offspring and

plot sizes required for progeny testing. Silvae Genetica, 33,203-209.

Crockford, K.J., Birks, J.S. & Barnes, R.D. 1989. Family

within provenance trials of Pinus caribaea, P oocarpa and Ppatula ssp. tecunurnanii - early results and implications for

breeding strategy In: Breeding tropical trees, edited by G.L.

Gibson, AR. Griffin & AC. Matheson, 176-186. Oxford: Oxford

Forestry Institute/Arlington, Virginia: Winrock International.

Cromwell, E. 1990. Seed difhision mechanisms in small

farmer communities. (ODI Network Paper, no. 21.) London:

Overseas Development Institute.

El-Kassaby, Y.A., Davidson, R. & Webber, J.E. 1986.

Genetics of seed orchards: a Douglas fir case study In:Proceedings of IUFRO Conference, edited by A.V Hatcher &R.J. Weir, 440-456. Raleigh, North Carolina: North CarolinaState University

Franklin, E.C. 1971. Statistical validity of single-tree plot in

forest genetic research. Silvae Genetica, 20, 73-75.

Franklin, E.C. 1986. Estimation of genetic parameters through

four generations of selection in Eucalyptus grandis. In: Proceedrigs

of IUFRO Conference, edited by AV Hatcher & RJ. Weir, 200-209.Raleigh, North Caro ltira: North Carolina State University

Franklin, E.C. 1989. Selection strategies for eucalypt treeimprovement - four generations of selection in Eucalyptus

grandis demonstrate valuable methodology In: Breedingtropical trees, edited by GL. Gibson, AR Griffin & AC.Matheson, 197-209. Oxford: Oxford Forestry Institute/Arlington,Virginia: Wiruock International.

Gregorius, H.R. 1980. The probability of losing an allele

when diploid genotypes are sampled. Biometrics, 36, 643-652.

Griffm, A.R., Raymond, C.A. & Lindgren, D. 1986. Effect

of inbreeding on seed yield and height growth of Pinusradiata. In: Proceedings of IUFRO Conference, edited by A.VHatcher & R.J. Weir, 603. Raleigh, North Carolina: NorthCarolina State University

Griliches, Z. 1957. Hybrid corn: an exploration in theeconomics of technological change. Econometrica, 25, 501-522.

Hamrick, J.L. & Godt, M.J.W. 1989. Allozyme diversity in

plant species. In: Plant population genetics, breeding, andgenetic resources, edited by A.H.D. Brown, M.T Clegg, AL.Kaliler & B.S. Weir, 1-19. Sunderland, Massachusetts: Sinauer.

Haugerud, A. & Collinson, M.P. 1990. Plants, genes and

people: improving the relevance of plant breeding in Africa.

Experimental Agriculture, 26, 341-362.

Hughes, C.E. 1987. Biological considerations in designing a

seed collection strategy for Gliricidia sepium. CommonwealthForestry Review 66, 31-48.

Hughes, C.E. 1989. New opportunities in Leucaeana genetic

improvement. In: Breeding tropical trees, edited by G.L.Gibson, AR. Griffin & A.C. Matheson, 218-226. Oxford: Oxford

Forestry Institute/Arlington, Virginia: Winrock International.

Hughes, C.E. & Styles, B.T. 1987. Benefits and potential

risks of woody legume introductions. International nee Crop

Journal, 4, 209-248.

Jahn, S.A.A. 1989. Moringa oleifera for food and waterpurification - selection of clones and growing of annual short

stem. Entwicklung Landlicher Raurn, 23, 22-25.

Johnstone, R.C.B. & Samuel, C.J.A. 1974. Experimental

design for forest tree progeny tests with particular reference to

plot size and shape. In: Proceedings of the Joint IUFRO MeetingStockholm, edited by E. Andersson, 357-376. Stockholm:Department of Forest Genetics, Royal College of Forestry

Ledig, F.T. 1974. An analysis of methods for the selection of

trees from wild stands. Forest Science, 20, 2-16.

Lindgren, D. 1986. How should breeders respond tobreeding values? In: Proceedings of IUFRO Conference, edited

by AV Hatcher & R.J. Weir, 361-372. Raleigh, North Carolina:

North Carolina State University

Liyanage, L.V.K. & Jayasundera, H.RS. 1988. Gliriciclia as a

multipurpose tree for coconut plantations. Coconut Bulletin, 5, 1-4.

Liyanage, M., Jayasundera, H.P.S. & Liyanage, L.V.K.1991. Evaluation of Gliricidia sepium provenances for lowcountry humid zone of Sri Lanka. International Tree Crops

Journal, 7, 83-94.

Luego, J.N. 1989. Calliandra stemborer infestation at a seed

orchard in Cebu, Philippines. Nitrogen Fixing Tree ResearchReports, 7, 76.

MacDicken, K.G. & Bhumibharnon, S. 1990. Barefoot

tree breeders: the use of farm resources for multipurpose

tree species improvement. In: Multipurpose tree species

research for small farms: strategies and methods, 149-154.

Arlington, Virginia: Winrock International.

101

Matheson, A.C. & Raymond, C.A. 1984. Effects of thinningin progeny tests on estimates of genetic parameters in Pinusradiata. Silvae Genetica, 33, 4-5.

Namkoong, G. 1984. Strategies for gene conservation inforest tree breeding. In: Plant gene resources: a conservationimperative, edited by C.W Yeatinan, D. Kafton & G. Wilkes,79-92. (AAAS Selected Symposium, no. 87.) Boulder,Colorado: Westview Press.

Olong'otie, P.A.S. 1992. Acacia tortilis (Forsk) Hayne: a studyof genetic stnicture and breeding systems. DPhil thesis, OxfordUniversity

Palmberg, C. 1989. Research needs in forest tree breedingand improvement in developing countries. AgroforestrySystems, 9, 29-35.

Reddy, K.V., Rockwood, D.L. & Meskimen, G.F. 1986. Astrategy for conversion of an Eucalyptus grandis basepopulation into a seedling seed orchard. In: Proceedings ofIUFRO Conference, edited by A.V Hatcher & R.J. Weir, 613-621.Raleigh, North Carolina: North Carolina State University

Scherr, S.J. 1992. The role of extension in agroforestrydevelopment: evidence from western Kenya. AgroforestrySystems, 18, 47-68.

Shuie, C.J. & Pau ley, S.S. 1961. Some considerations on thestatistical design for provenance and progeny tests in treeimprovement programs. Forest Science, 7, 116-122.

102

Simmonds, N.W. 1979. Principles of crop improvement.London: Longman.

Simons, A.J. 1992. Genetic improvement of non-industrialtrees. Agroforestry Systems, 18, 197-212.

Stewart, J.L. & Dunsdon A.J. 1991. Evaluation of CentralAmerican multipurpose hardwood tree species for dry zones.(Final Report ODA Research Scheme Project R.4454.) Oxford:Oxford Forestry Institute.

Tothill, J.D. 1940. Agriculture in Uganda. london: OxfordUniversity Press.

Ventakesh, C.S. 1988. Genetic improvement of multi-purposetree species. International Tree Crops Journal, 5, 109-124.

Waugh, R. & Powell, W. 1992. Using RAPD markers forcrop improvement. Trends in Biotechnology, 10, 186-191.

Wei, R.P. & Lindgren, D. 1991. Selection effects ondiversity and genetic gain. Silva Fennica, 25, 229-234.

Willan, R.L., Hughes, C.E. & Lauridsen, E.B. 1990. Seedcollection for tree improvement. In: Tree improvement ofmultipurpose species, edited by N. Glover & N. Adams, 11-38.(Multipurpose Tree Species Network Technical Series, vol. 2.)Arlington, Virginia: Winrock International.

Zobel, B.J. & Talbert, J. 1984. Applied forest treeimprovement. New York: Wiley

Clonal development and deployment: strategies toenhance gain while minimising risk

G S Foster' & F L G Bertolucci21US Department of Agriculture Forest Service, Southern Forest Experimental Station, PO Box 1328, Normal,Alabama 35762, USA2Aracruz Florestal SA, Rua Professor Lobo 112B, 29190 Aracruz ES, Brazil

ABSTRACTClonal tree improvement programmes contain two complementary aspects, clonal development and clonal deployment. Clonaldevelopment entails strategies to provide clones which not only express superior growth and yield, but are also bufferedgenetically against varying environments, including pests. Selection of clones from diverse populations and establishment of alarge breeding population enhance the chance of including substantial genetic diversity especially for developmental (individualtree) homeostasis toward environmental change. A multistage testing programme ensures that the clones in the productionpopulation are superior for growth at the forest stand level, as well as being genetically diverse for life history traits and pestresistance. Once clones are assigned to the production population, forest stand composition can be designed by the tree breederand forester to utilise genetic (population level) homeostasis to take advantage of diverse pest resistance among the clones ordiverse competitive strategies to occupy fully the niches within the ,stand. In addition, production traits which do not affectclone/environment interactions, such as wood properties and tree size, can be structured to be highly uniform within the stand inorder to enhance product development.

INTRODUCTION

Clonal reforestation is practised operationally withboth temperate (Kleinschrnit 1974; Shelbourne,Carson & Wilcox 1989) and tropical (Leakey1987) species worldwide. Interest among forestmanagers in clonal reforestation has increaseddramatically within the past 15-20 years inresponse to the development of operationalcloning systems (usually rooted cuttings) and toenhanced awareness of the benefits of suchsystems. To date, clonal technologies have beenapplied to a limited number of species. Thisreview therefore, draws primarily fromexperience with Eucalyptus, Populus andtemperate-zone conifers.

Some of the most important benefits of clonalreforestation include:• faster availability of genetically improved

planting stock for reforestation compared witha traditional seed orchard/seedling-based treeimprovement programme (Matheson &Lindgren 1985);

• higher genetic gains per generationcompared with a seedorchard/seedling-based tree improvementprogramme (Chaperon 1991);

• greater uniformity of the resultant stands andforest products (Campinhos &Claudio-da-Silva 1990; Zobel, Campinhos &Ikemori 1983); and

• greatly reduced dependence on seeds forplanting stock production (Leakey 1987).

Clonal tree improvement and reforestationprogrammes must be designed to optimiserealised genetic gain. With any tree improvementprogramme, there is a risk that realised geneticgain (ie forest stand yield) will be less thanpredicted genetic gain. The worst-case scenariowould be the loss of the stand prior to finalharvest. The goal of a tree improvementprogramme is to develop one or morepopulations of planting stock which are adaptedto the environments in which they areestablished, including resistance or tolerance todiseases, pests, and other adverse environmentalfactors, so that the risk of loss of stand growthand yield is minimised.

In a clonal tree improvement programme, thefinal yield of the forest stand (or other array oftrees, ie agroforestry) is strongly influenced bythe adaptation of the trees, individually andcollectively to the site. This adaptation has twocomponents: (i) clonal development, and (ii)clonal deployment or stand establishmentprocess. Clonal development approaches,including breeding, testing, and selection, largelyaffect the genetic quality of the resultant clonalpopulation available for reforestation.

Clones must be developed which are highlyselected for growth and productivity traits, butwhich display substantial developmental(individual basis) homeostasis as well as genetic(population basis) homeostasis. Developmentalhomeostasis refers to a genotype's ability to alter

103

its physiology physical development, and growthto adapt to its changing environment (Lerner1954). Clones with highly stable performanceover a range of environments (Burdon 1971;Bentzer et al. 1988) exhibit greaterdevelopmental homeostasis. Genetic homeostasisis 'the property of a population to equilibrate itsgenetic composition and to resist suddenchanges' (Lerner 1954). Lerner proposed thatheterozygosity provides the basis for both typesof homeostasis. In a clonal reforestationprogramme, the fate of a particular ramet of aclone- or clone . within a stand_is not as importantas the growth and yield of the entire stand.Some mortality is expected during standdevelopment, and trees have the ability to takeadvantage of additional growing space and otherresources so that the growth of the survivorsmay compensate for the missing trees. Clonaldeployment must strike a balance between theneed for efficiency of forest management,harvesting and converting into economicproducts and the need to deploy populationswhich are genetically buffered againstenvironmental changes (including pests).

CLONAL DEVELOPMENT

Clonal development relies on three processes:breeding, testing, and selection. Breeding createsnew genetic variability; testing exposes the newgenotypes to appropriate environments; andselection favours certain genotypes for populationdevelopment. At least three populations areconceptualised: the gene resource population, thebreeding population, and the productionpopulation (Libby 1973; Foster 1986; Shelbourneet al. 1989). The gene resource population isoften the wild or unimproved population fromwhich new selections can be derived. Thebreeding population is the somewhat improvedpopulation of all genotypes which are used tocreate the next generation of crosses. A widerange of genotypes may be kept in thispopulation as long as each one has at least onecharacteristic of possible future interest. Theproduction population consists of the highlyselected genotypes which are used forreforestation.

Breeding

Breeding is used to create genetic diversity fromwhich new selections can be made. To befunctionally useful, genetic diversity is needed intraits which aid survival, growth, anddevelopment of genotypes in the population.Functional genetic diversity results in enhanceddevelopmental and genetic homeostasis, andhence in adaptability

Several breeding methods can be used to breedfor enhanced genetic variability especially forcreating extreme genotypes useful for clonal

104

reforestation (Foster 1986, 1993). These methodsinclude the development of multiple populationsor sublines; use of specific mating designs; useof positive assortative mating; breeding fornon-additive and additive genetic variation; andinterspecific hybridisation. A brief summary ofeach follows.

Multiple populations, nucleus breeding andsublines

Multiple populations differ from sublines in thatthe former seek to enhance or at least maintaingenetic differences among populations, whilesublines are random subsets-of- the population(Burdon & Namkoong 1983). Genetic differencesamong sublines are due to the effects of initialsampling, 'founder' effects, and random drift overgenerations of breeding.

Nucleus breeding is a variant of the multiplepopulations approach in which a small subset ofindividuals (eg 40) is copied from the mainbreeding population into a nucleus population(Cotterill et al. 1989). One or more nucleuspopulations can be developed, each with thesame or different goals of selection. Breedingefforts continue at a lower level in the mainbreeding population, while the major breedingeffort is concentrated on the nucleuspopulation(s). The goal of multiple or nucleuspopulation development is to enhance geneticdifferentiation. Interpopulation crosses could beused as terminal crosses to create extremegenotypes for one or more traits. AracruzFlorestal SA has incorporated the multiplepopulation concept in its tree improvementprogramme with 20 populations (20 genotypesper population) of Eucalyptus grandis and 20populations of E. urophylla. Ten crosses will bemade inside each population, forming ten full-sibfamilies per population. In the full-sib families,the best tree will be selected and interspecificcrosses made to seek 'super hybrids' for theclonal programme (Campinhos & Ikemori 1989).

Mating designs

Once the genetic basis (type of gene action:additive, dominance or epistasis) is known forthe superiority of traits in specific genotypes,then mating designs, heretofore unique to animalbreeding, can be used. The goal of thesedesigns, such as breed complementation orspecific three-breed terminal crosses(Hohenboken 1985), is to develop specific typesof individuals in the offspring generation for theproduction population. The parents are oftensuperior for different traits, which are thencombined in the offspring.

Positive assortative mating

Positive assortative mating entails mating like'parents in the population rather than randommating. For example, the top-ranked individual is

mated to the second-ranked, third-ranked ismated with fourth-ranked, etc. This pattern ofmating is known to increase genetic variation inthe population while also increasinghomozygosity for the selected trait (Crow &Felsenstein 1968). Much of the effect is seen inthe first few generations of positive assortativemating. In a simulation study, Mahalovich andBridgwater (1989) found an increase in geneticgain of 0-6% from positive assortative matingand selection through 14 generations of mating.In addition, they found an increase in thenumber of extreme genotypes.

Breeding for additive and non-additivegenetic variation

Most tree improvement programmes emphasisebreeding and selection for additive geneticvariation or breeding value (recurrent selection).However, additional gain may be obtained bybreeding for both additive and non-additive(dominance and epistasis) genetic variation.Improvement plans such as reciprocal recurrentselection (Comstock: Robinson & Harvey 1949)can be modified for use in forestry (McKeand,Foster & Bridgwater 1986). These types ofimprovement plans require several generations todevelop, and to date remain untested in forestry

Interspecific hybridisation

Interspecific hybridisation is known to createlarge amounts of variability in the offspringgenerations. This genetic variability can then beused within a tree improvement programme,especially with clonal reforestation (eg Teissierdu Cros 1984). This process often contains manyproblems, eg cross-incompatibility and low-vigouroffspring, but occasionally some outstandingoffspring are produced. Results from suchcrossing vary tremendously depending on thespecies involved. One goal of interspecificcrossing is to join the desirable characteristics oftwo different species. A good example is crossesbetween E. grandis and E. urophylla, whosehybrids have shown great stability and highquality of wood for pulp production.

Testing

Tree breeders have expended considerable efforton researching the effects of genetic test designon genetic gain. There is a trade-off betweenaccuracy of genetic value estimation andintensity of selection (ie greater accuracy is atthe expense of numbers of families, individualsper family or clones). The optimum strategydepends on heritability of the trait and availabletesting resources. For clonal tree improvementprogrammes, Shaw and Hood (1985) and Russelland Libby (1986) found that, for fixed resources,as many clones as possible should be testedwith relatively few (ie 2-6) ramets per clone.

The tree improver must stratify the potential

reforestation environments into more or lessuniform (non-interactive) subsets. Initially theentire breeding population must be tested acrossthe potential reforestation environments. The treesin the breeding population are used as abioassay to delineate these 'breeding zones'.Simultaneously the breeding population and theenvironments are subdivided into subsets ofgenotypes and matching environments. Within abreeding zone, genotypes are expected toexpress a non-significant interaction (ie rankchange) with environments. Trees are matchedwithin but not among breeding zones. Resultantclones are planted within the zone.

The tree improver must also assess the potentialenvironment apparent within the range ofreforestation environments, and must establishappropriate genetic tests to assess genotypicperformance in the various environments. In mosttree improvement programmes, multiple traits (egsurvival, volume, height, disease resistance, bolestraightness) are being improved simultaneously.Different types of tests may be used to assesseach trait. For example, tree growth and yield,bole form, and wood properties may beassessed in field tests, while rapid screening fordisease resistance may be accomplished in agreenhouse environment (eg Matthews & Rowan1972). Disease screening in artificialenvironments allows collection and use of a widerange of pathogen races to screen simultaneouslyfor host resistance. However, strong correlationmust exist between host reaction in the field andin the artificial environment.

Genetic improvement for yield per hectare inforestry will undoubtedly require a series oftests. For clonal forestry Libby (1987) hasproposed four levels of testing. The four levelsare: initial screening (Level I) with large numbers(eg thousands) of genotypes and only a singleseedling or few ramets per clone; candidacytesting (Level II) with large numbers ofgenotypes (fewer than with Level I, hundreds orlow thousands) and low numbers (2-6) of rametsper clone; clonal performance (Level III) withmoderate numbers of genotypes (eg <200) andlarge numbers (eg 0.1 ha plots) of ramets perclone; and compatibility trials (Level IV) withsmall numbers of genotypes (eg 20-50) withvery large plot sizes. This scheme would resultin the choice of clones with detailed growth andyield performance. As such a scheme willrequire a long time period, clonal maturationmust not be an insoluble problem.

Selection

Initial selections are made from the generesource population. Selection intensity may behigh, but low heritabilities in natural standstypically result in relatively low genetic gain fromthis initial step in a selection programme, if the

105

resource population is a natural stand. Animportant exception may be genetic gain fordisease resistance when disease-free individualsare selected from heavily diseased forest stands.Superior select trees should be derived from awide range of locations within the appropriateportion of the species' range. General treeimprovement principles (Zobel & Talbert 1984;Wright 1976) can be used to guide the choiceof area to sample within the natural range. Awide dispersion of sample locations within thenatural range will enhance the chance ofcapturing a large range of genetic diversity inthe breeding population.

Once initial selections are made, genetic tests,either seedling or clonal, are established.Superior selections for the breeding populationand the production population are made from thegenetic tests. Selection for the breedingpopulation focuses on superiority for additivegenetic value. Because the breeding populationis relatively large (hundreds or even thousands ofgenotypes), a large number of genotypes isselected to reconstitute the population, and henceselection intensity and genetic gain are modest.

Clonal selection

Clonal trials can be established directly (byfelling the select trees and rooting cuttings fromthe stump sprouts, which is feasible with someEucalyptus species, for example), orseedling-based progeny trials can be establishedwith either open-pollinated or controlled-crossseedlings. In a clonal programme, the productionpopulation contains relatively few clones (usuallyfewer than 100; Aracruz Florestal SA currentlyhas 80 clones in operational plantations) whichare selected for genotypic value, not breedingvalue. The exact source of the geneticsuperiority is unimportant for this population. Infact, if non-additive gene action is important in apopulation, many of the genotypes in theproduction population may not be included inthe breeding population (Foster 1986, 1993).Genetic gain in the clonal production populationis greater, and possibly much greater, than inthe breeding population (Foster 1986).

Multiple trait selection

Most tree improvement programmes emphasiseselection for multiple traits. Unfortunately as thenumber of traits increases, the number ofgenotypes superior for all traits diminishesrapidly; selection intensity must therefore increase,or genetic gain will plummet. For this reason,only the few most economically important traits(typically growth and productivity traits) shouldbe emphasised. In environments where diseaseor drought are important, tolerance of theseconditions may also be strongly emphasised.Debate rages among pathologists and plantbreeders as to the best breeding strategy for

106

disease resistance (Marshall 1977). The use ofsingle major resistance genes (vertical resistance)has proven to be only a brief solution inbreeding agronomic crops, as the pathogens canquickly mutate to overcome the host's resistantgene. More promising strategies include breedingmultiple single genes for different resistancemechanisms into the same variety or utilisingpolygenic resistance genes (horizontal resistance)(Marshall 1977). Multiple resistance mechanismsfor the same disease are well known in foresttrees; such procedures are therefore also feasiblein forestry

Genetic conservation

Agronomic plant breeders have been somewhatcareless about allowing genetic diversity withintheir crop species to erode as a trade-off forgreater genetic gains (Marshall 1977). Treebreeders must be careful to maintain widegenetic variability for all traits that are not beingselected during directional selection, because thelatter narrows the genetic base for those traitsunder selection. This is especially true for traitswhich control adaptation. The breedingpopulation is a good place to maintain thisdiversity Traits of interest for developmentalhomeostasis (adaptability) include: growth rhythm,response to moisture deficits, disease resistance,pest resistance, tolerance of different irradiances,rate of photosynthesis and respiration, live crowndimensions, leaf shape, coppicing ability etc. It isvery important for tree breeders to know geneticcorrelations among these traits and the traits ofmore economic interest. This information will helpthem to monitor the inadvertent effects of indirectselection for the homeostatic traits and evencounteract the effects if 'they are detrimental.Given that the set of genotypes in the productionpopulation is dynamic, it is necessary to maintainclone banks as a repository for germplasm. Thebreeding population will also serve as arepository for genetic variability

CLONAL DEPLOYMENT

Despite its overwhelming importance, littleresearch has been conducted to investigate theeffects of different deployment strategies on thehealth, growth and yield of forest stands. Bynature, this type of research requires a largeamount of resources and a long period ofassessment. The major questions are as follows.

• How many clones should be in theproduction population, and how many shouldbe deployed to a single site?

• Should the clones be deployed at a singlesite as a mixture or as mosaics ofmonoclonal stands?

• What are the key attributes of the clonesthemselves that cause them to be usedeither as mixtures or as monoclonal plots?

Number of clones

The question of number of clones to bedeployed contains two aspects: the number ofclones within the production population, and thenumber of clones planted per site. Severalresearch papers have addressed the theoreticalaspects of these issues, but few experimentalresults have been reported.

Number of clones in the productionpopulation

Consensus among tree breeders and forestgeneticists indicates that production populationsizes of 100 genotypes or fewer are acceptable.Seed orchard/seedling-based tree improvementprogrammes utilise 20-50 parental genotypes inthe rogued seed orchard. Owing to geneticrecombination in the seed production process,the variety of offspring genotypes from such aprogramme is very large, especially given therelatively large amount of genetic heterozygositywithin most populations of forest trees.

Theoretical studies have sought to assess eitherthe risk of plantation failure or the stability ofperformance when using varying numbers ofclones in the production population. Although thegoal of these studies was to examine the effectof varying numbers of clones per Site, the resultscan be extended to gain insight into theminimum number of clones needed in theproduction population. Libby (1982) proposed arange of between seven and 30 clones as arelatively safe number, while Huehn (1988) foundthat a safe number of clones in a clonal mixwas generally between four and 50, dependingon assumptions and the area to be forested.

Actually the absolute number of clones is lessimportant than the range of genetic diversityamong the clones. Ten clones which share thesame alleles for a particular trait would expressno genetic diversity as contrasted with fiveclones each of which had different alleles for thesame trait. Hence, the tree breeder mustemphasise genetic diversity for traits associatedwith survival and adaptation, while exercisingstrong selection pressure on production traits.

Aracruz Florestal SA currently uses 80 Eucalyptusclones in its production population, with more inthe initial stage of multiplication. The plantingarea is divided into three regions with 65 clonesused in Aracruz, 40 in Sao Mateus, and 29 inthe south of Bahia State. Several of the clonesare common to all three areas because of theirsuperior developmental homeostasis. Followingrigorous multisite testing, new clones are addedto the production population each year.

Number of clones planted at each site

The goal of deploying superior clones toreforestation on a site is optimum production(solid wood products, fibre, food, or other forest

products) with minimal risk of failure to meet theproduction potential. Libby (1982) utilised theconcept of Yisk to a random genotype' (RRG) ina theoretical study investigating acceptablenumbers of clones to be deployed in clonalreforestation. Furthermore, he ufilised the conceptof 'maximum acceptable loss' (MAL) as the levelof stand loss below which a forest managerwould be unwilling to accept the loss. Based onprobability theory he found that seven to 30clones, depending on RRG and MAL, providereasonable probability of achieving an acceptablefinal harvest. In their theoretical study, Lindgren,Libby and Bondesson (1989) investigated theinteraction between clonal diversity withinmixtures and genetic gain. A key, and somewhatcontroversial, assumption of their analysis wasthat interactions between ramets of the sameclone are negative and, conversely, thatinteractions between ramets of different clbnesare positive. Therefore, fewer clones in thepopulation conferred an increasingly greaterdisadvantage, or reduction in genetic gain. Usingthis assumption, they developed a model whichrelated effective population size to the truncationpoint, due to selection, on the normal distributioncurve. Their analysis demonstrated that, byvarying the frequency of clones in the mixture,rather than using equal proportions, genetic gaincould be enhanced by selecting more clones,with the lower ranked clones occurring atincreasingly lower frequencies.

For operational plantings with Eucalyptus clones,Aracruz Florestal SA utilises at least four differentclones for each 150 ha. Also with Eucalyptus, theIndustrial Afforestation Unit of the Congo (UAIC)operationally plants their clones in 50 hamonoclonal plots around Pointe-Noire, Congo(Leakey 1987).

Well-designed empirical studies are needed toaddress the question of the number of cloneswhich should be deployed per site, to guideclonal reforestation. Such studies mustincorporate a variety of clones with a range oflaiown growth and yield performance, as well assusceptibility to disease, moisture deficits, andother adverse environmental factors. The testsshould include a wide diversity of site quality

Clonal mixtures versus mosaics ofmonoclonal stands

Most tree improvers and forest geneticists agreethat single-clone plantations do not offer anychance for genetic homeostasis, and so the riskof plantation failure owing to the failure of asingle clone is too high. If more than one cloneis deployed on a site, then the two extremealternatives are a complete mixture of all clonesat the site, termed 'widespread intimately mixedplantations', or 'mosaics of monoclonal stands'(Libby 1987). A large body of research results

107

on this issue with agronomic crops has beenpublished (Trenbath 1974; Harper 1977).Unfortunately the results have been mixed, withsome reports showing superiority of the purevarieties over the mixtures, while others haveindicated that the mixtures have significantlyoutperformed the average of the pure plots. Onoccasion, the mixtures have even outperformedthe best pure variety In his review onnon-woody plant biomass yields, Renbath (1974)found that, of 344 mixture comparisons reviewed,the mixtures over-yielded the best purecomponent 24% of the time, and under-yieldedthe worst component 13% of the time, with theremainder being intermediate between the twopure components. In his review Marshall (1977)showed only two out of 328 mixtures whichover-yielded the best pure component plot.

The only published study with forest treescomparing pure versus mixed-clone plots ofwhich we are aware was published by Markovicand Herpka (1986). The authors reportedfourth-year field results with five clones ofPopulus deltoides. Large plot sizes (100 trees)enabled valid estimations of volume productivityper hectare. For volume per hectare, averageheight, diameter at breast height (dbh), andbasal area, the mixed-clone plot performanceWas intermediate compared with the mean of thepure-clone plots. There was slightly higher(though non-significant) volume per hectare(12.8%), mean height, mean dbh, and basal areafor the mixed-clone plot compared with theaverage of the pure-clone plots. Hypotheticallystand development may have enabled thefaster-growing, dominant clones eventually tocontrol the stand so that the stand attributeswere raised closer to that expected of thesuperior clones.

In a previously unpublished study with Pdeltoides, G S Foster (USDA Forest Service,Huntsville, Alabama, USA), R Rousseau (VVestvacoCo, Wickliffe, Kentucky USA), and W Nance(USDA Forest Service, Gulfport, Mississippi, USA)established a mixing study with two locations(James River Co, Vicksburg, Mississippi, andWestvaco Co, Wickliffe, Kentucky USA), fourreplications per location, and seVen clones. Theplot size was 64 trees in a square plot with theinner 36 trees as the measurement plot. Aspacing of 3.66 m x 3.66 m was used, yieldingmeasurement plot sizes of 0.048 ha. Monoclonalplots were planted for each of the seven clones,and binary-clone plots were established forseven pairs of clones. Each binary combinationwas represented by two plot types: (i) 75%Clone Al25% Clone B and (ii) 25% Clone A/75%Clone B. With four plot types (two monoclonalplots and the clone mixture at two ratios), areplacement series graph of volume per plot

108

could be drawn. Total height and dbh weremeasured at ages two, three, and four, andwhole-tree volume was totalled on a plot basis.r1Vvo of the seven pair combinations arepresented in Figure 1. These two are presentedhere because of the contrasting results obtained.Different clones were represented in Mixes 1 and2. In Mix 1, the results were opposite at the twolocations, yet the mixed-plot yields are virtuallylinear across the replacement seri6s. In Mix 2,the ranldng of the two clones was the same atboth locations; however, the mixed plot 75%C:25% D over-yielded the best pure-clone (C)plot by a large margin. These results show thatfield results for mixed-clone stands versusmosaics of monoclonal stands may vary

4.0

3.5

3.0

2.5

2.0

1.5

1.0

0.5

4.0

3.5

E 3.0a)

2.0

2.5

'5 1.5

1.0

0.5

0

C/o /le 0.25o

MIX 1

0.25Cione

Propottion0.75

MIX 2

0.75Proponyon

1.00

1.00 2

2

[ISite 2

Site 1

4.0

3.5

3.0

2.5

2.0

1.5

1.0

0.5

0

4

3PCP

4

4.0

3.5

3.0

2.5

2.0

1.5

1.0

0.5

Figure 1. Plot volume production of Populus deltoides clonesat ages 2, 3, and 4 at two planting sites in a replacementseries study with binary-clone pairs. Mix 1 has clones A andB, and Mix 2 has clones C and D. At each of the foursample proportions, the ratios sum to 100% (ie 0% A/100% B)

dramatically by site and may be specific to theparticular set of clones.

Theoretical considerations argue against mosaicsof monoclonal stands; yet experimental resultsgenerally have failed to support the expectedsuperiority of mixtures. The potential advantagesof mixtures are: `(i) increased yields through theexploitation of favorable interactions betweencomponent genotypes; (ii) greater stability ofperformance in diverse environments; and (iii) areduction in disease and insect damage'(Marshall 1977). Mixed genotype stands shouldafford more resistance or tolerance of pest attackas non-susceptible genotypes will help slow thespread of the pest (Marshall 1977; Heybroek1982), but mosaics of monoclonal stands may bean acceptable compromise with certain diseases(Heybroek 1982). Actual results with agronomiccrop species (Trenbath 1974; Marshall 1977) andtree species (Markovic & Herpka 1986) generallyhave shown little or no greater yield of mixturesover the average of the monoclonal plots. If suchstudies are conducted in pest-infestedenvironments, the mixtures might fare muchbetter (Kroll, Moore & Lacy 1984).

Practical arguments can be made to supporteither deployment strategy From a managementstandpoint, the mosaics of monoclonal standsoffer the forest manager the chance to observethe performance of single clones. Thisobservation will lead to extremely valuableinformation on clonal growth and yield, clonalwood properties and log characteristics,clone/site interaction, and resistance to adverseenvironmental factors (including pest resistance).If a single clone succumbs to an adverseenvironmental factor prior to harvest, the affectedstand can be removed and replanted. However, ifthe clones are mixed, then a certain percentageof all trees in the stand may succumb, droppingthe stand yield below an economic threshold.Forest managers may take more risk withshort-rotation intensive management oragroforestry plantings, and . may thereforeexclusively plant mosaics of monoclonal stands.Managers worldng with species withmedium-length rotations should be more cautiousand may want a portion of their reforestationwith each of the two alternatives. Managers withlong rotations may want to use mixed-clonalstands as an insurance against losing the standsafter many years of investment and growth.

Another aspect which can influence the choiceof strategy for clonal deployment is the quality ofwood produced by the stand. In the case ofpulp production, for example, somecharacteristics should be maintained within anarrow band of variation in the productionpopulation (Demuner et al. 1991). Uniform woodproperties in pulp optirnise paper qualityproduced by a mill. This leverage point

influences the adoption of mosaics of monoclonalstands because, within each stand, thecharacteristics of the wood produced will behomogeneous and completeiy predictable.

Attributes of clones which influencedeployment

Few theoretical or empirical results have beenpublished regarding the relationship betweenphysical attributes and clonal performance inmonoclonal or mixed-clone blocks. Markovic andHerpka (1986) noted a slight, yet non-significant,growth advantage for the mixed plot over theaverage of the monoclonal plots. G S Foster, RRousseau and W Nance (unpublished) found alinear relation between plot volume andproportion of each clone in a two-clonereplacement series, while in a second two-clonereplacement series the same relation wascurvilinear (Figure 1).

Why do clones interact in a variety of ways,additively over-yielding, or under-yielding? Crownor root system architecture of the clones mayhelp explain their interaction. Clones that havenarrow crowns and root systems and otherwiseuse their allocated growing space efficiently(termed 'crop ideotypes') may interact in anadditive fashion in a replacement series. Clonesthat have broad crowns and otherwise take morethan their allocated growing space (termedIsolation ideotypes') may perform poorly ineither pure plots or mixtures with other clonesthat are isolation ideotypes. The most efficientmixtures may be between clones which are cropand isolation ideotypes, thereby more completelyusing all available resources. Attributes whichmay serve as mechanisms for these interactionsinclude live crown length, crown radius, leaf areaindex, height, dbh, depth of root system, andhorizontal spread of root system.

Another general mechanism for inter- andintraclonal interaction is through diversity ofresistance to adverse environmental factors.Performance of monoclonal stands of susceptibleclones could be severely reduced, even to zeroby complete mortality Disease resistanceprovides an interesting case of this class ofinteraction. Mixtures of clones with varyingdegrees of resistance could result in lower netloss than mosaics of monoclonal stands, asresistant trees tend to slow the rate and severityof disease infestation among susceptiblegenotypes (Heybroek 1978, 1982). The negativeside of the mixtures is that salvage of thesusceptible trees would be much more difficultthan with monoclonal stands.

CONCLUSION

A successful clonal tree improvement andreforestation programme depends on the

109

development of appropriate clones and thedeployment of those clones, based on lmowledgeof inter-tree competition and interaction with theenvironment. Clones must be developed whichhave substantial developmental homeostasis inorder to interact favourably with changingenvironments. Breeding methods are available toenhance genetic variability within the offspringgenotypes for traits which determine adaptationto the environment, while at the same timeencouraging development of extreme genotypesfor productivity traits. Appropriate genetic testsmust be used to assess these somewhat differenttypes of traits: survival and adaptation versusproductivity A series of tests combined withselection to test fewer and fewer clones withincreasing numbers of ramets is recommended.

Clonal deployment varies with rotation lengthand types of adverse environmental factors (egdiseases or drought) that the forest stand willencounter. A high degree of genetic homeostasismust be built into forest stands by careful clonaldeployment. Mosaics of monoclonal stands maybe particularly useful for short-rotationmanagement in the absence of adverseenvironmental factors. As the rotation length andthe probability of adverse environmental factorsincrease, either a mixture of mosaics ofmonoclonal stands and clonal mixtures or entirelyclonal mixtures should be used. Monoclonalstands are particularly useful for gaininginformation on specific clones.

These strategies for clonal development anddeployment, based on experience with a limitednumber of tree species, should be applicable tothe large number of tropical tree species whichare largely undomesticated. In the case of suchspecies, where tree improvement programmeshave often not yet been initiated, the first prioritymay be tO identify rapidly superior individuals inwild populations (see Ladipo et al., pp239-248;Lawson, pp112-123). Where the users of clonaltechnologies are small-scale farmers rather thanindustrial concerns (eg see Mesén, Boshier &Cornelius, pp249-255), novel strategies may haveto be developed to mitigate risk. The applicationof clonal techniques in such situations offersgreat prospects for increasing gains andstimulating reforestation in the future.

REFERENCESBentzer, B.G., Foster, G.S., Hellberg, A.R. & Podzorski,A.C. 1988. Genotype x environment interaction in Norwayspruce involving three levels of genetic control: seed source,clone mixture, and clone. Canadian Journal of Forest Research,18, 1172-1181.

Burdon, R.D. 1971. Clonal repeatabilities and clone-siteinteractions in Pinus radiata. Silvae Genetica, 20, 33-39.

Burdon, R.D. & Namkoong, G. 1983. Multiple populationsand sublines. Silvae Genetica, 32, 221-222.

Campinhos, E. & Claudio-da-Silva, E. 1990. Development

110

of the Eucalyptus tree of the future. In: Proceedings of theESPRA Spring Conference, 1-22. Seville: ESPRA.

Campinhos, E. & Ikemori, Y.K. 1989. Selection andmanagement of the basic population of Eucalyptus grandisand E. urophylla established at Aracruz for the long termbreeding programme. In: Breeding tropical trees: populationstructure and genetic improvement strategies in clonal andseedling forestry, edited by G.L. Gibson, AR. Griffin & AC.Matheson, 169-175. Oxford:. Oxford Forestry Institute.

Chaperon, H. 1991. Open-pollinated seed orchards: testedfamilies or tested clones for tomorrow's forests? In:Proceedings FRI/NZFP Forests Ltd., Clonal Forestry Workshop,edited by J.T Miller, 15-24. Rotorua: Forest Research Institute.

Comstock, R.E., Robinson, H.F. & Harvey, P.H. 1949. Abreeding procedure designed to make maximum use of bothgeneral and specific combining ability Agronomy Journal, 41,360-367.

Cotterill, P., Dean, C., Cameron, J. & Brindbergs, M.1989. Nucleus breeding: a new strategy for rapidimprovement under clonal forestry In: Breeding tropical trees:population structure and genetic improvement strategies inclonal and seedling forestry, edited by G.L. Gibson, A.R.Griffin & AC. Matheson, 39-51. Oxford: Oxford ForestryInstitute.

Crow, J.F. & Felsenstein, J. 1968. The effect of assortativemating on the genetic composition of a population. Eugenics,15, 85-97.

Demuner, B.J., Vianna Doria, E.L., Claudio-da-Silva, E.& Manfredi, V. 1991. The influence of eucalypt fibercharacteristics on paper properties. In: Proceedings of 1991International Paper Physics Conference (TAPPD, Konna,Konna: TAPPI.

Foster, G.S. 1986. Making clonal forestry pay: breeding andselection for extreme genotypes. In: Proceedings of 1UFROConference, Joint Meeting of Working Parties on BreedingTheory, Progeny 7bsting, and Seed Orchards, 582-590. Raleigh,North Carolina: North Carolina State University

Foster, G.S. 1993. Selection and breeding for extremegenotypes. In: Clonal forestry I. Genetics and biotechnology,edited by M.R. Ahuja & WJ. Libby 51-67, Berlin: Springer.

Harper, J.L. 1977. Population biology of plants. London:Academic Press.

Heybroek, H.M. 1978. Primary considerations: multiplicationand genetic diversity Unasylva, 30, 27-33.

Heybroek, H.M. 1982. Monoculture versus mixture:interactions between susceptible and resistant trees in amixed stand. In: Resistance to diseases and pests in foresttrees, edited by H.M. Heybroek, B.R. Stephan & K. vonWeissenberg, 325-341. Wageningen: PUDOC.

Hohenboken, WD. 1985. Genetic structure of populations 2.Matings among distantly related individuals. In: World animalscience, A4. General and q-uantitative genetics, edited by A.B.Chapman, 251-274. Amsterdam: Elsevier,

Huehn, M. 1988. Multiclonal mixtures and number of clones.I. Number of clones and yield stability (deterministicapproach without competition). Silvae Genetica, 37, 67-73.

Kleinschmit, J. 1974. Use of vegetative propagation forplantation establishment and genetic improvement. Aprogramme of large-scale cutting propagation of Norwayspruce. New Zealand Journal of Forest Science, 4, 359-366.

Kroll, T.K., Moore, L.D. & Lacy, G.H. 1984. Interplantingsusceptible and resistant radish cultivars reduces colonizationby Plasmodiophora brassicae. Horticultural Science, 19, 403-404.

Leakey, R.R.B. 1987. Clonal forestry in the tropics - areview of developments, strategies and opportunities.Commonwealth Forestry Review 66, 61-75.

Lerner, I.M. 1954. Genetic homeostasis. London: Oliver andBoyd.

Libby, W.J. 1973. Domestication stiategies for forest trees.Canadian Journal of Fbrest Research, 3, 265-276.

Libby, W.J. 1982. What is a safe number of clones perplantation? In: Resistance to diseases and pests in forest trees,edited by H.M. Heybroek, B.R Stephan & K. von Weissenberg,342-360. Wageningen: PUDOC.

Libby, W.J. 1987. Testing for clonal forestry Anna lesForestales, 13, 69-75.

Lindgren, D., Libby, W.J. & Bondesson, FL. 1989.Deployment to plantations of numbers and proportions ofclones with special emphasis on maximizing gain at aconstant diversity Theoretical and Applied Genetics, 77,825-831.

Mahalovich, M.F. & Bridgwater, F.E. 1989. Modeling elitepopulations and positive assortative mating in recurrentselection programs for general combining ability In:Proceedings of 20th Southern Forest Tree ImprovementConference, Charleston, South Carolina, 43-49. Springfield,Virginia: National Technical Information Service.

Markovic, J. & Herpka, I. 1986. Plantations in shortrotations. In: Poplars and willows in Yugoslavia, 182-198. NoviSad: Fbplar Research Institute.

Marshall, D.R. 1977. The advantages and hazards ofgenetic homogeneity In: The genetic basis of epidemics inagriculture, edited by PR. Day, 1-20. New York: New YorkAcademy of Sciences.

Matheson, A.C. & Lindgren, D. 1985. Gains from theclonal and the clonal seed-orchard options compared for treebreeding programs. Theoretical and Applied Genetics, 71,242-249.

Matthews, F.R. & Rowan, S.J. 1972. An improved methodfor large-scale inoculations of pine and oak with Gronartiumfusiforme. Plant Disease Report, 56, 931-934.

McHea.nd, S.E., Foster, G.S. & Bridgwater, F.E. 1986.Breeding systems for pedigree-controlled productionpopulations of loblolly pine. In: Proceedings of Workshop onAdvanced Generation Breeding: Current Status and ResearchNeeds, 53-61. (Southern Cooperative Series Bulletin, no. 309.)Baton Rouge, Louisiana: Louisiana State University.

Russell, J.H. & Libby, W.J. 1986. Clonal testing efficiency:the trade-ofEs between clones tested and ramets per clone.Canadian Journal of Forest Research, 16, 925-930.

Shaw, D.V. & Hood, J.V. 1985. Maximizing gain per effortby using clonal replicates in genetic tests. Theoretical andApplied Genetics, 71, 392-399.

Shelbourne, C.I.A., Carson, M.J. & Wilcox, M.D. 1989.New techniques in the genetic improvement of Radiata pine.Commonwealth Forestry Review, 68, 191-201.

Teissier du Cros, E. 1984. Breeding strategies with poplarsin Europe. Forest Ecology and Management, 8, 23-39.

Trenbath, B.R. 1974. Biomass productivity of mixtures.Advances in Agronomy, 26, 177-210.

Wright, J.W. 1976. Introduction to forest genetics. London:Academic Press.

Zobel, B.J. & Talbert, J.T. 1984. Applied forest treeimprovement. New York: Wiley

Zobel, B., Campinhos, E. & Ikemori, Y. 1983. Selectingand breeding for desirable wood. 7'appi Journal, 66, 70-74.

111

Regeneration of new forest resources

Indigenous trees in West African forest plantations:the need for domestication by clonal techniques

G J LawsonOffice National de Dévéloppement des Foréts/Overseas Development Administration Forest Managementand Regeneration Project, BP 163, Mbalmayo, Cameroon •(On secondment from the Institute of Terrestrial Ecology*, Bush Estate, Penicuik, Midlothian EH26 OQB, UK)*A component of the Edinburgh Centre for Tropical Forests

ABSTRACT

Forests in West Africa have recently been disappearing at 1.2 million ha (4%) per annum, yet the area planted annually withtrees was only 36 000 ha in the mid-1980s, and is now very much less. Many tree planting schemes, particularly in Cameroon,have failed because of inadequate site clearance and subsequent neglect of maintenance. Yet the yield of several species inresearch plots has been found to be acceptably high, and a number of species, such as 7trminalia ivorensis, T superba,Triplochiton scleroxylon and Aucoumea klaineana, can achieve an average diameter of 60 cm in 25-30 years. The economicreturns from rotations of 30 years or less are highly attractive, even with poor markets for thinnings. Internal rates of returnbetween 9% and 20% have been calculated for such stands in Cameroon. However, private sector and farmer confidence in thewisdom of tee planting is only likely to be encouraged by: (i) financial incentive schemes; and (ii) the development oftechniques for selection and improvement of faster-growing and more valuable planting stock than has been available up to nowThere is a particular need for an international hardwood improvement programme in West and Central Africa.

INTRODUCTION

Whilst reasonable information is available fromWest Africa on the yield and profitability ofexotic species of Pinus, Tectona, Eucalyptus andGmehha (Allison et al. 1986; Home 1967; Lowe1976; Kio et al. 1989; Ball 1992), it is moredifficult to obtain reliable data on the nativehardwood species which are being'domesticated' and used in timber plantations.Three indigenous species have been plantedwidely in West Africa: Triplochiton scleroxylon(obeche, samba, wawa, ayous), Terminaliaivorensis (black afara, framiré), and Terminaliasuperba (white afara, fraké, limba). However,many of these plantations were established on asmall scale, and few analyses exist of theirestablishment costs, yields and likely profitabilityThis paper therefore:

• considers the need for tree plantations inWest Africa;

• summarises existing information on plantationyields of indigenous species;

• comments on the likely profitability of theseplantations;

• describes the efforts being made by theCameroon Forest Management and RegenerationProject to improve the management and yieldsof indigenous tree plantations; and

• speculates on the potential for vegetativepropagation and clonal selection to increasethe growth rates and marketability of nativehardwood species.

112

DEFORESTATION AND REFORESTATION INWEST AFRICA

The average annual rate of deforestation in WestAfrica from 1981 to 1985 (the most recentreliable data) was 2.2% overall, and 4.1% inclosed forest. These are the highest regionalfigures in the world, and compare with anaverage annual loss of closed forest of 0.6% inAfrica and a global average of 0.3% (Table 1).

Official statistics also indicate that 36 000 ha ofplantations were established annually in WestAfrica during the mid-1980s. This figure is verysmall compared with the 1.7 million ha of forestwhich were being lost annually at the time, andit is even less significant when one considers thepoor maintenance and survival of theseplantations, and the fact that they were usuallyestablished in areas of existing forest.Furthermore, there is little doubt that the rate ofplantation establishment has declinedconsiderably since the mid-1980s.

Thus, although West-Africa has recently sufferedthe greatest proportional loss of forests in theworld, it has an almost insignificant replantingprogramme. Considerable recognition exists ofthe economic, environmental and social costs ofunabated removal of forests. In Nigeria, forexample, it is calculated that the decrease in soilfertility and water quality caused by deforestationcosts more than US$5 billion annually (WorldBank 1992b). However, there is no evidence oftree planting programmes being supported on asufficiently large scale to relieve the pressure on

Table 1. Rates of deforestation and tree planting in West Africa (source: World Resources Institute 1992)

Conservation schemes

undegraded natural forests, neither is there muchconsensus on how forest regeneration is bestencouraged.

There are five approaches to forestrymanagement in the humid forest zone.

These schemes often require that logging ceases,and place severe constraints on villagers'activities. Governments may be granted loanremission, and villagers given new facilities, butconservation 'exclusion zones', which do notallow local populations to utilise timber or game,are unlikely to encourage a genuine commitmentto conservation among the community (Besong1992).

Sustainable management of the naturalforest

This ideal has many adherents, but fewgenuine practitioners. Techniques for theeffective management of natural forests arepoorly understood, and there are manymanagement and economic problems. Somerecent results, involving careful inventory andselection of trees to be felled, together withselective poisoning of competing low-valuetrees, have enhanced the growth of commercialspecies (by up to 6-7 m3 ha-1 yr-1) withoutexcessive environmental damage (Jonkers 1987;Maitre 1987). However, the cost of fellingcompeting trees is high, natural regeneration isoften unreliable, and the yield from evenwell-managed natural forest in West Africa isunlikely to exceed 5 m3 ha-1 yr-1 (Kio &Ekwebam 1987; Nwoboshi 1987), which is lessthan half the yield Of most plantation species inthe area.

Enrichment planting

This is the process of under-planting valuablespecies in partially cleared natural forest. It isappropriate if the forest has insufficient naturalregeneration for natural forest management tosucceed, but is too valuable (economically orecologically) to be completely replaced byplantations. It also offers the opportunity to usegenetically selected planting stock, or specieswhich were not present in the original forest.Under-planting can take place in lines, as scatteredgroups, or in a heavily cleared forest where onlymiddle-aged timber species of good form havebeen retained. Unfortunately enrichment plantinghas seldom been carried out well: light-demandingspecies have often been established in excessiveshade and weeding operations have usually beenneglected (Anon 1989).

Industrial plantation after completeclearance

This has been the principal technique used inlarge plantation schemes fmanced byinternational lending institutions. Typical examplesare the Subri industrial plantations in Ghana(Plumptre & Earl 1986), the World Bankafforestation programmes in Nigeria (Kio et al.1989), and the state forestry plantations in Côted'Ivoire (SODEFOR) (Dupuy 1985). Unfortunatelyall these plantations were established inreasonably productive forest reserves, partly togain revenue from the forest clearance, andpartly because this was the only land freelyavailable to the state. Development banks arenow less keen on this type of scheme, especiallyif the plantations are run by state forestryorganisations. There have been too manyexamples in the past of inefficient management,

113

over-emphasis on planting targets, and grossneglect of maintenance operations (World Bank1992a). The banks and aid agencies are,therefore, shifting the balance of their fundingaway from industrial forestry schemes (theforest-first approach') and towards environmentalconservation, natural forest management andcommunity forestry (World Bank 1991). However,it is in plantation programmes such as that ofSODEFOR that there are genetic improvementschemes for indigenous hardwoods (see Ladipoet al., pp239-248).

Community tree plantations

These have not been much practised in thehumid forest zone. They are likely to take placein conjunction with farming, either as 'taungya' orin mixture with fruit trees or other perennialcrops like coffee or cocoa, and therefore involvereasonably complete forest clearance. Degradedforest and derived savanna zones are mostappropriate. However, community plantationschemes have often been limited by anunenthusiastic uptake of tree planting byvillagers, particularly where it is difficult to defmethe ownership of trees on communal land (WorldBank 1992a; Flint 1992). These 'people-first'schemes have met many of their targets buthave not established large areas of forestbecause of their intrinsically small-scaleapproach, problems of land tenure, and a lack ofmotivation for farmers to plant tree crops whichbring little financial benefit for many years. Thus,it is extremely encouraging that the World Bankis about to launch a pilot 'tree planting incentivescheme' in Nigeria which envisages thereimbursement (after successful establishment) of75% of standard planting and maintenance coststo companies, communities and individuals whoplant at least 5 ha of land (one ha in cases ofenvironmental degradation). Individuals mustpossess the land title before funds can beprovided, but assistance 'will be given with theprocess of land registration (World Bank 1992b).

Large-scale afforestation in future is likely torequire a mixture of community and industrialplanting. Both approaches have been hindered inthe past by human, fiscal and institutionalproblems. This paper will not consider thesedifficulties further, but will concentrate on thetechnical questions which impede theintroduction of a significant indigenous hardwoodplanting programme in West Africa, namely:

• the growth rates of plantations of indigenousspecies;

• appropriate silvicultural methods;

• the likely economics of plantations; and

• improvements which can be gained from themultiplication and selection of geneticallysuperior planting stock.

114

YIELD OF PLANTATIONS

Information on the yield of plantations ofindigenous species in West and Central Africa issparse. Several hundred sample plots have beenestablished in research and commercialplantations, but have often been destroyed byfire, logging and/or neglect. Results are extremelyvariable, with differences in soil type, plantingstock, planting methods, disease attack andsubsequent thinning making it difficult to drawconclusions. Most plantation research has beendevoted to a small number of faster-growingspecies.

Terrninalia ivorensis

Much of the reliable information on T ivorensis(framiré) comes from Côte d'Ivoire, whereinfrequent measurements exist from 13 plotsinstalled by the colonial forest service, and moreregular data from 209 plots established since1965 by the Centre Technique Forestier Tropical(CTI-'1) (Beligne 1985). The results are heavilyinfluenced by initial shading, planting density andsubsequent thinning, with establishment andgrowth being more rapid in plots which hadreceived intensive manual or mechanicalclearance of the original forest cover. T ivorensisis a species which is light-demanding formaximum growth, and begins to suffer growthchecks even before the tree crowns touch (Lowe1974), indicating the need for a heavy thinningregime.

Before 1972, the C11-1 plots received only partialclearance of the original forest, either using the'line planting' technique (with 8-25 m betweenlines, sensu Aubreville 1953) or 'manual recru'(cutting the undergrowth to knee height, andpoisoning the overstorey - but leaving some ofthe large trees in place - which was notrecommended in Catinot's (1965) definition of therecru system). These plots have grown moreslowly than the later 'intensive plantations' andaverage 8.3 m3 ha-1 yr-1 by 30 years, with amaximum annual volume increment at 15 yearsof 9.7 m3 ha-1 yr-1. The minimum rotation lengthwith these 'extensive plantations', to reach anaverage diameter of 50 cm, is 43 years.

In 1972, C11-'1introduced more intensive forestclearance methods. These plots recorded amaximum annual diameter increment of 5.5 cm

- 1yr at year 3, and the average annual diameterincrement is projected to remain more than 2 cm

- 1yr until the age of 20 (Beligne 1985). Themaximum annual wood volume increment of 24.7m3 ha-1 yr- 1 was achieved at year 6, and the 70final crop trees alone were thincreasing eirvolume at more than 10 m3 ha-1 yr-1 by year10. Rotation lengths can be predicted from these'intensive plantations' of 20 years to 50 cmaverage diameter or 36 years to 60 cm diameter.

The average productivity of T ivorensis (framiré)in SODEFOR is reputed to be 11.5 m3 ha 1 yr 1,although there is some doubt about whether ithas been achieved in practice (Aitken et al.1992).

T ivorensis (idigbo) has also been widely plantedin Nigeria. Sanders (1953) and Horne (1962)quote annual diameter increments ranging from2.2 cm yr 1 for an 11-year-old plot to 1.9 for a30-year-old trial at Sapoba. No recent yieldinformation has been found except for anine-year-old plantation in Imo State which hasan average diameter at breast height (dbh) of

16.1 cm (Mbakwe 1990). The Nigerian Forestry

Sector Review (World Bank 1992b) gives theexpected yield of T ivorensis in plantations as10-12 m3 ha-1 yr- 1, which is the highest of the

species quoted.

T ivorensis (emire) was widely planted in Ghana

between 1930 and 1960, but yield informationdoes not appear to have been collated (WorldBank 1988). However, it is no longer regarded asa safe plantation species in Ghana because ofwidespread mortality observed in middle-ageplantations (Ofosu-Asiedu & Cannon 1976).

In Cameroon, T ivorensis (framiré) has beenused in both research and commercialplantations. The ex-CTFT 1972 researchplantations at Bilik (Mbalmayo Forest Reserve)have an average diameter in excess of 60 cmafter 20 years for the 70 largest trees perhectare (Lawson, Perem & Foahom 1993), despitedamage from a serious fire in 1983. Previousmeasurements had reported an average diameterincrement of 4 cm yr 1 during the first six years(Grison 1979) and 3 cm yr 1 over the first tenyears (Foahom 1982).

The Office National de Dévéloppement de Foréts(ONADEF) is responsible for 22 500 ha of

i. Tree height

70

60

50

40

30

20

10 e fkAT ttl/

00 10 20 30 40 50

moist-forest zone plantations in Cameroon, whichdate back to 1937 but were mainly plantedbetween 1972 and 1989. Unfortunately at least

40% of these plantations have completely failed,and much of the remainder has suffered severegrowth checks due to inadequate weeding, or toplanting in heavy shade. A recent provisionalinventory of these plantations (ONADEF 1992;Lawson & Ngeh 1993) has enabled tentativeestimates of yield to be derived. The 4120 ha of

T ivorensis plantations demonstrate a wide rangeof survival and growth rates (Figure 1).

Terminalia superba

Terminalia superba has been widely planted inCongo, where a clonal selection programme hasbeen implemented (Koyo 1983, 1985), and Zaire(limba), usually in association with plantains. It isalso an important plantation species in Cbted'Ivoire and Cameroon (take), but to a lesser

extent in Ghana and Nigeria (afara). Diametergrowth is usually 2-3.5 cm yr 1 up to year 10, andis maintained above 2 cm yr 1 in the final croptrees to at least year 20 (Groulez & Wood 1984).Height growth is around 1.1-1.5 m yr-1 during thefirst ten years. However, rates of volume productionmay be lower than those of T ivorensis, with 8.4

3 1 —1m ha -yr measured from an 18-year-oldplantation in Congo, and around 7.5 m3 ha-1 yr 1predicted at year 40. Yields are particularlyinfluenced by soil fertility In Côte d'Ivoire,experimental plantations are reported to yieldaround 10 m3 ha 1 yr 1 by year 15, but Aitken et

al. (1992) quote 12.3 m3 ha 1 yr 1 as the expectedyield of the SODEFOR commercial plantations, witha final diameter of 45 cm at 25 years.

Triplochiton scleroxylon

T scleroxylon comprises the majority of all woodexported from Ghana (wawa); it has top place in

Years since planting

ii. Tree diameter

70

60

50

40

30

20

10

o:

Plantation7.1Belabo

Bonepoupa::]EdjaghamniKienke.K;Kumba[1Makak• MbalmayoEMungoNOttotomo

10 20 30 40 50

Figure 1 'free height (i) and diameter (ii) of 118 plantations of Ternilhalia ivorensis (framiré), inventoried by the Office National de

livéloppement des Foréts in 1990 (ONADEF 1992; Lawson & Ngeh 1993). These data were collected by a large number of

teams, without adequate standardisation of methods, and should therefore be regarded as provisional

115

i. Tree height

116

60

50

40

30

20

10

10 20 30 40 50 0

Years since planting

Cameroon (ayous) and is a major component inwhat remains of the Nigerian (obeche) andIvoirian (samba) forests, It has been used as aplantation species in all these countries, butnever on a large scale, probably because of thedifficulty in obtaining reasonable quantities ofseed (Jones 1974).

In Ghana, research plots of T scleroxylon (ForestProducts Research Institute (FPRI) 1972) haveproduced high diameter increments, ranging from12 cm yr 1 in a ten-year-old trial at Pra Annumto 3.0 cm yr 1 in a six-year-old trial at Subri.Generally growth rates are in excess of 2 cm

-1yr .

In Nigeria, research plots have achieved 27.1 cmdbh by 29 years at Sapoba and 27.1 cm after31 years at Cambari (Abayombi & Nwaigbo1985). This growth rate does not seem rapid butthe 100 largest stems ha l are projected to growat an average 1 cm yr 1 over the first 60 years(Ball 1975).

Many more experimental plantations have beenestablished in Côte d'Ivoire, and annual diameterincrements average 1.2-2.1 cm during the first20 years for trials in partial shade, and 1,9-3.2cm in completely cleared areas (Dupuy &Bertault 1985). The older plantations had meanannual volume increments ranging from 4-7 m3ha -1 yr -1 for the sparse plantations establishedunder cover (100-150 stems ha 1), to 7-10 m3

—1 —1 •ha yr in the taungya plantations (250-350stems ha 1). The younger plantations establishedafter complete clearance averaged 13-18 m3

-ha -1 yr - 1(175-300 stems ha 1) after 13 years.

No extrapolations have been made from thesetrials to the final expected yield and rotation

ii. Tree diameter

60

50

40

30

PlantationBelaboBonepoupa

1lEdjaghamAKienke

KumbaMakakMbalmayo

llMungoZOttotomo

12 24 36 48 60

Figure 2. Tree height (i) and diameter (ii) of 112 plantations of Lovoa trichilioides (bibolo), inventoried by the Office National deDévéloppement des Forets in 1990 (ONADEF 1992; Lawson & Ngeh 1993). These data were collected by a large number ofteams, without adecpate standardisation of methods, and should therefore be regarded as provisional

length of plantations, but a yield of 12.3 m3 ha 1yr- is predicted for the SODEFOR commercialplantations, with a final diameter of 45 cm at 25years (Aitken et al. 1992). Initial gains of 14-29%have been reported with genetically selectedplanting stock used in SODEFOR plantations(Verhaegen et al. 1992).

At Makak in central Cameroon, two plots ofT scleroxylon were established in 1936-37 at aspacing of 5 m x 20 m. After 46 years, the treeswere on average 34 m high and 52.8 cmdiameter. At Mbalmayo, also in centralCameroon, five experimental T scleroxylon trialswere established between 1966 and 1969. Theserecorded a mean diameter increment of 1.7 cm

- 1yr during their first 15 years, and at rear 10the best of these trials achieved 9.5 m ha 1

- 1yr (Pesme 1986). A further trial ofT scleroxylon has been planted at SouthBakiindu (SW Cameroon), using rooted cuttingswithout genetic selection, and has achieved anaverage diameter of 14 cm after only six years(N Songwe, personal communication).

Lovoa trichilioides

Around 6200 ha of Loves trichilioicles have beenplanted in Cameroon (Lawson & Ngeh 1993), butthe average yield achieved is -particularlydisappointing (1.8 m 3 ha -1 yr -): partly becauseit was often planted in unfavourable areas, butmainly because of excessive shade and neglectof subsequent maintenance (Figure 2). In theresearch trials at South Bakundu, a diameterincrement of 1.8 cm yr -1 was achieved after 18years (P Shiembo, personal communication), andequally good growth has also been observed for

this valuable timber at Sapoba in Nigeria, whereit appeared to grow more rapidly after the polestage, and an average diameter increment of1.0-1.8 cm yr-1 was recorded after 23 years(Nwoboshi 1987).

Other species

Many other native species trials exist in thehumid forest zone of West and Central Africa.Nauclea diderrichii has been grown in Nigeriasince 1918, as a nurse species for more valuablemahoganies, and as a useful species in its ownright. Plantations in Sapoba, Ona and Agbodiachieved between 1.2 and 1.6 cm yr-1(Nwoboshi 1987), whilst a plantation at SouthBakundu has recently recorded 3.1 cm ha-1 yr-1up to year 8 (P Shiembo, personalcommunication).

Mansonia altissima has also grown reasonablysuccessfully at South Bakundu, with an averagemean diameter increment of 1.5 cm yr-1 at 18years (P Shiembo, personal communication),although it is rather susceptible to attacks fromdefoliating caterpillars (Foahom 1989). Picnanthusangolensis (ilornba) is fast-growing in some ofthe ONADEF plantations in Cameroon (averagemean diameter increment frequently more than 2

-1cm yr, , even to the age of 40), but has notbeen the subject of detailed study Aucoumeaklaineana (okoumé) has been planted sucessfullyin Ghana (FPRI 1974), and in Cameroon (Arnoux1958; Bibani 1989), with an average diameterincrement of 1.7 cm yr-1 after 12 years inresearch plots at Mangombé, and of 1.8-2.5 cm

-1yr in commercial plantations near Kribi(Lawson & Ngeh 1993).

Of the non-native hardwood species, Gmelinaarborea, Tectona grandis, Cedrela odorata andMelia composita have been the most successful,and should not be excluded from future plantingprogrammes, even if the preference is for nativespecies.

THE CAMEROON FOREST MANAGEMENTAND REGENERATION PROGRAMME (FMRP)

The FMRP is a bilateral Cameroonian/Britishprogramme with a general aim of improving thecapacity of Cameroon to carry out wisemanagement and conservation of its moisthigh-forest areas. It started in July 1991, with anemphasis on the use of improved indigenoushardwood species in artificial plantations, and hasthe following specific aims.

• To examine and demonstrate the suitability ofstrongly contrasted silvicultural techniques forforest regeneration after logging.

• To develop a management plan for theMbalmayo Forest Reserve.

• To provide genetically improved plantingstock of selected humid forest species.

• To acquire cost data for establishment andmanagement of plantations using differentsilvicultural techniques.

• To acquire data on biological changesresulting from different silvicultural treatments.

• To acquire socio-economic data on theeffects of different silvicultural treatments onlocal populations.

• To establish an in-service training capacity inhigh-forest management techniques and todisseminate results to the forest sector.

The FMRP's planting programme provides anopportunity to demonstrate and study a numberof silvicultural options for regenerating degradedforest. Currently these options consist of theestablishment of T ivorensis, T scleroxylon orL. trichilioides using the:

• manual regrowth method, with completemanual clearance of the original canopy andcutting of undergrowth at knee height (sensuCatinot 1965);

• enrichment planting method, with somewell-formed individuals of high-value speciesin the 20-50 cm diameter class retained toform part of the final harvest of theunder-planted fast-rotation species;

• mechanical complete clearance method, wherelarge trees are felled by chainsaw andpushed by bulldozer, together with smallertrees, into windrows at a spacing of 40-50 m;

• line plantihg method, where V-shaped linesare cut at wide spacing (determined by theultimate crown diameter of the planted trees)and existing trees are heavily cleared (sensuDawkins 1966);

• taungya method, where FIVERP undertakesforest clearance and burning, and thevillagers hoe, plant and cultivate crops for aninitial period of two to four years.

This study to compare silvicultural treatmentsfor the growth of indigenous hardwoods buildson preliminary work at the same site toinvestigate the effects of some site preparationtechniques on the physical and chemicalproperties of the soil (Ngeh 1989), the sporepopulations of endomycorrhizal fungi (Musoko1991; Mason, Musoko & Last 1992) and thephysiology of the planted trees (Eamus et al.1990). The preliminary work formed part of aprogramme linked to the vegetativepropagation of indigenous hardwoods and thedevelopment of an appropriate silviculturalsystem for planting clonal material (Wilson &Leakey 1990; Leakey 1991). The currentinvestigation includes a study to determine theeffects of silvicultural treatments on thepopulations of insects, especially the potentialpests and their predators and parasites (Watt &Stork 1992). The philosophy behind this project

117

is that, in order to maximise °the productivityfrom genetically selected planting stock, it isimportant to minimise risk by the retention ofbiological diversity in the plantation (Leakey etal. 1993).

PROFITABILITY OF INDIGENOUS TREEPLANTATIONS IN WEST AFRICA

Ideally forestry plantations should demonstrate ahigher return on capital investment than othercommercial alternatives, and/or a more rewardinguse of farmers' time and land than alternativecrops. There are complications, however. What isa realistic rate to discount future expenditure andrevernie? How will the market for timber andthinnings change in the future? Can one accountfor the risk of fire or disease? Should thebalance sheet include environmental conservationand secondary products like bush meat, wildfruits, medicines, rattan, and rafia? An initialattempt has been made at the macro-economicscale in Nigeria to estimate the annualenvironmental costs of forest clearance (WorldBank 1992b). The study concluded that Nigeriansustainable net national productivity has beenreduced annually by US$750 million due to losttimber production, $2 billion due to soildegradation, $1 billion due to watercontamination, $10 million because of the loss ofwildlife, $1 million due to the loss of fisheries,$100 million owing to gully erosion, and $150million through coastal erosion.

It is not easy to fmd good records of the cost ofplantation establishment and management in Westor Central Africa (Skoup & Co 1989; Djomo1987). Reasonable informatibn is often collectedfor short periods, but changes of personnel anda lack of standard accounting and codingsystems limit the continuity of data collection.

The FMRP aims to collect financial information inparallel to yield and environmental data, and hasmade an initial estimate of the profitability offour of the silvicultural systems which are beingexamined experimentally together with threeoptions under consideration (Table 2).Comparisons of profitability can be made usingthe internal rate of return (IRR) on capitalinvestment (the aggregate return of all revenuesand expenditures over the lifetime of aplantation). This measure emphasises thatgreatest profitability is achieved'when initial costsare minimised and incomes are generated asrapidly as possible (either from early commercialthinnings or by shortening rotation length -perhaps through genetic (clonal) selection).

All the figures derived in Table 2 are speculative:cost keeping has been unreliable; the densityand richness of the existing forest are averaged;the effect of silviculture on yields has not beenadequately described; and much uncertainty

118

'Francophone', sensu Aubreville (1953) (egGrand Layon), may be carried out wheresome of the existing high-value trees arepreserved to generate thinning revenue andcontribute to the final yield. This method isappropriate if the exisiting forest is relativelyrich, but management exclusively for naturalregeneration is not appropriate.

• 'Anglophone', sensu Dawkins (1966), isappropriate where it is carried out indegraded forests. This system aims toproduce a virtual monoculture of theplantation species (ie removing all treeswhich are taller than the arms of aright-angled 'V' centred on each of theplanting lines).

The mechanical complete clearance method isunlikely to be applicable where much forest coverremains. It involves the complete destruction ofexisting forest, and the scraping by bulldozer offelling debris into windrows. Planting operations arecheaper and easier after this type of groundpreparation, but large windrows may make 10-20%of the area unplantable. Edens (1992) calculated afavourable IRR (Table 2) based on optimistic yieldassumptions from early studies in Côte d'Ivoire(Dupuy 1986). However, more recent work theredoes not confirm that the SODEFOR plantationscreated using this method have achieved greateryields that those from older manual regrowthplantations (Aitken et al. 1992).

•

surrounds future timber prices. Nevertheless,estimated IRRs appear to be in the order:manual regrowth < mechanical regrowth< complete clearance < shade trees and cocoaline planting (sensu Aubreville 1953) < taungya< short-rotation exotics. The reliability of theseassumptions will improve as the FIVIRP gathersmore information. It is not yet possible toeliminate any of the silvicultural options oneconomic grounds as different options are moreappropriate in differing circumstances.

Short rotations of exotic species, like Gmelinaarborea, have high apparent IRRs, and couldpossibly be integrated with current fallowrotations. However, they are dependent onmarkets developing for small roundwood, whichwill take some time ina country like Cameroonbecause forest products are still plentiful, there islittle market for exotic species, and as yet thereis no operational pulp mill.

Taungya offers the advantages of reducedweeding costs and a possible early enhancementto growth as the trees respond to the nutrientpulse provided by wood ash. However, taungyasystems require a good relationship betweenforesters and villagers, and particularly strictcontrol against fires. Taungya is often held towork best when land is in short supply

Line planting

Table 2. Preliminary estimate of the profitability of different silvicultural options for the humid forest zone of Cameroon (source:Edens 1992)

Options

*MAI, mean annual increment: MD, man days

Assumptions Total costs Total(costs in Cameroon CFAs) (000 CFAs) revenue

la Short rotation and Gmelina arborea planted at 153fallow cycle (single 3 m x 3 m density,rotation) seedling cost 70, 10 days

ha-I for prep & planting,no thinning, survival 75%,8 yr rotation, 2-3 coppicerotations possible, MAI* 18m3 yr- I, 2 MD* weeding

ha-I yr-I, sawing efficiency22%, logging cost 30%,sawing & marketing cost40% on ex-mill price of61 000 1-11-3

lb Two rotations 10 days ha-I pruning &weeding at end of 1stcycle

2a Taungya (no 7brminalia superba plantedpayment to farmers) at 9 m x 9 m spacing,

delivered seedling cost100, supervision costs10 000 in yr 1, 5000 inyrs 2-5 & 2000 thereafter,average MAI 9 m3 yr-I,standing value 20 720 m-3,35-year rotation withthinning at yr 22, farmers'revenue not included, butweeding costs excluded

2b l'aungya (medium As above but with farmerpayment to farmers) payments of 25 000 for

planting and maintenanceof 5000 yr-I for 1st 8 years

3 Shade tree Cocoa which -yields from yr 10-50.Plantation on fieldassumed. 1st thinning at14, 2nd thinning at 25, 3rolthinning at 35 (to 50 treesha-I)

4a Manual recru (fully Planting germinaliamanual) ivorensis at 5 mx 5 m

spacing. MAI of 9 m3 yr-I,man days assumed (lI1 yr)are 51(yr1), 3(yr2), 7(yr3),7(yr4), 6(yr5), 6(yr6), & 1thereafter, thinning at yr 25

4b Manual recru Brushcutters reduce labour(semi-mechanical) input by 66%

5 Mechanical recru Caterpillar undertakesha-initial partial clearance @

30 000 ha (assuming 100ha minimum)

6 Line planting Lines planted at 10 x 5 m(grand layon) spacing with overstorey

largely cleared but 3valuable species retainedin 40-60 cm diam class, 3in 20-40 cm diam class &6 in 5-20 cm diam class.Thinning at 20 yrs with 10

3 •m of high-value species& 20 m'' of whitewood,final cut at 40 yrs gives360 m3 & 5 m3 ofmahogany

7 Complete clearance Mechanical clearance ofsite, windrows burnt andre-raked, total machinecosts 290 000 ha-I (basedon Ivory Coastexperience), intensiveweed control. Totalestablishment costs506 000 ha-I. MAI 12.5

3 -1 •m yr . Thinnings at yr 3,6, 9, final cut at yr 20

530

'RR(A)

18.6

208 106 20.9 20% lower yield gives 1RRof 17.7%

140 664 15.5 This spacing requires verygood planting stock9 m x 4.5 m spacingwould allow earlier thinning

221 6640 13.2 This payment may ensurebetter weeding

Approx 10

Comments

20% lower yield reducesIRR to 15.1%Market for small sawlogsexists in Nigeria, but notyet in Cameroon. Thisoption could fit into thenormal bush/fallow cycle

Cocoa needs early shade.Many thinning and farmercompensation options canbe costed

Removal of total overstoreybut maintenance ofcontrolled shrub layer

Labour costs are high inCameroon

Establishment costs arelower because of widespacing. Early thinningrevenue valuable

597 6453 10.3 ERR highly sensitive topreparation cost and yieldchanges. More risky option

119

Shade trees and perennial crops could be anattractive option for community forests, and mayalso provide a means of establishing treesamong illegal cocoa plantations in forestreserves. Shade trees are best above cocoa, butmoderate shade can also be tolerated by coffeeand a number of important crops like yams,cocoyams and pineapple. The growth incrementof widely spaced timber trees will beconsiderable, though regular pruning is requiredto maintain good form in species which are notnaturally self-pruning.

A modified form of the manual regrowth methodhas been the main method of forest regenerationused recently in Cameroon. Non-commercialspecies and all trees under 60 cm dbh werenormally felled to waste or poisoned. Plantingoperations were, therefore, extremelylabour-intensive, often involving up to 200 man

-1days ha-1 or £400 ha (Lawson 1993). Inpractice, resources were often not available forsufficient initial poisoning or subsequent weeding,and the majority of plantations have failed orbeen severely checked. Nevertheless, the manualregrowth method may remain the favoured optionin circumstances where salvage felling is carriedout immediately before planting, and where theenvironmental and economic risks of mechanicalclearance are not acceptable.

The estimated internal rates of return areencouraging, particularly for the short-rotation(18.6%), taungya (15.5%) and line planting systems(13.6%). However, the models include only modestassumptions for supervision and overheads. Beforethe dissolution of the previous state forestregeneration agency in Cameroon (ONAREF),plantation overheads were running at 350%, notincluding headquarters costs (Lawson 1993). Whilstthe new state regeneration organisation (ONADEF)may be more efficient than its predecessor, it islikely that funding agencies will in future favourprivate sector planting rather than supportingbureaucracy Unfortunately there are few spareresources in the private sector of most West andCentral African countries, and little industrial orfarmer confidence in tree planting. Even thefavourable IRRs given below are unlikely to besufficiently large, or free from risk, to encouragelarge-scale planting without the impetus of carefullydirected incentive schemes.

IMPROVEMENTS IN YIELD ANDPROFITABILITY THROUGH CLONALSELECTION

Ladipo et al. (pp239-248) have reviewed thedevelopment of forest tree improvementprogrammes in West and Central Africa overthe past 25 years. Considerable advances haveclearly been made with the development ofvegetative propagation techniques for importantplantation species, and some. native

120

multipurpose trees. Clonal seed orchards havebeen established in several countries, and effortsare being made to develop disease-resistantclones and juvenile tests which predict the likelybranching pattern of clones in later life.

Populus, Pinus and Eucalyptus lead the way indemonstrating yield improvements through theselection of superior clones: a good examplebeing the Aracruz Eucalyptus plantations inBrazil (Zobel, Van Wyk & Stahl 1983), wherethe 100 best clones have more than doubledthe production of unimproved stock. However,few clonal trials have yet been established andprotected to demonstrate these potential gainswith West African species. With Triplochitonscleroxylon, Leakey and Ladipo (1987) suggestthat selecting clones with above-averagevolume and stem form would improve yieldsby 31%. Choosing the top 10% would increaseyields by 81%. Research in clonal propagationhas been underway in Côte d'Ivoire since the1960s (Bonnet-Masimbert 1970), but the firstcontrolled comparisons of the yield and formof clonal versus unselected material wereestablished in 1987, and now cover some 24ha (Verhaegen et al. 1992). After only twoyears, the clonal material averaged 14%greater height increment, but theabove-average clones averaged a 40% heightadvantage. These apparent genetic gains aresimilar to those attained at the same age inNigeria (Ladipo et al., pp239-248).

In Cameroon, the tree improvement componentof the Forest Management and RegenerationProject has restored the Parc de Bouturagefacilities established under a previous WorldBank programme at a 7 ha site in theMbalmayo Forest Reserve. An area containing240 m2 of mist beds, 40 non-mist propagators(120 m2), coppice-bed areas, and a dripirrigation system for up to 4 ha of stockplantshas now been established. Approximately 0.4ha has been planted with 357 clones ofTriplochiton scleroxylon, 129 clones of Lovoatrichilioides and 24 clones of Terminaliaivorensis; 1700 T ivorensis plants have beenestablished as stockplants.

Following the ideas of Russell and Libby (1986),and the advice of Leakey et al. (1993), the F'MRPis adopting the following low-technology strategyfor clonal collection, selection and testing.

Collection of germplasm

Seed collection has concentrated on Triplochitonscleroxylon, and 16 provenances are nowrepresented, including two from Nigeria.Collections will be extended to other speciesand countries. In Cameroon, a bounty paymentto villagers is made for seed (from identifiedtrees) when it is delivered to forest offices. Inaddition, cuttings have been collected from the

stumps of 20 T scleroxylon plus-trees identifiedin plantations of different ages. Attempts havealso been made to locate shoots growing on thecut stumps of superior trees which were selectedby exploiters. However, the inventory informationcurrently kept by logging companies does notlocate stumps sufficiently accurately to allowthem to be relocated.

Nursery screening

The first stage of a clonal selection programmeis underway using a test which measures theresponse of seedlings to decapitation at 60 cmand the standardisation of leaf number to twoleaves. The number of lateral shoots produced instandardised conditions and the time required toreassert apical dominance appear to be wellrelated to the growth and form of the tree afterfour years (Ladipo, Leakey & Grace 1991a, b),and it is hoped that they will also be so in laterlife. This predictive test is being used to select2% of the 14 000 T scleroxylon seedlingscurrently in the nursery which will beestablished as stockplants and used in the fieldscreening phase.

Field screening

The FMRP's 1993 planting programme forT scleroxylon (50 ha) is being used as asecond-stage screening for clones identified inthe predictive test. A 14 ha 'research zone' willbe planted with 20 clones (from ten differentprovenances), replicated t‘wice in each of fivesilvicultural treatments. The 46 ha 'demonstrationzone' will also be used for clonal planting butnot on a replicated planting pattern. Thus, alarger number of clones, with fewer ramets perclone, will be screened for field performance inlines of 20 trees. The lines will be labelled atthe time of planting.

Field testing

This phase, which has not yet commenced, willinvolve replicated clonal plantings with those cloneswhich have performed well in the previous phases.Field testing requires a significant land area andneeds large numbers of ramets per clone in arandomised experimental design.

Clone to silviculture matching

Clones may respond differently to planting inpartial overhead shade (enrichment planting), infull sun (complete clearance), or where sideshade from herbaceous vegetation is encouraged(line planting, manual recru, taungya). The FMRPwill gain some information on possibleclone/silvicultural interactions from the statisticallydesigned planting in the 'research zone'.

Clonal demonstration lines

Small, and unreplicated, plantings are planned

close to the FMRP's headquarters to demonstrateto farmers and other visitors the differences ingrowth rate and stem form between selectedclones and unselected seedlings.

Germplasm bank

Clones and plants of seedling origin which arenot incorporated into the project's clonal plantingprogramme will, however, be established andmaintained as a gene bank.

Clonal seed orchards

A tree breeding programme is not planned atthis stage under FMRP However, in order topermit genetic recombination between selectedclones, it is the intention to establish a clonalseed orchard of main species (Terminaliaivorensis, Triplochiton scleroxylon andL. trichilioides) using five to ten ramets of eachof the most successful clones.

CONCLUSIONS

It is clear that good genetic material iscontinuously being lost as forests are selectivelycleared of the most valuable species andindividuals. Natural regeneration is ofteninadequate to sustain continuous production.Viable seed is difficult to obtain for somespecies, and of doubtful genetic quality forothers. However, considerable scope exists forextraction licences to be linked to regenerationactivities. It is hoped that, through the results ofFMRP logging companies will see thecommercial potential of establishing plantations ofdomesticated hardwoods, and distributingselected seed and improved clones to villagers.

Given the threat to the forest gene pool, thepromise of a renewed private plantingprogramme, and our ignorance of propagationtechniques and growth potential of manypotentially important species, it is more than evernecessary to co-ordinate national germplasmcollection and breeding efforts on• nativehardwood trees through a regional hardwoodsimprovement programme, similar to thatproposed by Leakey and Grison (1985). Thisprogramme should not be narrowly based ontwo or three fast-growing species, but shouldinclude a range of timber and multipurposespecies (eg Baillonella toxisperma, Garcinia kolaand Prunus africana).

ACKNOWLEDGEMENTS

The assistance of many members of theUK/Cameroon Forest Management andRegeneration Project are gratefully aclmowledged.However, the views expressed in this paper aresolely those of the author and are notnecessarily shared by the Overseas DevelopmentAdministration (UK) or the Office National deDévéloppement des Foréts (ONADEF).

121

REFERENCES

Abayombi, J.O. & Nwaigbo, L.C. 1985. Forest plantationyield studies in the high forest zones of Nigeria. lbadan: ForestResearch Institute of Nigeria.

Aitken, J., Donaldson, D., Gardiner, L., Harrison, M.,Massey, P., Sirrat, R. & Tout, G. 1992. Evaluation of theSODEFOR Forestry Project, Ivory Coast (1979-1990).(Evaluation report EV520.) London: Overseas DevelopmentAdministration.

Allison, C.E., Umeh, L.I., Omoluabi, A.C. & Fawusi,A.K. 1986. Handbook of forest plantation techniques forNigeria. Boadan: Federal Department of Forestry

Anon. 1989. Pre-project study report on enrichment plantingReport by Japanese Overseas Forestry Consultants & CentreTechnique Forestier nopical for the International TropicalTimber Organization. (Unpublished.)

Arnoux, M. 1958. Enrichissement en Okoumé de la regionde Kribi. Nkolbisson: Centre Technique Forester Tropical.(Unpublished.)

Aubreville, A. 1953. L'experience de l'enrichissement parlayons en Côte d'Ivoire. Bois et Forêts des Tropiques, 29,3-9.

Ball, J.B. 1975. Re-assessment of the potential of obeche inplantations. lbadan: Food and Agriculture Organisation/FederalDepartment of Forests. (Unpublished.)

Ball, J.B. 1992. Forest plantations and the wise managementof tropical forests. In: Wise management of tropical forests,edited by F.R. Miller & KL. Adam, 97-110. Oxford: OxfordForestry Institute.

Beligne, V. 1985. Croissance et productivité du framire(Terminalia ivorensis) en plantations. Abidjan: Centre deRécherche Forestière, Centre Technique Forester Ropical.(Unpublished.)

Besong, J.B.B. 1992. Co-existence of forest management forforest production and wildlife. In: Wise management of tropicalforests, edited by ER. Miller & KL. Adam, 19-26. Oxford:Oxford Forestry Institute.

Bibani, M. 1989. Silviculture de l'okoume (Aucoumeaklaineana). Nkolbisson: Centre de Recherche Forestiere, CentreTechnique Forestier Tropical. (Unpublished.)

Bonnet-Masirnbert, M. 1970. Essai samba. Types de plants,Mopri 1968. Abidjan: Centre de Recherche Forestière, CentreTechnique Forester Tropical. (Unpublished.)

Catinot, R. 1965. Sylviculture tropicale en forét denseafricaine. Bois et Forêts des Tropiques, 100, 5-18, 101, 3-16,102, 3-16, 103, 3-16, 104, 17-30.

Dawkins, H.C. 1966. 'Line planting: a definition andprinciples for the technique on tropical high-forest land.(Silvinote, no. 60.) Oxford: Oxford Forestry Institute.

Djomo, L. 1987. Etude du coat de plantation sylvicole enzone de foret et en zone de savane dans les ChantiersONAREE (Rapport de Stage.) Gabon: L'Université OmarBongo, Ecole Nationale des Eaux et Foréts.

Dupuy, B. 1985. Principales régles de sylviculture pour lesplantations a vocation bois d'oeuvre. Abidjan: Centre deRecherche Forestière, Centre Technique Forestier Tropical.(Unpublished.)

Dupuy, B. & Bertault, J.G. 1985. Reg les sylvicoles pour lesamba en plantation. Abidjan: Centre de Recherche Forestière,Centre Technique Forestier Tropical. (Unpublished.)

Eamus, D., Lawson, G.J., Leakey, R.R.B. & Mason, P.A.1990. Enrichment planting in the Cameroon moist deciduousforest: microclimatic and physiological effects. In: Proceedingsof the 19th IUFRO World Congress, Montreal, Canada, 258-270.Montreal: IUFRO.

122

Edens, J. 1992. The economic viability of different forestrytreatments to be established under the Cameroon ForestManagement and Regeneration Project, together with suggestedagroforestry treatments. (FMRP Socio-economic survey part B.)London: Overseas Development Institute. (Unpublished.)

Flint, M., ed. 1992. Forestry synthesis evaluation study(Evaluation report, EV541.) London: Overseas DevelopmentInstitute. (Unpublished.)

Foahom, B. 1982. Etude du comporternent general du fiamire(Terrninalia ivorensis) a Bilik Nkolbisson: Centre de RechercheForestière, Centre Technique Forestier Tropical. (Unpublished.)

Foahom, B. 1989. Sylviculture de bete, (Mansonia altissima).Nkolbisson: Centre de Recherche Forestère, Centre TechniqueForestier Tropical. (Unpublished.)

Forest Products Research Institute. 1972. Annual report(1971-72). Kumasi: FPRI.

Grison, F. 1979. Note sur les plantations de framiré (T.ivorensis) en forêt semi-decidue (Station de Bilik). Nkolbisson:Office National de la Recherche Forestière/IRAE

Groulez, J. & Wood, P.J. 1984. Terminalia superbamonographie. Nogent-sur-Marne: Centre Technique ForestierTI-opical/Oxford: Commonwealth Forestry Institute.

Horne, J.E.M. 1962. Growth rates in the timber plantations ofWestern Nigeria. (Nigerian Forestry Information Bulletin (NS),no. 12.) lbadan: Federal Department of Forestry

Horne, J.E.M. 1967. 7bak in Nigeria. (Nigeria ForestryInformation Bulletin (NS), no. 16.) Ibadan: Federal Departmentof Forestry

Jones, N. 1974. Records and comments regarding theflowering of Triplochiton scleroxylon K. Schum. CommonwealthForestry Review, 53, 52-56.

Jonkers, W.B.J. 1987. Vegetation structure, logging damageand silviculture in a tropical rain forest in Suriname.Wageningen: Department of Silviculture, AgricUltural University

Kio, P.R.O. & Ekwebam, S.A. 1987. In: Natural managementof tropical moist forests, edited by F Mergen & J.R. Vincent,150-176. Newhaven, Connecticut: Yale University Press.

Kio, P.R.O., Abayomi, 1.0. & Okorie, P.E. 1990. Growthand productivity of Gmelina arborea in West Africa. Ibadan:Forestry Research Institute of Nigeria. (Unpublished.)

Kio, P.R.O., Ladipo, D.O., Ola-Adams, B.A. & Oguntala,A.B.O. 1989. Forest resources development and managementin Nigeria. Washington, DC: World Bank. (Unpublished.)

Koyo, J.P. 1983. Le bouturage au service des plantationsameliorés de 7brminalia superba en République Populaire duCongo. Silvicultura, 8, 331-333.

Koyo, J.P. 1985. Current status of genetic improvement workwith 7brminalia superba in the People's Republic of Congo. In:Provenance and genetic improvement strategies in tropicalforest trees, edited by R.D. Barnes & GL. Gibson, 547-552.Oxford: Commonwealth Forestry Institute/Harare: ForestResearch Centre.

Ladipo, D.O., Leakey, R.R.B. & Grace, J. 1991a. Clonalvariation in apical dominance of Trip loch/ton scleroxylon inresponse to decapitation. Silvae Genetica, 40, 135-140.

Ladipo, D.O., Leakey, R.R.B. & Grace, J. 199 lb. Clonalvariation in a four-year-old plantation of Triplochitonscleroxylon K. Schum. and its relation to the predictive test forbranching habit. Silvae Genetica, 40, 130-135.

Lawson, G.J. 1993. Yield and profitability of tree plantationsin southern Cameroon. In: The future of ONADEF's plantations,84-103. Yaoundé: Office National de Dévéloppement desForéts.

Lawson, G.J. & Ngeh, R. 1993. Analysis of the ONADEF

1990 plantation evaluation. (Forest Management andRegeneration Project R&D Paper, no. 2.) Mbalmayo: OfficeNational de Développement des Forets.

Lawson, G.J., Perem, A. & Foahom, B. 1993. Re-evaluationof a 1972 framire plantation in the Mbahnayo Forest Reserve.(Forest Management and Regeneration Project R&D Paper,no 3.) Mbalmayo: Office National de Développement desFortis.

Leakey, R.R.B. 1991. Clonal forestry: towards a stidtegy

Some guidelines based on experience with tropical trees. In:Tree breeding and improvement, edited by J.E. Jackson, 27-42.Tring: Royal Forestry Society

Leakey, R.R.B. & Grison, F. 1985. Regional programme forthe improvement of tropical hardwoods - West and CentralAfrica. Preparatory phase report to the European DevelopmentFund, CECINERC/CW1'. Penicuik, Midlothian: Institute ofTerrestrial Ecology

Leakey, R.R.B. & Ladipo, D.O. 1987. Selection forimproved tropical hardwoods. In: Improving vegetativelypropagated crops, edited by A.J. Abbott & R.K. Aitken,229-242. London: Academic Press.

Leakey, R.R.B., Wilson, J., Newton, A.C., Watt, A.D. &Deans, J.D. 1993. Indigenous topical tree species: clonalapproaches and maintenance of biodiversity as keys tosustainable development. In: Proceedings of the PlanningMeeting of the Commonwealth Science Council BiologicalDiversity and Genetic Resources Network. London:Commonwealth Science Council. In press.

Lowe, R.G. 1974. Some effects of stand density on thegrowth of individual trees of several plantation species inNigeria. PhD thesis, University of lbadan.

Lowe, R.G. 1976. Teak (7bctona grandis) thinning experimentin Nigeria. Commonwealth Forestry Review, SS, 189-202.

Lowe, R.G. 1977. Development of taungya in Nigeria. In:Agroforestry: realities, possibilities and potentials, edited byH.L. Gholz. Dordrecht: Nijhoff.

Maitre, H.F. 1987. Natural high forest management in Côted'Ivoire. Unasylva, 39, 53-60.

Mason, P.A., Musoko, M.O. & Last, F.T. 1992. Short-termchanges in vesicular-arbuscular mycorrhizal spore populationsin 7brininalia plantations in Cameroon. In: Mycorrhizas inecosystems, edited by D.J. Read, D.H. Lewis, A.H. Fitter &I.J. Alexander, 261-267. Wallingford: CAB International.

Mbakwe, R.C. 1990. A note on the yield of 7brminaliaivorensis in Umuagwo, Nigeria. Journal of Tropical Forest

Science, 3, 88-89.

Musoko, M.O. 1991. The ecology of endomycorrhizas insome Cameroon forests with respect to species of Terminalia.PhD thesis, University of Edinburgh.

Ngeh, C.P. 1989. Effects of land clearing methods on atropical forest ecosystem and the growth of Terminaliaivorensis (A. Chev). PhD thesis, University of Edinburgh.

Nwoboshi, L.C. 1987. Regeneration success of naturalmanagement, enrichment planting and plantations of nativespecies in West Africa. In: Natural management of tropicalmoist forests, edited by F Mergen & J.R Vincent, 72-91.Newhaven, Connecticut: Yale University Press.

Ofosu-Asiedu, A. & Cannon, P. 1976. 7brminalia ivorensisdecline in Ghana. PANS, 22, 239-242.

Office National de Dévéloppement des Forêts. 1992.Evaluation des plantations des reserves forestières. Yaounde:ONADEF

Pesme, X. 1986. L'ayous (Triplochiton scleroxylon) enplantation. Nkolbisson. Centre de Recherche.Forestière, CentreTechnique Forester nopical. (Unpublished.)

Plurnptre, R.A. & Earl, D.A. 1986. Integrating smallindustries with management of tropical forest for improvedutilisation productivity Journal of World Forest ResourceManagement, 2, 43-55.

Russell, J.H. & Libby, W.J. 1986. Clonal testing efficiency:trade-offs between clones tested and ramets per clone.Canadian Journal of Forest Research, 16, 925-930.

Sanders, M. 1953. Provisional volume table for T ivorensis.(Nigeria Information Bulletin, no. 6.) lbadan: FederalDepartment of Forestry

Skoup & Co Ltd. 1989. Report on the forest revenue systems studyfor Nigena. lbadan: FORMECU, Federal Department of Forestry

Verhaegen, D., Kadio, A., Boutin, B., Delaunay, J. &Legaré, D. 1992. La samba. Selection phenotypique d'arbres'+' et production industrielle de boutures en C6te d'Ivoire.

Bois et Forêts des Tropiques, 234 (4), 13-27.

Watt, A.D. & Stork, N. 1992. Ecology of insects inCameroon plantation forests. (Second Consultancy report toUK Overseas Development Administration/GOC Office Nationalde Dévéloppement des Forets.) Mbalmayo: ONADEF, ForestManagement and Regeneration Project.

Wilson, J. & Leakey, R.R.B. 1990. Repairing the damage:re-establishing trees in the tropics. Crown Agents Review, 1,14-20.

World Bank. 1988. Ghana Forest Resource ManagementProject. (World Bank report 7295-GH.) Washington, DC: WorldBank.

World Bank. 1991. Forestry - the World Bank's experience.Washington, DC: World Bank, Evaluation Department.

World Bank. 1992a. Project completion report - Cameroon.(Forestry Project loan 2092-CM.) (World Bank Report10476.FS.) Washington, DC: World Bank.

World Bank. 1992b. Forestry sector review, Federal Republicof Nigeria. (WB 10744-UNI.) Washington, DC: World Bank.

World Resources Institute. 1992. World resources 1992-93.Oxford: Oxford University Press.

Zobel, B.J., Van Wyk, G. & Stahl, P. 1983. Growing exoticforests. New York: Wiley

123

The role of tree domestication in agroforestry

F L Sinclair 1, I Verinumbe 2& J B Hall'

1School of Agricultural and Forest Sciences, University of Wales, Bangor; Gwynedd LL57 2UW UK2Department of Biological Sciences, University of Maiduguri, PO Box 1069, Maiduguri, Nigeria

ABSTRACTAgroforestry is both an interclisciplinary approach to land use and a set of integrated land use practices. The maintenance anduse of a broad genetic base have been a central theme in the development of agroforestry in terms of utilising undomesticatedspecies for productive purposes and of conserving biodiversity through the use of agroforestry practices in the protection ofnatural habitat.

The importance of domestication depends upon both the extent to which undomesticated species are utffised and the extent towhich their use is controlled. Despite the development of data bases and directories of multipurpose trees, it is not easy toappraise the current status of domestication. While there are ubiquitous reports of localised tree utilisation, it is often unclearwhether they refer to the exploitation of wild vegetation or controlled husbandry of the resource.

More than a thousand multipurpose tree species are cited as being utilised, but there are only about 25 species for which thereis sufficient knowledge to support promotion at farm level. Despite this vast potential for new domestication, most agroforestryresearch and development projects concentrate on a few fast-growing exotic tree species, grown in regular arrangements. Thesetechnologies do not enjoy high rates of adoption among resoume-poor farmers. In contrast, traditional agroforestry practices ofteninvolve spatially and floristically complex mixtures of indigenous species.

There is a need to collect and collate basic information on the propagation, ecology silvicultural management and cropcompatibility of more of the thousand or so species that are currently utilised locally but which are not well documented. Farmersshould be encouraged and supported in their efforts to grow and manage the tree species that they value locally where theyconsider appropriate within their agricultural systems.

INTRODUCTION

Agroforestry is a relatively new subject forscientific study but a traditional practice with along history in many parts of the tropics (King1968; Nair 1989). Existing reviews (Huxley 1983a,b, 1985; Burley Huxley & Owino 1984; Burley1987) and practically oriented texts (Wood &Burley 1991) discuss general aspects ofmultipurpose tree improvement, and other papersin this volume consider the progress being madein this area with specific examples (see Simons,MacQueen & Stewart, pp91-102). This paperconcentrates on the use of woody perennialspecies within agricultural practice and theconsequences of their genetic improvement forthe interactive role of agroforestry trees. Thisfocus is appropriate because the multipurposerole of agroforestry trees distinguishes them andtheir improvement from general forestry activity Itis essential for effective progress in agroforestrydevelopment that these aspects are broughttogether and combined with the local experienceof farmers.

Examples of two widespread but contrastingagroforestry practices are used to illustrate thegeneral points made in the text. The first ofthese practices is the use of dispersed trees incrop fields, a system which is particularly

124

important in the drylands of Africa, but hasparallels in other continents. The second is themultilayered homegarden, which occursworldwide. Both these systems have beenreviewed and put in context by Von Maydell(1987) and Soemarwoto (1987), respectively Forconvenience, the word tree' is used in this paperto refer to all woody perennials, rather thanbeing restricted to those of a certain stature.

DEFINITION OF AGROFORESTRY

Agroforestry has been defined as the set of landuse practices which involve the deliberatecombination of woody perennials and herbaceouscrops andJor animals on the same landmanagement unit, in some form of spatialarrangement or temporal sequence, such thatthere are significant ecological and economicinteractions between woody and non-woodycomponents (adapted from Lundgren 1987).Although the prevailing definition now in use, itdoes not distinguish between agroforestry as aninterdisciplinary approach to land use and as aset of integrated land use practices, and, whileimplying consideration of agronomic aspects atthe field scale, it is not explicit about theconsideration of the impacts of agroforestrypractices on whole-farm systems and their

aggregated effects at landscape and regionalscales. The agroforestry approach combines thestudy of woody perennials, herbaceous plants,livestock and people, and their interactions withone another in farming and forest systems. Itembraces an ecosystem focus considering thestability sustainability and equitability of land usesystems, in addition to their productivity (seeConway 1987; Marten 1988). Consideration ofsocial as well as ecological and economicaspects is implied.

AGROFORESTRY AND BIODIVERSITY

The maintenance and use of a broad geneticbase have been a central theme in thedevelopment of agroforestry in terms of utilisingundomesticated species for productive purposes(Pickersgill 1983; Burley 1987; Prance 1990a) andof conserving biodiversity through the use ofagroforestry practices in the protection of naturalhabitat (Oldfield 1988; Prance 1990b). While ithas been suggested that for undomesticatedtrees with potential for exploitation the bestplace for conservation is in cultivation'(Simmonds 1985), agroforestry may involve areduction in genetic diversity when comparedwith the natural vegetation it may replace. Intropical rainforest, up to 300 species of tree andliana have been reported from one hectare(reviewed in Prance 1989, 1990b). Althoughbetween 70% and 100% of rainforest treespecies may be utilised by local people (Balée1986, 1987; Broom 1990), 20-30 is the typicalnumber of tree species where they aredeliberately cultivated in forest gardens ororchards (Padoch & De Jong 1987; Perera &Rajapakse 1991). The large number of treespecies utilised should not be confused with themuch smaller number cultivated. While humanselection in cultivation may modify genotypes indesirable ways, extraction of wild resources mayresult in negative selection pressure for desirabletraits.

Although multilayered garden systems are clearlymuch more diverse in terms of number ofspecies of higher plants than either conventionalagriculture or plantation forestry muchagroforestry research and development hasinvolved mixtures of only two species, with theintroduction of a single tree or shrub speciesinto cereal cropping and the woody componentoften spatially confmed, as exemplified inhedgerow intercropping (Kang & Wilson 1987).

DOMESTICATION OF NEW SPECIES

Analyses of the vegetation structure anddynamics of natural ecosystems (Oldeman 1983),and the indigenous Imow ledge of localpractitioners (Walker, Sinclair & Muetzelfeldt1991) might be expected to be principal sourcesof information about tree species and how they

might be utilised in agroforestry The importanceof making progress in the domestication ofagroforestry trees depends upon the extent towhich undomesticated species are currentlyufilised, the extent to which their use iscontrolled, and the potential for genetic gains.Despite the development of data bases anddirectories of multipurpose trees, it is not easy toappraise the current status of domestication.While there are ubiquitous reports of localisedtree utilisation, it is often unclear whether theserefer to the exploitation of wild vegetation or tocontrolled husbandry of the resource.

It is logical to distinguish, as far as possible,utilisation, protection and management. Theacquisition of materials from trees without regardto their survival is no more than utilisation.Protection advances the process by involvingmore restrained intervention with the consciousintent that the individual plant should survive andrecover sufficiently to allow re-exploitation at afuture date, including measures intended toenhance natural regeneration. Management, incontrast, is manipulative rather than opportunistic,by actively promoting regeneration and plantestablishment.

Domestication implies that the managementprocess influences the genotypes being utilised,but, for agroforestry tree species mentioned here,only the most preliminary steps towardsdomestication have been taken, and with aminimal number of species.

THE EXTENT OF LOCAL DOMESTICATION

Attention has recently focused on the indigenousmanagement of forest and woodland resourcesand individual trees on farms (see reviews byMathias-Mundy et al. 1991; Prance 1991;Shepherd 1992).

Dry land Africa is a convenient case to illustratethe sort of documentation available of indigenousmanagement activity made possible byShepherd's (1992) analysis of more than 100bibliographic summaries. There are ubiquitousexamples of utilisation of a wide range ofspecies, and some reports of protection, such asselective pollarding of valuable trees by theTurkana in northern Kenya (Barrow 1988). Raison(1988) reports that in Mandara, Cameroon, a shiftfrom stand to individual tree management occursas parkland agroforestry practices are intensified.

Protection of naturally regenerated seedlings fromgrazing, even during the dry season, has beenreported from parkland agroforestry practices inBurkina Faso (Marchal 1983), but there are veryfew documented examples of indigenous treeplanting. It has been suggested that deliberateplanting of trees has not constituted a significantpart of the development of various parklandagroforestry systems in the Sudano-sahelian zone

125

of Africa (Raison 1988), which have developedby selective cutting of natural forest cover andmanagement of natural regeneration (World Bank1985). Barrow (1988) suggests that the impactsof increasing human and animal populations arenot being incorporated into indigenous resourcemanagement strategies by the Turkana, indicatingthat, despite a need for more interventionistmanagement, it has not developed spontaneously

In a global context, the extent of undocumentedindigenous knowledge of the ecology of treespecies which are utilised by local people isthought to be significant, as evidenced by theburgeoning interest in ethnoecology (Prance1991). This information requires novel methods tobe adequately appraised; such methods are onlynow being developed and applied (Walker et al.1991).

AGROFORESTRY TREE SPECIES

Present concern is with the members of a newgeneration of tropical multipurpose trees. Over1000 multipurpose tree species are listed by VonCarlowitz, Wolf and Kemperman (1991), butpresent knowledge of the vast majority is toosuperficial to enable their routine management asresources. In terms of the tropics, for fewer than50 of these species is there sufficient knowledgeto support promotion at farm level, half of thembeing familiar because of long usage inconventional forestry or horticulture. The newgeneration of species represents the remainder ofthis small group. Only about 25 species areinvolved (Table 1), all of which either have been

Table I. Tropical multipurpose trees with a comprehensive knowledge base generated by agroforestry interests

Species

Acacia mearnsiiA. niloticaA. senegalA. tortilisAlbizia lebbeckAlnus acuminataBalanites aegyptiacaBombacopsis quinataCalliandra calothyrsusCasuarina equisetifoliaErythrina poeppigianaFaid.herbia albidaGliricidia sepiumGreyillea robustaGuazuma ulmifoliaLeucaena leucocephalaMimosa scabrellaParkia biglobosaProsopis cinerariaP chilensisP tamarugoSesbania grandilloraS. sesbanTreculia africanaVitellaria paradoxaZiziphus mauritiana

126

neglected by forestry and horticulture or are nowmore prominent in agroforestry roles than inother, longer-standing roles. It is only during thelast 20-25 years that enough information hasbeen amassed to justify including them amongoptions for extensive use as planted resources inagroforestry systems. In many cases, efforts topublicise their value and potential have beenreflected in the appearance of monographic orother broad treatments bringing together existingknowledge (Table 1). For the other specieslisted, the principal geographical region of use isindicated, within which published documentationis accessible and widely known and would allowpreparation of unified accounts, although nonehas yet appeared.

Consideration of the details of collating andinterpreting information on additionalmultipurpose tree species is outside the scope ofthis account, but it is relevant to consider howsuch species should be identified. In this processthe impressive advances in international liaisonco-ordinated by the Food and AgricultureOrganisation (FAO) through the Panel of Expertson Forest Gene Resources should be noted. ThePanel drew up a priority list of tree species forimprovement and conservation in 1968 (FAO1969), and at subsequent meetings reviewed andupdated it. The list, worldwide in scope, hasgradually absorbed multipurpose trees since theoriginal version appeared. For example, the initiallist omitted all West African savanna trees exceptnaya senegalensis. Subsequently multipurposetrees were added to the lists, including Acacia

Noteworthy references/Geographical region where native

Sherry (1971)Fagg & Greaves (1990b)Cheema & Qadir (1973); Philip (1975)Fagg & Greaves (1990a)IndiaCentral AmericaHall & Walker (1991)Central AmericaNational Research Council (1983)Midge ley et al (1983); National Research Council (1984a)Nitrogen Fixing Tree Association (1993)Felker (1978); Centre Technique Forestier nopical (1988)Withington, Glover & Brewbaker (1987)Harwood (1989)Central AmericaNational Research Council (1984b); Lulandala & Hall (1991)BrazilBonkoungou (1986)Mann & Saxena (1980)Simpson (1977)FAO (1981); Habit (1985)Nitrogen Fixing Tree Association (1990a, b)Nitrogen Fixing 'free Association (1990a, b)Bijtterbier (1981)Bonkoungou (1987)India

nilotica, A senegal, A tortilis, Dalbergiamelanoxylon, Faidherbia albida and Vitellariaparadoxa (FAO 1977), Parkia biglobosa (FAO1984) and Acacia seyal (FAO 1988). Additionalspecies may be expected to be included in thefuture, such as Balanites aegyptiaca and Ziziphusmauritiana and members of the Capparidaceae.

THE ROLE OF AGROFORESTRY SPECIESIN RELATION TO TREE IMPROVEMENT

Agroforestry trees have often been consideredsynonymous with multipurpose trees (IVIVI) andthe concept of simultaneous production ofdifferent above-ground parts, such as fruit, leavesfor fodder, stems and branches for wood (Huxley1985; Burley 1987; Wood & Burley 1991).However, for the few agroforestry species forwhich there have been selection and breedingprogrammes, there has been a tendency toselect divergent genotypes for different purposes.A notable example is Leucaena leucocephala,where a shrubby heavily branching form hasbeen bred for fodder as opposed to arborealtypes for wood production (Brewbaker 1987).Similar strategies are being actively applied forProsopis species (Felker, pp 183-188) and Acaciaauriculiformis (Booth & Ranbull, pp189-194). Thisevidence suggests that, as domesticationprogresses, the multipurpose nature ofagroforestry trees will be replaced bysingle-purpose genotypes selected for yield ofpriority parts of the plant along conventionalforestry lines, although alternative approacheshave been proposed (Simons et al., pp91-102).The logical outcome of selecting forsingle-purpose genotypes is to negate themultipurpose nature of the species.

This scenario ignores the fact that agroforestrytrees are part of an agricultural system; they mayhave important roles that contribute to sustainabilityparticularly in relation to soil protection andimprovement, and will compete with agriculturalcrops for light, water and nutrients. In this context,agroforestry trees are always multipurpose,regardless of what is harvested and extracted fromthe system. Consequently the farmer is presentedwith a series of trade-offs determined by theallocation of plant biomass to different parts of thetree, which is the primary aspect manipulatedduring selection and breeding. These trade-offs arediscussed below What is clear is that theinteractive behaviour of the tree within theagricultural system represents an obligatorymultipurpose character of agroforestry genotypes,even if they are selected with only a singleharvestable output in mind. The significantinteractive roles of such trees need to be explicitin the development of selection criteria, and thespecification of the associative ideotypes proposedby Huxley (1985). At present, it is difficult to beexplicit about many of the selection criteria

required because of a lack of knowledge aboutfunctional relationships below-ground.

HARVESTED BIOMASS VERSUS RECYCLING

The economic reasons for incorporating trees infarming systems are many, often representing aneed to produce material on-farm that waspreviously available from a now-dwindling forestresource. A major reason for the internationalprominence of agroforestry is the perceivedability of trees to contribUte to the sustainabilityof staple food production in low-input farmingsystems.

Yield improvements through the breeding ofagricultural crops have been achieved largely byincreasing the harvest index, and secondarily byincreased adaptation to agronomic environments(Carmen 1985). The fundamental efficiency of theuse of absorbed radiation is conservative acrossspecies and environments, and there is noevidence that it has been altered throughdomestication (Russell, Jarvis & Monteith 1989).Increasing the harvest index of agroforestry treesmay lead to higher returns in the short termfrom the harvested product, but may causeproblems of system sustainability in the longerterm, if less plant biomass is recycled in the soil(Young 1989). Because fine root and mycorrhizalturnovers have been calculated to contribute twoto four times more N and six to ten times moreP and K than above-ground litterfall in treeplantations (Bowen 1984), a reduction in rootbiomass in favour of the harvested shoot couldhave signifiOant effects on the soil nutrient statusin agroforestry systems.

SOIL AMELIORATION VERSUSCOMPETITIVE EFFECTS

Farmers may often face a trade-off betweenlonger-term benefits of trees, such as soilstructural amelioration (Van Noordwijk, Widianto& Kurniatun-Hairiah 1991), and immediatecompetition with the crop for resources.Threshold levels of staple crop production maybe a high priority and it may be desirable toselect trees as much in relation to their effect oncompanion crops as to their productivity Thisselection can be achieved by evaluatinggenotypes on-farm rather than on-station (seeClement & Villachica, pp230-238), whichrepresents a shift from conventional forestryapproaches and has obvious cost implications forselection and breeding programmes.

SHOOT VERSUS ROOT AND THEIMPORTANCE OF NICHEDIFFERENTIATION

An important ecological basis for yieldadvantages through the intimate integration oftrees with agricultural crops in agroforestry

127

practices, as opposed to growing trees and cropsseparately is that the combined system willutilise environmental resources more efficientlythan monocultures through niche differentiation inspace and time. It has been demonstrated bycontrolled experimentation that, if twonon-interbreeding populations are grown togetherat high density then natural selection may favoura divergence in behaviour, such that eachpopulation ceases to make such heavy demandson resources needed by the other (Harper1977). Marked differentiation within a singleTrifolium (Legurninosae) species has been foundover small distances, with a variety of formseach adapted to live with different associatedgrass species (Ilirkington 1975). The strikingtemporal niche differentiation of the Faidherbiaalbida canopy which is leafless during the rainyseason when intercropped cereals are developingtheir canopy is readily observed and well known(Felker 1978). However, much agroforestry ispredicated on an assumption of nichedifferentiation below-ground.

The extent to which this differentiation actuallyoccurs in agroforestry is not clear. Tree rootsystems typically have lower root length densitiesin upper soil layers than herbaceous crops, aswell as greater distances between roots anddeeper root penetration (Bowen 1985).Examination of vertical distributions of fine rootsof several tree species used in agroforestry inMorogorro, l'anzania (including Cassia siamea,Leucaena leucocephala and Prosopis chilensis),however, showed them to be similar to that ofmaize (Jonsson et al. 1988). Ong et al (1991) haveshown strong competition for water and nutrientsbetween Leucaena leucocephala and millet inhedgerow intercropping on alfisols in semi-aridtropical conditions.

Bowen (1985) has suggested that theproduction and turnover of fine roots andmycorrhizas are general adaptations of treeroot systems that confer flexibility permittingphenotypic adaptation to a wide range of siteconditions. Soil conditions may consequently bemore significant in determining root distributionat a particular site than the actual speciesplanted. For example, Hall and Walker (1991),in reviewing literature on Balanites aegyptiaca,found a wide range of architectural patterns ofthe gross root system from studies at differentsites. Agricultural intercrops are likely to altersoil conditions through resource consumptionand, therefore, to affect tree root distribution.There are few comparative data on tree rootdistributions in relation to different intercrops. Insilvopastoral experiments, the allocation ofbiomass to roots in Prunus avium increased,and the vertical distribution of roots in Fraxinusexcelsior was downwardly displaced, as a result

128

of competition with aggressive swards (Campbell& Mackie- Dawson 1991; Tomlinson 1992).Similarly the density of pigeon pea roots wasreduced in the top 15 cm of soil when grownwith sorghum in India (unpublished data ofY S Chauhan, cited in Ong et al. 1991). Theseobservations indicate a response to reducedwater and nutrient availability in upper soil layersas a result of resource consumption bycompanion crops, and suggest that mixtures oftrees and crops may exhibit more extensive nichedifferentiation than expected from their rootdistributions in monoculture.

There is a danger that selection based onaltering the partitioning of carbon towardsparticular above-ground organs could result inlower partitioning to roots, and that this mayinfluence the root competition between the treeand the crop. Conversely there is an opportunityfor selecting trees specifically to reduce rootlength density in upper soil layers and toincrease rooting below 30 cm, thereby selectinggenotypes likely to exhibit enhanced ecologicalcombining ability with staple crops.

ABOVE- AND BELOW-GROUNDCONNECTIVITY

Selection of genotypes in an agroforestry contextis further complicated by interactions betweencanopy and root systems which may beinfluenced by silvicultural manipulations such aspruning, and by effects of past development oftree root systems on subsequent rooting patterns.The possibilities for manipulating canopyarchitecture through pruning have beendiscussed by Cannell (1983), and the importanceof phenology and entrainment in themanagement of multipurpose trees have beenstressed by Huxley (1983c, 1985).

By way of example, Peltophorum pterocarpumwas identified, through screening variousindigenous species, as suitable for alley croppingon acid soils in the humid tropics in Sumatra,Indonesia (Van Noordwijk et al. 1991). The taproot and horizontal 'sinkers' developed frombranch roots some distance from the stem, andpenetrated to below 1 m into a zone of ironconcretion, but lower pruning height resulted inmore, smaller-diameter branch roots in the top10 cm of soil.

P pterocarpum also provides a convenientexample of the importance of successionalphases in agroforestry when planted intoImperata cylindrica grassland (Webb et al. 1984).The tree develops a deep root system inresponse to competition from the grass which itthen shades out, leaving what may be acomplementary rooting pattern for subsequentintercropping with cereals. Selection criteria for asuitable tree in this case might involve:

• deep rooting combined with a strongdownward displacement response in verticalroot distribution to competition in surface soil- suitable in the first phase; and

• a reduced surface root proliferation responseto shoot pruning - suitable in the secondintercropping phase.

Clearly, the seleclion requirements for woodyspecies that will be optimal through severalintercropping phases are complex, and trade-offsin the functionality of the genotype in differentstages are likely

BASIC DATA FOR MODELLINGAPPROACHES

The way in which trees interact with crops andanimals in given environmental contexts is centralto their suitability as agroforestry species. Keyprocesses that determine yield in combinationsof trees and crops (eg light and rainfallinterception, nutrient and water uptake) andmeasurable tree variables (eg leaf area and rootlength density) can be combined inphysiologically based models to investigate theeffect of incorporating a range of treefrequencies in various arrangements on farmland(reviewed in Muetzelfeldt & Sinclair 1993). Recentdevelopments in sensing equipment permiteasier and less destructive measurement of leafarea in individual trees (Lang & McMurtrie 1992)and of root structure and dynamics (Atkinson &Mac lde-Dawson 1990). There is, therefore, anopportunity to collect basic information on manynew agroforestry species, by makingnon-destructive measurements through seasonalcycles. This collection could be done for speciescurrently retained at low frequencies on farmland,and the data then used to investigate the effectsof higher tree frequencies and the impact ofpossible changes in tree form and phenologythat might help in specifying appropriateassociative ideotypes.

COMPARATIVE EVALUATION OFAGROFORESTRY TREES IN DRYLANDAFRICA: A CASE STUDY

This section presents growth data of agroforestrytree species grown in cultivated field situationson vertisolic soil in dryland Africa, and willillustrate and develop some of the general pointsmade earlier. The experimentation is reportedmore fully elsewhere (Verinumbe et al. 1992).

Context

As population pressure increases throughout thesudano-sahelian zone of Africa, forcing adecrease in the length of the bush/fallow cycleand the progressive removal of trees from thelandscape, farmers in the region perceive adecrease in soil fertility and a reduction in water

available for clop growth. These conditions resultin lower crop yields and people having to walkfurther to collect firewood, fodder and fruit.Agroforestry practices involving the planting andretention of trees on crop land have beenproposed as simple solutions to these problems,building on a long tradition of farmers retaininglow densities of trees (10-30 ha-1) on farm land(Nye & Greenland 1960). Current reseamhexplores the hypothesis that the turnover ofbiomass from trees, planted at stocking densities inthe order of ten-fold higher than those traditionallyretained on crop land, can improve soil structure,leading to increased infiltration and penetration ofwater and subsequent improvement in theproductivity and sustainability of food crop and neeproduction. This rests on the premise, investigatedhere, that locally used tree species can beestablished in cultivated fields and can realisesubstantive growth when intercropped.

Location

The New Marte experimental site (12°15'N,13°52'E) is situated in north-eastern Nigeriabetween the town of Maiduguri and Lake Chadat the northern limit of the Sahel transition zone(White 1983). The soils are fme-textured chromicvertisols derived from alluvial and lacustrinedeposits. Mean annual rainfall is just under 650mm; mean monthly precipitation exceeds 50 mmonly for the period June-September.

Objectives

The experimental objectives involved thequantitative appraisal of the establishment andgrowth of five locally used tree species identifiedin a survey of traditional agroforestry practicesand markets in the region (Verinumbe 1991).These five species, Acacia nilotica, A. seyal, A.senegal, Balanites aegyptiaca and Prosopis juliflora,were evaluated over five successive years ofplanting, in relation to varying annual rainfallpattern and soil conditions, and an assessmentwas made of the impact of intercropping on treeestablishment and performance.

Experimental design

The experimental design incorporated treatments of:

• five tree species, including a control with notrees;

• five successive years of planting; and

• plus or minus intercropping.

The trees were planted in a randomisedsplit-split plot design, with double replication atthe lowest level that allowed sufficient residualdegrees of freedom to make comparisonsbetween species (Figure 1 i). The trees weredispersed throughout cultivated plots at astocking rate of 400 trees ha-1.

129

iii. Tree survival (%)species and year effects removed

Figure 1. Plans of the experimental site at New Marte

i. Layout of experimental plots

ii. Spatial variation of sand content across the site

iii. Spatial variation of tree survival across the site measured in 1991 after effects of species and year of planting have beenremoved

iv Spatial variation of total cross-sectional area of stem measured in 1991 after effects of species and ear of planting have beenremoved

130

i. Layout of experimental plots

1 Control plots (no trees)2 Acacia nilotica3 Acacia seyal4 Acacia senegal5 Balanites aegyptiaca6 Prosopis juliflora

Plots intercropped with sorghum

> 9990-99

[ _I 81-9072-81

Mt' 63-7254-6345-5436-4527-3618-27

11111 < 9

I I

Block II Year Block I

Other information:Size of plots = 25 m x 25 mNo. of trees per plot = 25No. of plots per sp. per yr = 16No. of trees per sp. per yr = 400Total no. of trees per yr = 2000Total no. of trees planted = 10000

> 9488-94

I 183-8877-83

rtr,- 72-7766-7261-66

r 155-6150-5544-50

139-44< 39

ii. Sand content (% dry wt soil)

> 3330-3327-3024-27

Ai 21-2418-2115-1812-159-126-93-6< 3

iv. Mean total cross-sectional areaadjusted for species and year

"z

Data were analysed using an angular (forsurvival data) or logarithmic transformation to

equalise the variance. Results are presented as

back-transformed means, with effective standard

errors, where appropriate. The variability of

surface soil properties was obtained fromanalysis of surface (0-10 cm) soil samples (200

g), collected by auger over the experimentalarea. Two samples were taken in each plot, one

direct/y under the tree canopy and one mid-way

between adjacent trees. While a full set of

analyses have been completed on these samples,

only data on particle size (measured using a

Micrometrics X-ray sedigraph machine) are

presented here. Quadratic interpolation was used

to map tree and soil data spatially (Figure 1), to

identify systematic trends and localised variationat the site.

Results and discussion

A distinct change is apparent from clay-rich soils

in the south-west (SW) to sand-rich in the

north-east (NE) (Figure I ii), reflecting the varying

content of lake-deposited material tosuperimposed and intercalated wind-blown sand.

Blocks 1 and 2, with a higher clay content at the

surface, were subject to more severe seasonal

ponding and cracking and, therefore, longer

periods of waterlogging in the wet season. Other

soil parameters show similar trends; for example,

pH, measured in water, varies from <5 in the

SW to >7.5 in the NE (median=6.3), indicating

the . introduction of calcareous material associated

with the wind-blown sand. glee survival (Figure

I iii), plotted across the site, after species and

year effects were removed, follows the trend in

soil texture, with high survival in Blocks 3 and 4,

where soil was lighter-textured at the surface,

and poor establishment in the clay-rich part ofthe site. Observations at the site suggest that

survival was most affected by a few particularly

severe flooding events, which occurred in the

wetter years and were aggravated by the soil

conditions in the southern blocks. The total

cross-sectional area of tree stems (TCSA), when

effects of year and species were removed, did

not vary appreciably across the site (Figure 1 iv),

suggesting that, while the degree of waterlogging

was critical for the survival of newly planted

seedlings, it did not affect subsequent tree

growth.

The species fall into three groups in terms of

their TCSA after four years' growth (Figure 2i).

Prosopis juliflora and Acacia nilotica had asignificantly greater TCSA than the other species,

and both A. seyal and A. senegal showedsignificantly faster above-ground growth than

Balanites aegyptiaca. P juliflora grew faster initially

being larger than all other species at both two

and three years after planting. A. nilotica was

larger than all other species except P juliflora

300

0å-)

ca"6

c.)

50

150

100

0

0

1 2

2 .3Age (yrs)

3 4

4

Figure 2. Comparative growth of five tree species planted overfour years at New Marte. The graphs show back-transformed

mean values and effective standard errors of measurementsmade on the nine central trees in each plot of (i) total

cross-sectional area of stem (measured as the sum of thethree largest stems per tree); (ii) height; and (iii) mean crown

diameter

three years after planting, and not significantly

different from P julifora after four years. Tree

height (Figure 2i1) followed a similar pattern, but

the mean crown diameter of A. nilotica remained

significantly less than P juliflora, even at four

years after planting, causing less interference

with the sorghum intercrop (Figure There

was no significant effect of intercropping on tree

survival or growth.

While the means and effective standard errors

calculated on a transformed scale are

appropriate for making species comparisons, the

raw means and standard deviations (liable 2)

reflect high intraspecific variability of tree growth

131

Table 2. Within-plot raw mean and standard deviation (SD) of total cross-sectional area of stems (TCSA), height and mean crowndiameter (MCD) by species for five agroforestry tree species grown in cultivated plots at New Marte, NE Nigeria, planted in1987 and measured in 1991 (data from Verinumbe et al. 1992). For full species names, see text

within plots, which is relevant to tree planting incultivated fields by farmers. The coefficient ofvariation (CV) of stem cross-sectional area atfour years ranged from 60% to 75% for mostspecies, but was roughly twice as high for B.aegyptiaca than the other species. Height andcrown diameter were less variable (CVs typically25-40%, although approaching double this rangefor B. aegyptiaca).

Despite faster initial growth, P juliflora was lesssuitable for use in fields than A. nilotica becauseit had a more spreading crown, which interferedwith the intercrop. B. aegyptiaca was by far theslowest-growing species in the experiment, but iscommonly protected and managed by farmers incultivated fields. While the reason may be largelythat the tree produces highly valued fruit, even inlow-rainfall years, a high root/shoot partitioningratio in early years of growth may be associatedwith ecological combining ability in intercroppedcereals. Research now in progress at New Marte(Sinclair et al. 1992) involves detailedmeasurement of water uptake and the growth oftrees and intercropped sorghum, and will allowcomparison of the tree species in relation tostaple crop yield and water use efficiency ofindividual components and the system as awhole.

There is a marked contrast between theslower-growing fruit and gum-producing speciesused in fields by farmers and the faster-growing,often exotic, species that have been the mainfocus of agroforestry research and developmentefforts. Much of the basis for expectations ofsustainability benefits from trees on crop landhas been associated with tree foliage biomass, ofconsiderable nutrient content, being recycled inthe system. High above-ground biomassproduction was suggested by Young (1989) as afundamental characteristic of soil-improving trees.There is clearly little scope with theslow-growing fruit trees favoured by farmers insemi-arid conditions: foliage was never more than5% of total tree biomass in B. aegyptiaca treesranging from 10 kg to 600 kg total biomass,measured in Senegal and Burkina Faso (Bil le1978, 1980).

Agroforestry practices involving the slow-growing,deep-rooting, drought-resistant species that are

132

valued locally for their fruit and thefaster-growing legumes may be appropriate forthe region, and would combine the acceptedeconomic value of the fruit with the service rolesof carbon and nitrogen addition to the soil.Mixtures of species would also reduce the riskof loss of production from trees, in relation toboth pest and disease attack and drought.

The high intraspecific variability recordedsuggests that there may be considerable scopefor improvement through selection and breedingin terms of above-ground growth, although spatialvariability in soil conditions and otherenvironmental factors could be expected toaccount for more than half of the phenotypicvariation measured. Farmers typically retain orplant small numbers of trees in their fields tominimise competition with intercrops, leaving littlescope for selection after tree establishment;therefore, the reliable performance of individualsis particularly important.

One means of increasing the reliability ofperformance is to maximise the suitability of theplanting site. This has often been achieved byimproving the soil by cultivation and the additionof inorganic fertilizers to produce more uniformfield conditions; it requires substantial inputs ofcapital and support energy More recently it hasbeen suggested that traditional farming methodsmay be effective in the utilisation ofheterogeneous resource bases, such as that inFigure 1, because mixtures of species andgenotypes are used in such a way that differentgenotypes are matched to microsite conditions(Anderson & Sinclair 1993). Differences in soilconditions, and higher crop productivity havefrequently been reported under crowns ofFaidherbia albida trees growing in crop fieldsacross its natural range, and in northern Nigeriain particular (Verinumbe 1991). As with muchsoils research in agroforestry however, soil hasusually been sampled at one point in time fromnearby sites with different tree cover, so that it isunclear to what extent trees have improved soilconditions by their presence or have grown inpatches of better soil in the first place (Sanchez1987). If farmers are, in fact, able to optimiseuse of a spatially variable soil system, then theymust be able to:

• recognise and characterise different microsites;

• know the micro-environmental requirements ofdifferent species; and

• match species to microsites.

While there are many examples from variousparts of the world of detailed local classificationsof soil, vegetation and site type, and of their

correspondence with scientific categories(Brokensha, Warren & Werner 1980; Richards1985), there is less documented evidencerelating to indigenous knowledge of species'requirements. One of the most frequentlyrepeated claims in the agroforestry literature is

that the seemingly random arrangement andcomposition of multilayered homegardens inAfrica, Asia and Central America are, in fact, the

end result of a complex process of matchingspecies to microsites by farmers (Michon 1983;Fernandes, O'kting'ati & Maghembe 1984; Jacob& Al les 1987). No evidence supporting theseassertions has been presented, but research inprogress (Walker et al. 1991) addresses this type

of question directly Farmers are asked about the

ecological interactions occurring in theagroforestry practices that they operate, and their

answers are then formally represented incomputerised knowledge bases that aretransparent and can be interrogated. Recentfieldwork involving interviews with farmers inKandy District, Sri Lanka, yielded over 3000factual statements about species' interactions inforest gardens (Walker & Southern 1992). Theresulting knowledge base contains information of

ecological relevance to 180 species, with asignificant amount of information for about 15 of

them (about 200 statements on each).

CONCLUSIONS

There is a need to collect and collate basicinformation on the propagation, ecologysilvicultural management and crop compatibilityof more of the 1000 or so agroforestry speciesthat are currently utilised locally but are not well

documented. Priorities have been identifiedinformally by farmers in what they have selected

to grow and on a formal international basis bythe FAO. Interdisciplinary co-operation is required

to capture and combine knowledge from localpeople, the existing ecological literature, andexperimentation complementing the collection,

evaluation and conservation of germplasm withthe development of agroforestry systems thatextend rather than supplant local practice.Farmers should be encouraged to grow and

manage the tree species they value locallywhere they consider appropriate, within theiragricultural systems. Recent developments insensing equipment for analyses of canopy androot structures of individual trees should enableinteractive processes in agroforestry to be

explained in terms of measurable tree variables.

Modelling efforts based on this approach couldenable the investdgation of the likely effect of

different tree frequencies on system performance

from basic data that could be readily collected

for a wide range of new species.The extension of technology packages has

proved less effective in agroforestry development

than incremental improvements building on local

practice (Buck 1990; Bunderson et al. 1990).

Fundamental research that establishes yield

responses to parameter changes has been

recognised as complementary to adaptive

research trials comparing alternative technology

designs on-farm (Scherr 1990). However, the

design phase in the well-known agroforestry

diagnosis and design procedures developed at

the International Centre for Research inAgroforestry (ICRAF) (Raintree 1990) has not

been justified by field experience. This casts

doubt on the common assertion that there is a

primary requirement in agroforestry development

for adaptive research, leading to rapid extension

of technology packages that can be extended to

farmers (Beer, Borel & Bonnemann 1990; Van den

Hoek & Bekkering 1990), rather than a need to

support farmers in making decisions about

incorporating trees within their farming systems,

based on the results of more fundamental

research. In the context of the domestication of

new species, the priority may turn out to be the

collection of basic information on the canopy and

root systems of the many trees currently used by

farmers, to provide better advice on the likely

impact of introducing or increasing their density

on farmland, rather than concentrating on the

development, on-station, of a necessarily small

number of improved genotypes.

ACKNOWLEDGEMENTS

The research projects referred to on comparative

evaluation of agroforestry trees in dryland Africa

(Projects R4181, R4850 and R4858) and on the

formal representation of indigenous knowledge

(Projects R4594 and R4731) are funded by the

UK Overseas Development Administration (ODA),

P Adder ley and D Jenkins were responsible for

the analysis of soil samples from New Marte and

produced the maps in Figure 1, at the University

of Wales, Bangor. The paper was written while

the principal author was collaborating with L G

Andeison of the Institute of Terrestrial Ecology

Bangor, and R Muetzelfeldt of the Institute of

Ecology and Resource Management, University of

Edinburgh, on a review of ecological interactions

and modelling in agroforestry commissioned by

the ODA, and this work has influenced the

content of the paper.

REFERENCESAnderson L.S. & Sinclair F.L. 1993. Ecological interactions

in agroforestry systems. Agroforestry Abstracts, 6, 57-91.

133

Atkinson, D. & Mackie-Dawson, L.A. 1990. Root growth:methods of measurement. In: Soil analysis: physical methods,edited by K.A. Smith & C. Mullins, 447-509. New York: Dekker,

Ba lée, W. 1986. Andlise preliminar de inventório florestal e aeinobotånica Ka'apor (Maran-hão). Boletin do Museo ParaenseEmilio Goeldi, Ser Botånica, 2, 141-167.

Balée, W. 1987. A etnobotånica quantitativa dos indios Tembe(Rio Gurupi, ParA). Boletin do Museo Paraense Emilio Goeldi,Ser Botånica, 3 (1), 29-50.

Barrow, E. 1988. Trees and pastoralists: the case of the Pokotand Turkana. (ODI Social Forestry Network Papers, no. 6b.)London: Overseas Development Institute.

Beer, J., Borel, R. & Bonnemann, A. 1990. On-farmagroforestry research planning in Costa Rica. In: Planning foragroforestry edited by WW Budd, I. Duchhart, L.H. Hardesty& F Steiner, 58-78. Arnsteiclam: Elsevier.

Bille, J.C. 1978. Role des arbres et arbustes en tant quesources de proteines dans la gestion des pâturages d'Afriquetropicale. In: Proceedings of the Eighth World ForestryCongress, Jakarta, 1978, vol. 3, 1369-1384. Jakarta: WorldForestry Congress Organizing Committee.

Bille, J.C. 1980. Measuring the primary palatable productionof browse plants. In: Browse in Africa: the current state ofknowledge, edited by H.N. Le Houérou, 185-195. AddisAbaba: International Livestock Centre for Africa.

Bijtterbier, J. 1981. Agroforestry and fight against penury offood in the Third World. Communication presented at the 17thIIJFRO World Congress, Kyoto, Japan, 1981.

Bonkoungou, E.G. 1986. Etude mongraphigue du néré:Par lda biglobosa (lacg.) Benth. (Leguminosae: Mimosoideae).Ouagadougou: Centre National de la Recherche Scientifiqueet Technologique.

Bonkoungou, E.G. 1987. Monographie du karit6(Butyrospermum paradoxum (Gaertn. t)) Hepper, espéceagroforestiere å usages multiples. Ouagadougou: Institut deRecherche en Biologie et Ecologie Tropicale, Centre Nationalde la Recherche Scientifique et Technologique.

Bowen, G.D. 1984. rIl-ee roots and the use of soil nutrients.In: Nutrition of plantation forests, edited by CD. Bowen &E.KS. Nambiar, 147-179. London: Academic Press.

Bowen, G.D. 1985. Roots as a component of treeproductivity In: Attnbutes of trees as crop plants, edited byM.G.R. Cannell & J.E. Jackson, 303-315. Abbots Ripton,Huntingdon: Institute of Terrestrial Ecology

Brewbaker, J.L. 1987. Leucaena: a multipurpose tree genusfor tropical agroforestry In: Agroforestry: a decade ofdevelopment, edited by H.A. Steppler & PK.R. Nair, 289-323.Nairobi: International Centre for Research in Agroforestry

Brokensha, D., Warren, D.M. & Werner, 0. 1980.Indigenous knowledge systems and development. Lanharn:University Press of America.

Broom, B. 1990. Useful plants of the Panare Indians ofVenezuelan Guayana. Advances in Economic Botany, 8, 41-56.

Buck, L. 1990. Planning agroforestry extension projects, theCare International approach in Kenya. In: Planning foragroforestry, edited by WW Budd, I. Duchhart, L.H. Hardesty& F Steiner, 101-131. Amsterdam: Elsevier.

Bunderson, T.W., Wakeel, A. el, Saad, Z. & Hashim, I.1990. Agroforestry practices and potentials in western Sudan.In: Planning for agroforestry edited by WW Budd, I. Duchhart,L.H. Hardesty & F Steiner, 227-246. Amsterdam: Elsevier.

Burley J. 1987. Exploitation of the potential of multipurposetrees and shrubs in agroforestry In: Agroforestry: a decade ofdevelopment, edited by HA. Steppler & PK.R. Nair, 273-287.Nairobi: International Centre for Research in Agroforestry

134

Burley, J., Huxley, P.A. & Owino, F. 1984. Design, managementand assessment of species, provenance and breeding trials ofmultipurpose trees. In: Provenance and genetic improvementstrategies in tropical forest trees, edited by RD Barnes & G.L.Gibson, 70-80. Oxford: Commonwealth Forestry Institute.

Campbell, CD. & Mackie-Dawson, L.A. 1991.Below-ground competition between trees and grass in asilvopastoral system. Agroforestry in the UK, 2 (3), 39-43.

Cannell, M.G.R. 1983. Plant management in agroforestry:manipulation of trees, population densities and mixtures oftrees and herbaceous crops. In: Plant research andagroforestry, edited by PA. Huxley 455-487. Nairobi:International Centre for Research in Agroforestry

Cannell, M.G.R. 1985. Dry matter partitioning in tree crops.In: Attributes of trees as crop plants, edited by M.G.R Cannell& J.E. Jackson, 160-193. Abbots Ripton, Huntingdon: Institute ofTerrestrial Ecology

Centre Technique Forestier Tropical. 1988. Faidherbiaalbida: monographie. Nogent-sur-Marne: C11-1.

Cheema, M.S.Z.A. & Qadir, S.A. 1973. Autecology ofAcacia senegal (L.) Willd. Vegetatio, 27, 131-162.

Conway, G.R. 1987. The properties of agroecosystems.Agricultural Systems, 24, 95-117.

Fagg, C.W. & Greaves, A. 1990a. Acacia tortilis. '(AnnotatedBibliography F41.) Farnham Royal: CommonwealthAgricultural Bureaux.

Fagg, C.W. & Greaves, A. 1990b. Acacia nilotica. (AnnotatedBibliography F42.) Farnham Royal: Commonwealth AgriculturalBureaux.

Felker, P. 1978. State-of-the-art: Acacia albida as acomplementary permanent intercrop with annual crops.Riverside: University of California.

Fernandes, E.C.M., O'kting'ati, A. & Maghembe, J.1984. The Chagga homegardens: a multistoreyed agroforestrycropping system on Mount Kilimanjaro (northern Tanzania).Agroforestry Systems, 2, 73-86.

Food and Agriculture Organisation. 1969. Report of thefirst session of the FAO Panel of Experts on Forest GeneResources. Rome: FAO.

Food and Agriculture Organisation. 1977. Report of thefourth session of the FAO Panel of Experts on Forest GeneResources, Rome: FAO.

Food and Agriculture Organisation. 1981. Prosopistamarugo: fodder tree for arid zones. FAO Plant Productionand Protection Paper 25, 1-110.

Food and Agriculture Organisation. 1984. Report of thefifth session of the FAO Panel of Experts on Forest GeneResources. Rome: FAO.

Food and Agriculture Organisation. 1988. Report of thesixth session of the FAO Panel of Experts on Forest GeneResources, Rome: FAO.

Habit, M.A. 1985. The current state of knowledge onProsopis tamarugo. Rome: Food and Agriculture Organisation.

Hall, J.B. & Walker, D.H. 1991. Balanites aegyptiaca: amonograph. Bangor: School of Agricultural and ForestSciences, University of Wales.

Harper, J.L. 1977. Population biology of plants. London:Academic Press.

Harwood, C.E. 1989. Grevillea robusta: an annotated bibliographyNairobi: International Centre for Research in Agroforestry

Huxley, P.A. 1983a. Some characteristics of trees to beconsidered in agroforestry In: Plant research and agroforestryedited by PA. Htudey 3-12. Nairobi: International Centre forResearch in Agroforestry

Huxley, P.A. 1983b. The role of trees in agroforestry: somecomments. In: Plant research and agroforestry, edited by PAHuxley 257-270. Nairobi: International Centre for Research inAgmforestry.

Huxley, P.A. 1983c. Phenology of tropical woody perennialsand seasonal crop plants with reference to their managementin agroforestry systems. In: Plant research and agroforestry,edited by PA Huxley 503-525. Nairobi: International Centrefor Research in Agroforestry

Huxley, P.A. 1985. The basis of selection, management andevaluation of multipurpose trees - an overview. In: Attributesof trees as crop plants, edited by M.G.R Cannell & J.E.Jackson, 13-35. Abbots Ripton, Huntingdon: Institute ofTerrestrial Ecology

Jacob, J. & Al les, W.S. 1987. The Kandyan gardens of SriLanka. Agroforestry Systems, 5, 123-137.

Jonsson, K., Fidjeland, L., Maghembe, J.A. & Hogberg,P. 1988. The vertical distribution of fine roots of five treespecies and maize in Morogoro, Tanzania. AgroforestrySystems, 6, 63-70.

Kang, B.T. & Wilson, G.F. 1987. Alley cropping as apromising agroforestry technology In: Agroforestry: a decade ofdevelopment, edited by HA. Stepp ler & PK.R. Nair, 227-243.Nairobi: International Centre for Research in Agmforestry

•King, K.F.S. 1968. Agri-Silviculture. (Bulletin 1.) Ibadan:Department of Forestry University of lbadan.

Lang, A.R.G. & McMurtrie, R.E. 1992. Total leaf areas ofsingle trees of Eucalyptus grandis estimated from transmittanceof the sun's beam. Agricultural Meteorology 58, 79-92.

Lulandala, L.L.L. & Hall, J.B. 1991. Leucaenaleucocephala: potential role in rural development. (InternationalCouncil for Reseach in Agroforestry Working Paper, no. 65.)Nairobi: International Centre for Research in Agroforestry.

Lundgren, B.O. 1987. Institutional aspects of agroforestryresearch and development. In: Agroforestry: a decade ofdevelopment, edited by HA. Stepp ler & PK.R. Nair, 43-51.Nairobi: International Centre for Research in Agroforestry

Mann, H.S. & Saxena, S.K. 1980. nejri (Prosopis cineraria)in the Indian desert - its role in agroforestry. (Central AridZone Research Institute Monograph, no. 11.)

Marchal, J.Y. 1983. Yatenga, nord Haute-Volta: la dynamiquefun espace rural soudano-sahlien. (Yatenga, northern BurldnaFaso: the dynamics of a Sudano-Sahelian rural area.) ORSTOM,Travaux et Documents, 167, 170-175.

Marten, G.G. 1988. Productivity stability sustainabilityequitability and autonomy as properties of agroecosystemassessment. Agricultural Systems, 26, 291-316.

Mathias-Mundy, E., Muchena, 0., Mc Kiernan, G. &Mundy, P. 1991. Indigenous technical knowledge of privatetree management. Ames, Iowa: Iowa State University Centerfor Indigenous and Rural Development.

Michon, P. 1983. Village-forest-gardens in West Java. In: Plantresearch and agroforestry, edited by PA. Huxley 13-24. .Nairobi: International Centre for Research in Agroforestry

Midgeley, S.J., Turnbull, J.W. & Johnston, R.D. 1983.Casuarina ecology, management and utilization. Melbourne:Commonwealth Scientific and Industrial Research Organization.

Muetzelfeldt, R.I. & Sinclair, F.L. 1993. Ecologicalmodelling of agroforestry systems. Agroforestry Abstracts, 6(3). In press.

Nair, P.K.R., ed. 1989 Agroforestry systems in the tropics.Dordrecht: Kluwer.

National Research Council. 1983. Calandra: a versatilesmall tree for the humid tropics. Washington, DC: NationalAcademy Press.

National Research Council. 1984a. Casuarinas:nitrogen-fixing trees for adverse sites. Washington, DC: NationalAcademy Press.

National Research Council. 1984b. Leucaena: promisingforage and tree crop for the tropics. Washington, DC: NationalAcademy Press.

Nitrogen Fixing Tree Association. 1990a. PerennialSesbania production and use, a manual of practical informationfor extension agents and development workers. Wairnanolo:Nitrogen Fixing Tree Association.

Nitrogen Fixing Tree Association. 1990b. PerennialSesbania species in agroforestry systems. Waimanolo: NitrogenFixing Tree Association.

Nitrogen Fixing Tree Association. 1993. Erythrina in theNew and Old Worlds. In: Proceedings of the InternationalConference, Turrialba, Costa Rica, 1992. In press.

Nye, RH. & Greenland, D.J. 1960. The soil under shiftingcultivation. (Technical Communication 51.) Harpenden:Commonwealth Bureau of Soils.

Oldeman, R.A.A. 1983. The design of ecologically soundagroforests. In: Plant research and agroforestry edited by PA.Huxley 173-207. Nairobi: International Centre for Research inAgroforestry

Oldfield, S. 1988. Buffer zone management in tropical moistforests: case studies and guidelines. Gland: International Unionfor the Conservation of Nature and Natural Resources.

Ong, C.K., Corlett, J.E., Singh, R.P. & Black, C.R. 1991.Above and below-ground interactions in agmforestry systems.Forest Ecology and Management, 45, 45-57.

Padoch, C. & De Jong, W. 1987. Traditional agroforestrypractices of native and ribereno farmers in the lowlandPeruvian Amazon. In: Agroforestry: realities, possibilitiesand potentials, edited by H.L. Gholz, 179-195. Dordrecht:Nijhoff.

Perera, A.H. & Rajapakse, R.M.N. 1991. A baseline studyof Kandyan Forest Gardens in Sri Lanka: structure,composition and utilization. Forest Ecology and Management,45, 269-280.

Philip, M.S. 1975. Gum arabic and Acacia senegal. Nigeria

Forestry Information Bulletin (N.S.), 29, 1-12.

Pickersgill, B. 1983. Aspects of evolution in herbaceous andtree crops relevant to agroforestry In: Plant research andagroforestry edited by PA. Huxley 309-321. Nairobi:International Centre for Research in Agroforestry

Prance, G.T. 1989. American tropical forests. In: Tropicalrainforest ecosystems edited by H. Leith & M.J.A. Werger,99-132. Amsterdam: Elsevier.

Prance, G.T. 1990a. Fruits of the rainforest. New Scientist,125 (1699), 42-45.

Prance, G.T. 1990b. Future of the Amazonian Rainforest.FUTURES, November, 891-903.

Prance, G.T. 1991. What is ethnobotany today? Journal of

Ethnopharmacology, 32, 209-i216.

Raintree, J.B. 1990. Theory and practice of agroforestrydiagnosis and design. In: Agroforestry classification andmanagement, edited by K.G. MacDicken & N.T Vergara,

58-97. New York: Wiley

Ralson, J.P. 1988. Les 'Parcs' en Afrique: état desconnaissances et perspectives de recherches (Africanparklands': the state of knowledge and perspectives fromresearch). (Document de Trevail.) Paris: Centre d'Etudes

Africaines, EHESS.

Richards, P. 1985. Indigenous agricultural revolution. London:

Hutchinson.

135

Russell, G., Jarvis, P.G. & Monteith, J.L. 1989.Absorption of radiation by canopies and stand growth. In:Plant canopies: their growth, form and function, edited by G.Russell, B. Marshall & PG. Jarvis, 21-39. (Society ofExperimental Biology Seminar Series, no. 31.) Cambridge:Cambridge University Press.

Sanchez, P.A. 1987. Soil productivity and sustainability inagroforestry systems. In: Agroforestry: a decade ofdevelopment, edited by HA. Stepp ler & PK.R. Nair, 205-223.Nairobi: International Centre for Research in Agroforestry

Scherr, S.J. 1990. The diagnosis and design approach toagroforestry project planning and implementation: examplesfrom western Kenya. In: Planning for agroforestry, edited byWW Budd, I. Duchhart, L.H. Hardesty & F Steiner, 132-160.Amsterdam: Elsevier.

Shepherd,G. 1992. Managing Africa's tropical dry forests: areview of indigenous methods. (ODI Agricultural Paper, no.14.) London: Overseas Development Institute.

Sherry, S.P. 1971. The black wattle (Acacia mearnsii DeWild.). Pietermaritzburg: University of Natal Press.

Simpson, B.B. 1977. Mesquite - its biology in two desertscrub ecosystems. (United States International BiologicalProgramme Synthesis Series.) Stroudsberg: Dowden,Hutchinson & Ross.

Simmonds, N.W. 1985. Perspectives on the evolutionaryhistory of tree crops. In: Attributes of trees as crop plants,edited by M.G.R. Cannell & J.E. Jackson, 3-12. Abbots Ripton,Huntingdon: Institute of Terrestrial Ecology

Sinclair, F.L.,Verinumbe, I., Adder ley, W.P., Anderson,L.S., Grime, V.L., Hall, J.B., Hassan, B., Jenkins, D.A.& Stevens, P.A. 1992. Soil amelioration, water use and treecrop productivity in dryland agroforestry in Africa.Agroforestry Forum, 3 (2), 21-24.

Soemarwoto, 0. 1987. Homegardens: a traditional agroforestrysystem with a promising future. hr. Agroforestry: a decade ofdevelopment, edited by HA. Stepp ler & PK.R. Nair, 157-170.Nairobi: International Centre for Research in Agroforestry

Tomlinson, H.E 1992. The mineral nutrition of isolated treesin pasture. PhD thesis, University of Bristol.

Turldngton, R.A. 1975. Relationships between neighbours amongspecies of pennanent gra,ssland PhD thesis, University of Wales.

Van den Hoek, A. & Bekkering, T. 1990. Planning ofagroforestry programmes in Java, Indonesia. In: Planning foragroforestry, edited by WW Budd, I. Duchhart, L.H. Hardesty& F Steiner, 79-100. Amsterdam: Elsevier.

Van Noordwijk, M., Widianto, M.H. &Kurniatun-Hairiah. 1991. Old tree root channels in acidsoils in the humid tropics: important for crop root penetration,water infiltration and nitrogen management. In: Plant-soilinteractions at low pH, edited by R.J. Wright, 423-430.Domlrecht: Kluwer.

136

Van Noordwijk, M., Kurniatun Hairiah, Syekhfani, M.& Flach, E.N. 1991. Peltophorum pterocarpa (D.C.) Back(Caesalpiniaceae), a tree with a root distribution suitable foralley cropping on acid soils in the humid tropics. In: Plantroots and their environment, edited by B.L. McMichael & H.Persson, 526-532. Amsterdam: Elsevier.

Verinumbe, I. 1991. Agroforestry development in northeastern Nigeria. Forest Ecology and Management, 45,309-317.

Verinumbe, I., Sinclair, EL., Jenkins, D.A. & Adderley,W.P. 1992. Comparative evaluation in Africa of agroforestrytrees on vertisolic soils under low rainfall. (Site report.)Bangor: School of Agricultural and Forest Sciences, Universityof Wales.

Von Carlowitz, P.G., Wolf, G.V. & Kemperman, E.M.1991. Multipurpose tree and shrub database - an informationand decision-support system. User's Manual, Version 1.0.Nairobi: International Centre for Research in Agroforestry

Von Maydell, H.G. 1987. Agroforestry in the dry zones ofAfrica: past, present and future. In: Agroforestry: a decade ofdevelopment, edited by H.A. Steppler & PK.R. Nair, 157-170.Nairobi: International Centre for Research in Agroforestry

Walker, D.H. & Southern, A.J. 1992. Report on fieldwork inSri Lanka April-May 1992. Bangor: School of Agricultural andForest Sciences, University of Wales.

Walker, D.H., Sinclair, F.L. & Muetzelfeldt, R.I. 1991.Formal representation and use of indigenous ecologicallaiowledge about agroforestry: pilot phase report, June 1991.Bangor: School of Agricultural and Forest Sciences, Universityof Wales.

Webb, D.B., Wood, P.J., Smith, J.R & Henman, G.S. 1984.A guide to species selection for tropical and subtropicalplantations. (T1-opical Forestry Papers, no. 15, 2nd ed.) Oxford:Commonwealth Forestry Institute.

White, F. 1983. The vegetation of Africa. (UNESCO NaturalResources Research, 20.) Paris: UNESCO.

Withington, D., Glover, N. & Brewbaker, J.L. 1987.Gliricidia sepium (jacq.) Walp.: management and improvement.(Nitrogen Fixing Tree Association Special Publication 87-01.)Waimanolo: Nitrogen Fixing Tree Association.

Wood, P.J. & Burley J. 1991. A tree for all reasons: theintroduction and evaluation of multipurpose trees foragroforestry (Science and Practice of Agroforestry no. 5.)Nairobi: International Centre for Research in Agroforestry

World Bank. 1985. Desertification of the Sahelian andSudanian Zones of West Africa. Washington, DC: World Bank.

Young, A. 1989. Agroforestry for soil conservation. Wallingford:CAB International.

Use and potential of domesticated trees for soilimprovement

E C M Fernandes', D P Garrity2 , L T Szott3 & C A Palm4

1Project on Alternatives to Shifting Cultivation in the Western Amazon, Dept of Soil Science, North Carolina

State University and Centro de Pesquisa Agroflorestal da Amazonia Ocidental (EMBRAPA-CPAA), CP 455,

Manaus 69001, AM Brazil2International Centre for Research in Agroforestry, SE Asian Regional Research Programme, Forest

Research & Development Centre, PO Box 382, Bogor 16001, Indonesia3Agroforestry Program, International Development Research Centre, Centro Agronomico Tropical de

Investigacion y Enserianza 7170, Turrialba, Costa Rica4Tropical Soil Biology & Fertility Programme, UNESCO, PO Box 30592, Nairobi, Kenya

ABSTRACTThe ancient and widespread use of tree and bush fallows to restore soil productivity following several years of cropping has

resulted in research aimed at identifying, planting and managing trees for soil improvement. The goal is either to reduce the

traditional fallow periods (normally between four and 25 years), or to eliminate the fallow period by introducing systems in which

'soil-improving' trees are managed together with crops or pasture species (agroforestry). nees commonly used by farmers for

soil improvement include Acioa barteri, Calliandra calothyrsus, Cassia reticulata, Erythrina poeppigiana, Faidherbia albida, Gliricidia

sepiurn, 1nga edulis, Leucaena leucocephala, Prosopis cineraria, and Sesbania sesban. Although many of these species are largely

undomesticated, several are currently the focus of genetic improvement programmes (Owino, Oduol & Esegu, pp205-209; Simons,

Mac Queen & Stewart, pp91-102; Milimo, Dick & Munro, pp210-219). The possibility exists that in such programmes, selections

could be made for tree characteristics to significantly improve soils. A synthesis of the research data from work on acid and high

base status soils in the humid, subhumid, and semi-arid regions of Central and Latin America, Africa and Asia is used to

evaluate critically 'tree/soil improvement' hypotheses related to nutrient cycling, organic matter additions, and nitrogen fixation,

These hypotheses are discussed in the context of developing domestication strategies for multipurpose species, taking account of

the influence of the species on the soil environment.

INTRODUCTION

The claim that trees have the potential toimprove soils probably originates from the factthat forest fallows have been used to restore soil

productivity as shown in the traditional andwidespread practice of shifting cultivation(Greenland & Nye 1959). In addition, there existseveral traditional land use systems in whichfarmers retain trees on farmland. Recognition ofthe importance of the existing, but little studied,integration of trees in tropical land use systemsled to the term 'agroforestry' (Bene, Beall & Cote

1977).

Increased yields have been reported for cropsand forage grasses in close proximity to trees,and this increase has been partly attributed toimproved chemical and physical properties ofsoil as a result of tree growth (Nair 1984).Several hundred semi-domesticated tree speciesare currently being used by farmers in thetropics. If these species can be utilised tocounteract the soil-degrading effects of croppingand animal husbandry farmers in the tropics whoare currently practising shifting cultivation may

not need to slash-and-burn additional forestsevery two to four years. The objective of thispaper is to examine critically 'tree/soilimprovement' hypotheses in relation to some ofthe semi-domesticated tree species that have notyet been used to any significant extent inplantation forestry Because multipurpose treespecies with potential for soil improvement arecurrently the focus of considerable scientificresearch in the context of agroforestry (Young1989), we will examine primarily agroforestrydata sets to test the key tree/soil improvementhypotheses related to organic matter additions,nutrient cycling, and nitrogen fixation, and willconsider the ways in which the domestication of

tropical forest trees can enhance soil improvement.

TREE-BASED CROPPING SYSTEMS IN THETROPICS

Trees in rotation with crops (forest fallows)

Shifting or 'slash-and-burn' cultivation is a majorforrii of land use in the tropics (Greenland 1974).Farmers typically clear about one hectare of

137

primary or secondary forest, burn it, and thenplant crops for one or more years, takingadvantage of the nutrients released in the ash.Once yields decline or the weeds are no longereasily controlled, the site is abandoned. Thenutrient accumulation by the forest fallow, theshading out of weeds, and the action of treeroots and associated micro-organisms and faunaare the processes by which the potential soilproductivity is gradually improved to a stateapproaching that of a primary forest.

In areas of high population pressure, such asNigeria, farmers are no longer able to leave theland fallow for a long period and resort to'accelerated or biologically enriched fallows',involving tree species such as Acioa barteri,Alchornea cordifolia, Anthonotha macrophylla,Gliricidia sepium and Harungana madagascariensis(Okafor 1982; Oldgbo 1983). In SE Asia, farmersuse Calliandra calothyrsus (National ResearchCouncil (NRC) 1983) and Leucaena leucocephala(NRC 1984) as fallow species.

Trees in association with crops

In the sahelian and sudanian zones of WestAfrica, farmers maintain and manage around 20different tree species on crop lands. Of thesespecies, Faidherbia albida (Charreau & Vidal1965) and Parkia biglobosa (Kessler 1992) havebeen reported to improve soils. Farmers in thesemi-arid zones of India commonly retain andmanage tree species such as Prosopis cineraria(Aggarwal et al. 1976), P juliflora and Delonixelata (Jarnbulingam & Fernandes 1986) forfuelwood, fodder and soil improvement. Innorth-eastern Thailand, where multipurpose treesare maintained or planted on paddy fields,significantly higher contents of organic matterand soil macronutrients were found in fields withthe nitrogen-fixing Samanea saman than in fieldswith non-leguminous trees (Sae-Lee, Vityakon &Prachaiyo 1992). Other tree species used toimprove soils include Sesbania sesban (Evans &Rotar 1987), Erythrina berteroana and E.poeppigiana (Ramiréz et al. 1990).

TREE CHA.RACTERISTICS FOR SOILIMPROVEMENT IN THE TROPICS

Fast-giowing tree species are major componentsof agroforestry technologies (Nair, Fernandes &Wambugu 1984). With the exception of Leucaenaleucocephala (Brewbaker 1990), the greatmajority of tree species have only just begun tobe scientifically tested for their biologicalpotential, adaptability to site conditions, andresponses to management factors such aspruning, fertilization, and competition withagronomic crops.

On the more fertile soils, the presence ofapparently sustainable agroforestry systems

138

indicates that the tree component contributes toreduced erosion and efficient nutrient cycling(Alpizar et al. 1988). On acid, infertile soils, thepotential of trees to increase nutrient inputs orreduce losses is limited by chemical constraintsto root expansion. High aluminum (Al) saturationin the subsoil, low concentrations of calcium (Ca)and phosphorus (P), and a lack of weatherableminerals can severely limit plant productivity andthereby reduce nutrient capture and recycling,compared with more fertile sites (Szott,Fernandes & Sanchez 1991). Adaptability to soilchemical and physical conditions is an obviousbut vital requirement for any tree species tocontribute to soil improvement. Desirablecharacteristics for soil-improving trees include thefollowing.

Above-ground biomass production: around 2 Mg(megagram) hC1 yr-1 for semi-arid regions,4 Mg hC1 yr-1 for subhumid regions and 8-10Mg hC1 yr-1 for the humid tropics (Table 1).These values are based on estimates by Young(1989) of organic residues required to maintainsoil organic carbon at 7.5, 15 and 30 Mg hC1for each region, respectively

Nitrogen fixation potential: 10-200 kg N hC 1 yr- 1,

based on estimates from various species in thetropics (Peoples & Herridge 1990). The lowervalues are likely to be encountered for trees inthe semi-arid tropics and also on acid soils withlow P availability in the humid tropics.

Low fme root (<2 mm diameter) density in thetopsoil: for example, no more than 30% of fineroots in the top 15 cm of the soil, and thesepreferably concentrated close to the stem (egAcioa barteri, as described by Ruhigwa et al.1992).

Capacity to form effective mycorrhizal associationswith native populations of vesicular-arbuscular(VA) mycorrhizal fungi or ectomycorrhizal fungi(see Lapeyrie & H6gberg, pp158-164; Mason &Wilson, pp165-175)

Moderate to high nutrient concentrations in leafybiomass (eg 2.0-3.5% N, 0.2-0.3% E 1-3%potassium (K), and 0,5-1.5% Ca). Interpretingdata for micronutrient concentrations in tree andshrub biomass is still difficult.

Appreciable nutrient concentrations in the rootsystems. For comparison, fine and coarse rootsfrom two.primary rainforest sites in Asia(Andriesse 1987) had the following meanconcentrations: 0.66% N, 0.04% E 0.55% K, 0.85%Ca, and 0.18% magnesium (Mg). Obviouslythese values are likely to change with soil type,vegetation and root sizes.

Different rates of leaf litter decay: rapid litter decay(1-3 weeks) where tree biomass is used toprovide nutrients to associated crops, or slow (2-6months) decay when tree biomass is used asmulch for weed Suppression and soil protection.

Table 1. Biomass production by nee species in alley cropping on fertile and infertile soils in the humid, subhumid, andsemi-arid zones of the tropics (source: 1. Singh et a). 1989; 2. Hocking & Rao 1990; 3. Kang, Wilson & Sipkens 1981; 4.Fernandes 1990; 5. Ruhigwa et al. 1992; 6. Evensen 1989; 7. Kass et al. 1989; 8. Rosecrance, Brewbaker & Fownes 1992)

Species

Hyderabad, India; rainfall 750 mm yr-1; alfisol, pH 7.0, P=8 ppm (Olsen)

Leucaena leucocephala

lbadan, SW Nigeria; rainfall 1280 mm yr-1; alfisol, pH 6.2, P=25 ppm

Leucaena leucocephala

Yurirnaguas, Peru; rainfall 2200 mm yr-1; ultisol, pH 4.2-4.6, P=8 ppm (Olsen)

Inga edulis

Gliricidia sepiurn 14/84

Gliricklia sepium 34/85

Onne, SE Nigeria; rainfall 2400 mm yr-1; ultisol, pH 4.0, P=50 ppm (Bray-1)

Acioa barteri

Alchornea cordifolia

Cassia siamea

Gmelina arborea

Sumatra; rainfall 2575 mm yr-1; oxisol, pH 4.1, P=4.8-6.8 mg kg (Melich 1)

Paraserianthes falcataria

Calliandra calothyrsus

Gliricidia sepium

Costa Rica; rainfall 2640 mm yr-1; inceptisol, PH 4.3-4.8, P=8-15 ppm (Olsen)

Glincidia sepium 6 666

Erythrina poeppigiana 555

-.Western Samoa; rainfall 3000 mm yr 1, mod. fertile inceptisol, no

Calliandra calothyrsus

Gliricidia sepium

/ ND, not determined2 1, leaves and green shoots; w, woody material

Trees(nos ha-1)

8 888

5 000

5 000

2 500

2 500

2 500

2 500

Absence of toxic substances in the foliage or rootexudates. For example, in the acid savannas(cerrado) of Brazil, some species have been shownto accumulate 4000-14 000 mg kg-1 of Al in thefoliage (Haridasan 1982). Non-accumulator speciestypically have Al concentrations <200 mg kg-1.

Other desirable characteristics of soil-improvingtrees remain to be identified, and there isconsiderable potential for exploiting theintraspecific genetic variation in thecharacteristics listed. Several tree species withsoil-improving characteristics can be identified:

'ee age Prunings Dry matter(month) (nos yr-1) (Mg ha-1 yr-1) Ref

2 000 48 ND 1.41

833 48 ND 7.4 l+w2

10 000 36 6 6.51

soil data

11

11

11

48

48

48

48

ND

ND

ND

ND

3 9.6 l+w

3 8.1 l+w

3 1.8 l+w

13.8 l+w

14.9 l+w

122 l+w

12.3 l+w

19 900 09 4 4.9 l+w

21 4 9.7 l+w

19 900 09 4 6.8 l+w

21 4 10.7 l+w

10 000 09 4 0.6 l+w

21 4 1.4 l+w

24 2 9.6 l+w

60 2 15.2 l+w

24 2 7.4 l+w

60 2 11.1 l+w

5 000 48 3 12.1 l+w

3 333 48 3 7.6 l+w

5 000 48 3 10.7 l+w

3 333 48 3 6.5 l+w

• acid soils, humid tropics: I. edulis, Cassiareticulata, Cassia spectabilis, Erythrinaberteroana, G. sepium (On 14/84) andParaserianthes falcataria;

1

2

3

4

5

6

7

8

• non-acid soils, humid and subhumid tropics:Calliandra calothyrsus, Erythrina fusca, E.poeppigiana, G. sepium and Leucaenaleucocephala;

• non-acid soils, semi-arid tropics: F: albida,Cassia siamea, Prosopis cineraria, P julifloraand Sesbania sesban.

139

TREE/SOIL IMPROVEMENT HYPOTHESES

Sanchez (1987) formulated the agroforestryhypothesis that 'appropriate agroforestry systemshave the potential to control erosion, maintain soilorganic matter and soil physical properties,augment nitrogen fixation, and promote efficientnutrient cycling'. What distinguishes agroforestryfrom other cropping systems is the presence oftrees, either in association or in rotation withannual crops and/or animals. The hypothesisedadvantages of agroforestry are thus directlyrelated to the presence, function, andmanagement of the tree component and theinteractions with the crop and/or animalcomponents (Lundgren & Raintree 1983).

The rationale for the hypothesis that soilimprovement in agroforestry is superior to that inother cropping systems is that the nutrientexports via the harvest of crop and animalproducts, erosion, leaching and volatilisation, andthe deterioration in soil physical properties dueto cropping or grazing, can be counteracted bythe tree component. This process may occur via:• nutrient uptake by deep-rooted trees allowing

for capture and surface deposition ofnutrients beyond the reach of crop roots(more efficient nutrient cycling);

• increased amounts of organic inputs to thesoil maintaining soil organic matter and thusimproving soil structure and nutrient status;and

• increased nutrient additions to the soil fromN fixation and dust or aerosol interceptionby tree canopies.

These hypotheses are examined below

Nutrient cycling and nutrient pumping

According to the 'nutrient pumping' hypothesis,nutrients absorbed from the subsoil are used intree growth and eventually deposited in or onthe surface soil via root and leaf litterproduction, nutrient leaching from foliage, andpruning of leaves or branches (Nair 1984; Young1989). Nutrient pumping is one of the mostfrequently cited potential advantages ofincorporating trees into cropping systems(agroforestry), and is based on the general beliefthat trees have deep spreading root systemscapable of exploiting a greater soil volume thanthose of annual crop plants. The carefulcombination of trees and crops with differentrooting characteristics offers the possibility ofincreasing the productivity of admixtures byoptimising the complementarity of soil nutrientexploitation. Factors such as the spatialdistribution and temporal patterns of root growthare clearly important, and should influence thechoice of species or genotypes to be used.

Research on a variety of soil types and indifferent climates has involved the injection of 32P

140

at different soil depths and varying distancesfrom the stems of banana (Musa spp.), cocoa(Theobroma cacao), coffee (Coffea arabica),coconut (Cocos nucifera) and oil palm (Elaeisguineensis). Results showed that 30-85% of Puptake occurred within a depth of 30 cm and adistance of up to 1 m from the stems of banana,cocoa and coffee. In coconut and oil palm,however, P uptake from similar zones accountedfor less than 20% of the total applied. Theuptake of P from a depth of 45 cm or greaterand 1.5 m or further from the stem was lessthan 4% for the majority of tree crops, but up to15% for coconut (International Atomic EnergyAgency 1975). Indirect evidence of nutrientuptake by trees from the subsoil is provided bystudies on Eucalyptus and Prosopis that showedwater uptake from soil depths of 5 m or greater(Stone & Kalisz 1991). Leucaena leucocephala inan alley cropping system depleted soil moistureto a depth of 60 cm (Singh, Ong & Saharan1989).

In spite of the potential for deep rootingpossessed by many tree species, most roots donot reach great depths, because of the presenceof physical or chemical barriers in the soil.Barriers to root expansion restrict the absorptionof nutrients, such as P and K, that diffuse slowlyin the soil and are depleted rapidly fromlocalised pockets around roots, or those such asCa, which is absorbed mainly throughunsuberised root apices. However, rootpenetration of these barriers does occur.Adaptations to high soil Al concentrations exist inmany plants (Marschner 1986). Furthermore,localised discontinuities in these barriers, causedby changes in soil texture, mineralogy or bioticfactors such as soil macrofauna, may permitroots of non-tolerant plants to grow through. Theimportance of these factors to nutrient uptakeand cycling on infertile soils deserves furtherattention.

In summary the greatest opportunity for nutrientpumping by deep-rooted trees exists onrelatively fertile soils where physical barriers toroot penetration are absent.

Organic inputs and soil organic matter

It has been hypothesised that the presence oftrees in agroforestry systems results inconcentrations of soil organic matter (SOM) thatare adequate for maintaining soil fertility andproductivity (Young 1989). Organic inputs to thesoil consist of the above- and below-ground litter,crop residues, mulches, green manures andanimal manures. Soil organic matter is the resultof the partial or complete transformation in thesoil of these inputs. Trees may be managed toprovide large quantities of biomass for use asorganic inputs. Furthermore, considerable scopeexists for selecting between and within tree

species for different chemical compositions anddecomposition rates of the biomass produced.This option offers a means of manipulatingnutrient release from biomass inputs, soilmicrobial activity and the rate of conversion ofthese organic inputs to SOM.

The quality of organic inputs is one of thefactors that influences the rate of decompositionand release of nutrients from organic residues.Research on tropical plant species has shownthat indices of quality (eg C/N or lignin/N ratios)which are adequate for temperate zones (Melillo,Aber & Muratore 1982) may not be the bestpredictors of N mineralisation from leguminousmaterials in the tropics. N release patterns maybe more closely related to tissue polyphenoliccontents than to lignin or N contents or lignin/Nratios (Palm 1988). In general, on fertile soils,leaf litter quality is higher and decompositionfaster than on acid, infertile soils. Very little isknown about the quality of root litter. Root litterfrom trees may differ from that of annual cropsin proportions of fme to coarse roots, the amountof lignins and polyphenols present, quantities ofsecondary or allelopathic compounds, andnutrient concentrations. Conventional indices ofquality may not, therefore, be good predictors oftree root decomposition.

Organic phosphorus is the most importantsource of P in highly weathered tropical soils(Tiessen, Stewart & Cole 1984). Incorporation offast-growing trees into tropical farming systemsoffers the potential to increase this organic Ppool. Tiessen, Salcedo and Sampaio (1992) founda 28% increase in organic P in the soil but onlya 9% increase in total P during a ten-year bushfallow There is even some evidence that somespecies are able to utilise insoluble forms of soilP not used by crop species (Ae et al. 1990).

The role of soil organic matter in contributing tosoil fertility and productivity was summarised byAlison (1973) as:

• a source of inorganic nutrients to plants;

• a substrate for micro-organisms;

• an ion exchange material;

• a factor in soil aggregation and rootdevelopment; and, consequently

• a factor in soil and water conservation.

Despite these potential effects of SOM, thereexist several cases where no consistentcorrelation has been found between total SOMand existing or potential soil productivity(Sanchez & Miller 1986). As a result, researchhas focused on several conceptual fractions ofSOM based on different turnover times in thesoil, as a function of chemical structure andassociation with soil particles (Parton et al. 1987).

A key issue involves the possibility of influencingsoil improvement via the selection and

management of tree species or genotypes withdiffering litter qualities. Litter that decomposes ata slow rate (low quality) is likely to lead to theformation of slow and passive fractions of SOMwhich are less important in short-term nutrientavailability but which play an important role inmaintaining soil structure. In managed fallow orplantation systems where trees are used toregenerate soil productivity the use ofdomesticates with low-quality slowlydecomposing materials would build up theorganic pools of nutrients and buffer long-termnutrient availability and release. In systems wheretrees and crops are grown simultaneously (egalley cropping), the use of a mixture of speciesor genotypes with different litter qualities may beadvisable.

N fixation by trees

There are around 650 species of nitrogen-fixingtrees and shrubs (see Sprent, pp176-182). Sometree species appear to have a higher capacityfor nitrogen fixation than others. Publishedestimates of the amount of N fixed by N-fixingtrees are in the range 10-500 kg N ha-1 yr-1(Young 1989). The proportion of tissue nitrogenin trees derived from fixation can vary between4% and 80% (Peoples & Herridge 1990; Danso,Bowen & Sanginga 1992). In addition todifferences in N fixation among tree species,large differences exist among genotypes within aspecies. Sanginga, Bowen and Danso (1990)reported two- or three-fold differences in Nfixation among genotypes of E albida and L.leucocephala after 12 weeks and 70% differencesafter 36 weeks.

The potential N contributions of leguminous treespecies to the soil are often not realised,however, because of high soil acidity low soilnutrient contents (Szott et al. 1991), and low soilmoisture contents (Venkateswarlu, Korwar &Singh 1990). Studies of leguminous annuals haveshown that low concentrations of P in acid soils

can severely limit nitrogen fixation (Barea &Azcon-Aguilar 1983). There is, however, a lack ofexperimental data for leguminous tree species. InP-deficient soils, the planting of N-fixing treespecies with crops may actually result incompetition for P between the trees and theassociated crops. The addition of moderateamounts of P fertilizer to the trees could alleviatethe competition for phosphorus. Effectivesymbiosis with compatible vesicular-arbuscular(VA) mycorrhizal fungi (see Mason & Wilson,pp165-175) will also be important.

N-fixing trees probably contribute little N directlyto the soil environment in excess of their own Nrequirements for growth, but this situation can beovercome by shoot pruning and the applicationof this biomass to associated crops (eg in alleycropping). In addition to the release of N and

141

other nutrients from the decomposing biomass,shoot pruning brings about an additionalcontribution of N to the soil via the death of fineroots and nodules. In a pot experiment with I.edulis, shoot pruning resulted in significantreductions in fine root biomass, total root lengthinfected with VA mycorrhizal fungi, and numbersof active nodules, relative to non-pruned controlplants (Fernandes 1990). The renewal ofnodulation and nodule function may be furthersuppressed by the release of N from thedecomposing prunings, fine roots and nodules(Danso et al. 1992).

In summary a number of factors can affect Nfixation in trees. They include soil acidity low soil 1?VA mycorrhizal fungi, high soil N, rhizobial strains,soil moisture and temperature, tree age, shootpruning, and tree genotype. Optimisation of Nfixation by tree species will require not onlyselections among and within tree species foradaptability to soil factors adveisely affecting Nfixation, but also selections for enhanced efficiencyof tree/rhizobial/VA mycorrhizal associations.

SIGNIFICANCE OF TREES AS NUTRIENTEXPORTERS

While trees may contribute significantly to soilimprovement, they can also have adverse effectson soils and associated crops or pasture species.These potential negative effects include biomassand nutrient export from the site through treeharvests, as well as competition for nutrients andmoisture between the trees and associated cropspecies. We review these issues throughdiscussion of three types of agroforestry systems:domesticated timber tree-based systems,tree/perennial crop systems, and prunedhedgerow intercropping systems.

Timber-based systems

Tabora (1991) compared three perennialagroforestry species in terms of their impact onannual nutrient removal: an abaca-based (manilahemp, Musa sp.) system, a coconut-based system(Cocos nucifera), and a Paraserianthes falcatariapulpwood production system. The abaca systemwas found to sustain the least nutrientexportation, with only 4 kg of macronutrients(sum of elemental N, p and K) removed ha-1

-1yr in the fibre harvest. Abaca fibre is a highlycellulosic material composed of only 1.0% ash.The coconut-based system removed double thisquantity of nutrients, while macronutrientexportation of the Paraserianthes wood productionsystem was an order of magnitude higher(Table 2). In a 6.8-year-old plantation, Halenda(1989) estimated that harvesting A. mangiumstems alone would haire resulted in the exportfrom the site of 284 kg N, 15 kg P, 225 kg Caand 24 kg Mg ha-1 respectively Clearly although

142

Table 2. Estimated nutrient export per hectare per year ofthree agroforestry systems (source: Tabora 1991)

the trees accumulated considerable biomass andnutrients, tree harvesting for two or threerotations would result in a nutrient-depleted site.

Tree/perennial intercropping

Alpizar et al. (1986) and Fassbender et al. (1988)inventoried the nutrient uptake and flows in twoagroforestry systems composed of cocoa plantedat 1111 trees ha-1, in association with eitherErythrina poeppigiana or Cordia alliodora servingas shade and timber trees spaced at 278 treesha-1. Biomass accumulation in stems of C.alliodora was 2.5 times higher than in E.poeppigiana. The latter accumulated the majorityof its biomass in branches and leaves, whichrecycled in the system through pruning andlitterfall. The total annual net primary productivityof the two systems, however, was fairly similar(21.6 versus 21.3 t ha-1).

Nutrient accumulation in the standing biomasswas much higher for the timber trees than forthe cocoa. This fact might have induced seriouscompetition for nutrients, were it not for therelatively limited nutrient removal demands of thesystem, in comparison with the abundant nutrientcycling occurring within it. Annual nutrientextraction from the system through the harvest ofcocoa pods and husks was relatively low(compared with annual crop systems) at 19-25kg N, 4.0-4.3 kg P and 28-27 kg K. There wasalso a low proportion of nutrient exportationcompared with the total nutrient concentration ofthe standing biomass (5% N, 8% P and 8% K).The recycling of nutrients by residues returningback to the soil system exceeded removal by5.9-6.8 times in the case of N, and by 3.8-4.8times for P, and about 2 times for K. Only one ofthe overstorey trees was a nitrogen-fixing species(E. poeppigiana). The cocoa/E. poeppigianacombination registered a total increase of 80 kgN ha-1 in the soil/plant system over a five-yearperiod, indicating an annual contribution fromfixation equivalent to 18 kg N ha-1.

Pruned free hedgerows

Direct nutrient exportation in alley cropping orhedgerow intercropping systems may occur bythe use of leaves and green stem prunings asanimal fodder, or by the harvest of poles,fuelwood, or fruits. Indirect export occurs throughthe harvest of annual crops obtaining nutrientsfrom hedgerow litterfall and prunings. Theharvest of economic products results in directnutrient exports but contributes to theproductivity and profitability of the system.Farmers practising alley cropping often find thatthe economic value of hedgerow prunings whenused as animal fodder may often exceed theirvalue -as green manure (Kang, Reynolds* &Atta-Krah 1990). This practice may, however,result in an accelerated decline in soil fertilityThe increased quantity of animal manure thatresults from greater fodder production can bereturned to the field, along with additionalinorganic fertilizer, to at least partly offset thelosses. Otherwise, the increased rate of soilnutrient removal may jeopardise the long-termsustainability of the system.

SIGNIFICANCE OF TREES ASCOMPETITORS FOR RESOURCES INAGROFORESTRY SYSTEMS

The very nature of agroforestry dictates thatresource competition largely controls systemviability and productivity The development ofsystems involves the search for tree domesticatesto fill complementary or synergistic niches thatminimise resource competition. Limited researchduring the past decade has begun to explorethe nature of tree/crop interactions (Ong et al.1991). The concepts of resource sharing inagroforestry systems were reviewed by Buck(1986).

The tree component of agroforestry tends tohave a competitive advantage over an annualcrop; owing to its dominant stature, both above-and below-ground. As in any intercroppingsystem, competition may increase the productionof the system as a whole, or help stabiliseoutputs when resource supply is erratic(Monteith, Ong & Corlett 1991).

Below-ground competition

In water-limited environments, the potential ofagroforestry is strongly influenced by the majortree/crop differential in ability to capture soilmoisture. Severe reductions in crop yields inalley cropping are often observed in thesemi-arid tropics, drawing the viability of thesesystems into question, eg in India. Studies withroot barriers have indicated that these reductionsare mainly due to competition for water (Singhet al. 1989). The presence of a shallow rootbarrier (to 0.5 m depth) was sufficient to

eliminate a yield reduction in cowpea andsorghum yield. Ong et al. (1991) found that rootbarriers eliminated crop yield reductions intree/annual crop associations, and concluded thatcompetition for water was the driving factor.Roots of Leucaena leucocephala were abundant inthe top 30 cm of soil in adjacent alleys, contraryto the assumption that tree roots do not competewith cropsMalik and Sharma (1990) observed thatEucalyptus tereticornis extracted five times morewater than mustard from a 150 cm soil profile atthe tree/crop interface, when grown in ahedgerow cropping system in a semi-aridenvironment. Mustard and wheat yields werereduced by 47% and 34%, respectively inadjoining 10 m wide strips. Likewise, boundaryplantings of L. leucocephala, Eucalyptuscamaldulensis, Albizia procera, and Morus albareduced yields of wheat by 15-70% up to 4 maway in adjoining fields (Akbar et al. 1990).These results emphasise the importance oftaking water use characteristics into accountduring any domestication programme.

Because uptake of N, P and K is most stronglyregulated by plant demand and plant demand isstrongly related to growth, it follows thatcompetition between soil-improving trees andassociated crop or pasture species may beminimised if the trees and associated speciesdiffer in phenology In many tropical tree species,-rapid growth and high nutrient demand occurduring the rainy season. An exception to thispattern is provided by Faidherbia albida, whichloses its leaves at the beginning of the wetseason. Presumably this timing would result inless nutrient demand and less nutrientcompetition with annual crops during the wetseason.

Demand for nutrients is generally highest duringthe period of growth in foliage and fme roots,which usually occurs within the first ten to 20years of tree growth, although root accumulationmay continue to occur for longer periods oftime. Because internal re-translocation is limitedin young trees, high demand is placed on soilstocks (Miller 1984). The implication is thatcompetition for nutrients and water is likely tobe most severe during the early stages of treegrowth and will decrease as the trees get older,because of the increasing importance of internalre-translocation and the expansion of the treeroots into previously unexplored soil. If treeproducts are being harvested, however, thenutrient exports in the harvested fraction willreduce internal re-translocation and increasecompetition between trees and associatedvegetation, as the trees seek to compensate for

the exported nutrients (Atkinson 1986).

Although the soil volume explored by tree rootsmight be expected to increase as the trees age,

143

144

Hedgerow 12 months old0 No root prune (-R)• Root prune (+R)Hedgerow 6 months old0 No root prune• Root prune

1200

»a)c.)

600

300

Hedgerow

4-0.75 m

4-0.5 m

1 2

14-0.5 m 4,1

differences in rooting volume vary amongspecies, provenances and individuals of aspecies (Jenik 1978), and may therefore be thesubject of selection during tree improvement.More equitable sharing of below-groundresources between trees and associatedvegetation could be accomplished by:

• selection of appropriate genotypes orideotypes;

• varying plant spacing and time of planting;

• and the use of above- or below-groundpruning, shade, and liming.

In an alley cropping experiment on acid soils(Fernandes 1990), root pruning of hedgerowroots significantly increased crop yields in thesecond year of tree/crop interaction (Figure 1).

Perhaps the most cost-effective technique wouldbe to use species and provenances that havecomplementary patterns of nutrient uptake. Thispossibility presupposes knowledge regarding thespatial and temporal patterns of nutrient uptakeand/or rooting patterns. While data on rootingpatterns of the more recently domesticated treesare very scarce, genetically based differences inrooting patterns exist even among provenances(Vandenbeldt 1991). In a study of 15

3 4Crop rows

4 m

+R: Grain he = 0.1 + 575.7 row - 82.9 (row)2R-square = 0.989

-R: Grain ha' = 146.5 + 574.3 row - 80.8 (row)2R-square = 0.989

0

5 6

Root pruning of hedgerows to a depth of 20 cm

Hedgerow

Tgure 1. Rice grain yield with distance from hedgerows of Inga edulis, as affected by age of hedgerow and pruning ofhedgerow roots (source: Fernandes 1990)

provenances of G. sepium, there were significantlydifferent patterns of vertical root distributionamong provenances 20 months after growth inthe field (Fernandes 1990). Such differences offerthe potential of selecting adapted germplasm foroptimising the desired effects of trees on soils.

Tree species used in agroforestry systems arecommonly selected for their ease ofestablishment, fast growth, ability to coppicevigorously high foliage production, N-fixing abilityand high rates of production of useful productssuch as wood or fruits (Kang et al. 1990). Thesetrees are typical of the 'competitive' ideotype asdescribed by Donald and Hamblin (1976).Invariably high tree productivity is associatedwith high nutrient demand and high nutrientuptake efficiency On infertile soils, however,these tree characteristics will increasecompetition for nutrients and water betweentrees and associated crops, thereby negating anybeneficial effects that the trees have on soilchemical or physical properties. Trees of the'crop' ideotype, however, are not strongcompetitors, and make efficient use of those siteresources to which they have access (Cannell1978). Ti-ees with a crop ideotype are likely tobe more suitable than trees with a competitive

ideotype for agroforestry systems in which treesand crops are present simultaneously

Above-ground competition

Theoretical aspects of above-ground interactionsin agroforestry systems were reviewed byMonteith et al. (1991). The taller-tree componenttends to exert changes in four types ofmicroclimatic influence on an associated annualcrop: light interception, rainfall interception,saturation vapour pressure in the canopy, andcrop temperature. The effect on the crop may benegative or positive, depending on theenvironment, and particularly on the degree ofwater deficit.

Shading effects tend to be particularlypronounced on short-statured crops grown inassociation with tree hedgerows, such as uplandrice. Solera (1992) inserted' root barriers next tohedgerows of GlinCidia sepium and C. spectabilisin the southern Philippines, and observed thatthe dry matter yield reduction that occurred inthe rows of rice nearest to the trees was notcaused by below-ground competition. The 50 cmtall hedges apparently caused a deleteriousshading effect on the rice, even though the treeswere pruned frequently during the growingseason. These results contrast with observationsfrom semi-arid areas, where above-groundinteractions were found to be of minorimportance compared with below-groundcompetition for soil moisture (Ong et al. 1991).The importance of tree architecture and ofmatching trees to their light environments arediscussed elsewhere in this volume (seeOldeman & Sieben-Binnekarnp, pp25-33; Fasehun& Grace, pp148-157).

Allelopathy

There is evidence that at least some tree speciesobserved in agroforestry systems producephytotoxins that can directly affect germinationand growth of common agricultural crops (Bhatt& Todaria 1990). The species include Adniacordifolia, Alnus nepalensis, Celtis australis andPrunus cerasoides. The significance of thisallelopathy at the field level, however, is not yetknown, and there is little information concerningintraspecific variation in this characteristic.

CONCLUSIONS

The mechanisms by which trees may improvesoils are now reasonably well understood. Thatcertain tree species can improve soils is shownby the regeneration of soil productivity in follows.Pioneer species that possess adaptivecharacteristics to overcome soil constraints canplay a key role in modifying the site for latesuccessional species. The timeframe for thismodification is usually long (10-20 years). It ishypothesised that it is possible to combine trees

and crops in certain proportions so that cropyield is increased by soil-improving trees.Evidence so far has shown that, on fertile soilswhere nutrients are not limiting, this result canbe achieved. On nutrient-poor soils, however,tree/crop competition is usually so severe thatthe intensive management of both components isrequired before any soil improvement is obtainedfrom the associated trees. The use ofwell-adapted tree domesticates in managedfollows, together with minimum levels of inorganicinputs, could significantly shorten the time forregenerating soil productivity on acid soils.

ACKNOWLEDGEMENTS

Financial support for travel to present this paperwas provided by Aracruz forestal Internationaland by the Rockefeller Foundation, and isgratefully acknowledged.

REFERENCESAe, N., Arihara, J., Okada, K., Yoshihara, T. &Johansen, C. 1990. Phosphorus uptake by pigeon pea andits role in cropping systems of the Indian subcontinent.Science, 248, 477-480.

Akbar, G., Ahrnad, M., Rafigue, S. & Babar, K.N. 1990.Effect of trees on the yield of wheat crop. AgroforestrySystems, 11, 1-10.

Aggarwal, R.K., Gupta, J.P., Saxena, S.K. & Muthana,K.D. 1976. Some physico-chemical and ecological changes ofsoil under permanent vegetation. Indian Forester, 102,862-872.

Alison, F.E. 1973. Soil organic matter and its role in cropproduction. New York: Elsevier.

Alpizaz, L., Fassbender, H.W, Heuveldop, J., Folster, H.& Enriguez, G. 1988. Modelling agroforestry systems ofcacao (Theobroma cacao) with laurel (Cordia alliodora) andporo (Erythrina poeppigiana) in Costa Rica. Part II. Cacao andwood production, litter production and decomposition.Agroforestry Systems, 6, 37-48.

Andriesse, J.P. 1987. Monitoring proj6ctof nutrient cycling insoils used for shifting cultivation under various climaticconditions in Asia. (Final Report: Part I.) Amsterdam: RoyalTropical Institute.

Atkinson, D. 1986. The nutrient requirements of fruit trees:some current considerations. In: Advances in plant nutrition,edited by B. Tinker & A. Lauchli, vol. 2, 93-128. New York:Praeger.

Barea, J. & Azcon-Aguilar, C. 1983. Mycorrhizas and theirsignificance in nodulating nitrogen-fixing plants. Advances inAgronomy, 36, 1-54.

Bene, J.G., Beall, H.W. & Cote, A. 1977. Trees, food andpeople: land management in the tropics. Ottawa: InternationalDevelopment Research Centre.

Bhatt, B.P. & Toclaria, N.P. 1990. Studies on the allelopathiceffects of some agroforestry tree crops of Garhwal Himalaya.Agroforestry Systems, 12, 251-255.

BreWbaker, J.L. 1990. Nitrogen fixing trees. In: Fast growingtrees and nitrogen fixing trees, edited by D. Werner & PMtiller, 253-262. Stuttgart: Fischer.

Buck, M.G. 1986. Concepts of resource sharing inagroforestry systems. Agroforestry Systems, 4, 191-203.

145

Cannell, M.G.R. 1978. Improving per hectare forestproductivity In: Proceedings of the 5th North American ForestBiology Workshop, edited by CA. Hollis & A.E. Squillace,120-148. Gainesville, Florida: School of Forest Resources,University of Florida.

Charreau, C. & Vidal, P. 1965. Influence de l'Acacia albidaDel. sur le sol, nutrition minérale et rendements des milsPennisetum au Sénégal. Agronomie Tropical, 20, 600-626.

Danso, S.K.A., Bowen, MD. & Sanginga, N. 1992.Biological nitrogen fixation in trees in agro-ecosystems. Plantand Soil, 141, 177-196:

Donald, C.M. & Hamblin, J. 1976. The biological yieldand harvest index of cereals as agronornic and plantbreeding criteria. Advances in Agronomy, 28, 361-405.

Evans, D.O. & Rotar, P.P. 1987. Sesbania in agriculture.(Westview Tropical Agriculture Series, no. 8.) Boulder,Coloradb: Westview

Evensen, C.L.I. 1989. Alley cropping and green manuringfor upland crop production in West Sumatra. PhD dissertation,University of Hawaii.

Fassbender, H.W., Alpizar, L., Heuveldop, J., Folster, H.& Enriquez, G. 1988. Modelling agroforestry systems ofcacao (Theobroma cacao) with laurel (Cordia alliodora) andporo (Erythrina poeppigiana) in Costa Rica. III. Cycles oforganic matter and nutrients. Agroforestry Systems, 6, 49-62.

Fernandes, E.C.M. 1990. Alley cropping on acid soils. PhDdissertation, North Carolina State University

Greenland, D.J. 1974. Evolution and development of differenttypes of shifting cultivation. In: Shifting cultivation and soilconservation in Africa, 5-13. (FAO Soil Bulletin, no. 24.) Rome:Food and Agriculture Organisation.

Greenland, D.J. & Nye, RH. 1959. Increases in the carbonand nitrogen contents of tropical soils under natural fallows.Journal of Soil Science, 10, 284-299.

Halenda, C. 1989. Nutrient content of an Acacia mangiumplantation. Nitrogen Fixing Tree Research Reports, 7, 46-48.

Haridasan, M. 1982. Aluminum accumulation by somecerrado native species of central Brazil. Plant and Soil, 65,265-273.

Heuveldop, J., Fassbender, H.W., Alpizar, L.,Enriquez, G. & Folster, H. 1986. Modelling agroforestrysystems of cacao (Theobroma cacao) with laurel (Cordiaalliodora) and poro (Erythrina poeppigiana) in Costa Rica. I.Inventory of organic matter and nutrients. AgroforestrySystems, 4, 175-189.

Hocking, D. & Rao, D.G. 1990. Canopy managementpossibilities for arboreal Leucaena in mixed sorghum andlivestock small farm production systems in semi-arid India.Agroforestry Systems, 10, 135-152.

International Atomic Energy Agency. 1975. Root activitypatterns of some tree crops. (Technical Report Series, no.170.) Vienna: IAEA.

Jambulingam, R. & Fernandes, E.C.M. 1986.Multipurpose trees and shrubs on farmlands in Tamil NaduState, India. Agroforestry Systems, 4, 17-32.

Jenik, J. 1978. Roots and root systems in tropical trees:morphologic and ecologic aspects. In: Tropical trees as livingsystems, edited by PB. Tomlinson & M.H. Zimmermann,323--349. Cambridge: Cambridge University Press.

Kang, B.T., Reynolds, L. & Atta-Krah, A.N. 1990. Alleyfarming. Advances in Agronomy, 43, 315-359.

Kang, B.T., Wilson, G.F. & Sipkens, L. 1981. Alleycropping maize (Zea mays L.) and leucaena (Leucaenaleucocephala Lam.) in southern Nigeria. Plant and Soil, 63,165-179.

146

Kass, D.L., Barrantes, A., Bermudez, W, Campos, W.,Jimenéz, M. & Sanchez, J. 1989. Resultados de seis ariosde investigacion de cultivo en callejones (alley cropping), en"la montaria", Ilirrialba, Costa Rica. El Chasqui, 19, 5-24.

Kessler, J.J. 1992. The influence of kaarité (Vitellariaparadoxa) and néré (Parkia biglobosa) trees on sorghumproduction in Burkina Faso. Agroforestry Systems, 17,97-118.

Lundgren, B.L. & Raintree, J.B. 1983. Sustainedagroforestry In: Agricultural research for development:potentials and challenges in Asia, edited by B. Nestel, 37-49,The Hague: ISNAR.

Malik, R.S. & Sharma, S.K. 1990. Moisture extraction andcrop yield as a function of distance from a row of Eucalyptustereticornts. Agroforestry Systems, 12, 187-195.

Marschner, H. 1986. Mineral nutrition of higher plants.London: Academic Press.

Melillo, J.M., Aber, J.D. & Muratore, J.F. 1982. Nitrogenand lignin control of hardwood leaf litter decompositiondynamics. Ecology, 63, 621-626.

Miller, H.C. 1984. Dynamics of nutrient cycling in plantationecosystems. In: Nutrition of plantation forests, edited by G.D.Bowen & E.KS. Nambiar, 53-78. London: Academic Press.

Monteith, J.L., Ong, C.K. & Corlett, J.C. 1991.Microclimatic interactions in agroforestry systems. In:Microclimatic interactions in agroforestry systems, edited byPG. Jarvis, 31-43. Amsterdam: Elsevier.

Nair, P.K.R. 1984 Classifications of agroforestry systemsAgroforestry Systems, 3, 97-128

Nair, RK.R., Fernandes, E.C.M. & Warnbugu, P. 1984Multipurpose trees and shrubs for agroforestry AgroforestrySystems, 2, 145-163

National Research Council. 1983. Calliandra: a versatilesmall tree for the humid tropics. Washington, DC: NationalAcademy Press.

National Research Council. 1984. Leucaena: promisingforage and tree crop for the tropics. 2nd ed. Washington, DC:National Academy Press.

Okafor, J.C. 1982. Promising trees for agroforestry insoutheastern Nigeria. In: Agro-forestry in the African humidtropics, edited by L.H. MacDonald, 103-107. Tokyo: The UnitedNations University

Okigbo, B.N. 1983. Plants and agroforestry in land usesystems of West Africa. In: Plant research and agroforestry,edited by PA. Huxley 25-42. Nairobi: International Centre forResearch in Agroforestry

Ong, C.K., Corlett, J.E., Singh, R.P. & Black, C.R. 1991.Above- and belowground interactions in agroforestry systems.Forest Ecology and Management, 45, 45-47.

Palm, C.A. 1988. Mulch quality and nitrogen dynamics in analley cropping system in the Peruvian Amazon. PhDdissertation, North Carolina State University

Parton, W.J., Schimel, D.S., Cole, C.V. & Ojima, D.S.1987. Analysis of factors controlling soil organic matter levelsin Great Plains grasslands. Soil Science Society of AmericaJournal, 51, 1173-1179.

Peoples, M.B. & Berridge, D.F. 1990. Nitrogen fixation bylegumes in tropical and subtropical agriculture. Advances inAgronomy, 44, 155-223.

Ramiréz, C., Sanchez, G., Kass, D., Viquez, E.,Sanchez, J., Vasquez, N. & Ramirez, G. 1990. Advancesin Erythrina research at CATIE. In: Fast growing trees andnitrogen fiking trees, edited by D. Werner & P Mailer, 96-105.Stuttgart: Fischer.

Rosecrance, R.C., Brewbaker, J.L. & Fownes, J.H. 1992.Alley cropping of maize with nine leguminous trees.Agroforestry Systems, 17, 159-168.

Ruhigwa, B.A., Gichurn, M.P., Marnbani, B. & Tariah,N.M. 1992. Root distribution of Acioa barteri, Alchorneacordifolia, Cassia siarnea and Gmelina arborea in an acid

Agroforestry Systems, 19, 67-78.

Sae-Lee, S., Vityakon, P. & Prachaiyo, B. 1992. Effects oftrees on paddy bund on soil fertility and rice growth innortheast Thailand. Agroforestry Systems, 18, 213-223.

Sanchez, P.A. 1987. Soil productivity and sustainability inagroforestry systems. In: Agroforestry: a decade ofdevelopment, edited by HA. Stepp ler & PK.R. Nair, 205-223.Nairobi: International Centre for Research in Agroforestry

Sanchez, P.A. & Miller, R.H. 1986. Organic matter and soilfertility management in acid soils of the tropics. Transactionsof the 13th International Congress of Soil Science, 6, 609-625.

Sanginga, N., Bowen, G.D. & Danso, S.K.A. 1990.Assessment of genetic variability for N2 fixation between andwithin provenances of Leucaena leucocephala and Acaciaalbida estimated by I sN labelling techniques. Plant and Soil,127, 169-178.

Singh, R.R, Ong, C.K. & Saharan, N. 1989. Above andbelow ground interactions in alley cropping in semi-arid India.Agroforestry Systems, 9, 259-274.

Solera, C. 1992. Determinants of competition betweenhedgerow and alley species in contour hedgerow intercroppingsystems. PhD thesis, University of the Philippines at Los Bafios,

Stone, E.L. & Kalisz, F.J. 1991. On the maximum extent oftree roots. Forest Ecology and Management, 46, 59-102.

Skott, L.T., Fernandes, E.C.M. & Sanchez, P.A. 1991.Soil-plant interactions in agroforestry systems. Forest Ecologyand Management, 45, 127-152.

Tabora, P.C., Jr. 1991. Analysis and evaluation ofagroforestry as an alternative environmental design in thePhilippines. Agroforestry Systems, 14, 39-63.

Tiessen, H., Salcedo, I.H. & Sampaio, E.V.S.B. 1992.Nutrient and soil organic matter decomposition dynamicsunder shifting cultivation in semi-arid northeastern Brazil.Agriculture, Ecosystems and Environment, 38, 139-151.

Tiessen, H., Stewart, J.W.B. & Cole, CV. 1984. Pathwaysof phosphorus transformations in soils of differingpedogenesis. Soil Science Society of America Journal, 48,853-855.

Vandenbeldt, R.J. 1991. Rooting systems of western andsouthern African Faidherbia albida (Del.) A. Chev (syn. Acaciaalbida Del.) - a comparative analysis with biogeographicimplications. Agroforestry Systems, 14, 233-244.

Venkateswarlu, B., Korwar, G.R. & Singh, R.P. 1990.Studies on nitrogen fixation and nutrient addition by Leucaenaleucocephala in a semi-arid alfisol. Leucaena Research Reports,11, 65-67.

Young, A. 1989. Agroforestry for soil conservation. (Scienceand Practice of Agroforestry no. 4.) Wallingford: CABInternational/Nairobi: International Centre for Research inAgroforestry

147

Matching trees to their light environments

F E Fasehur0 & J Grace 2

'Department of Forestry & Resource Management, University of lbadan, Oyo Road, lbadan, Nigeria2Institute of Ecology and Resource Management, Schools of Ecological Science and Forestry, University ofEdinburgh*, Darwin Building, Mayfield Road, Edinburgh EH9 31U, UK*A component of the Edinburgh Centre for Tropical Forests

ABSTRACTRadiant energy striking the earth's surface is in the waveband 310-2300 nm, of which the visible part (light') and also thephotosynthetically active radiation (PAR) fall between 400 and 700 nm. There are differences in the incident solar radiation andpatterns of PAR in the open, in dense forest, and in gaps. The spectral shifts and light absorbance at different wavelengths,especially the red/far-red ratio, play significant roles in seed germination and plant photomorphogenesis. Plants are classified aslight-demanders and shade-tolerators based on their response to irradiance. Pioneers are light-demanders, having photosyntheticattributes similar to weeds. Compared to late successional species, they display higher stomatal conductances, light compensationpoints and dark respiration rates. Most tropical tree species used in plantations and agroforestry systems are pioneers. Researchneeds to focus on the ecophysiology of tropical trees, the influence of red/far-red ratio and sunflecks on plants growing on theforest floor, so as to match trees effectively to their light environments.

INTRODUCTION

An essential part of the domestication process inplants as well as animals is the accumulation ofknowledge on the characteristics of species, sothat strengths and weaknesses peculiar to eachgenotype may be revealed. In the case of trees,this implies an understanding of growth inrelation to the environment, and in relation tosilvicultural practices. The growth of trees in theforest is determined by the expression ofphysiological processes within specificenvironmental regimes. Of all the above-groundenvironmental factors affecting the growth anddevelopment of the tropical rainforest plants, lightis the most variable, complex and least readilyquantified. Only part of the solar radiationincident on earth (waveband 400-700 nm) isactive in photosynthesis. This waveband isreferred to as photosynthetically active radiation(PAR).

Light climates vary widely between forests,between habitats within a forest, and within asingle habitat. Nevertheless, some generalisationscan be made that are relevant to the dynamicsof tropical forests. The solar radiation incident onthe forest canopy is composed of two differentforms: direct beam, and radiation diffused by theearth's atmosphere, usually referred to simply as'direct' and 'diffuse' respectively (Chazdon 1988).These two forms penetrate the forest canopy indifferent ways, as a result of their differentangular properties (Anderson 1964a, b;Hutchinson & Matt 1977). When the sun isshining, predominantly direct-beam sunlight

148

passes through holes in the canopy arriving atthe forest floor as sunflecks. If the holes aresmall, the sunfleck is less intense than fullsunlight, but nevertheless the light is ten or 20times as bright as in the adjacent shaded region.As the solar angle changes, each sunflecktraverses across the forest floor, exposing seedsand seedlings to a large stepwise change inphoton flux density (PFD), which may in somecases be beneficial and in others deleterious(Chazdon 1988).

Diffuse light incident on the forest canopypenetrates all canopy openings, although notalways equally As both direct beam and diffuseskylight pass through vegetation, the spectralquality is altered by selective absorption,transmission and reflection of the wavelengths byfoliage, and to a lesser extent by branches andtree trunks. Thus, the light climate at a particularlocation in the forest is determined by threemajor factors, as enumerated by Chazdon andFetcher (1984a), namely:

• position of the sun (solar elevation andazimuth);

• atmospheric condition (cloudiness andaerosol content); and

• vegetation structure (pattern in vertical andhorizontal plants; leaf angle distribution).

Both the photon flux density and spectraldistribution are significantly affected by the forestvegetation. Blue and red wavelengths areselectively absorbed by chlorophyll, andshadelight is consequently enriched in green andfar-red. Sunflecks contribute significantly to the

total PAR available in the forest understorey(Chazdon 1988), typical values for Australianrainforest being 50% of the total availablephotons. Similar figures have been reported as70% for Nigerian forest (Evans, Whitmore &Wong 1960), 50% for Singapore forest (Whitmore& Wong 1959) arid 60% for forest in Ecuador(Grubb & Whitrnore 1967). Direct measurementsof photosynthesis under sunflecks have shownthat sunflecks make an important contribution tothe daily photosynthesis (Pearcy & Calkin 1983).

The light climate in canopy gaps .is highlyvariable. Much of the variability in light betweengaps can be attributed to gap size (Chazdon &Fetcher 1984b; Denslow 1987). Gap orientation,the height of the surrounding forest, topographyand solar angle are other important factorsaffecting gap light climates. The increased lightavailability in gaps is extremely important for thedistribution and the regeneration of many tropical

plant species.

The difference in red/far-red (R:FR) ratiosbetween gaps, forest canopy and shadedunderstorey may affect seed germination andseedling growth. Spatial patterns in light andshade may be explored by traversing sensorsthrough the forest, passing through gaps ofvarying size (Figure 1). Photon flux densities inthe largest gaps in a forest in Nigeria werearound 2000 pmol mm-2 s-1, whilst in thedensest shade they were as low as 50 i.unol m-2

-1s . The R:1-E. ratios were strongly correlated withphoton flux densities in this example (Figure 1).Values of R:FR ratio exceeding 1 occurred indirect sunlight, and much lower values, around0.2, occurred in the densest shade as a result of

1.20

-0 0.80P2

cc

0.40

0

2000

-ox 1200

400

0 10 20 30 40 50 60Distance (m)

Figure 1. Signatures of a disturbed forest made by traversing

light sensors along a 60 m transect of a secondary tropical

forest at Ibadan, Nigeria. The transect passes through a small

gap at 7-10 m, and larger ones at 30-42 m and 46-53 m.

The upper graph shows the red/far-red ratio, and the lower is

the photon flux density (rmol photons ITC2 s-1) (data of

J Grace, I Riddoch, F E Fasehun & D Ladipo, unpublished)

the low absorptivity of leaves to far-red radiation.Similar data from other types of tropical foresthave been reported by Chazdon and Fetcher(1984a).

PHOTOSYNTHETIC PERFORMANCE INCONTROLLED CONDITIONS

Photosynthetic adaptability to shade has beenwidely studied in herbaceous species and intemperate trees (Björkrnan & Holmgren 1963;Björkman 1981; Bazzaz & Carlson 1982). Recently,with the advent of field-portable gas exchangeequipment, there has been an upsurge of interestin tropical trees (Fasehun 1980; Langenheim etal. 1984; Mooney, Field & Vazquez-Yanes 1984;Kwesiga, Grace & Sanford 1986; Oberbauer &Strain 1984; Pearcy Osteryoung & Calkin 1985;Pearcy 1987; Thompson, Stocker & Kriedemann1988; Hogan 1988; Ramos & Grace 1990;Riddoch et al. 1991). Gas exchange parametershave been shown to depend on the light climateduring growth and leaf development, with strongevidence for a morphogenetic effect in responseto a reduced red/far-red ratio (Kwesiga et al.1986). In most of the studies, increases in theamount of light led to increases in rates ofstomatal conductance, photosynthesis and darkrespiration, and ultimately to higher growth rates.

There is considerable variation in the extent towhich the light regime during growth influencesphotosynthetic characteristics. For example,Ramos and Grace (1990) found rather smalldifferences between light response curves ofleaves grown in high and low light in four SouthAmerican species, the largest difference being inthe fast-growing Cordia alliodora (Figure 2). Asimilar variation in responsiveness between fast-and slow-growing species was noted by Riddochet al. (1991) for West African species. Thefast-growing Nauclea diderrichli displayed anextra layer of palisade cells when grown inbright light, and its leaves were increased inthickness, but the relatively slow-growingEntandrophragma angolense did not developextra palisade, and the rate of photosynthesis athigh light was less responsive than in Nauclea.Much of the observed acclimation is the result ofan increase in the amount of photosyntheticmachinery per unit area of leaf, which mustdepend on the capacity of the leaf to an. dergo aphotomorphogenetic response to a changing lightregime, as well as on the availability of nitrogen,required particularly for synthesis of thecarboxylating enzyme, RUBISCO. Theserelationships are best investigated by reference tothe relationship between photosynthetic rate, A,and internal 002 concentration, Ci, as the ratesof photosynthesis can then be observed in theabsence of stomatal limitations to the diffusion 'of

002 into the leaf. From such relationships (Figure3), the activity of RUBISCO under non-limiting

149

8

6

E 43 2

150

pmolo 800• 80

i. Brosimum alicastrum ii. Swietenia macrophylla

Cedrela odorata iv. Cordia alliodora

- a

_

fl

•

•-

•°- 2

—20 250 500 750 0 250 500

Photosynthetic photon flux density (pmol m

Figure 2. Light response curves of photosynthesis of fourtropical tree species grown under 'sun' (open squares,800 umol m-2 S-1) and 'shade' (filled squares, 80 pmol I11-2

S 1) . Each line is the non-rectangular hyperbola obtained fromthe model using the parametem in Table 1 and setting gs tothat found at the highest photon irradiance. Data points are themeans of five replicates (source: Råmos & Grace 1990). Notethat the initial slope. (quantum efficiency) is not much changed;shade-grown plants show saturation at lower photon fluxdensity; and the late successional species (i) and (ii) are verylittle affected by growth conditions

- -

0 Low light and nutrientsco • High light and nutrients•c7)a) 22C",u) 18(n'r2 E 14 •

0 oo „, • •-co.8 10 •15— 6 •

a)—

0 0 0 0f ?0 m. 2 • 0 °

Nauclea

••

• 00

• • • • Entandrophragma

I.

750 1000

0 00 00

200 600 1000 200 600 1000

Internal CO2 concentration (pmol mor)

Figure 3. A/Ci curves obtained for Nauclea diderrichii andEntandrophragma angolense at four treatments. The data arefrom the median case of three runs. Note the similaritybetween AlCi curves of the two species when the plants aregrown in low light. In high light, the Nauclea curve shows asteeper initial slope than the Entandrophragma curve, showingan enhanced capacity to fix CO2 (source: Riddoch, Lehto &Grace 1991)

002 can be found (as diI/dCi), as well as thephotosynthetic rates at 002 saturation. Therelatively fast-growing Nauclea diderrichii takesup no more N per leaf area than theslower-growing Entandrophragma when grownunder high light, but apparently uses it moreefficiently as judged by cli-I/c1Ci and Amax

(Table 1). Entandrophragma has more mass perunit area of leaf, presumably associated with thetough and long-lived characteristics of its leaves.

Leaves raised under one light regime mayacclimate when moved to a new light regime,even when they are fully expanded in the firstinstance. For example, in the Asian speciesBischofia javanica, leaves developed under highlight showed a large negative net photosyntheticrate when transferred to low light, but withinseven days the rate increased and became equalto that of leaves which had been grown underlow light (Kamaluddin & Grace 1992a). Thesechanges involved considerable re-organisation ofphotosynthetic machinery including a reductionin mass per unit leaf area, an increase inchlorophyll content and a decline in thechlorophyll a:b ratio.

The transition from low to high light is perhapsthe more interesting one, as it occurs in manysilvicultural manipulations, and also when thecanopy is opened by the falling of a tree. In thiscase, leaves which have developed at extremelylow photon fluxes are exposed suddenly to fulldaylight. It is well known that exposure ofshade-adapted leaves to high light results inphotoinhibition of photosynthesis (Kok 1956;Critchley & Smillie 1981; Langenheim et al. 1984;Oberbauer & Strain 1984). This photoinhibition isusually manifested by a reduced quantum yieldand light-saturated capacity of photosynthesis, areduced electron transport activity, and alteredchlorophyll fluorescence characteristics. It is likely

-that shade understorey seedlings suddenlyexposed to high light during gap formation inthe forest canopy always suffer photoinhibitionand chlorophyll destruction. Such processes havebeen studied in controlled environments.

Kamaluddin and Grace (1992b) examined thepotential for light acclimation in shade-grownseedlings of Bischofia javanica: they grew theseedlings under simulated forest shade light

nble I. Enhancement of RUBISCO activity Am., mass andnitrogen content in leaves developed under high (700 umolI/1-

2S-1) as opposed to low (30 umol m-2 S-1) light. All data

are on a leaf area basis (source: Table 1 of Riddoch et al.(1991)

RUBISCOactivity

per areaof leaf Amax Mass(urnol

m-25-1)

(Iunolrn- 2 s- 1)

per area per area(g nn- 2) (g,rn - 2)

Nauclea diderrichii

Low light 0.0118 7.3 23.2 0.64

High light 0.050 20.8 62.5 1.31

Enhancement factor x4.2 x2.8 x2.4 x2.0

Entandrophragma angolense

Low light 0.0125 5.6 41.6 0.69

High light 0.036 16.9 71.4 1.40

Enhancement factor x2.8 x3.0 xl .7 x2.0

r-s

1.0

0.8

0 5 10 15 20 25 30 35 40

Time (d)

Figure 4. Acclimation pf shade-grown leaves of Bischofiajavanica on transfer from simulated shade (PAR = 40 Limo! m - 2

-S 1, R:FR = 0.15) to simulated open conditions (PAR = 1200pmol mm-2 s-1, R:FR = 1.95). The chlorophyll fluorescenceratio (top graph) is a measure of photoinhibition. The isolatedpoint at the right-hand side of each graph is the value forleaves which formed in bright light, whereas the connectedvalues are leaves formed at low light which are acclimating tobright light (source: Kamaluddin & Grace 1992b)

(40 gmol m2 s-1) and, after the transfer of theseedlings to a higher irradiance (1200 grnolmm-2 S-1), they observed changes in chlorophyllinduction kinetics, net photosynthesis, leafchlorophyll and leaf anatomy. They reported thatthe leaves grown under low light displayedphotoinhibition immediately after transfer to highlight. The photoinhibition lasted for some daysbefore the leaves acclimated to the higher light(Figure 4). Although the light-saturatedphotosynthetic rate declined from 3 to 2 iimolIII-

2S-1 over seven days, and chlorophyll

content declined by 50% over 15 days, theleaves survived and acclimated. Over 20 days,leaf mass per unit area increased to almost thevalue of leaves grown in high light, and palisade

increasedthickness doubled. Photosynthetic rateto 6, grnol m 2 5-1

It is clear from this discussion that leaves haveconsiderable potential to acclimate to thechanges in photon flux that occur naturallyalthough there may well be important differencesbetween species in the ease with which thisacclimation can occur.

PHOTOSYNTHETIC RATES OBSERVEDUNDER FIELD CONDITIONS

Photosynthetic rates of nine species were studiedat the Garnbari Forest Reserve, Ibadan, Nigeria(Riddoch, Lehto & Grace 1991). This study wasdesigned to test the hypothesis of Bazzaz andPickett (1980) and Bazzaz and Carlson (1982)that pioneer species have an enhanced capacityto utilise the high photon flux density associatedwith gap formation. Species chosen included twoherbaceous weeds (Chromolaena odoratum andPiper umbellatum), as well as pioneer and latesuccessional trees. Light response curves wereobtained by aggregating measurements ofphotosynthesis obtained over the course of theday and statistical comparison was made afterfitting the non-rectangular hyperbola model ofJarvis, Miranda and Muetzelfeldt (1985) to thedata (Figure 5; Table 2). This model provides

Table 2. Mean photosynthetic parameters of a range oftropical tree species: dark respiration Rd (1=01 CO2 M -2 s-1),

quantum efficiency a (mol CO2 per mol photon), mesophyllconductance gm (mol m - 2 s-1) and light compensation pointQo (ltmol photons m - 2 s-1) (source: Table 2 from Riddoch,

Lehto & Grace 1991)

151

152

11

9

7

5

3

1

11

9

7 •

5

3

1

11

9

7

5

3

1

i. Chromolaena odoratum

iii. Ceiba pentandra

•

• •

• •

•

• •

•

v. Strombosia pustulata

•

400 800. 1200

estimates of the apparent quantum efficiency a,the mesophyll conductance, gm (synonymouswith RUBISCO activity in this analysis), darkrespiration, Rd, and the light compensation pointas 'Rd/a. No species displayed particularly highrates of photosynthetic gas exchange, the highestbeing Piper at 11 gmol m2 s-1. Chromolaena(Eupatorium) odoraturn displayed rates of only6 innol m2 s-1, despite being a fast-growingand very troublesome weed. Pioneer trees(Ceiba pentandra) performed in a similar way toChromolaena. When the parameters of the modelare compared, it is clear that, despite somevariation between the two sites, there are distinctdifferences between species: late successionaltrees have low rates of dark respiration and lowmesophyll conductances, as well as a low lightcompensation point (Rd/a).

In most studies of sun and shade acclimation, ithas been found that the apparent quantumefficiency (the initial slope of the light responsecurve) is insensitive to growing conditions

ii. Piper umbellatum•

• •

•• • •

iv. Milida excelsa

e ••

vi. Blighia sapida

Photon flux density (pmol m-2 s-')

•

•

•• •

•

•

•

•

• •

•

400 800 1200

•

Figure 5. Light response curves of seedlings at plot 2 of the study area in a tropical rainforest in Nigeria (source: Riddoch et al. 1991)

(Boardman 1977; Björkman 1981; Kwesiga et al.1986; Ramos & Grace 1990; Riddoch et al. 1991).Shade-tolerant species have low lightcompensation points, not as a result of a highquantum efficiency but because they have lowrates of dark respiration.

Several studies have been designed toinvestigate the response of leaves to sunflecks.Woodward (1981) pointed out that shade leavesmay find it difficult to tolerate the sudden energyloading associated with a sunfleck. First, theremay be a rapid rise in leaf temperature and aconsequent high transpiration rate and the onsetof water deficits. Second, the chlorophyllantennae will receive an increased photon flux ofa factor of x50 over normal, operating conditions,perhaps causing photoinhibition as suggestedfrom the laboratory experiments mentionedabove. Indeed, some species may display subtleleaf movements to reduce the energy loading ofsunflecks, as demonstrated in a South AmericanPiper by Chiariello, Field and Mooney (1987).

Some species may respond in a positive way tosunflecks, opening stomata to photosynthesiseand transpire rapidly and hence to maximisecarbon gain whilst keeping the leaf cool(Chazdon & Pearcy 1986). However, such apositive response was not found in the Nigerianstudy (Riddoch, -Lehto & Grace 1991). Instead,seedlings of late successional species on theforest floor were unresponsive, displaying little orno tendency to open their stomata in responseto strong illumination.

RESPONSE TO RED/FAR-RED RATIO

It is now widely appreciated that many of thewell-known growth responses to vegetationalshadelight are elicited not by shade per se butby the reduced red/far-red (R:FR) ratio thatvegetational shadelight contains. Shade isdetected in the plant by phytochrome, amolecule that changes state according to themagnitude of the ratio. Smith (1981) pointed outthat this arrangement is more reliable fordetecting shadelight than simply detecting areduction in photon flux density as the latterchanges with weather conditions and time ofyear, whereas reduced R:FR unambiguouslysignals the presence of other vegetation.According to Kwesiga and Grace (1986), tropicalseedlings use this detection method to sense thepresence of gaps in the canopy Consider howimportant it is for a seedling to sense when agap occurs: a late successional species mayhave survived for years as a member of theseedling bank, with a very low respiration rateand a near-zero growth rate. When a gap forms,the seedling needs to effect a metabolic changeto begin active growth. This hypothesis wassupported by the study by Kwesiga and Grace(1986) on the pioneer species Terminaliaivorensis and the relatively shade-tolerant speciesKhaya senegalensis. The former responded to areduced R:FR ratio by accelerating growth, andparticularly by leaf area expansion. This responsemay be interpreted as the correct strategy for afast-growing species of open sites in the face ofcompetitors. Khaya, on the other hand, showedthe opposite behaviour, growing in response to ahigh R:1-E ratio, such as might occur at the forestfloor when a gap forms. For any seedling on theforest floor, growth in response to a low R:FRratio would be suicidal, as fast' growth with itsassociated high rate of respiration would simplyconsume valuable resources with no hope ofattaining an emergent position in the canopy

Recent experiments suggest that the detectionsystem is in the young stem, which is especiallyresponsive to lateral illumination (Banal-6, Scopel& Sanchez 1990), As growth in height can bedetected with great precision, it is possible todesign a simple screening test for sensitivity toR:FR (Kamaluddin 1991). The phytochrome

system may also be used by seeds to detectdepth of soil and gaps in the canopy(Vazquez-Yanes et al. 1990). Many pioneerspecies produce large numbers of small seedswhich remain dormant in the surface layers ofsoil, where the R:FR ratio is very low. Speciessuch as Musanga cecropioides germinate in largenumbers when the canopy is cleared and soil isdisturbed; presumably many species rely on theabrupt change in R:FR ratio as a cue forgermination.

SURVIVAL IN SHA.DELIGHT

Light compensation points of leaves are in therange 10-50 umol photons 111-2 s-1 (Table 2); theshade under dense canopies may be below 10umol photons rn-2 s-1 for much of the day. Forthe seedling as a whole, light compensationpoints are low because much of the plantconsists of tissues which respire but have anegligible rate of photosynthesis. Most seedlingshave chloroplasts in the cells of the stem cortex,and so there may be significant recycling ofsome respired carbon. Even so, it is surprisingthat seedlings survive in such conditions. It ispresumed that such dense shade stronglydiscriminates between species, allowing survivalof only the most shade-tolerant, and that failureoccurs as a result of a negative carbon balance.

However, there are few actual studies of survivalin relation to carbon balance. Augspurger (1984)assessed the survival over one year of 18species of tree seedling at Barro ColoradoIsland, Panama (Figure 6). The seedlings weregrown in plant pots, in open and shadedlocations within a glasshouse to simulate sun andshade locations within the forest. No speciessurvived better under shade than under sun, andsurvival under sun and shade was positivelycorrelated. Generally those species known to betolerant of shade survived, with almost no lossesunder both sun and shade. Light-demandingspecies grew rapidly in sun, and successivelyyounger leaves matured at larger sizes, but therewere always mortalities. In our own study atlbadan, we have recorded a similar result usinga smaller number of species growing in theforest as wildlings. It is perhaps a general andimportant result that very shade-tolerant speciesare able to survive best in almost any conditions,whereas 'light-demanders' with high respirationrates and high light compensation points mayeasily succumb to adverse environmental factors,including attack by insects and fungi as well asabiotic factors.

One intriguing possibility as yet not examinedcritically in the field, is that mycorrhizalconnections between parent trees and theiroffspring may provide a significant carbon sourcefor seedlings (Read 1992). If this hypothesis istrue, growth of seedlings ,under shade may to

153

154

0 10 20 30Time (weeks)

40 50

Figure 6. Survival of three contrasting species placed in openand shaded positions. Ochroma pyramidale is a veryshade-intolerant species which requires large gaps; Ceiba

pentandra is a pioneer that tolerates semi-shade in smallgaps; and Myroxylon balsamum is very shade-tolerant. This isa sample of three from the 18 species, redrawn fromAugspurger (1984)

some extent be decoupled from the light climatemeasured at the seedling.

RECOMMENDED SCREENING TRIALS ASAN AID TO DOMESTICATION

At present, there is a worldwide effort tomaximise the benefit of forest resources throughselection, domestication and breeding, as well asthrough silvicultural practices aimed atsustainable land use. For tree species whichhave been exploited in the past, there may besome 'reputation' among foresters and ecologistsregarding their behaviour in relation to light, anddesignations such as 'shade-loving' or'light-demanding' can be found in the literature(Taylor 1960). However, these designations maynot be sufficient, and certainly do not cover all

species, not to mention new selections or clonesproduced in domestication programmes bygenetic selection or engineering.

It would clearly be beneficial to screen samplesof seedlings for their reaction to shade.Shade-tolerant genotypes, for example, may berequired in enrichment planting. Rather fewauthors have discussed the experimentalproblems of screening (Lee 1988). The literatureis not always helpful, as most authors haveexamined a small group of species, and eachstudy has had its own characteristic way ofsimulating shade. This problem is especiallyacute in the case of the dipterocarp speciesfrom the forests of south-east Asia. They areregarded as delicate species, but laiowledge oftheir response to canopy opening in the seedlingand sapling phase is scant (Seibert 1990).

In this section we make some recommendationsconcerning the implementation of a screeningproject in the nursery

Measurement of microclimate

Quantum sensors and sensors are requiredto measure photon flux density and R:FR ratio.rIWo sensors of each type would enablecomparison of the experimental shade conditionswith an open situation. To measure temperatureand humidity a supply of thermometers,thermocouples, psychrometer units or relativehumidity sensors are required. To record thedata, a sensitive microvoltmeter with data loggershould ideally be available.

Manipulation of shadelight

It is not sufficient simply to reduce photon fluxby slats or nets, as much of the responseexhibited by seedlings is to R:FR ratio. Inlaboratory conditions, we use combinations ofcelluloid filters, designed for use with theatrelights. This material is much cheaper than opticalglass filters, and seems to be able to withstandcontinuous exposure to high-intensity lighting.However, it has not been tested under tropicalconditions, and it may respond adversely to highultraviolet radiation and rain. Screening trialsmight, therefore, be better undertaken in aprotected environment, possibly a well-ventilatedpolythene tunnel or frame.

Experimental units

In the glasshouse, we use plastic tubes to growsmall seedlings and to maintain individual controlover light climate and microclimate. Normal plantpots are avoided as they provide insufficientrooting volume and thus cause drought. Thisarrangement is only suitable for very smallseedlings. For plants beyond the three-leaf stage,the seedlings outgrow the tubes and must thenbe grown under large filters supported in frames.

This arrangement is less satisfactory as problemsof edge effects arise, and statisticalindependence is sacrificed. If plants are to begrown beyond the six-leaf stage, it is better toallow them to root in the natural soil. Otherwise,they will outgrow their root containers, and thelargest plants .will suffer first, thus obscuringtreatment effects.

Nutrient supply

In most of our experiments, we have usedperlite/vermiculite as a suitable inert mixture towhich we add nutrient solutions in a controlledmanner. For screening trials, this may not bepractical, as the medium may not be locallyavailable, or be too expensive to use on a largescale. Forest soil is very variable, and itsproperties change rapidly with depth. It does,however, contain spores of mycorrhizal fungi, andthis may be preferred (cf Lapeyrie & Hogberg,pp158-164; Mason & Wilson, pp165-175). Toovercome problems of variability large suppliesof soil should be collected from one place, andmixed well. The supply should not be storedwhere it will be subject to the leaching effect ofrain.

Variables to be measured

Growth may be assessed non-destructively bymeasuring stem height and counting leaves atsuccessive intervals. Experience suggests that 20replicates per treatment are required for 95%confidence limits to be ±10% (assuming allreplicates survive). Where there is sufficientmaterial to permit destructive sampling, andwhere there are adequate drying and weighingfacilities, growth analysis may be used as it ismuch more informative (see Hunt 1978). Evenvery limited destructive harvesting of leaves, toobtain the specific leaf area (area of leaf permass of leaf), can be very valuable, as thisvariable is particularly sensitive to changes inthe light climate. Photosynthetic response curvesrequire expensive instrumentation and a muchhigher level of staff training to operate theequipment successfully

Analysis of data

Large data sets are generated in this kind ofwork, and it is unwise to rely on manualmethods of data processing. Data should becollected in a notebook and transferred to aspreadsheet program on a computer. The dataanalysis can then be accomplished with anappropriate statistical package. It may beadvantageous to collect such data in a standardmanner, and to manage data bases through aninternational network. As well as data on plantgrowth as a function of the light regime,specimen data on diurnal and seasonal lightclimates would also be valuable, as they would

permit the matching of species to light regime,and enable models of plant growth to beimplemented using real data.

CONCLUSIONS

Forest trees display a considerable capacity toadapt to the changes in light climate that arebrought about by treefalls and canopymanipulations in silviculture. They do, however,vary substantially in their growth responses to achange in light regime, and in their absolute andrelative growth rates. Only part of -this variationhas anything to do with photosynthesis, althoughit is photosynthesis that has mostly been studied,usually without much regard to its relationship toassimilate allocation, respiration and the wholeplant carbon balance.

Several research priorities emerge from thisreview.

• There are relatively few studies on theDipterocarpaceae, despite the economicvalue of this group • and the need tounderstand their light requirements in relationto current logging practices in Malaysia andIndonesia.

• Seedlings have received much more attentionthan saplings, and it is important to lmow thequalitative and quantitative differences inplant response to light as individuals passthrough the different stages of their life cycle.

• The relationship between photosyntheticperformance, growth and the supply ofmineral nutrients is not clear. Seedlings mayhave a critical dependency on mycorrhizas,and disturbance of the forest floor duringlogging may affect the nutrient supply

REFERENCESAnderson, M.C. 1964a. Studies of woodland light climate I,the photographic computation of light conditions. Journal ofEcology, 52, 27-41.

Anderson, M.C. 1964b. Studies of the woodland climate II,seasonal variation in the light climate. Journal of Ecology, 52,643-663.

Augspurger, C.K. 1984. Light requirement of neohopicaltree seedlings: a comparative study of growth and survival.Journal of Ecology, 72, 777-792.

Ballaré, C.L., Scopel, A.I. & Sanchez, R.A. 1990. Far-redradiation reflected from adjacent leaves: an early signal ofcompetition in plant canopies. Science, 247, 329-332.

Bazzaz, F.A. 1979. The physiological ecology of plantsuccession. Annual Review of Ecology and Systematics, 10,351-371.

Bazzaz, F.A. & Carlson, LW. 1982. Photosyntheticacclimation to variability in light environment of early and latesuccessional plants. Oecologia, 54, 313-316.

Bazzaz, F.A. & Pickett, S.T.A. 1980. Physiological ecologyof tropical succession. A comparative review. Annual Review ofEcology and Systematics, 11, 287-310.

Björkman, 0. 1981. Responses to different quantum fluxdensities. In: Encyclopedia of plant physiology: vol. 12a,

155

physiological ecology, edited by al,. Lange, PS. Nobel,C.B. Osmond & H. Ziegler, 57-107. Berlin: Springer.

Björkman, 0. & Holmgren, P. 1963. Adaptation ofphotosynthetic apparatus to light intensity in ecotypes fromexposed to shaded habitats. Physiologia Plantarurn, 16,889-914.

Boardman, N.K. 1977. Comparative photosynthesis of sunand shade plants. Annual Review of Plant Physiology, 28,355-377.

Chazdon, R.L. 1988. Sunflecks and their importance toforest understory plants. Advances in Ecological Research, 18,1-63.

Chazdon, R.L. & Fetcher, N. 1984a. Photosynthetic lightenvironments in a lowland tropical rainforest in Costa Rica,Journal of Ecology, 72, 553-564.

Chazdon, R.L. & Fetcher, N. 1984b. Light environments oftropical forests. In: Physiological ecology of plants of the wettropics, edited by E. Medina, HA. Mooney & C.Vazquez-Yanes, 27-36. The Hague: Junk.

Chazdon, R.L. & Pearcy, LW. 1986. Photosyntheticresponse to light variation in rainforest species II. Carbongain and photosynthetic efficiency during lightflecks.Oecologia, 69, 524-531.

Chiariello, N.R., Field, C.B. & Mooney, H.A. 1987.Mid-day wilting in a tropical pioneer tree. Functional Ecology,1, 3-13.

Critchley, C. & Smillie, R.M. 1981. Leaf chlorophyllfluorescence as an indication of high light stress(photoinhibition) on Cucumis sativus, L. Australian Journal ofPlant Physiology, 67, 1161-1165.

Denslow, J.S. 1987. Tropical rainforest and tree speciesdiversity Annual Review of Ecology and Systematics, 18,431-451.

Evans, G.C., Whitmore, T.C. & Wong, T.K. 1960. Thedistribution of light reaching the ground vegetation in atropical rainforest. Journal of Ecology, 48, 193-204.

Fasehun, F.E. 1980. The effects of irradiance on growthrespiration, and nitrate reductase activity of 7brminaliaivorensis and 7brminalia superba. Plant Physiology, 48, 574-577.

Grubb, P.J. & Whitmore, T.C. 1967. A comparison ofmontane and lowland forest in Ecuador III. The light reachingthe ground vegetation. Journal of Ecology, 55, 33-57.

Hogan, K.P. 1988. Photosynthesis in two tropical palmspecies. Functional Ecology, 2, 371-377.

Hunt, R. 1978. Plant growth analysis. London: Arnold.

Hutchinson, B.A. & Matt, D.R. 1977. The distribution Ofsolar radiation within a deciduous forest. EcologicalMonographs, 47, 185-207.

Jarvis, P.G., Miranda, H. & Muetzelfeldt, R.I. 1985.Modelling canopy exchanges of water vapour and carbondioxide in coniferous forest plantations. In: The forestatmosphere interaction, edited by BA. Hutchinson & B.B. Hicks,521-542. Dordrecht: Reidel.

Kamaluddin, M. 1991. The growth and physiology oftropical tree seedlings in relation to light. PhD thesis,University of Edinburgh.

Kamaluddin, M. & Grace, J. 1992a. Acclimation inseedlings of a tropical tree Bischofia javanica following astepwise reduction in light. Annals of Botany, 69, 557-562.

Kamaluddin, M. & Grace, J. 1992b. Photoinhibition andlight acclimation in seedlings of Bischofia javanica, a tropicaltree from Asia. Annals of Botany, 69, 47-52.

Kok, B. 1956. On the inhibition of photosynthesis by intenselight. Biochimica et Biophysica Acta, 21, 234-244.

156

Kwesiga, F.R. & Grace, J. 1986. The role of red/far-redratio in the response of tropical tree seedlings to shade.Annals of Botany, 57, 283-290.

Kwesiga, F.R., Grace, J. & Sanford, A.P. 1986. Somephotosynthetic characteristics of tropical timber trees asaffected by the light regime during growth. Annals of Botany,58, 23-32.

Langenheim, J.H., Osmond, C.B., Brooks, A. &Ferar, P.J. 1984. Photosynthetic responses to light inseedlings of selected Amazonian and Australian rainforest treespecies. Oecologia, 63, 215-224.

Lee, D.W. 1988. Simulating forest shade to study thedevelopmental ecology of tropical plants: juvenile growth inthree vines in India. Journal of Tropical Ecology, 4, 281-292.

Mooney, H.A., Field, C. & Vazquez-Yanes, C. 1984.Photosynthetic characteristics of tropical forest plants. In:Physiological ecology of plants of the wet tropics, edited byE. Medina, HA. Mooney & C. Vazquez-Yanes, 113-128. TheHague: Junk.

Oberbauer, S.F. & Strain, B.R. 1984. Photosynthesis andsuccessional status of Costa Rican rainforest trees.Photosynthesis Research, 5, 227-232.

Pearcy, R.W. 1987. Photosynthetic gas exchange responsesof Australian tropical forest tree in canopy gap and understorymicro-environments. Functional Ecology, 1, 169-178.

Pearcy, R.W. & Calkin, H.W. 1983. Carbon dioxideexchange of C3 and C4 tree species in the understory of aHawaiian forest. Oecologia, 58, 26-32.

Pearcy, R.W., Osteryoung, K. & Calkin, H.W. 1985.Photosynthetic responses to dynamic light environments byHawaiian trees. Plant Physiology, 79, 896-902.

Ramos, J. & Grace, J. 1990. The effects of shade on thegas exchange of seedlings of four tropical trees from Mexico.Functional Ecology, 4, 667-677.

Read, D.J. 1992. Mycorrhizal fungi in natural andsemi-natural plant communities. In: Ecophysiology ofectomycorrhkal forest trees, vol. 7, 27-53. Falun: MarcusWallenberg Foundation,

Riddoch, I., Lehto, T. & Grace, J. 1991. Photosynthesis oftropical tree seedlings in relation to light and nutrient supplyNew Phytologist, 119, 137-147.

Riddoch, I., Grace, J., Fasehun, F.E., Riddoch, B. &Ladipo, D.O. 1991. Photosynthesis and successional status ofseedlings in a tropical semi-deciduous rainforest in Nigeria.Journal of Ecology 79, 491-503.

Seibert, B. 1990. Light requirements of seedlings of someDipterocarpaceae in the field and in green- and shadehouse. Aliterature review and annotated bibliography. (GFG Report, no.16.) Samarinda: Indonesian-German (GTZ) Forestry Project.

Smith, H. 1981. Light quality as an ecological factor. In:Plants and their atmospheric environment, edited by J. Grace,E.D. Ford & PG. Jarvis, 93-110. Oxford: Blackwell Scientific.

Taylor, C.J. 1960. Synecology and silviculture in Ghana.Edinburgh: Nelson.

Thompson, W.A., Stocker, G.C. & Kriedemann, P.E.1988. Growth and photosynthetic response to light andnutrients of Flindersia brayleyama. F Micell, a rainforest treewith broad tolerance of sun and shade. Australian Journal ofPlant Physiology, 15, 299-315.

Vazquez-Yanes, C., Orozco-Segovia, A., Rinco, E.,Sanchez-Coronado, M.E., Huante, P., Toledo, J.R. &Narradas, V.L. 1990. Light beneath the litter in a tropicalforest: effect on seed germination. Ecology, 71, 1952-1958.

Whitmore, T.C. & Wong, T.K. 1959. Patterns of sunfleckand shade in tropical rainforest. Malayan Forester, 22, 50-62.

Woodward, EL 1981. Shoot extension and water relations ofCirmea lutetiana in sunflecks. In: Plants and their atmosphericenvironment, edited by J. Grace, ED. Ford & PG. Jarvis, 83-91.Oxford: Blackwell Scientific.

157

Harnessing symbiotic associations: ectomycorrhizas

F Lapeyriel & P Högberg 2

1Institut National de Recherches Agronomiques, Centre de Recherches Forestières de Nancy Laboratoirede Microbiologie, 54280 Champenoux, France2Section of Forest Science, Department of Ecological Botany, University of Umeå, S-901 87 Umeå, Sweden

ABSTRACTEctomycorrhizas are widespread in native tropical forests, and compatible ectomycorrhizal fungi will usually quickly colonise anynursery as long as no adverse treatments are applied such as soil fumigation or very high applications of fertilizers. It is now

possible. for the forester to control the nursery rnicroflora and to introduce selected efficient fungi which will then be carried onto the plantation site. The control of mycorrhizal microflora may be viewed as another tool (like breeding or fertilization) toimprove tree survival and wood production, and in this way may comprise part of the domestication process.

Most experimental results published so far concern Pinus, Eucalyptus and Pseudotsuga species. rIWo years after planting,depending on site, tree species or fungal species used, wood volume production can be increased by a factor of 1.3-1.9.Survival after plantation may be multiplied by a factor of 1.3, and up to 80% of the fertilizer normally applied can be saved.These field results have mainly been published during the 1980s, and none were published before 1976, so that we are still inthe research phase of ectomycorrhizal application to forestry Even when using well-domesticated tree species, we are stillexperiencing both successes and failures, so the outcome of inoculating relatively undomesticated species is even lesspredictable, although the potential for improvements in growth and survival is clearly high.

INTRODUCTION

Mycorrhizas are in fashion. Referring to theenthusiasm for mycorrhizas among youngforesters in the tropics, J Harley wrote that their'popularity reflects almost a belief in magic, noless; a belief that some expedient, incompletelyunderstood by most foresters, will right manyfaults in tree growth and increase production' (LeTacon & Harley 1990). The aim of this paper isto replace concepts of magic with the factsabout the technologies available and the potentialof controlled ectomycorrhizal inoculation topromote the domestication of tropical tree species.

DISTRIBUTION OF MYCORRHIZAS

Virtually any tree growing in native forest bearseither ectomycorrhizas or endomycorrhizas.Some trees, eg Eucalyptus species, somemembers of the Dipterocarpaceae belonging tothe genus Hopea, and Afzelia africana,simultaneously bear both types of mycorrhiza onthe same root system. In Eucalyptus roots, bothtypes have even been described on the sameroot apex, one over the other (Lapeyrie &Chilvers 1985). The interactions between bothsyrnbionts on the same tree are still to beunderstood; they may be complementary or evenantagonistic, depending on environmentalconditions. Indeed, according to Arnorim andMuchovej (1993), endomycorrhizas inhibit thegrowth of Eucalyptus in the nursery

Vesicular—arbuscular (VA) endomycorrhizas are

158

more common among species of tropical treethan ectomycorrhizas (Janos 1983; Newman &Redden 1987). However, ectomycorrhizal speciessometimes dominate quite extensive forestecosystems, such as rainforests withDipterocarpaceae in SE Asia or rainforest andsavanna woodland, with Caesalpinioideae inAfrica (Alexander 1989; Hogberg 1989). It isfrequently stated (eg Malloch, Pirozynski & Raven1980) that such forests are typically monospecific.Closer examinations have revealed that VAendomycorrhizal species may sometimes alsoform low-diversity stands (Hogberg 1989, 1992),and that systems dominated by ectomycorrhizalspecies may be quite diverse, as for examplesome African miombo woodlands, which containboth Caesalpinioideae and Dipterocarpaceae, aswell as additional genera forming ectomycorrhizas(Hogberg & Piearce 1986). In Tanzanian miombowoodland, approximately 40% of species werefound to be endomycorrhizal and 20% wereectomycorrhizal, with 50% of the trees bearingendomycorrhizas and 20% bearingectomycorrhizas. In this instance, 60% of treevolume was associated with ectomycorrhizalfungi, while only 20% was associated withendomycorrhizal fungi (Hogberg 1992).

It has been speculated that a lower diversityof forest stands dominated by ectomycorrhizalspecies could be a result of the lower numberof hosts being compatible with theectomycorrhizal mycelial network (eg Hogberg1992). Many species, not only woody are

capable of sharing the benefits of a VAendomycorrhizal network, and this shouldpromote a higher diversity of VAendomycorrhizal stands. Specificity is crucial; tosupport the hypothesis, there have been fewreports of the extent of specificity among tropicaltree species and ectomycorrhizal strains. Smits(1983, 1985) claimed that some dipterocarps onlydeveloped ectomycorrhizas with inoculumisolated from the same host species. Inoculationtests on Caesalpinioideae, by Ba and Thoen(1990) using fungi from Uapaca spp., have notindicated any compatibility Based on results fromthe temperate zone (Duddridge 1986), there hasbeen some controversy about this issue. Cases ofstrict specificity concerning the wholeectomycorrhizal network could be ratherexceptional, and 'specific aggressivity' and/or'metabolic specificity' may be much moreprevalent. Cases of specific aggressiveness havebeen reported where, among compatible treesand fungi, only some pairings are able todevelop quickly the proper infection structure(Lei et al. 1989, 1990; Malajczuk, Lapeyrie &Garbaye 1990). Evidence of 'metabolic specificity'has been given by Dell et al. (1989), showingthat a given fungus could operate differentmetabolic pathways, depending on the host plantit is infecting.

It is now possible for foresters to manipulate thesymbiotic fungal microflora in nurseries, andconsequently in plantations. Prime candidates forinoculation tests among indigenous tropicalectomycorrhizal species are members of theDipterocarpaceae from the rainforests of SE Asiaand of the Caesalpinioideae from the rainforestand savanna woodlands of Africa.

REVIEW OF EXPERIMENTAL PLANTATIONSWORLDWIDE

r1Wenty-seven published accounts of results fromforest plantations have been reviewed to judgethe potential of ectomycorrhizal inoculations(update of the review by Garbaye 1991). Fromthis review (see Appendix), it is obvious that weare still in an early phase of mycorrhizalresearch applied to forestry as these studieshave been published only in the past 10-15years. In addition, 66% of them concerned onlythree tree genera (Pinus, Eucalyptus andPseudotsuga), and only seven of the studiescontained comparative plantations established intropical dry or humid climates. Furthermore, thefungal strains tested belonged to only 18 fungalspecies, with 60% of the results obtained withonly three species (Pisolithus tinctorius, Laccarialaccata and L. bicolor) and often the same isolate.Therefore, whether using well-domesticated treespecies or new undomesticated tropical trees, westill have much to learn from both success andfailure.

Of great interest from this survey is the growthstimulation following inoculation of seedlings inthe nursery recorded one to eight years afterplanting. Depending on the site, the tree species,and the fungus used, wood volume productionhas been multiplied in many cases by 1.5-2.5.This fact is of particular interest considering thatthe extra cost due to nursery inoculation is onlya few per cent of total plantation cost. Theeffects are even more dramatic on very poorand/or toxic sites, on which wood volumeproduction was multiplied six- or even 40-fold.Survival after plantation was increased by 30% inmost cases and much more (up to 39-fold) ontoxic sites.

Interesting conclusions can be drawn from acloser look at two field experiments.

Philippines

In 3-year-old experimental Eucalyptus degluptaplantations on industrial plantation sites, the extrawood production attributable to inoculation was154% (De la Cruz & Lori lla 1990). Comparisonof wood production from non-inoculated andinoculated plots, with increasing application ratesof a complete fertilizer, showed that, to reachmaximum production, mycorrhizal inoculationcould replace about 86% of the fertilizersrequired by uninoculated trees. If fertilizers arereadily available at a reasonable cost, then thesame experiment shows that, even at highfertility the trees on the inoculated plots madebetter use of the fertilizers.

Congo

Pisolithus tinctorius stimulated the growth ofEucalyptus urophylla x E. kirtoniana (Garbaye,Delwaulle & Diangana 1988). The most importantgrowth stimulation was recorded at 27 months,with 61% extra wood production per hectare.After 50 months, the increase in wood productionhad dropped to only 29%, apparently becausethe introduced strain had disappeared from theroot system. This result may indicate that theinoculated strain had poor 'specificaggressiveness' (Malajczuk et al, 1990) onEucalyptus roots, making it a poor competitoragainst indigenous unselected strains. Therefore,a selected strain must be efficient, andaggressive, to be competitive.

STRATEGIES FOR STRAIN SELECTION

Very different techniques are available for fungalinoculum production. Some can be producedonly by industry and applied in largecommercial nurseries, others require theequipment of a basic microbiology laboratorywhile some can be applied directly at thenursery level for no real cost. The strategy forstrain selection and multiplication depends on thesituation and rationale for inoculating the species.

159

On mining sites, for example, there may be noinoculum available locally and any fungus can beintroduced, with a great chance of success interms of growth stimulation, so long as it isadapted to the new ecosystem and able to infectthe tree being planted. In some situaticins,theremay be no indigenous inoculurn compatible withthe exotic tree being introduced, as was thecase with Pinus introductions into Australia andAfrica. In this situation, any inoculum collected inthe area of origin of the exotic tree should be asatisfactory solution: it could be soil, spores orpure culture.

On very poor toxic soils, dry land or calcareousareas, a fungal strain adapted to this environmentand able to correct the physiological disordersuffered by the tree should be selected. Thissituation is more complex, as the physiologicalprocesses involved are in general still poorlyunderstood (eg Lapeyrie 1990). Basic research,as well as time-consuming strain screening, isrequired to achieve success.

Looking for a fungal strain which is moreefficient than locally abundant strains that arewell adapted to their environment and to theirhost plant is a difficult task. Further, there isoften a lack of correlation or even a negativecorrelation between the efficiencies of strains inthe nursery and in the field (eg Grove et al.1991). Newly introduced fungal strains should beefficient and competitive. They can be wildisolates or even genetically improved strains,such as the efficient strain of Laccaria laccatarecently obtained through a breeding programme(W Yan et al., unpublished data).

The case of dipterocarps is interesting, asplantations are being established (see Smits, deFraiture & Yasman, pp267-272), and it has beensuggested that the poor growth and survival ofthese species in logged forest are due to theabsence or. death of the mycorrhizal fungi,resulting from logging operations and soiltemperature elevation (Lee 1990). Investigation ofthe mycorrhizal status of wildlings and seedlingsin nurseries and after plantation establishmentmakes it difficult to accept this hypothesis, as allof them are infected by numerousectomycorrhizal fungi (Lee 1990). Mycorrhizasadapted to the nursery environment areabundant, but their ability to adapt to conditionsin the field after planting could affect their abilityto survive, to colonise the soil, and to extractnutrients from mineral and organic matter(Nguyen et al. 1992; Dighton 1991). Furthermore,their metabolic interaction with the host plant isnot laiown and could be critical.

Few research programmes aimed atunderstanding the mycorrhizal status of seedlingsor cuttings in the nursery have been initiated.The inoculum usually used is either soil fromnative forest, pieces of carpophores (eg Hadi &

160

Santoso 1988), or pure cultures (eg, Lapeyrie,Lee & Sanip 1993). As stated previously onlypure-culture mycelia will allow the selection anduse of permanently efficient strains. Recentresearch testing the specificity of the mycorrhizalassociations in dipterocarps has found that youngseedlings of Hopea odorata and H. helferibedome well infected after inoculation with aSouth American strain of Pisolithus tinctoriuscollected under Eucalyptus (Y Sanip, S S Lee &F Lapeyrie, unpublished). In this instance, heightgrowth was enhanced by a factor of 1.5 afterfour months. In addition, Hopea odorata cuttingswere extensively infected (80% of root apices)by a strain of Hebeloma crustuliniforme collectedin France under Picea (Lapeyrie et al. 1993).These promising preliminary results suggest that,in terms of inoculum specificity dipterocarpsprobably behave like temperate tree species.However, germplasm banks of mycorrhizal fungiin pure culture, isolated from native dipterocarpforest, are now urgently needed.

CONCLUSION

Inoculation of seedlings with selected fungi canpromote symbiotic ectomycorrhizal associations,and improve the establishment and growth oftree seedlings. In this way ectomycorrhizalinoculation may contribute to the domesticationprocess, so long as it is integrated in a soundprogramme of nursery and reforestationtechnologies.

REFERENCESAlexander, I.J. 1989. Mycorrhizas in tropical forests. In:Mineral nutrients in tropical forest and savanna ecosystems,edited by J. Proctor, 169-188. Oxford: Blackwell Scientific.

Amorim, E.F.C. & Muchovej, R.M.C. 1993. Developmentand effect of endo- ancl/or ectomycorrhizal fungi on seedlingsof Eucalyptus grandis Plant and Soil. In press.

Ba, A.M. & Thoen, D. 1990. First syntheses ofectomycorrhizas between Afzelia africana Sm.(Caesalpinioideae) and native fungi from West Africa. NewPhytologist, 114, 99-103.

Berry, C.R. & Marx, D.H. 1978. Effects of Pisolithustinctorius ectomycorrhizae on growth of loblolly and Virginiapine in the Tbnnessee Copper Basin. (USDA Forest ServiceResearch Note SE-264.) Washington, DC: USDA Forest Service.

De la Cruz, R.E. & Lorilla, E.B. 1990. Ectomycorrhizaltablets for Eucalyptus species. In: Fast-growing andnitrogen-fixing trees, edited by D. Werner & P Miller, 371.Stuttgart: Fischer.

Dell, B., Botton, B., Martin, F. & Le Tacon, F. 1989.Glutamate dehydrogenases in ectomycorrhizas of spruce(Picea excelsa L.) and beech (Fagus sylvatica L.). NewPhytologist, 111, 684-692.

Delwaulle, J.C., Diangana, D. & Garbaye, J. 1987.Augmentation de la production du pin des caraibes dans larégion côtière du Congo par introduction du champignonectomycorrhizien Pisolithus tinctorius. Revue ForestièreFrancaise, 39, 409-417.

Delwaulle, J.C., Garbaye, J. & Okombi, G. 1982.Stimulation de la croissance initiale de Pinus caribaea Morelet

dans une plantation du Congo par contn5le de lamycorhization. Bois et Fbrets des Tropiques, 196, 25-32.

Dighton, J. 1991. Acquisition of nutrients from organicresources by mycorrhizal autotrophic plants Experientia, 47,362-369.

Dixon, R.K., Garrett, H.E., Cox, G.S., Johnson, P.S. &Sander, T.L. 1981. Container- and nursery-grown black oakseedlings inoculated with Pisolithus tinctorius: growth andectomycorrhizal development following outplanting on anOzark clear-cut. Canadian Journal of Forest Research, 11,492-496.

Duddridge, J.A. 1986. Specificity and recognition inmycorrhizal associations. In: Physiological and geneticalaspects of mycorrhizae, edited by V Gianinazzi-Pearson &S. Gianinazzi, 45-58. Paris: Institut National de RecherchesAgronomiques.

Garbaye, J. 1990. Les problèmes posés par la mycorhizationcontrélée du chéne. Revue Forestière Francaise, 42, 233-239.

Garbaye, J. 1991. Utilisation des mycorhizes en sylviculture.In: Les rnycorhizes des arbres et des plantes cultivées, editedby D.G. Strut lu, 197-248. Paris: Technique et DocumentationLavoisier.

Garbaye, J., Delwaulle, J.C. & Diangana, D. 1988.Growth response of eucalypts to mycorrhizal inoculation inthe Congo. Forest Ecology and Management, 24, 151-157.

Grove, T.S., Malajczuk, N., Burgess, T., Thomson, B.D.& Hardy, G. 1991. Growth responses of plantation Eucalyptusto inoculation with selected ectomycorrhizal fungi. In:Symposium on intensive forestry: the role of Eucalyptus,edited by A.PG. Schdnau, 86-93, Durban, South Africa.

Hadi, S. & Santoso, E. 1988. Effect of Russula spp.,Scleroderma sp., and Boletus sp. on the mycorrhizaldevelopment and on the growth of five dipterocarp species.In: Agricultural and biological research priorities in Asia.Proceedings of the IFS Symposium of Science Asia 87, editedby S. Mohinder Singh, 183-186. Kuala Lumpur: ForestResearch Institute of Malaysia.

Hogberg, P. 1989. Root symbioses of trees in savannas. In:Mineral nutrients in tropical forest and savanna ecosystems,edited by J. Proctor, 121-136. Oxford: Blackwell Scientific.

Hogberg, P. 1992. Root symbioses in African dry tropicalforests. Journal of Vegetation Science, 3, 393-400.

Hogberg, P. & Piearce, G.D. 1986. Mycorrhizas in Zambiantrees in relation to taxonomy vegetation communities andsuccessional patterns. Journal of Ecology, 74, 775-785.

Janos, D.P. 1983. Tropical mycorrhizas, nutrient cycles andplant growth. In: Tropical rain forest: ecology and management,edited by S.L. Sutton, TC. Whitmore & AC. Chadwick,327-345. Omford: Blackwell Scientific.

Krugner, T.L. 1976. Development of ectomycorrhizae, growth,nutrient status, and outplanting performance of loblolly pineseedlings grown in soil infested vdth Pisolithus tinctorius and

Thelephora terrestris under different fertilization regimes. PhDthesis, North Carolina State University

Krugner, T.L. & Pinto, J.R.J.E. 1982. Efeitos de inoculacaocom Pisolithus tinctorius e de fertilizacao mineral em Pinuscaribaea var. hondurensis no litoral norte da Bahia. In: Anais

da I Reuniao Brasileira de Micorrizas, edited by J.0. Siqueira& E. Oliveira, 207. Lavras, MG: Edicoes FAEPE.

Lapeyrie, F. 1990. The role of ectomycorrhizal fungi incalcareous soil tolerance by 'symbiocalcicole' woody plants.Anna les des Sciences Forestières, 21, 579-589.

Lapeyrie, F.F. & Chilvers, G.A. 1985. Anendomycorrhiza-ectomycorrhiza succession associated withenhanced growth of Eucalyptus dumosa seedlings planted incalcareous soil. New Phytologist, 100, 93-104.

Lapeyrie, F., Lee, S.S. & Sanip, Y. 1993. Controlledectomycorrhizal inoculation of Hopea odorata(Dipterocarpaceae) cuttings with Hebeloma crustuliniforme.In: Proceedings of BIOREFOR Workshop, Tsukuba, Japan. Inpress.

Le Tacon, F. & Harley, J.L. 1990. Deforestation in thetropics and proposals to arrest it. Ambio, 19, 372-378.

Le Tacon, F., Garbaye, J., Bouchard, D., Chevalier, G.,Olivier, J.M., Guirnberteau, J., Poitou, N. & Frochot, H.1988. Field results from ectomycorrhizal inoculation in France.In: Canadian Workshop on Mycorrhizae in Forestry, Quebec,edited by M. Lalonde & Y. Piché, 51-74. Quebec: UniversitéLaval.

Lee, S.S. 1990. The mycorrhizal association of theDipterocarpaceae in the tropical rain forests of Malaysia.Arabic), 19, 383-385.

Lei, J., Ding, H., Lapeyrie, F., Piche, Y., Malajczuk, N.& Dexheimer, J. 1989. Ectomycorrhizal formation on theroots of Eucalyptus globulus and Pinus caribea with twoisolates of Pisolithus tinctorius: structural and cytochemicalobservations. In: Fourth International Colloquium onEndocytobiology and Symbiosis, edited by P Nardon, Gianinazzi-Pearson, A.M. Grenier, L. Margulis & D.C. Smith,123-126. Paris: Institut National de Recherches Agronomiques.

Lei, J., Lapeyrie, F., Malajczuk, N. & Dexheimer, J.1990. Infectivity of pine and eucalypt isolates of Pisolithustinctorius (Pers.) Coker & Couch on roots of Eucalyptusurophylla S.T Blake in vitro. 2. Ultrastructural and biochemicalchanges at the early stage of mycorrhizal formation. NewPhytologist, 116, 115-122.

Malajczuk, N., Lapeyrie, F. & Garbaye, J. 1990.Infectivity of pine and eucalypt isolates of Pisolithus tinctorius(Pers.) Coker & Couch on roots of Eucalyptus urophyllaST. Blake in vitro. 1. Mycorrhizal formation in model systems.New Phytologist, 114, 627-631.

Ma lloch, D.W., Pirozynski, K.A. & Raven, RH. 1980.Ecological and evolutionary significance of mycorrhizalsymbioses in vascular plants (a review). Proceedings ofNational Academy of Sciences USA, 77, 2113-2118.

Marx, D.H. 1977. The role of mycorrhizae in forestproduction. In: TAPPI Annual Meeting Conference Papers,151-161. Atlanta, Georgia: TAPPI.

Marx, D.H. & Artman, J.D. 1979. Pisolithus tinctoriusectomycorrhizae improve survival and growth of pineseedlings on acid coal spoils in Kentucky and Virginia.Reclamation Review, 2, 23-31.

Marx, D.H., Bryan, W.C. & Cordell, C.E. 1977. Survivaland growth of pine seedlings with Pisolithus ectomycorrhizaeafter two years on reforestation sites in North Carolina andFlorida. Forest Science, 23, 263-273.

Marx, D.H., Cordell, C.E. & Clark, A. 1988. Eight-yearperformance of loblolly pine with Pisolithus ectomycorrhizaeon a good quality forest site. Southern Journal of AppliedForestry 12, 275-290.

Marx, D.H., Hedin, A. & Toe 1V S.F.P. 1985. Fieldperformance of Pinus caribaea var. hondurensis seedlings withspecific ectomycorrhizae and fertilizer after three years on asavanna site in Liberia. Forest Ecology and Management, 13,1-25.

Momoh, Z.O. & Gbadegesin, R.A. 1980. Fieldperformance of Pisolithus tinctorius as a mycorrhizal fungus ofpines in Nigeria. In: Tropical mycorrhizal research, edited byP Mikola, 72-79. Oxford: Clarendon Press.

Newman, E.I. & Reddell, P. 1987. The distribution ofmycorrhizas among families of vascular plants. NewPhytologist, 106, 745-751.

161

Nguyen, C., Yan, W, Le Tacon, F. & Lapeyrie, F. 1992.Genetic variability of phosphate solubilizing activity bymonocaryotic and dicaryotic mycelia of the ectomycorrhizalfungus Laccaria bicolor Plant and Soil, 143, 193-199.

Ofosu-Asiedu, A. 1980. Field performance of Pinus canbaeainoculated with pure cultures of four mycorrhizal fungi. In:Tropical mycorrhizal research, edited by P Mikola, 82-87.Oxford: Clarendon Press.

Ruehle, J.L., Marx, D.H., Barnett, J.P. & Pawuk, W.H.1981. Survival and grOwth of container-grown and bare-rootshortleaf pine seedlings with Pisolithus and Thelephoraectomycorrhizae. Southern Journal of Applied Forestry 5, 20-24.

Smits, W.Th. 1983. Dipterocarps and mycorrhiza. Anecological adaptation and a factor in forest regeneration. FloraMalesiana Bulletin, 36, 3926-3937.

162

Smits, W.Th. 1985. Specificity of dipteiocarp mycorrhiza. In:Proceedings of the 6th North American Conference onMycorrihizae, edited by R. Molina, 364. Corvallis, Oregon:Oregon State University

Strenstram, E., Ek, M. & Unestam, T. 1986. Prolongedeffect of initially introduced mycorrhizae on pine plants afteroutplanting. In: Physiological and genetical aspects ofmycorrhizae, edited by V Gianinazzi-Pearson & S. Gianinazzi,503-506. Paris: Institut National de Recherches Agronomiques.

Theodorou, C. & Bowen, MD. 1970. Mycorrhizal responsesof radiata pine in experiments with different fungi. AustralianForester, 34, 23-26.

Villeneuve, N., Le Tacon, F. & Bouchard, D. 1991.Survival of inoculated Laccaria bicolor in competition withnative ectomycorrhizal fungi and effects on the growth ofoutplanted Douglas fir seedlings. Plant and Soil, 135, 95-107.

Appendix. Main results of controlled ectomycorrhizal inoculation applied to forestry (update of the review by Garbaye 1991)

Reference

Ofusu-Asiedu(1980)

Delwaulle,Diangana& Garbaye (1987)

Krugner (1976) Pinus taeda

Berry & Marx(1978)

Ruehle et al.(1981)

Marx, Cordell& Clark (1988)

Marx &Artman (1979)

Marx (1977)

Marx (1977)

Theodorou &Bowen (1970)

Stenstrbm,& Unestam (1986)

Le Tacon et a/.(1988)

Le l'acon et al.(1988)

Le Tacon et al.(1988)

Le Tacon et al.(1988)

Villeneuve, LeTacon & Bouchard(1991)

Garbaye, Delwaulle Eucalyptus& Diangana (1988) urophylla

x kirtoniana

Tree species Cotmtry

Krugner & Pinto PMI's caribaea Brazil(1986)

Momoh & Pimis caribaea NigeriaGbadegesin (1981)

Pinus caribaea Ghana

Delwaulle, Garbaye Pinus caribaea Congo& Okombi (1982)

Pinus caribaea Congo

Marx, Hedin Pinus caribaea Liberia& Toe (1985)

Pinus taedaPinus virginiana

USA(N Carolina)

USA (Tennessee)

Marx, Bryan Pinus taeda USA (Florida && Cordell (1977) Pinus elliottii North Carolina) subtropical

Pinus virginiana

USA (Arkansas &Oklahoma)

Anus echinata

Pinus taeda USA (Georgia)

Pinus taeda USA (Kentucky)Pinus echinata

Pinus virginiana USA (Kentucky)

Pinus taeda USA (Georgia)

Pinus radiata

Pinus sylvestris Sweden

Pseudotsugamenziesii

Pseudotsugamenziesii

Pseudotsugamenziesii

Pseudotsugamenziesii

Pseudotsugamenziesii

Western Australia Mediterranean

France

Rance

France

France

France

Congo

Climse Soil

Dry tropical Poor acidic- non-fertilized- fertilized NPK

Semi-humid ? (six differenttropical plantation sites)

Semi-humid ? (two differenttropical plantation sites)

Humid Poor sandytropical

Humid Poor sandytropical

Humid Cleared nativetopical forest

Hot temperate Sandyacidic

Hot temperate Very poor acidicforest soil (2 sites)

Hot temperate to

Hot temperate Forest soil(two sites)

Hot temperate Fertile . forest soil

Hot temperate Coal mining spoil(very acidic andtoxic)

Hot temperate

Hot temperate Kaolin quarry spoil

Ibmperate

Temperate

Temperate

Temperate

rIbmperate

Wet tropical

•

Forest soil ofaverage fertility(two sites)

Coal mining spoil(very acidic andtoxic)

Acidic soil

Cold temperate Forest soil

Poor acidic(heathland)

Acidic (grassland)

Acidic (bracken)

Acidic forest soil

Acidic sandyforest soil

Poor sandy soil

Growth parameter

Heightafter twoyeas

Height after30 months

Height after18 months

Height after20 months

Height after12 months

Volume ha-'after three years

Height afterone year

Stem volume aftertwo years

Stem volumeafter two years

Stem volume aftertwo years

Volume ha-Iafter eight years

Stem volume aftertwo years

Stem volume% survival (aftertwo years)

% survival afterone year

Height after oneYear

Stem volumeafter two years

Height after fiveyears

Height after fouryears

Height after threeyears

Height after one

Year

Height after twoyearsStem mass aftertwo years

Vohnne ha-' afterbur years

Fungal speciesintroduced

Pisolithus tinctorius

Pisolithus tinctorius

Pisolithus tinctoriusRhizopogon luteolusThelephora ten t,stri.

Pisolithus tinctorius

Pisolithus tinctorius(spores)

Pisolithus tinctorius x 1.5Thelephora terrestris x 1unknown (from x 1.7humus)

Pisolithus tinctorius x 1.6

Pisolithus tinctofius

Pisolithus tinctorius

Pisolithus tinctorius

Pisolithus tinctorius

Pisolithus tinctorius

Pisolithus tinctorius

Pisolithus tinctorius x 1.3Cenococcum x 1.2granifonne

Rhizopogon luteolus

Lactarius rufusTricholomaalbobrunneumAmanita muscaria

Hebelomacylindrosporum

Laccaria laccata

Laccaria laccataLaccaria bicolor

Laccaria laccataLaccaria bicolor

Laccaria laccata

Pisolithus tinctoriusSclerodermaaurantium

Growth increasecompaied withnon-inoculated

plant

x 1.6x 2.4

x 1.2-1.8(depending on site)

x 0.6-1x 1x 0.8-1.2(depending on site)

x 1.2-1.5(depending onstrain)

x 1.2

x 1.5-1.9x 1.3-1.9(depending on site)

x 1.2-1.5x 1.2-1.5x 1.2-1.5(depending on site)

x 1.9-2.4(depending on site)

x 1.5

x 6.5x 5.1

x 4.3x 3.9

x 1.5

x 1.4x 1.3

x 1.2

x 0.8

x 1.2

x 1.3x 1.1

x 1.4x 1.4

x 1.5x 4.0

x 1.6x 1.5

163

Appendix. continued

Reference 'Free species Country Climate

Dixon et al. Quercus velutina USA (Missouri) Hot temperate(1981)

Garbaye (1990) Quercus robur France Temperate

De la Cruz Pinus caribaea Philippines Wet tropical& Lori lla (1990)

De la Cruz Eucalyptus Philippines Wet tropical& Lori lla (1990) deglupta

Grove et al. Eucalyptus Western Australia Mediterranean(1991) diversicolor

Eucalyptus globulus

164

Soil Growth parameter

Acidic forest soil

Acidic forest loam(three sites)

- degraded &infertile grassland- relatively goodsite

Relatively goodsite

'1Wo fertilized sites(cleared forest):- red loam- gravelly redpodzolic soil

Dry matter massafter one year

Height increaseduring first andsecond year

Stem volume:- after three years- after two years

Fungal speciesintroduced

Pisolithus tinctorius

Paxillus involutus

HebelomacrustuliniformeLaccaria laccata

Pisolithus tinctorius(spores)

x 1

x 1.4x 3.2

Stem volume Pisolithus tinctorius x 2.5after three years (spores)

Above-groundbiomass (drymass) after oneyear

Growth increasecompared withnon-inoculated

plant

x 1.2-1.4(depending on site)x 1.1

Laccaria sp. x 1.0-1.8Setchelliogaster spAmanita sp. (depending onProtubera sp. EucalyptusHydnangiurn sp. species, fungalScleroderma sp. stain and site)Hymenogaster sp.Hysterangium sp.Descolea sp.

Harnessing symbiotic associations:vesicular-arbuscular mycorrhizas

P A Mason & J WilsonInstitute of Terrestrial Ecology*, Bush Estate, Penicuik Midlothian EH26 OQB, UK

*A component of the Edinburgh Centre for Tropical Forests

ABSTRACTAlthough the majority of tropical trees form mycorrhizas with vesicular-arbuscular mycorrhizal (VA1v1) fungi, most mycorrhizal

research on tropical trees has been on ectomycorrhizal fungi associated with a few tree species which have been widely planted

as exotics in plantations. Despite this, the presence of appropriate and adequate VAM populations is likely to have a significant

impact on tree performance, as many tree koecies are highly dependent upon their mycorrhizal associatiOnsfor nutrient uptake,

particularly in infertile soils.

Although naturally occurring inoculum is widespread in soils, site disturbance (as occurs when land is cleared for planting) can

reduce the number of infective propagules, and change the proportions of different fungal species. Hence, the potential for

mycorrhizal formation may be lowered at the critical time of tree establishment. The extent to which mycorrhizal populations are

depleted and changed is related to the method of site clearance used. Short-term studies indicate that mycorrhizal inoculation

improves tree growth in nurseries. Only a few studies have followed the progress of inoculated trees after outplanting. In these

studies, improvements in both tree survival and growth rate have been found, but more extensive studies are needed.

A number of options exist for the management of mycorrhizal associations: minimising the change in indigenous populations by

use of appropriate site preparation techniques; manipulating damaged populations to increase their inoculum potential; and

inoculating trees in nurseries either with single isolates or mixtures. Where the performance of particular tree/symbiont

combinations is being assessed, the criteria for selection need to be carefully defined.

INTRODUCTION

Mycorrhizal fungi are an integral part ofpractically all plant communities, natural ormanaged, and form the link by which mineralnutrients are transferred from the soil to theplant, while carbon compounds are transportedin the opposite direction. Thus, they have afundamental role in determining plant productivityand in the functioning of ecosystems. Despitetheir importance, they tend to be ignored byforesters and agriculturalists, although anunderstanding of their role is essential to thedevelopment of sustainable land use systems(Bethlenfalvay & Linderrnan 1992).

Vesicular—arbuscular mycorrhizal (VAM) fungipossess a worldwide distribution. Publishedrecords indicate that 71% of tropical plantspecies are VAM (Sieverding 1991), and about95% of the tree species in natural tropical forestecosystems are mycorrhizal exclusively with VAMfungi (Le Tacon, Garbaye & Carr 1987). AlthoughVAIVI trees predominate in the tropics, they haveuntil recently received comparatively littleattention from mycorrhizal researchers, who havefocused instead on a few ectomycorrhizal trees,such as members of the Pinaceae, Casuarina andvarious Eucalyptus species which are widelyused in plantations (Le Tacon et al. 1987; seealso Lapeyrie & Hogberg, pp158-164). The VAmycorrhizas of the great majority of tropical

forest trees, which include many valuablespecies, are virtually unstudied. The failures ofectomycorrhizal trees as a result of mycorrhizaldeficiencies when planted as exotics are welldocumented (eg Mikola 1970). By comparison,failures or poor performance of VAM trees afterplanting have rarely been attributed tomycorrhizal problems. The more widespreaddistribution of VAM fungi might suggest thatmycorrhizal deficiencies would be unlikely tooccur; however, as this paper will indicate, thereare many circumstances in which VAMpropagules may be restricted in number or formineffective associations, resulting in treesperforming suboptimally

Consideration of VAM fungal populations is

important not only because of their role inecosystem function, but also because they have

direct and immediate effects upon treeperformance. Evidence from short-term studiessuggésts that many tropical trees are highlydependent upon their VAIVI associations forsurvival and growth. Although VAM fungi are

widely distributed, site disturbance, such asoccurs during land clearance, can adverselyaffect mycorrhizal populations, reducing theamount of inoculum available to infect rootsystems during the period when trees arebecoming established. Hence, the conservation of

indigenous VAM populations, or the inoculation of

165

trees with VAM fungi, may be needed to ensurethat adequate inoculum is present to supporttree growth.

Unfortunately while understanding of theecological role of these fungi has increased inrecent years, it has not been matched bydevelopments in their practical application ormanagement. A number of short-term studieshave demonstrated that inoculation can improvethe growth of trees in nurseries, but (with theexception of the Citrus industry) VAM inoculationof trees has not become part of nurserypractice. A serious limitation to progress is thelack of evaluations of the performance ofinoculated plants over periods longer than a fewmonths, which hampers any appraisal oflong-term effects in the field.

IMPORTANCE OF VAM FUNGI FOR PLANTGROWTH IN THE TROPICS

Function and distribution of VA mycorrhizas

The functions of VAM associations have beenreviewed extensively elsewhere (eg Harley &Smith 1983; Powell & Bagyaraj 1984; Barea 1991),and will only be summarised briefly here.However, it should be pointed out that themajority of experiments have been conducted inpots, with isolated plants. Observations of theactivities of mycorrhizas under field conditionsare far more limited, and much work remains tobe done to understand their functioning inecosystems (Fitter 1985).

The mycorrhizal fungus takes up mineralnutrients from the soil, via the extramatricalfungal mycelium which can reach for aconsiderable distance away from roots. Thepresence of a widely ramifying extramatricalmycelium is particularly important for the uptakeof minerals (P and others) that have a slow rateof diffusion (Cooper 1984). It has beendemonstrated that this mycelium can extendphosphate depletion zones up to at least 7 cmfrom roots, compared with the 1-2 mm depletionzones associated with non-mycorrhizal roots, andtherefore it can absorb nutrients from a muchgreater soil volume than a non-mycorrhizal root(Rhodes & Gerdemann 1975). The ability ofmycorrhizal fungi to access P is likely to beparticularly important to plants in dry conditions,as the diffusion coefficient for phosphate in soildecreases linearly with increasing soil dryness(Fitter 1985). In addition to the enhancement ofnutrient uptake through exploitation of increasedsoil volume, there is some evidence that VAMhyphae have the capacity to alter the weatheringrates of soil P and increase the pool of availableP (Bolan 1991). Mycorrhizal infection has alsobeen found to improve plant water relations, as aresult of either indirect effects upon plant nutrientstatus or more direct effects on water uptake

166

(Cooper 1984); improved resistance to pathogenshas also been noted (Sharma, Johri & Gianinazzi1992).

The distribution of VAM and ectomycorrhizalfungi in the tropics is reviewed briefly byLapeyrie and Hbgberg (pp158-164). In broaderterms, when the major plant communities of theworld are surveyed, it becomes apparent thateach has its distinctive population of mycorrhizaltypes: VA, ecto-, ericoid or arbutoid mycorrhizas(Read 1991a). The changes in doininance ofdifferent mycorrhizal types in contrastingecosystems have been attributed to variations intheir ability to access or mobilise the plantnutrients that are limiting growth in that particularecosystem (Read 1991b). Evidence suggests thatdomination by VAM plant species increases withdecreasing soil organic matter and increasingbase status of soil, where phosphorus replacesnitrogen as the major limiting nutrient (Read1991b), although there are exceptions to this rule(Allen 1991). Lowland tropical soils arecharacteristically lacking in organic matter. In theoxisols and ultisols which account for two-thirdsof the world's tropical soils by area, organicmatter decomposition is fairly rapid, and there isoften little accumulation of litter. These soils areacid clays which possess high potentialaluminium toxicity low cation exchange capacityand, as a consequence, low availability of manyminerals (Janos 1987). In this situation,phosphorus is present in soil solution at very lowconcentrations mainly because of its incorporationinto less soluble forms (Alexander 1989). In suchcircumstances, the formation of effectivemycorrhizal associations confers a strongcompetitive advantage, and mycorrhizalassociations with VAIVI fungi, which (with theirextensive extramatrical mycelium) are effectivescavengers for phosphate, predominate (Read1991a, b).

Under these conditions, the uptake of phosphorusby plants, especially tropical trees possessing thecoarse root morphology characteristic of obligatemycorrhizal plants (see below), will be limitedby the slow diffusion of phosphate ions in thesoil and by the development of depletion zonesaround roots. The presence of effective VAMassociations will reduce the constraints of soilnutrition. As it has been estimated (Sieverding1991) that plant production is limited by lowmineral nutrient availability on 30% and bydrought on 33% of tropical soils, thedevelopment of effective mycorrhizal associationsis likely to be beneficial to tree growth over awide range of tropical sites.

Mycorrhizal dependency

Most plants in natural ecosystems havemycorrhizas (Brundrett 1991). However, thebenefit which individuals are likely to receive

from the presence of VA mycorrhizas variesaccording to the plant species underconsideration, because plant species differ in theextent to which they depend on VAM fimgi forsurvival and growth at a given soil fertility (Janos1980a, 1987). They have been characterised aseither non-rnycorrhizal, obligately mycotrophic orfacultatively mycotrophic: obligately mycotrophicplant species cannot grow without mycorrhizaseven in the most fertile soil where they normallygrow, whereas facultative mycotrophs can surviveand grow without mycorrhizas, but performbetter when they are mycorrhizal, particularly inless fertile soils (Janos 1987). For tropical plants,dependence upon VA mycorrhizas is related totheir root morphological characteristics andphosphate requirements. Superficial, small rootsystems that have low orders and frequencies ofbranching, and coarse ultimate rootlets with fewor no root hairs, are characteristic of trees havinga high mycorrhizal dependency (Baylis 1975; StJohn 1980; Janos 1987; Manjunath & Habte 1991).Recent studies by Manjunath and Habte (1992)indicate that mycorrhizal dependency is alsorelated to phosphate usage: whennon-mycorrhizal plants of species possessing arange of degrees of mycorrhizal dependencywere grown in soils containing differentconcentrations of E the mycorrhizal-dependentspecies required higher concentrations of P inthe soil solution for maximum dry matterproduction, while the concentrations of P in theirshoots at maximum dry matter production werelower than for non-dependent species.

Occurrence of mycorrhizal-dependent species

In determining the importance of conserving andmanaging VA mycorrhizal populations inassociation with tropical trees, an appreciation ofthe frequency of occurrence of non-, facultativelyor obligately mycotrophic tree species is needed.

Certain families of plants, including tropical trees,appear to be characteristically non-mycorrhizal,or to have some non-mycorrhizal genera(Brundrett 1991). They include members of theLecythidaceae, Sapotaceae and Proteaceae. Whileextensive studies of mycorrhizal dependencyhave not been conducted, and more work needsto be done, the results of a study by Janos(1980b) and other studies reported in this reviewindicate that a high proportion of other trees arelikely to be dependent upon their mycorrhizas.In a study of Central American lowland tropicalmoist forest plant species (Janos 1980b), VAIVIinoculation improved growth and/or survival of 24out of the 28 species tested, when they weregrown in a sterilised soil in containers. Sixteen ofthese species ceased growth in the absence ofmycorrhizas and were found to be completelydependent on VAM for sustained growth.

The mycorrhizal dependency of plant species

appears to be related to their position in thesuccessional development of plant communities(eg Allen 1991; Brundrett 1991). On the basis ofhis data from the tropical moist forest, Janos(1980a, b) has suggested that early successionalspecies are likely to be facultatively mycotrophicor non-mycorrhizal, with a well-branched rootsystem and small easily dispersed seeds. Seralspecies tend to be facultative mycotrophs, whilemature forest canopy and subcanopy species areobligate mycotrophs, and large-seeded to providenutrient reserves until the root system hasdeveloped mycorrhizal infection to enhancenutrient uptake. Consequently plants that are lessdependent on their VA mycorrhizas or arenon-mycorrhizal are likely to becomeprogressively more dominant in disturbedecosystems where mycorrhizal inoculum islimited (Janos 1987; Miller 1987), although it willdepend upon nutrient availability (Allen & Allen1990). This evidence suggests that there is adirect mycorrhizal role in the determination ofthe structure and composition of plantcommunities. These patterns of change inmycorrhizal dependency with ecologicalsuccession have important practical implicationsfor the growth of VAM trees in plantations andfor the regeneration of degraded land.

Recently the level of VAM dependency of somefast-growing leguminous trees for sustainable landuse systems in the subhumid and semi-arid tropicshas been examined (Habte & Turk 1991). Amongthe plant species that have been suggested asimportant for reforestation and agroforestry systemsare Leucaena leucocephala, Cassia spectabilis andGliricidia sepium (see Brewbaker & Sorensson,pp195-204; Simons, MacQueen & Stewart,pp91-102). Short-term studies with pottedseedlings have shown that L. leucocephala is veryhighly dependent upon VAM fungi (Habte &Manjunath 1987), while C. spectabilis and G. sepiurnalso exhibit VAM dependency although to a lesserextent (Habte & Turk 1991).

Throughout the tropics, a clear assessment of themycorrhizal dependency of tree species wouldenable much more precise recommendations tobe made about the conditions under whichdifferent tree species are likely to succeed, andindicate a particular need for inoculation or carein conservation of indigenous mycorrhizalpopulations when planning for plantations ofmycorrhizal-dependent tree species. Thisinformation would aid the successful developmentof domestication strategies and promotepredictable harnessing of VAM symbioses inpractical forestry across the tropics.

EITECTS OF DISTURBANCE ON VAM FUNGI

Access to adequate VAM inoculum is, as alreadyindicated, a prerequisite for the growth of many

167

tropical trees because of their dependence uponmycorrhizas. Sources of potential infection areresting spores, fungal hyphae in the soil, andcolonised root fragments. These sources ofinoculum can be adversely affected by manyfactors, including fires, soil erosion, loss ofvegetation, cultivation, long fallow periods, soilcompaction, changes in soil temperature,moisture and chemistry colonisation bynon-mycorrhizal plants, and biocide application(Abbott & Robson 1991; Sieverding 1991). Hence,many of the processes involved in deforestationof tropical forest areas have serious implicationsfor mycorrhizal inoculum and, consequently theperformance of trees other than species whichare unresponsive to inoculation. However, there islittle quantitative information to indicate thedegree of inoculum availability which is requiredfor the growth of obligately or facultativelymycorrhizal tree species. Based on his extensivestudies of Manihot esculenta (cassava), which isobligately mycorrhizal, Sieverding (1991) hassuggested that field responses to inoculation ofthis crop may be obtained when there are lessthan 900 indigenous infective propagules per100 g of dry soil. If tree crops respond in asimilar manner, very extensive areas of thetropics are likely to be deficient in mycorrhizalinoculum for acceptable levels of productivity inmanaged systems.

The different forms of VAM propagules vary intheir longevity While spores can survive for upto several years in the soil, fungal mycelium onlyretains its ability to infect after separation fromthe host plant for two to four weeks in fieldconditions. Colonised root fragments can retaintheir viability for several months under dryconditions, but rapidly lose viability under moistconditions (Hetrick 1984; Sieverding 1991).Consequently the abundance of different types ofpropagule in the soil before disturbance maydetermine the impact of disturbance onmycorrhizal infection. If vegetation cover is lost,spores will become increasingly important asmycelium rapidly loses its infectivity resulting ina shift in mycorrhizal populations afterdisturbance, in favour of those fungi whichsporulated prior to the disturbance.

Effects of disturbance will vary with ecosystem.When Jasper, Abbott and Robson (1991)examined the impact of disturbance on soilsremoved from an annual pasture, forest andheathland growing in a mediterranean climate insouth-west Australia, they noted that the extent ofmycorrhizal infection on the roots of Trifoliumsubterraneum (subterranean clover) plantsgrowing in the pasture soil was little affected bydisturbance. In contrast, the extent of infection inboth disturbed jarrah forest and heathland soilswas reduced by almost half. Parallel soil dilutionexperiments indicated that VAM propagule

168

numbers were much lower in the undisturbedforest and heathland soils than in the pasture.Jasper et al. (1991) inferred that, wherepropagule numbers were naturally low as in theforest and heathland soils, disturbance wouldinevitably lead to a further reduction in thenumber of infective propagules and, as aconsequence, a reduction in the soils' potentialinfectivity However, in soils which harboured alarge number of propagules, like the nativegrassland soils, disturbance even of a highlydamaging nature might not reduce the overallnumber of VAM propagules below a level whichwould limit mycorrhizal development (see alsoJasper, Abbott & Robson 1992).

The studies described above indicate that thoseecosystems that support a high number of VAMpropagules will be more resilient, and morelikely to maintain infectivity and hence plantproductivity after disturbance. However, it shouldbe noted that, in these studies, the test plantswere pre-germinated and were planted into thesoils soon after disturbance, and grown infavourable conditions. The impact of disturbanceon mycorrhizal infection in the field is likely tobe greater, as plants will colonise disturbedground more slowly and environmental conditionsmay be more severe. Both these factors will leadto a decline in inoculum infectivity and are likelyto have an adverse effect on mycorrhizalinfection of colonising plants. Continuedmonitoring of the recovery of the jarrah forestsite indicated that mycorrhizal infectivitygradually recovered, but patches of high and lowinfectivity remained for several years. A uniformdistribution of infectivity equivalent to that of theundisturbed forest, was not achieved for sevenyears after vegetation was re-established (Jasperet al. 1992).

In ecosystems lacldng robust VAIVI propagules(spores), disturbance can rapidly destroymycorrhizal infectivity Jasper et al. (1989)observed changes in VAM infectivity in jarrahforest soil on land disturbed by bauxite mining.Spores were not found in this soil before thedisturbance. The proportion of root length onbait plants of Trifolium which became infectedby mycorrhizal fungi declined from 21% beforeclearance to 5% one month afterwards (althoughsoil disturbance was slight during this operation),and no mycorrhizal infection was found sixweeks after the soil was stripped and respreadon another site. It is likely that the cause of thisextreme loss of infectivity was related to theabsence of mycorrhizal spores in the originalsoil; consequently infection was dependent uponmycelial and root fragments which weredamaged during the processes of soil transfer.

Even slight surface erosion can have severeeffects upon all types of mycorrhizal propagules,as they tend to be concentrated close to the soil

surface (Abbott & Robson 1991). Erosion leadsto a loss not only of surface soil particles andnutrients, but also much of the mycorrhizalpopulation, thereby reducing the number offungal propagules available for colonising plants.

Maintaining vegetation cover is important formycorrhizal populations. Recent assessments ofVA mycorrhizal propagule densities in semi-aridbushland of Somalia (Michelsen & Rosendahl1989) have shown that, where there was goodcover (of Tbrminalia brevipes forest), theconcentration of propagules was 200 per 100 mlsoil. In neighbouring bushland degraded byheavy grazing, logging and clearing for shiftingagriculture, and where only 8% of the surfaceremained covered by vegetation, infective VAMpropagule numbers were only 30 per 100 mlsoil. The extent to which such damage candeplete semi-arid ecosystems of their active VAIVIinoculurn, and the impact of such a reduction,was clearly displayed following disturbance of awestern Colorado Teucrium (sage) community(Moorman & Reeves 1979). The root systems oftest maize seedlings were 77% infected whengrown in undisturbed soil but only 2% infectedwhen harvested from adjacent disturbed soil.

In lowland tropical rainforests, spore numbers canbe considerably lower than those found intropical agronomic and grassland ecosystems(Redhead 1977; Louis & Lim 1987; Sieverding1989; Musoko 1991), although there may be agreater diversity of species (Sieverding 1989). Innormal circumstances, the high degree ofvegetation cover found in such forests shouldensure all-year-round access to externalmycelium and mycorrhizal roots, and thereforelittle dependence upon the presence of spores.However, this heavy reliance on easily damagedinfected roots and extrarnatrical hyphae meansthat the stable equilibrium between plant andsoil microbes in undisturbed tropical moist forestcan become greatly disrupted when the forest isopened for cultivating agricultural crops or foresttrees.

In semi-deciduous moist forest at Mbalmayo,Cameroon, the extent of damage resulting fromdifferent methods of forest clearance wasexamined by following short-term changes in theVAM spore populations in plots subjected tomanual canopy opening, partial or completeclearance by bulldozer, and an undisturbedcontrol treatment. When the spore populationswere initially examined, the study site was foundto possess a homogeneous spore distribution.However, when spore numbers weresubsequently re-assessed in samples taken sixmonths later (three months after the differentmethods of site preparation had been applied),the picture changed considerably Althoughseasonal effects appear to have led to a fall inspore numbers in all four treatments, the

reduction in numbeis was greatest in thecompletely cleared plot, which had received themost disturbance. It retained only 34% of itsoriginal spore numbers, while the undisturbedforest retained 73% (Musoko 1991; Mason,Musoko & Last 1992). Additional assessments ofspore numbers, 18 months after the initialassessments, indicated that the decrease in sporenumbers had been a short-term effect oflogging, and that there had been a subsequentincrease in numbers on the disturbed sites, sothat they were higher than they had beenoriginally Results from a study of 7brminaliaplantations in Côte d'Ivoire (Wilson et al. 1992),which examined both undisturbed forest andforest which had been manually or mechanicallycleared and replanted up to 23 years previously,indicated that forest clearance and replanting canresult in long-term increases in spore numbers,lasting at least 16 years.

Decreases in spore numbers after logging havebeen reported from studies in Malaysian forest(Alexander, Ahmad & Su-See 1992), where therewas a 75% reduction as a result of severelogging. Not only were spore numbers reduced,but most probable number (MPN) tests, andobservations of root infection of plants alongtransects at the test sites both indicated that theoverall inoculum potential of the sites wasreduced where sites had been heavily logged,and adverse effects persisted for at least twoyears. Root systems of plants persisting on orcolonising heavily logged forest were only20-35% mycorrhizal, while those in theundisturbed forest were 70% mycorrhizal.Although there were large differences in speciescomposition of the colonising plant species onthe plots, which could affect the degree ofmycorrhizal formation, comparisons of the extentof mycorrhizal infection on the few specieswhich did occur in common between sitesconfirmed that mycorrhizal infection was loweron the plots which had been logged. Care mustbe taken with the interpretation of MPN testsbecause they themselves cause a lot ofdisturbance to inoculum; however, the rootinfection observations lend support to the viewthat logging reduced inoculum potential. It islikely that similar effects of logging on inoculumpotential also occurred at Mbalmayo. Quantitativeassessments (using field bioassays) of theimpacts of logging on all sources of inoculumare needed to evaluate fully the effects inCameroon, and complement the accumulateddata on spore numbers. However, examination ofthe root systems of outplanted krminalia ivorensisseedlings 18 months after plantation establishmentdid demonstrate that there were lower rates ofmycorrhizal infection on the completely clearedplot, compared with the partially cleared plot(Musoko 1991). Survival of outplanted seedlings

169

was also lowest on the completely cleared plot(Mason et al. 1989).

An additional facet of the studies in Mbalmayo,Cameroon (Musoko 1991; Mason et al. 1992), isthat, when spores proliferated after logging, thebalance of species was changed. On thecompletely cleared plot, numbers of Glomusoccultum/Acaulospora scrobiculata increasedsix-fold after disturbance. By contrast, the leastdamaging method of site preparation (manualcanopy opening) maintained a speciesdistribution similar to that in the undisturbedcontrol plot, where Glomus etunicatum was thedominant spore type. Because spores are likelyto be important sources of infection in disturbedsoil after root systems and fungal mycelium aredisrupted, such changes in species compositionmay have an impact upon the subsequentdevelopment of vegetation. Soil samplesremoved from the root zones of a range of plantspecies from Mbalmayo Forest indicated that thechange in species distribution of spores withinthe severely disturbed plots was related to therapid invasion of the herbaceous weedChromolaena odoratum (Siam weed), with whichspores of G. occultum and A. scrobiculata werestrongly associated (Musoko 1991). In C6ted'Ivoire (Lawson et al. 1991; Wilson et al. 1992),spore species diversity was increased by sitedisturbance, for at least 16 years, and long-termchanges in species dominance occurred, whichwere most marked on the mechanically cleared plots.

The combination of information from infectivitybioassays in Malaysia (Alexander et al. 1992)and the above observations on changes in sporespecies composition and number indicates thatforest clearance has major effects uponmycorrhizal populations which are likely to besignificant for the sustainability of forestplantations. Further studies are needed to assessthe impact of logging on all sources of inoculumin different tropical forest ecosystems. Althoughmost bioassays are effected under controlledconditions in pots, more precise information onhow logging affects the onset and spread ofinfection will be obtained from testing bait plantsof target tree species in the field, therebyavoiding the further disturbance to inoculumcaused by sample collection. These studies nowneed to be developed in order to assess theimportance of the extent and nature of the VAMinoculum remaining after logging for the survivaland growth of tropical tree seedlings, both innatural regeneration and plantations. In order toevaluate the consequences of changes in thespecies composition of mycorrhizal fungi, it isnecessary to compare the effectiveness of differentfungal species in promoting tree growth.

The importance of maintaining a highly effectivepopulation was demonstrated in Colombia, whereManihot esculenta is produced in either

170

monoculture or rotational systems (Sieverding1991). At one particular site, monoculture led tothe build-up of a VAM spore population of whichmore than 50% were members of speciesconsidered to be inefficient, whereas, when Mesculenta was rotated with other crops, a smallerproportion of spores belonged to this group. Undermonoculture, root yields were less than half thoseunder rotation. Because chemical soil fertility wasnot a limiting factor, it was concluded that thecomposition of the indigenous VAIVI population wasa crucial factor in determining the yield of cassava,which is obligately mycotrophic. Such observationshave profound implications for forest managementfor both timber and non-timber species, andindicate that quantities of VAM propagules, thespecies composition, and their effectiveness needto be evaluated in much more detail.

Although it is possible to predict that thoseecosystems which support a productive vegetationdominated by VAM plants, and which containmycorrhizal spores in addition to other sources ofinoculum, are more likely to maintain infectivityeven if disturbed, it is clear that many tropicalecosystems are likely to be highly vulnerable todamage, and therefore require careful managementto ensure that mycorrhizal populations aremaintained. The use of less destructive methods ofsite preparation to prevent damage occurring maybe both economically and ecologically preferableto attempting to rectify the damage by inoculationor site manipulation.

MANAGEMENT OF VAM FUNGI

With improved understanding of the damagingeffects of site disturbance, it may be possible inmany cases to reduce the amount of damage tothe natural mycorrhizal population that occursduring site preparation by selecting lessdamaging methods. However, where disturbanceis unavoidable, or where indigenous VAIVIpopulations are already damaged, action tocorrect mycorrhizal deficiencies may beadvantageous.

TWO methods can be used to correct VAMdeficiencies:• manipulation of indigenous VAM fungi;• inoculation with selected VAM fungi.Indigenous populations can be manipulated toenhance the overall levels of inoculum at a site,or to enhance a particular mycorrhizal species,and to increase or decrease VAM speciesdiversity Inoculation can be used to improvenursery or outplanting performance, or both, witheither single isolates, or mixtures.

Manipulation of indigenous VAM fungi

The use of suitable plant hosts to increasevegetation cover and restore VAM populations totheir normal levels is a management practice

which could be used as an alternative toinoculation (Dodd et al. 1990a).

Several studies have examined the effects ofpre-cropping and crop rotation on VAMpopulations in agricultural systems (Schenck &Kin loch 1980; Howe ler, Sieverding & Saif 1987;Jeffries & Dodd 1989; Dodd et al. 1990a, b).They indicate that cropping history cansubstantially influence the abundance andcomposition of the fungal community and can beused to manipulate mycorrhizal populations infavour of the target agricultural crop. Methods ofland management can have a major impact uponVAM species diversity which declines withincreasing intensity of land management(Sieverding 1989). Studies in forest plantations inCôte d'Ivoire indicate that VAM species diversitywas positively correlated with herbaceous plantspecies diversity (Lawson et al. 1991).

There is little information relating to the use ofrevegetation programmes for restoringmycorrhizal populations in non-agriculturaltropical ecosystems. However, in a severelydisturbed savanna in southern Venezuela, wherespontaneous recolonisation was very poor,revegetation with a range of plant speciesenabled the recovery of the VAM inoculum to anextent adequate to enhance plant growth(Cuenca & Lovera 1992). As a result, nativeplants were able to colonise the revegetatedareas at a higher rate than the non-revegetatedareas. These results indicate that revegetationprogrammes may be used to promote therecovery of VAIVI inoculum in order to aidrecolonisation by native plants.

In the same vein, intercropping of perennials andannuals may possess the hidden advantage thatwoody perennials may maintain VAM populationsof benefit to the annual crop. Studies ofintercropping systems in Kenya have indicatedthat mycorrhizal infection of maize grown in soiltaken from close (0.5 m) to Cassia siameahedges is greater than when it is grown in soiltaken from further away (2.0 m) (S McGreevy &K Wakanene Mbuthia, unpublished data),suggesting that trees may act as reservoirs ofVAM inoculum. Whether the species of fungimaintained by the tree are beneficial to the cropremains to be investigated.

Inoculation with selected VAM fungiThere are many examples in the literature whichindicate that inoculation of both tropical cropsand trees can be very successful in promotingthe growth of the host plant in short-termexperiments. The majority of studies on treeshave been targetted at fruit trees and atnitrogen-fixing species. Mycorrhizal inoculationhas been found to be an effective tool forimproving tree performance in the Citrusindustry (Menge 1983), and has reduced the

requirement for fertilizer. Sieverding (1991) listsstudies by 28 authors in short-term nurseryexperiments in Central and South Americancountries. Practically all the tree speciesinvestigated responded positively to mycorrhizalinoculation.

Positive responses to VAM inoculation have beenreported for seedlings of many leguminous treespecies, many of which require effective VAmycorrhizas to supply P for nodulation andnitrogen fixation (Manjunath, Bagyaraj &Gopala-Gowda 1984; De la Cruz et al. 1988).Acacia auriculifonnis (De la Cruz et al. 1988), A.holosericea (Cornet, Diem & Dommergues 1982),A. mangium (De la Cruz et al. 1988), A. nilotica(Michelsen & Rosendahl 1990; Reena & Bagyaraj1990a), A. scleroxyla (Borges & Chaney 1988),Albizia falcatana (Ahmad & Maziah 1988; De laCruz et al. 1988), Calliandra calothyrsus (Reena& Bagyaraj 1990a), Cassia spectabilis (Habte &Turk 1991), Leucaena leucocephala (Manjunath etal. 1984; Ahmad & Maziah 1988; Michelsen &Rosendahl 1990; Jagpal & Mukerji 1991),Sesbania grandiflora (Habte & Aziz 1985),Sesbania sesban (Jagpal & Mukerji 1991) andTamarindus indica (Reena & Bagyaraj 1990b)have all been shown to be responsive toinoculation in short-term studies.

Reports among non-legumes are less frequent;however, several tree species from a wide rangeof tropical habitats have been found to respondpositively to inoculation. They include speciessuch as naya grandifoliola (Redhead 1975),Terminalia superba (Blal 1985) and Theobromacacao (Chulan 1991). The list of Sieverding(1991) includes fruit trees such as Citrus spp.,Persea americana, Carica papaya, Psidium guajava,Annona muricata, Cyphomandra betacea, Passifloraspp., Solanum quitoense and Mangifera indica,and forest trees such as Hibiscus elatus, Cedrelamexicana, Ficus glabrata, Carludovica palmata,Terminalia oblonga, GemPa americana, Tabebuiaspp. and Vitex cooperi.

Results from a number of experiments in whichdifferent single-species inocula were compareddemonstrate that there is considerable variationin effectiveness between fungal species andbetween isolates (eg Bagyaraj, Byra-Reddy &Nalini 1989; Reena & Bagyaraj 1990a, b). Thisevidence suggests that, where inoculation isrequired, there is the opportunity to screen andselect those VAM fungi possessing greatestsymbiotic efficiency. Criteria for selection includespeed of colonisation, ability to take up nutrientsand water, improvement of plant growth,persistence and competitivity with indigenousfungi, and production of large numbers ofresistant propagules (Abbott, Robson & Gazey1992). Selection may be exploited particularlyeffectively in a tree domestication programmewhere high-performing clones of trees could be

171

matched with highly effective isolates of VAMfungi.

When inocula are selected, the researcher musthave a clear perception of the conditions forwhich the inocula are being screened — is theaim to achieve growth responses in potted plantsin the nursery (where comparisons may bemade with non-mycorrhizal control plants grownin sterile soil, or unsterile soil containing amixture of indigenous mycorrhizal fungi) or inplants grown in the field with a more dispersedroot system and a background indigenousinoculum, or is the researcher hoping for growthresponses in both the nursery and the field,despite the widely differing conditionsencountered? Selection for improved growthunder nursery conditions is easily achieved, asdemonstrated by the many papers on thissubject, but selection for long-term effects posesa far greater challenge which agriculturalmycorrhizal researchers do not have to meet.

Persistent effects of inoculation on growth in thefield have not yet been demonstratedconvincingly Long-term successful symbioticrelationships with trees will need to be effectiveunder a range of environmental conditions, fromnursery to outplanting site and maturingplantation (Wilson & Coutts 1985; Janos 1988),and with a changing background of indigenousfungi, which may have been deficient at planting,but which will recover as vegetation isre-established. Because of the changingconditions, and the diversity in VAM physiology(Brundrett 1991), a mixture of VAM fungi (as innatural soil inoculum of undisturbed sites) maybe a long-term advantage for trees, enhancingtheir performance under a range of conditions,and reducing the variation in host response. Asfungal species diversity is lower in intensivelymanaged systems, it may increase ecosysteminstability because of the more limitedphysiological range of the VAM fungi present,and the more intensively managed forestplantations may be in greater need ofmycorrhizal management than the less intensivelymanaged, more mixed plantations.

While the performance of fungal mixtures incomparison with single isolates does not appearto have been tested in long-term experiments,short-term studies have demonstrated that amixture of indigenous mycorrhizal fungi was aseffective in promoting growth of Albizia falcataria,Gmelina arborea, Intsia palembanica andLeucaena leucocephala as the best of sevenintroduced VAM species used as single-speciesinocula (Ahmad & Maziah 1988).

An important gap in the literature is informationrelating to performance of inoculated trees afteroutplanting. Practically all studies have beenrestricted to short-term observations in tree

172

nurseries, where inoculation is frequentlybeneficial. The limited information that existsfrom field plantings indicates that inoculation canimprove both survival and tree growth afteroutplanting, although this improvement has notbeen seen in all the experiments reported. Janos(1988) describes experiments in Costa Rica andPanama in which inoculated Pitheceflobiumlongifolium seedlings had faster growth rates thannon-inoculated seedlings in the nursery Afteroutplanting to over-grazed pasture, the inoculatedseedlings maintained their superior growth ratesfor seven months before the growth rates of thenon-inoculated plants caught up. In a furtheroutplanting trial of non-inoculated andmycorrhizal seven-month-old Inga edulis plantedon sites that had been cut three or 15 monthspreviously and then allowed to regrow theaverage height of myoorrhizal plants increasedten times faster than that of survivingnon-inoculated individuals for eight monthsfollowing transplanting on the more recentlycleared land, and two times faster on land thathad been cleared 15 months previously Thecontrasting responses on these two sites mayindicate the lower inoculum potential on themore recently disturbed site, compared with theolder site where revegetation may have resultedin recovery of inoculum potential.

Dart, Umali-Garcia and Almendras (1991) havereported that Acacia mangium seedlings grewbetter in the nursery when inoculated withmycorirhizal fungi, and that all the non-inoculatedseedlings had died two years after outplanting inthe field. In Kenya, four tree species (Acaciatorah's, Prosopis juliflora, Termihalia brownii and Tprunioides) were pre-inoculated in the nurserywith mixed VAM inoculants collected locally andwith rhizobia (Wilson et al. 1991; see alsoMiimo, Dick & Munro, pp210-219). Inoculatedplants generally grew better in the nurseryAssessments made 30 weeks after outplanting totwo different field sites indicated that inoculatedplants survived better than non-inoculated plants,with the exception of P juliflora at one site andT brownii at the other. When averaged over thetwo sites, inoculated Acacia tortilis, Prosopisjuliflora, Terminalia brownii and T prunioides hadsurvival rates at 30 weeks of 97%, 94%, 89%and 89% respectively while those ofnon-inoculated plants were 79%, 88%, 81% and57%. Further assessments two years afterplanting indicated that the beneficial effects ofinoculation upon survival were sustained (Wilsonet al., unpublished data). However, while survivalin the field was improved by inoculation, therewere no substantial differences betweentreatments in the growth of surviving plants.Similarly Cornet et al. (1982) found that, althoughinoculation improved the growth of Acaciaholosericea in the nursery and the absolute

differences in height growth observed at plantingwere sustained for at least seven months afterplanting in the field, the relative differencesbetween inoculated and uninoculated plantsdiminished. Ducousso and Colonna (1992) havealso reported that, although inoculation improvedthe growth of Acacia holosericea, A senegal andA. raddiana in the nursery the beneficial effectsof inoculation diminished after one or two yearsin the field. As mycorrhizal inoculation in thenursery often produces larger plants, differencesin survival between inoculated and uninoculatedtreatments may result from the initial differencesin plant size, from differences in mycorrhizalinfection, or a combination of both, and dataneed to be interpreted with care. Furthermore,differences in growth rate can be misleading,when initial plant size is different.

The effects of inoculation on field performancewill depend upon the quantity and efficacy ofnaturally occurring inoculum, which needs to beassayed at the time of planting and at intervalsthereafter to interpret the performance ofinoculated and control plants. Likewise,assessments of the extent of mycorrhizal infectionon root systems before planting and at intervalsafter planting are desirable, althoughobservations on field material are very difficult.Responses to inoculation are most likely to befound on sites which have low backgroundavailabilities of inocula, and withmycorrhizal-dependent tree species.

SYNOPSIS

Many tropical trees are highly dependent upontheir mycorrhizal associations for growth andsurvival. Mycorrhizal fungi are majorintermediaries between plant and soil, ofparticular importance in adverse environments,yet their role in tree domestication, and indeveloping sustainable forestry systems in thetropics, has been neglected.

Site preparation can have a great impact uponindigenous mycorrhizal populations, particularlythose communities in which spore inocula havebeen relatively unimportant. The use of lessdamaging methods of site preparation results inless disturbance to mycorrhizal populations, andlimited studies indicate that this factor is morefavourable to mycorrhizal infection on the rootsof planted tree seedlings. Where there is achoice in the method of site preparation to beapplied, consideration should be given to usingmethods which retain much of the ecologicaldiversity of the indigenous mycorrhizal flora, andwhich do not lead to its depletion, therebyproviding a good source of inocula for plantedtrees and potential benefits in long-termsustainability When sites are deficient inmycorrhizal fungi, they may be manipulated toenhance mycorrhizal populations, and, where

specific tree/VAM fungal isolate combinations areto be exploited, plants may be inoculated in thenursery Methodology is available for assessingthe mycorrhizal status of sites, and themycorrhizal dependency of plant species. Thesetools should be applied to determine appropriatestrategies for the treatment of different sites andtree species.

Short-term studies of the effects of VAM fungiupon plant growth have demonstrated that theuse of inoculants to improve seedling growth inthe nursery is readily achievable for a widerange of tree species. The use of selectedisolates of inocula can considerably enhancegrowth, and opportunities exist for theexploitation of tree clone/symbiont interactions, sothat the domestication of trees could run inparallel with the domestication of mycorrhizalfungi. However, the use of fungal mixtures asinoculants may be preferable to the use ofsingle isolates.

Inoculation can be used to improve survival ratesduring the establishment phase after treeplanting. Longer-term effects on post-plantingperformance have yet to be realisticallyevaluated. The development of longer-term effectswill depend upon the selection of inocula thatare appropriate for the site conditions (whichmay not necessarily be those that are mostbeneficial in the nursery), and the availability ofeffective indigenous inocula on-site. Plantsdestined for disturbed, nutrient-poor sites, withlow levels of background inocula, are most likelyto benefit from the application of inoculants inthe nursery Results so far indicate thatexpectations of long-term cumulative effects onmost sites should not be raised too high: theadvantages of inoculation are most likely tooccur during the establishment phase.

REFERENCES

Abbott, L.K. & Robson, A.D. 1991. Factors influencing theoccurrence of vesicular-arbuscular mycorrhizas. Agriculture,Ecosystems and Environment, 35, 121-150.

Abbott, L.K., Robson, A.D. & Gazey, C. 1992. Selection ofinoculant vesicular-arbuscular mycorrhizal fungi. In:Rchnigues for the study of mycorrhiza, edited by IR. Norris,D.J. Read & AK. Varma, 1-22. (Methods in microbiology vol.24.) London: Academic Press.

Ahmad, N. & Maziah, Z. 1988. Mycorrhizalexperimentation with some timber tree species in Malaysia.In: Trees and mycorrhiza: proceedings of the Asian seminar,edited by FS.P Ng, 127-132. Kuala Lumpur: Forest ResearchInstitute.

Alexander, I.J. 1989. Mycorrhizas in tropical forests. In:Mineral nutrients in tropical forest and savanna ecosystems,edited by J. Proctor, 169-188. Oxford: Blackwell Scientific.

Alexander, I.J., Ahmad, N. & Su-See, L. 1992. The role ofmycorrhizas in the regeneration of some Malaysian foresttrees. Philosophical Transactions of the Royal Society of LondonB, 335, 379-388.

173

Allen, E.B. & Allen, M.F. 1990. The mediation ofcompetition by mycorrhizae in successional and patchyenvironments. In: Perspectives in plant competition, edited by J.B.Grace & G.D. Tilman, 367-389. New York: AcademiC Press.

Allen, M.F. 1991. The ecology of mycorrhizae. Cambridge:Cambridge University Press.

Bagyaraj, D.J., Byra-Reddy, M.S. & Nalini, P.A. 1989.Selection of an efficient inoculant VA mycorrhizal fungus forLeucaena. Forest Ecology and Management, 27, 81-85.

Barea, J.M. 1991. Vesicular-arbuscular mycorrhizae asmodifiers of soil fertility Advances in Soil Science, 15, 1-40.

Baylis, G.T.S. 1975. The magriolioid mycorrhiza andmycotropy in root systems derived from it. In:Endomycorrhizas, edited by FE. Sanders, B. Mosse & PB.Tinker, 373-389. London: Academic Press.

Bethlenfalvay, G.J. & Linderrnan, R.G. 1992. Mycorrhizaein sustainable agriculture. (American Society of Agronomyspecial publication, no. 54.) Madison, Wisconsin: AmericanSociety of Agronomy

Blal, B. 1985. Etude sur les complexes mycorhiziensd'especes forestières tropicales: influence sur la croissance deTerminalia superba (frake). Diploma d'études approfondies (enecologie), Univeisité de Montpellier.

Bolan, N.S. 1991. A critical review on the role ofmycorrhizal fungi in the uptake of phosphorus by plants.Plant and Soil, 134, 189-207.

Borges, R.G. & Chaney, W.R. 1988. The response of Acaciascleroxyla aiss. to mycorrhizal inoculation. International TreeCrops Journal, 5, 191-201.

Brundrett, M. 1991. Mycorrhizas in natural ecosystems.Advances in Ecological Research, 21, 171-313.

Chu lan, H.A. 1991. Effect of fertilizer and endomycorrhizalinoculum on growth and nutrient uptake of cocoa(Theobroma cacao L.) seedlings. Biology and Fertility ofSoils, 11, 250-254.

Cooper, K.M. 1984. Physiology of VA mycorrhizalassociations. In: VA mycorrhizas, edited by C.L. Powell & D.J.Bagyaraj, 155-186. Boca Raton, Florida: CRC Press.

Cornet, F., Diem, H.G. & Dommergues, Y.R. 1982 Effetde l'inoculation avec Glomus mossae sur la croissanced'Acacia holosericea en pepinière et apres transplantation surle terrain. In: Les mycorrhizes.. biologie et utilisation, edited byS. Gianinazzi, V Gianinazzi-Pearson & S. llouvelot, 287-293.Paris: INRA Press.

Cuenca, G. & Lovera, M. 1992. Vesicular-arbuscularmycorrhizae in disturbed and revegetated sites from La GranSabana, Venezuela. Canadian Journal of Botany, 70, 73-79.

Dart, P., Umali-Garcia, M. & Almendras, A. 1991. Roleof symbiotic associations in nutrition of tropical acacias. In:Advances in tropical acacia research: proceedings of aWorkshop held in Bangkok, Thailand, 1991, edited by J.WTurnbull, 13-19. (ACIAR Proceedings, no. 35.) Canberra: ACIAR.

De la Cruz, R.E., Manalo, M.Q., Aggangan, N.S. &Tambalo, J.D. 1988. Growth of three legume trees inoculatedwith VA mycorrhizal fungi and Rhizobium. Plant and Soil, 108,111-115.

Dodd, J.C., Arias, I., Koomen, I. & Hayman, D.S. 1990a.The management of populations of vesicular-arbuscularmycorrhizal fungi in acid-infertile soils of a savanna ecosystem. I.The effect of pre-ciopping and inoculation with VAIVI-fungi onplant growth and nutrition in the field. Plant and Soil, 122, 229-240.

Dodd, J.C., Arias, I., Koomen, I. & Hayman, D.S. 1990b.The management of populations of vesicular-arbuscularmycorrhizal fungi in acid-infertile soils of a savannaecosystem. II. The effects of pre-crops on the spore

174

populations of native and introduced VAM-fungi. Plant andSoil, 122, 241-247.

Ducousso, M. & Colonna, J.P. 1992. Endomycorrhizalinfection in young Faidherbia albida: influence on growth anddevelopment. In: Faidherbia albida in the West Africansemi-arid tropics, edited by R.J. Vandenbeldt, 151-159.Patancheru: International Crops Research Institute for theSemi-Arid Tropics/Nairobi: International Centre for Research inAgmforestry

Fitter, A.R. 1985. Functioning of vesicular-arbuscularmycorrhiza under field conditions. New Phytologist, 99,257-265.

Habte, M. & Aziz, T. 1985. Response of Sesbania grandiflorato inoculation of soil with vesicular-arbuscular mycorrhizalfungus. Applied and Environmental Microbiology, 50, 701-704.

Habte, M. & Manjunath, A. 1987. Soil solution phosphorusstatus and mycorrhizal dependency in Leucaena leucocephala.Applied and Environmental Microbiology, 53, 797-801.

Habte, M. & Turk, D. 1991. Response of two species ofCassia and Gliricidia sepium to vesicular-arbuscularmycorrhizal infection. Communications in Soil Science and PlantAnalysis, 22, 1861-1872.

Harley, J.L. & Smith, S.E. 1983. Mycorrhizal symbiosis.London: Academic Press.

Hetrick, B.A.D. 1984. Ecology of VA mycorrhizal fungi. In:VA mycorrhiza, edited by C.L. Powell & D.J. Bagyaraj, 35-56.Boca Raton, Florida: CRC Press.

Howeler, R.H., Sieverding, E. & Saif, R. 1987. Practicalaspects of mycorrhizal technology in some tropical crops andpastures. Plant and Soil, 100, 249-283.

Jagpal, R. & Mukerji, K.G. 1991. VAM fungi inreforestation. In: Plant roots and their environment, edited byB.L. McMichael & H. Persson, 309-313. Amsterdam: Elsevier.

Janos, D.P. 1980a. Mycorrhizae influence tropical succession.Biotropica, 12 (supplement), 56-64.

Janos, D.P. 1980b. Vesicular-arbuscular mycorrhiza affectlowland tropical rainforest plant growth. Ecology, 61, 151-162.

Janos, D.P. 1987. VA mycorrhizas in humid tropicalecosystems. In: Ecophysiology of VA mycorrhkal plants, editedby ER. Safir, 107-134. Boca Raton, Florida: CRC Press.

Janos, D.P. 1988. Mycorrhiza applications in tropical forestry;are temperate-zone approaches appropriate? In: Trees andmycorrhiza: proceedings of the Asian seminar, edited by FS.PNg, 133-188. Kuala Lumpur: Forest Research Institute.

Jasper, D.A.,Abbott, L.K. & Robson, A.D. 1989. The lossof VA mycorrhizal infectivity during bauxite mining may limitthe growth of Acacia pulchella R. Br. Australian Journal ofBotany, 37, 33-42.

Jasper, D.A., Abbott, L.K. & Robson, A.D. 1991. Theeffect of soil disturbance on vesicular-arbuscular mycorrhizalfungi in soils from different vegetation types. New Phytologist,118, 471-476.

Jasper, D.A., Abbott, L.K. & Robson, A.D. 1992. Soildisturbance in native ecosystems - the decline and recoveryof infectivity of VA mycorrhizal fungi. In: Mycorrhizas inecosystems, edited by D.J. Read, D.H. Lewis, A.H. Fitter & I.J.Alexander, 151-155. Wallingford: CAB International.

Jeffries, P. & Dodd, J.C. 1989. The use of mycorrhizalinoculants in forestry and agriculture. In: Handbook of appliedmycology vol. 1, Soil and plants, edited by D.K. Arora, B. Rai,KG. Mukerji & G.R. Knudsen, 155-185. New York: Dekker.

Lawson, G.J., Wilson, J., Ingleby, K. & Mason, P.A.1991. Mycorrhizal and chemical changes after land clearancein the Mopri Forest Reserve, Côte d'Ivoire. (NaturalEnvironment Research Council contract report to the UK

Overseas Development Administration.) Penicuik, Midlothian:Institute of Terrestial Ecology

Le Tacon, E, Garbaye, J. & Carr, G. 1987. The use ofmycorrhizas in temperate and tropical forests. Symbiosis, 3,179-206.

Louis, I. & Lim, G. 1987. Spore density and root colonizationof vesicular-arbuscular mycorrhizas in tropical soil Transactionsof the British Mycological Society, 88, 207-212.

Manjunath, A. & Habte, M. 1991. Root morphologicalcharacteristics of host species having distinct mycorrhizaldependency Canadian Journal of Botany, 69, 671-676.

Manjunath, A. & Habte, M. 1992. External and internal Prequirements of plant species differing in their mycorrhizaldependency Arid Soil Research and Rehabilitation, 6,271-284.

Manjunath, A., Bagyaraj, D.J. & Gopala-Gowda, H.S.1984. Dual inoculation with VA mycorrhiza and Rhizobium isbeneficial to Leucaena. Plant and Soil, 78, 445-448.

Mason, P.A., Musoko, M.O. & Last, F.T. 1992. Short-termchanges in vesicular-arbuscular mycorrhizal spore populationsin 7brminalia plantations in Cameroon. In: Mycorrhizas inecosystems, edited by DJ. Read, D.H. Lewis, A.H. Fitter & I.J.Alexander, 261-267. Wallingford: CAB International.

Mason, P.A., Leakey, R.R.B., Musoko, M., Ngeh, C.P.,Smith, R.I. & Sargent, C. 1989. Endomycorrhizas andnutrient cycling in indigenous hardwood plantations inCameroon: effects of different systems of site preparation.(Natural Environment Research Council contract report to theUK Overseas Development Administration.) Penicuik,Midlothian: Institute of Terrestrial Ecology

Menge, J.A. 1983. Utilization of vesicular-arbuscularmycorrhizal fungi in agriculture. Canadian Journal of Botany61, 1015-1024.

Michelsen, A. & Rosendahl, S. 1989. Propagule density ofVA-mycorrhizal fungi in semi-arid bushland in Somalia.Agriculture, Ecosystems and Environment, 29, 295-301.

Michelsen, A. & Rosendahl, S. 1990. The effect of VAmycorrhizal fungi, phosphorus and drought stress on thegrowth of Acacia nilotica and Leucaena leucocephalaseedlings. Plant and Soil, 124, 7-13.

Mikola, P. 1970. Mycorrhizal inoculation in afforestation.International Review of Forestry Research, 3, 123-196.

Miller, R.M. 1987. Mycorrhizae and succession. In:Restoration ecology, edited by WR. Jordan III, M.E. Gilpin& J.D. Aber, 205-219. Cambridge: Cambridge UniversityPress.

Moorrnan, T. & Reeves, F.B. 1979. The role ofendomycorrhizae in revegetation practices in the semi-aridwest. H. A bioassay to determine the effect of landdisturbance on endomycorrhizal populations. American Journalof Botany, 66, 14-18.

Musoko, M.O. 1991. The ecology of endomycorrhizalassociations of trees in Cameroon, with special reference toTerrninalia. PhD thesis, University of Edinburgh.

Powell, C.L. & Bagyaraj, D.J., eds. 1984. VA mycorrhiza.Boca Raton, Florida: CRC Press.

Read, D.J. 1991a. Mycorrhizas in ecosystems - nature'sresponse to the law of the minimum'. In: Frontiers inmycology, edited by D.L.Hawksworth, 101-130. Wallingford:CAB International

Read, D.J. 1991b. Mycorrhizas in ecosystems. Experientia,47, 376-391.

Redhead, J.F. 1975. Endotrophic mycorrhizas in Nigeria:some aspects of the ecology of the endotiophic mycorrhizalassociation of Khaya grandifoliola C.DC. In: Endomycorrhizas,edited by FE. Sanders, B. Mosse & PB. Tinker, 447-459.London: Academic Press.

Redhead, J.F. 1977. Endotophic mycorrhizas in Nigeria:species of the Endogonaceae and their distribution.Transactions of the British Mycological Society, 69, 275-280.

Reena, J. & Bagyaraj, D.J. 1990a. Response of Acacianilotica and Caiandra calothyrsus to different VA mycorrhizalfungi. Arid Soil Research and Rehabilitation, 4, 261-268.

Reena, J. & Bagyaraj, D.J. 1990b. Growth stimulation ofTamarindus indica to selected VA mycorrhizal fungi. WorldJournal of Applied Microbiology and 7bchnology, 6, 59-63.

Rhodes, L.A. & Gerdemann, J.W. 1975. Phosphate uptakezones of mycorrhizal and non-mycorrhizal onions. NewPhytologist, 75, 555-561.

St John, T.V. 1980. Root size, root hairs and mycorrhizalinfection: a re-examination of Baylis's hypothesis with tropicaltrees. New Phytologist, 84, 483-487

Schenck, N.C. & Kin loch, R.A. 1980. Incidence ofmycorrhizal fungi on six field crops in monoculture on a

• newly cleared woodland site. Mycologia, 72, 445-456.

Sharma, A.K., Johri, B.N. & Gianinazzi, S. 1992.Vesicular-arbuscular mycorrhizae in relation to plant disease.World Journal of Microbiology and Biotechnology, 8, 559-563.

Sieverding, E. 1989. Ecology of VAM fungi in tropicalecosystems. Agriculture, Ecosystems and Environment, 29,369-390.

Sieverding, E. 1991. Vesicular-arbuscular mycorrhizamanagement in tropical agrosystems. Eschborn: GTZ.

Wilson, J. & Coutts, M.P. 1985. Exploiting treecrop-symbiont specificity In: Attributes of trees as crop plants,edited by M.G.R. Cannell & J.E. Jackson, 359-379. AbbotsRipton, Huntingdon: Institute of Terrestrial Ecology

Wilson, J., Munro, R.C., Ingleby, K., Mason, P.A.,Jefwa, J., Muthoka, P., Dick, J.McP. & Leakey, R.R.B.1991. Tree establishment in semi-arid lands of Kenya - role ofmycorrhizal inoculation and water-retaining polymer. ForestEcology and Management, 45, 153-163.

Wilson, J., Ingleby, K., Mason, P.A., Ibrahim, K. &Lawson, G.J. 1992. Long-term changes invesicular-arbuscular mycorrhizal spore populations in7brminalia plantations in C6te d'Ivoire. In: Mycon-hizas inecosystems, edited by D.J. Read, D.H. Lewis, A.H. Fitter & I.J.Alexander, 268-275. Wallingford: CAB International.

175

Harnessing symbiotic associations: the potentialsof nitrogen-fixing trees

SprentDepartment of Biological Sciences, University of Dundee, Dundee DD1 4HN, UK

ABSTRACTNitrogen-fixing trees are of two general types: (i) those nodulating with single-celled bacteria - rhizobia - all but one of whichare members of the family Legurninosae; (ii) those nodulating with the filamentous bacterium Franks, which are found scatteredin a range of families. Potentially these trees fix large amounts of nitrogen, and have multiple uses as crops (fuel, forage, etc) andfor controlling soil degradation. However, if this potential is to be realised, host and symbiont genotypes must be selected andmatched not only to each other, but to the environment in which they are grown. To illustrate both the problems and potentials ofnitrogen-fixing trees, examples will be drawn from the dry and humid tropics, concentrating on legumes and the non-legumeCasuarina.

INTRODUCTION

Of the approximately 18 000 species (in about650 genera) in the family Leguminosae, themajority are tropical or subtropical, and most aretrees. Many have the ability to form root nodulesand fix nitrogen, but there are also numerousexceptions. Research into nitrogen-fixing legumeshas been very heavily biased towardsherbaceous crop species, such as soybeans andTrifolium (clover), and hence there is a tendencyto regard these as the norm. Tree legumes maydiffer in almost all aspects of nodulation fromtheir herbaceous relatives (Sprent 1989). Onlythe basic enzymology of the conversion of N2 toammonia and of ammonia assimilation appear tohave been highly conserved during evolution.Thus, although the potential of leguminous treesfor nitrogen fixation is being actively investigatedin many countries, few facts are available,especially concerning mature trees. For thisreason, no attempt will be made here togeneralise, but rather selected examples arepresented to guide an assessment of thepotential for nitrogen fixation, and also to indicatehow much work remains to be done.

Although numerically legumes account for mostof the tropical nitrogen-fixing trees, there are twoother non-legume genera of interest, namelyParasponia and Casuarina. Parasponia is a genusof the Ulmaceae and to date the only oneoutside the legumes kriown to nodulate with asingle-celled bacterium belonging to the generalcategory of rhizobia. Parasponia species arerestricted to SE Asia and the Pacific islands.Their nodules can fix nitrogen at high rates, andthus they have considerable potential, but theyonly grow in a comparatively restricted areawhere daylengths are around 12 h and there is

176

a high photon flux density with temperatures inthe range 25-32°C. Parasponia spp. appear to beindifferent to both type and fertility of soils. Thegenus is not considered further here (for furtherdetails, see Trinick & Hadobas 1988; Becking1992).

Of the disparate group of non-leguminous plantsnodulating with the filamentous bacterium Frankia,Casuarina has been the most widely used intropical areas. Other actinorhizal plants occurless commonly in tropical countries and tend tobe shrubs rather than trees (see Schwintzer &Tjepkema 1990). The plants most commonlyreferred to as 'casuarinas' are members of thefamily Casuarinaceae, which has recently beendivided into four genera, Casuarina, Allocasuarina,Gymnostoma and Ceuthostoma (Johnson 1988).All have traits such as reduced needle-likebranches and cone-like fruits which make themsuperficially resemble conifers, although they arefirmly located in the angiosperms, with affmitiesto some other actinorhizal groups (Mullin,Swenson & Goetting-Minesky 1990). They haveproved to be useful in all tropical countries,being prized for fuelwood, construction and landrehabilitation. These uses have been welldocumented (National Research Council 1984;Diem & Dommergues 1990), as have theirstrategies for improving nitrogen fixation (Diem &Dommergues 1990).

Sanginga, Bowen and Danso (1990a) grew 11provenances each of Casuarina equisetifolia andC. cunninghamiana, either with nitrogen fertilizeror inoculation with a mixture of Frankia strains(Table 1). Great variability in nodule massproduced was recorded; not all nodules wereequally effective at fixing nitrogen, as the datafor shoot nitrogen content show (Table 1). Some

Table 1. Variation in nodulation and nitrogen fixation amongstprovenances of Casuarina cunninghannana, inoculated with amixture of Frankia strains (source: Sanginga et al 1990a)

Dry massof N in shoot"

Origin of nodules of nodulatedProvenance (mg plant I) plants (mg)

/Obtained by interpolation from histogram2All from Australia: NSW, New South Wales; QLD, Queensland;ACT, Australian Capital Territory

of these effects may be indirect, resulting fromdifferent potential growth rates of theprovenances under the conditions used (aglasshouse in Vienna). To allow for this factor,growth of nodulated plants was compared withgrowth of N-fertilized plants, although the form ofN (ammonium or nitrate) was not specified. Itcan be seen that some provenances assimilatedmore N when fixing N, and some when usingsoil N. The former is usually true of plants witheffective nodules (eg provenance 1), but thelatter is not as well correlated with ineffectivenodules. For example, provenance 11 had alarge nodule mass, but little N in shoots of eithernodulated plants or plants grown on fertilizer N.Provenance 7, with the same nodUle mass andthe same shoot N content in nodulated plants asprovenance 11, grew much better on soil N.Unfortunately we do not laiow how much N wasused for root growth, a factor which could bevery important for the ability of the provenancesto cope with certain field environments (asdescribed later for Faidherbia albida).

N FIXATION IN THE LEGUMINOSAE

Nodulation

A list of genera known to nodulate has recentlybeen published (Faria et al. 1989). There havebeen new reports of nodulating species withinthese genera, and some new reports (mainlynegative) on additional genera. However, theessential features of the list remain unchanged,as has the balance between nodulation in thedifferent subfamilies (see also Allen & Allen1981). In general, few caesalpinioid legumes cannodulate whereas most of the mimosoid andpapilionoid legumes are able to do so. There

are, however, notable exceptions in all groups.Even if a species has the potential to nodulate, itmay not do so in all environments.

NThere are many legumes which have proven

innodulated value or potential for domestication which cannot

shoots nodulate. Examples in the Caesalpinioideae are

N in Parkinsonia, Bauhinia, Caesalpinia and Cassiafertilized (sensu l.iicto; the former section Chamaecrista,

shoots now a separate genus, can nodulate). It ispossible to argue a case for using some ofthese in preference to nodulated tree legumes,as has been done recently for Caesalpiniaparaguariensis (Aronson & Toledo 1992).

Of the more important mimosoid legumes, thepantropical genus Adenanthera (not to beconfused with its nodulated relativeAnadenanthera) does not nodulate. Thenodulation status of Parka needs very carefulexamination, as early positive reports have notbeen confirmed. Similarly care is needed withAcacia spp., as some species, such asA. brevispica (Odee & Sprent 1992), may havelost the ability to nodulate.

A number of papilionoid trees appear unable tonodulate. They include some large trees oftropical America, such as Vatairea andVataireopsis. The large pantropical genusPterocarpus is unusual in that African, Asian andVenezuelan species nodulate, but Brazilian oneshave so far proved negative. There are manygenera, including some large trees from tropicalWest Africa (eg Amphimas, Angylocalyx) forwhich there is no information. Indeed, over 40%of all legume genera, mainly tropical ones, havenot been examined for nodulation.

Rhizobia

The bacteria which nodulate legumes, collectivelyknown as rhizobia, comprise at least threegenera. Although all are grouped in the asubdivision of the proteobacteria, they are lessclosely related to each other than to a wholerange of bacteria, including some animalpathogens (Young 1992). Although rhizobialgenera may be categorised (see Table 2), theydo not always fit easily into discrete categories.Growth may be fast, intermediate, slow or veryslow Slow growth may be combined with narrowhost range, and fast growth with broad hostrange (Table 3).

In the work of Tark and Keyser (1992), bacterialisolates from tree nodules were classified aseither Rhizobium or Bradyrhizobium according notonly to their growth rate, but also to variousstandard cultural criteria such as acid production.The data selected for Table 3 are for thosestrains which, when inoculated on to the hostspecies from which they were isolated, formedeffective nodules (shown by bold italic type onthe diagonal of the matrix). Considering first the

177

Table 2. Some characters of the three established genera of rhizobia known to nodulate legumes

Character

1 Growth rate in culture

2 N2 fixation ex planta

3 Assimilation products of N2 fixation ex planta

4 Major location for nod, nif genes

5 Host range

6 Geographical distribution

Table 3. Range of hosts and effectiveness of nodules formed from rhizobia isolated from tree legumes. Headings of columnsdesignate species as listed (source: auk & Keyser 1992)

Host species of isolation

Leucaena leucocephala

Gliricidia sepium

Calliandra calothyrsus

Sesbania grandiflora

Robinia pseuclOacacia

Acacia auriculiformils

Acacia mangiurn

Paraserianthes falcataria

Bradyrhizobium or Rhizobium

*E, effective; e, partially effective; I, ineffective; 0, nodules not formed; ?, inconsistent response

rows in the Table, it can be seen that strainsclassified as Rhizobium vary greatly in thenumber of species, other than that from whichthey were isolated, that they nodulate. Theformation of ineffective nodules on a number ofhosts could lead to problems if these strainswere used as inoculants for soils supporting arange of legume trees. The data set forBradyrhizobium is less extensive, but alsosuggests variability The isolation of largenumbers of Rhizobium strains from tropical trees,now found in many laboratories, counters theconclusions of some early workers that tropicalrhizobia are principally slow-growing forms (see,for example, discussion in Allen & Allen 1981).

In this study Leucaena leucocephala nodulatedonly with strains isolated from itself, Gliricidiasepium or Calliandra calothyrsus. This corner ofthe Table appears very tidy but we should notethat Gliricidia is a papilionoid genus, whereas theother two are mimosoids. Taxonomically Gliricidiais clOsely allied to Sesbania and Robinia; strainsfrom these genera do not nodulate theLeucaena/Gliricidia/Calliandra species tested,although the converse situation sometimes leadsto ineffective nodules. The most specificassociation in this work is that between Robiniapseudoacacia and its isolates, but even hereineffective nodules were formed with isolatesfrom several other genera.

It should be emphasised that the data in Table 3represent only the tip of the iceberg. Sometropical trees can nodulate with strains of bothRhizobium and Bradyrhizobium. A further

178

Rhizobium

Fast

Absent

Absent

Plasmid

Usually narrow

Global

Bradyrhizobium

Slow

Some strains

Absent

Chromosome

Often broad

Mainly tropical

Azorhizobium

Fast

Generic

Generic

?Chromosome

Sesbania rostrata

Tropical

Host species tested

Ll Gs Cc Sg Rp Aa Am Pf

E E e 0 I 0 I I*

E E E 0 I 0 0 I

E E E 0 I I 0 I

O 0 0 E I 0 I I

O 0 0 0 E 0 0 0

O 0 0 0 0 E 0 E

O 0 0 e 0 E E E

O I 0 I I ? E E

complication is that trees may be nodulated bystrains from crop plants and vice versa. The mostwidely studied example is the newly designatedspecies Rhizobium tropici, which forms noduleswith Phaseolus vulgaris and Leucaena spp.(Martinez-Romero et al. 1991). Finally it shouldbe noted that inconsistent results are notuncommon in this type of work. In acomprehensive study recently carried out inKenya (D W Odee et al., unpublished data), fourseedlings each of a range of tree species weregrown in soil from 12 different parts of thecountry Nodulation among these four seedlingswas often very variable, from zero to largenumbers of nodules formed. This could resultfrom genetic variation in the host, a point whichwill be considered later.

FACTORS AFFECTING NODULATION ANDNITROGEN FIXATION

Although most of the examples in this sectionare taken from legumes, similar arguments canusually be made for Casuarina. There are twomajor problems in attempting to estimatenodulation and N fixation in the field. First,finding nodules, even if they are present, may bedifficult for various reasons:

• the weak connection to the subtending rootin many species, especially Acacia, meansthat nodules readily become detached duringexcavation;

• nodules may be deep in the soil, especiallyin arid areas (see Johnson & Mayeux 1990).

Second, there is no universally applicable methodof measuring N2 fixation in the field for trees ofany size. Seedlings and young plants aredifferent from mature trees, because N balanceand 15N dilution methods can be used (seeGil ler & Wilson 1991). The only feasible methodfor mature trees is to use 15N at naturalabundance levels. The methodology has beenreviewed several times in recent years, eg byShearer and Kohl (1986). Recently, the methodhas been successfully applied to trees in a rangeof different environments, from arid (Mariotti,Sougoufara & Dommergues 1992 for Casuarinaeguisetifolia; Johnson & Mayeux 1990 for Prosopisglandulosa) to the Amazon floodplain (Martinelliet al. 1992 for unspecified mimosoid legumes).

Space precludes an assessment of all factorslikely to affect nitrogen fixation. Experiments aremost easily carried out in pots, and a range offactors are known to affect N fixation under theseconditions (Table 4). Rather surprisingly therehas been comparatively little work on the effectof combined N on N fixation by tree species. Itis generally assumed that the 'normal' situationfor crop species applies, namely that combinedN inhibits both nodulation and N fixation, but thatsome 'starter N' may be beneficial. There is afurther tacit assumption that legumes prefernitrate to ammonium as an N source, but this isnot necessarily true. In Acacia auriculiformis, arapidly growing Australian species being used inBrazil and elsewhere (see Booth & rffirnbull,pp189-194), plants grew very poorly on nitrate,becoming chlorotic, whereas on ammonium theygrew and nodulated well (Goi et al. 1993).

Table 4. Some factors which may affect estimates of nitrogenfixation potential in pot-grown plants. For access to relevantliterature, see Sprent and Sprent (1990) and Gil ler and Wilson(1991)

Age of plant WeeksMonthsYears

Growth conditions TemperaturePhotosynthetically active photon fluxdensityWater supply

Soil factors PhysicalChemicalBioticEndosyrnbiont (Frankia, rhizobia)Helper and other bacteria

A further complication which may be foundwhen plants are grown on different sources of Nconcerns the allocation of resources betweenroot and shoot. N M Gitonga of the University ofDundee is studying Faidherbia albida, and hasmeasured root growth and plant dry mass when

Table S. Growth of Faidherbia albida on different sources ofnitrogen (N), eight weeks after planting (source: unpublisheddata of N M Gitonga)

*0, control, uninoculated, no added N; N2, effectivelynodulated; NO3—, 5 mmol nitrate-N; NH4+5 mmol ammonium-NITap toot plus laterals2Values within a column with the same letter are notsignificantly different (P<0.05)

seedlings were given various forms of N. Thisspecies is well known for its ability to producedeep tap roots which may penetrate to the watertable - where, according to Dupuy and Dreyfus(1992), there may be plenty of compatiblerhizobia - as well as producing its leaves in thedry season. After eight weeks' growth in aDundee greenhouse, tap roots were almost 1 min length (Table 5), with those plants givennitrate being slightly shorter than control,ammonium-grown or nodulated plants. Note thatthis length of tap root was achieved over analmost two-fold range of root dry mass. In plantsgiven no N, shoot dry mass was very low, beingless than a third of those grown in ammonium.These differences in response led to a range ofvalues for root/shoot ratios and suggest thatcaution should be used when assessing a plant'sability to fix N by reference to shoot N contentalone (see below).

Obviously before using a plant for its ability tofix N, its nodulation potential in the soils where itis proposed to grow must be examined. Arecent study of this type assessed 44 legumespecies on two soils (one acid, one neutral) inChile (Aronson, Ovalle & Avendario 1992).Several species known to nodulate did not do so,and others nodulated only poorly which suggeststhat, if these species are to be grown in Chile,they should be inoculated, a topic consideredbelow. Many of the species studied by Aronson,Ovalle and Avendafio (1992) which nodulatedwell were considered to be potential weeds andtherefore excluded from further trials. This opensup the contentious matter of introducing exoticspecies into new areas. A risk/benefit analysis ofsuch procedures has been made by Hughes andStyles (1989). Introduced species often producelarge numbeis of seeds, and in the absence ofpredators they can spread rapidly InvasiveN-fixing plants may seriously affect nutrientcycling in natural vegetation (Witkowsld 1991).

179

Not surprisingly plants vary greatly in their abilityto fix N under adverse environmental conditions.Reports are now beginning to appear for treespecies. For example, Marcar, Dart and Sweeney(1991) studied three Australian species of Acaciaand ranked their salinity (NaC1) tolerance in theorder A. ampliceps > A. auriculiformis > A.mangium. Perhaps N-fixing plants will be evenmore useful under increased atmospheric 002concentrations: Norby (1987) found that Robinia

pseudoacacia and two actinorhizal plants (Alnusglutinosa and Elaeagnus angustifolia) grew morerapidly and nodulated more when grown at twicethe ambient 002 concentration.

Table 6. Variation within rhizobia isolated from Robiniapseudoacacia nodules (source: Batzli et al. 1992)

Factor

6 phosphogluconatedehydrogenase(nmol NADPH min-1 mg-1)

pH reaction in culture

Mean generation time (h)

Tolerance to NaC1

Table 7. Shoot N content in three provenances each of twolegumes inoculated (I) or grown with added fertilizer (N),harvested at 12 or 36 weeks after planting (source: Sangingaet al. 1990b)

Legume species

Provenance

Leucaena leucocephala

A 68

36

46

Faidherbia (Acacia) albida

A 25

20

23

180

6-394

Acid, neutral, alkaline

3.4-17.8

No growth on 120 mol m-3 togrowth on 256 mol IT1-3

12 weeks

Shoot N (mg p1ant-1)

88

54

63

30

38

32

Range found

36 weeks

289

214

363

107

80

60

287

188

232

96

97

79

1Based on shoot N only2Based on number of shared DNA fragments within a population3Sum of polymorphic loci detected by nine oligonucleotide primers plant-1

GENETIC VARIATION

Variation in rhizobia

This topic has been studied for many years andsome aspects have already been considered fortree syrnbionts (see Table 3). In another recentstudy Batzli, Graves and van Berkum (1990b)examined genetic diversity in 186 effectiveisolates from Robinia pseudoacacia, all obtainedfrom two sites in Allegany County Maryland, USA(Table 6). Clearly in spite of the host specificityfound by 'Dirk and Keyser (1992) (Table 3), therhizobia nodulating R. pseudoacacia are veryvaried.

Variation in host genotype

A number of studies on various species havebeen carried out at the International AtomicEnergy Authority's laboratories in Vienna. Datawere obtained by Sanginga et al. (1990b) fromthree provenances each of Leucaena leucocephalaand Faidherbia albida (liable 7), which showedthat harvests made after 12 and 36 weeks weredifferent because some provenances nodulatedslowly initially However, there is clear variationamong provenances of each species, especiallyfor inoculated plants harvested at 36 weeks. InL. leucocephala, variation was also foundbetween provenances grown on combined N(type unspecified). Plants had greater shoot Ncontent when inoculated. Whether this resultreflects greater total N content or is aconsequence of altered root/shoot ratio (seeTable 5) is not Imown.

Another study considered Gliricidia sepium(Sanginga, Manrique & Hardarson 1991). Theprovenances were obtained from the OxfordForestry Institute and some were the same asthose used for a molecular study of variation byChalmers et al. (1992) (Table 8). Considerablevariation can be seen both in the mass ofnodules produced and in the amount of N fixed(again based only on shoot N content). Theamount of N fixed per unit dry mass of noduleswas very similar in the lower three populationsin Table 8 (0.47, 0.46 and 0.43 mg, respectively).However, the amount fixed by the Nicaraguan

Table 8. Variation in nodulation and nitrogen fixation and genetic differences in populations of Gliricidia sepium (source: data from

Sanginga et al. 1991; Chalmers et al. 1992)

Dry mass of nodules N2 fixed' Polymorphic

Origin of population (mg plant-1) (mg plant-1) Genetic similarity2 loci3

Managua, Nicaragua 56 38 0.78 24

Mariara, Venezuela 84 23 0.93 3

El Roblar, Costa Rica 64 30 0.85 15

Playa Tamarindo, Costa Rica 48 22 0.78 24

Pedasi, Panama 72 31 0.94 2

population was higher (0.67) and by theVenezuelan lower (0.27). The authors commentedthat the coefficients of variation for their datawere much higher for inoculated than forN-fertilized plants. All these data suggestconsiderable variation between and withinpopulations of G. sepium in their ability tonodulate and fix N.

Individuals from a provenance may vary gTeatlyin their ability to nodulate in the same soil. Thispattern is not obtained with highly selectedcultures of crop legumes. Because tree speciesare generally grown from unselected seed (whenvegetative propagation is not employed), it isperhaps not surprising that individual plants varywidely Further, as nodulation requires thematching of genotypes from two diversepopulations, it is not surprising that thecoefficient of variation for N-fixing plants isgreater than for plants grown on combinednitrogen. How, then, can we attempt to selecthost genotype/rhizobial strain combinations fortrees, bearing in mind the problems of assessingvariation and breeding of woody plants?

Assessment of variation may be greatly simplifiedby use of modern molecular methods. The datain the two right-hand columns of Table 8 arebased on the technique known as RAPD(randomly amplified polymorphic DNA). Thistechnique was used by Chalmers et al. (1992) toexamine within- and between-population variationin Gliricidia sepium and G. maculata. Of thepopulations shown, those from Venezuela andPanama were genetically relatively uniform, andtherefore perhaps not good candidates for tryingto select for improved N fixation. On the otherhand, the Nicaraguan population, which had themost efficient nodules, also displayed the mostgenetic diversity and is therefore a morepromising source of material for improvement.These studies are currently being extended tolook for genetic markers more closely related toN fixation. Within the next few years, large-scalescreening of woody germplasm for N-fixingpotential (and, of course, many other traits)should be possible. All we then have to do is tomatch good host germplasm to effective, efficientand competitive rhizobia. This matching is likelyto be difficult. Inoculation, particularly ofpromiscuous genotypes, is fraught with problems.Now that the basis of recognition between hostsand rhizobia is being elucidated (see Brewin1991 for a review), at least for a few cropspecies, we have the possibility of geneticallyengineering rhizobia for particular host genotypes.

CONCLUSION

There is great potential for domestication ofN-fixing trees. Before this potential can berealised, basic studies on the relationshipsbetween host, endosymbiont and the environment

are needed. Modern molecular techniques shouldaid in the selection of -host material and inmatching it to suitable rhizobia/Franida. Targetingone or two selected species for an intensivestudy might be most effective in defining waysfor screening populations, avoiding the time- andlabour-consuming processes of factorial hostgenotype/rhizobial (or Frankia) strain experiments.

ACKNOWLEDGEMENTS

I thank my many colleagues for fruitful discussionson these topics and for allowing me to quoteunpublished data. Our laboratory has beengenerously funded for tree research by the UKOverseas Development Administration, EuropeanCommission, CNPq (Brazil), British Council, and theNatural Environment Research Council.

REFERENCESAllen, O.N. & Allen, E.K. 1981. The Leguminosae. Madison:University of Wisconsin Press/London: MacMillan.

Aronson, K. & Toledo, C.S. 1992. Caesalpinia paraguariensis(Fabaceae): forage tree for all seasons. Economic Botany, 46,121-132.

Aronson, J., Ovalle, C. & Avendafio, J. 1992. Early growthrate and nitrogen fixation potential in forty-four legumespecies grown in an acid and a neutral soil from centralChile. Forest Ecology and Management, 47, 225-244.

Batzli, J.M., Graves, W.R. & van Berkum, P. 1992.Diversity among rhizobia effective with Robinia pseudoacac a.Applied and Environmental Microbiology, 58, 2137-2143.

Becking, J.H. 1992. The Rhizobium symbiosis of thenon-legume Parasponia. In: Biological nitrogen fixation, editedby G. Stacey R.H. Bowen & H.J. Evans, 497-559. New York:Chapman and Hall.

Brewin, N.J. 1991. Development of the legume root nodule.Annual Review of Cell Biology 7, 191-226.

Chalmers, K.J., Waugh, R., Sprent, J.I., Simons, A.J. &Powell, W. 1992. Detection of genetic variation between andwithin populations of Gfiricidia sepium and G. maculata usingRAPD markers. Heredity 69, 465-472.

Diem, H.G. & Dommergues, Y.R. 1990. Current andpotential uses and management of Casuarinaceae in thetropics and subtropics. In: The biology of Frankia andactinorhizal plants, edited by C.R. Schwintzer & J.D. Tjepkema,317-342. New York: Academic Press.

Dupuy, D.C. & Dreyfus, B.L. 1992. Bradyrhizobiumpopulations occur in deep soil under the leguminous treeAcacia albida. Applied and Environmental Microbiology 58,2415-2419.

Faria, S.M. de, Lewis, G.P., Sprent, J.I. & Sutherland,J.M. 1989. Occurrence of nodulation in the Leguminosae.New Phytologist, 111, 607-619.

Giller, K.E. & Wilson, K.J. 1991. Nitrogen fixation intropical cropping systems. Wallingford: CAB International.

Goi, S.R., Sprent, J.I., James, E.K. & Jacob-Neto, J.1993. Influence of nitrogen form and concentration on thenitrogen fixation of Acacia auriculiformis. Symbiosis, 14,115-122.

Hughes, C.E. & Styles, B.T. 1989. The benefits and risks ofwoody legume introductions. In: Advances in legume biology,edited by C.H. Stirton & J.L. Zaructhi, 505-531. (Monograph inSystematic Botany no. 29.) Missouri: Missouri Botanical Garden.

181

Johnson, H.B. & Mayeux, H.S. Jr. 1990. Prosopisglandulosa and the nitrogen balance of rangelands: extent andoccurrence of nodulation. Oecologia, 84,-176-185.

Johnson, L.A.S. 1988. Notes on Casuarinaceae III. The newgenus Ceuthostoma. lblopea, 3, 133-137.

Marcar, N.E., Dart, P. & Sweeney, C. 1991. Effect ofroot-zone salinity on growth and chemical composition ofAcacia ampliceps B.R. Maslin, A. auriculiformis A. Cunn. exBenth. and A mangiurn Wil ld. at two nitrogen levels. NewPhytologist, 119, 567-573.

Mariotti, A., Sougoufara, B. & Dornmergues, Y.R. 1992.Estimation de la fixation d' azote atmospherique par letracage isotopique naturel dans une plantation de Casuarinaequisetifolia (Forst). Soil Biology and Biochemistry 24, 647-653.

Ma.rtinelli, L.A., Victoria, R.L., Trivelin, P.C.O., Devol,A.H. & Richey, J.E. 1992.15N natural abundance in plantsof the Amazon river floodplain and potential atmospheric N2fixation. Oecologia, 90, 591-596.

Martinez-Romero, E., Segovia, L., Mercante, F.M.,Franco, A.A., Graham, P.H. & Pardo, M.A. 1991.Rhizobium tropici: a novel species nodulating Phaseolusvulgaris L. beans and Leucaena sp. trees. International Journalof Systematic Bacteriology 41, 417-426.

Mullin, B.C., Swenson, S.M. & Goetting-Minesky, P.1990. Hypotheses for the evolution of actinorhizal symbioses.In: Mtrogen fixation: achievements and objectives, edited byPM. Gresshoff, L.E. Roth, G. Stacey & WE. Newton, 781-787.New York: Chapman and Hall.

National Research Council. 1984. Casuarinas:nitrogen-fixing trees for adverse sites. Washington, DC: NationalAcademy Press.

Norby, R.J. 1987. Nodulation and nitrogenase activity innitrogen-fixing woody plants stimulated by CO2 enrichment ofthe atmosphere. Physiologia Plantarum, 71, 77-82.

Odee, D.W. & Sprent, J.I. 1992. Acacia brevispica: anon-nodulating mimosoid legume? Soil Biology andBiochemistry 24, 717-719.

182

Sanginga, N., Bowen, G.D. & Danso, S.K.A. 1990a.Genetic variability in symbiotic nitrogen fixation within andbetween provenances of two Casuarina species using the15N-labelling methods. Soil Biology and Biochemistry, 22,539-547.

Sanginga, N., Bowen, G.D. & Danso, S.K.A. 1990b.Assessment of genetic variability for N2 fixation between andwithin provenances of Leucaena leucocephala and Acaciaalbida estimated by 15N labelling techniques. Plant and Soil,127, 169-178.

Sanginga, N., Manrique, K. & Hardarson, G. 1991.Variation in nodulation and N2 fixation by the Gliricidiasepium/Rhizobium spp. symbiosis in a calcareous soil. Biologyand Fertility of Soils, 11, 273-278.

Schwintzer, C.R. & Tjepkerna, J.D., eds. 1990. The biologyof Frankia and actinorhizal plants. New York: Academic Press.

Shearer, G. & Kohl, D.H. 1986. N2-fixation in field settings:estimations based on natural 15N abundance. Australian Journalof Plant Physiology 13, 699-756.

Sprent, J.I. 1989. Which steps are essential for the formationof functional legume nodules? (Tansley Review, no. 15.) NewPhytologist, 111, 129-153.

Sprent, J.I. & Sprent, P. 1990. Nitrogen-fixing organisms:pure and applied aspects. London: Chapman and Hall.

Trinick, M.J. & Hadobas, P.A. 1988. Biology of theParasponia-Bradyrhizobium symbiosis. Plant and Soil, 110,177-185.

Turk, D. & Keyser, H.K. 1992. Rhizobia that nodulate treelegumes: specificity of the host for nodulation andeffectiveness. Canadian Journal of Microbiology, 38, 451-460

Witkowski, E.T.F. 1991. Effects of invasive alien acacias onnutrient cycling of the coastal lowlands of the Cape Fynbos.Journal of Applied Ecology 28, 1-15.

Young, J.P.W. 1992. Phylogenetic classification ofnitrogen-fixing organisms. In: Biological nitrogen fixation, editedby G. Stacey RH. Bowen & H.J. Evans, 43-86. New York:Chapman and Hall.

Case studies

Capturing and managing the genetic variation inProsopis spp. for economically useful characters

P FelkerCenter for Semi-And Forest Resources, Caesar ICeberg Wildlife Research Institute, Tbxas A&I University,Kingsville, TX 78363, USA

ABSTRACTSelf-incompatibility in Prosopis spp. results in obligately outcrossed trees whose progeny from a single parent tree may vary by afactor of 100% or more in the production of pods or total biomass. Asexual propagation is, therefore, required in orcler to capturegenetic variation in Prosopis spp. Despite ten years of research on three continents, tissue culture of Prosopis spp. has not proveduseful. Rooting of stem cuttings is the most effective technique for mass propagation of clonal material, but requires intensivelymanaged stockplants. A solar-powered mist system for the propagation of stem cuthngs has been developed for use in thetropics where power supplies are often intermittent. Rooting of cuttings from mature trees is difficult, which hinders the capturingof elite, proven genotypes identified in the field. Although air layering is effective in this respect, grafting techniques arepreferable and often 100% successful. Clones of various Prosopis species are available that produce large (25 cm), nutritious(44% sugar content) pods; grow in 3.2% NaC1 (seawater); are thornless, high biomass producers; and are non-browsed. Theseclones could be used: (i) directly in high biomass-producing fuelwood plantations; (ii) as scions to upgrade existing stands forgreater pod production and thornlessness; (ii) to create clonal seed orchards; or (iv) for grafting on to rootstocks with atolerance of high alkalinity or salinity

INTRODUCTION

In self-incompatible (Simpson 1977), obligatelyoutcrossing genera such as Prosopis, the geneticvariability in progeny is extremely large. While itis useful to conserve genetic variability to meetfuture needs, such as an outbreak of pests ordisease, it has been found that the variabilitywithin any batch of seedlings is so high that itlimits the commercial exploitation of manyindividual traits. To overcome this problem, andto capture the attributes of individual genotypes,some form of asexual propagation is necessary.It would allow the vegetative multiplication ofelite individual trees found growing in naturalstands or in field trials.

The four basic types of asexual propagation fortrees, namely tissue culture, rooting of stemcuttings, air layering and grafting, have beenevaluated with respect to Prosopis spp. Despiteconsiderable effort by several laboratories, areliable tissue culture protocol for Prosopis spp.has not developed to the point where it can beused even for research-scale production of clonalmaterial. In contrast, rooting of stem cuttings, airlayering and grafting techniques have now beendeveloped to the stage where they could beused operationally However, each of thesemethods has its advantages and disadvantages,which are considered here in the context of agenetic improvement programme.

TECHNIQUES OF ASEXUAL REPRODUCTION

Tissue culture

Laboratories in Chile (Jordan & Balboa 1985),England (Batchelor et al. 1989), India (Goyal &Arya 1981; Ramawat & Nandwani 1991;V Dhawan, personal communication) and Texas(Green, `Pabone & Felker 1990; Tabone et al.1986) have been working to develop tissueculture propagation methods for Prosopis. Whilebudbreak and shoot growth have been achievedfrom Prosopis explants, in no case were shootsproduced other than from the initial explant. Thatis, shoots could not be excised from the originalmother explant and successfully subcultured.

Our laboratory has found that, while 95% of theaxillary buds produce 2 cm long shoots in sixweeks, these shoots senesced and died by the12th week in culture (the explants aresubcultured every three weeks) (Tabone et al.1986). The shoots begin to yellow and die fromthe terminal tip. The young leaves are the first todie, followed by the older leaves. This symptomis typical of a deficiency of an immobile plantnutrient, such as calcium, for whole plantsgrowing in soil. However, increasing the calciumconcentrations in the media failed to overcomethe senescence.

After having developed a reasonablecytokinin/auxin concentration to stimulate shoot

183

production, Tabone (1986) examined all 20inorganic nutrients and vitamins at low (20%),normal (100%) and high (500%) concentrationsof the basal Murashige-Skoog media, but foundno striking nutrient deficiency that could accountfor this necrosis. Coconut milk, liquid cornendosperm, charcoal, ascorbate, sulthydrylreagents and polyvinylpyrrolidone were added tothe cultures without avail (Green 1989). To testthe possibility that gases such as ethylene wereproduced that inhibited shoot development, ahumidified air exchange system was constructedin side-arm flasks. Air exchange did not preventthe mortality of the shoot explants (Green 1989).

Mature Prosopis trees secrete gums at the endsof cut branches, probably to reduce theincidence of secondary infections. Similar gumsmay develop during in vitro culture, forming abarrier that prevents nutrient uptake. Trimmingbrown surfaces from the explants has beenfound to stimulate the growth of the explants ,forabout a week (R Schmidt & P Felker,unpublished data). Further research is neededon nutrient uptake by explants, if tissue cultureof Prosopis spp. is to become an effective toolfor .routine asexual propagation.

Rooting of stem cuttings

Rooting of stem cuttings can produce greaternumbers of asexual propagules than any othertechnique: for example, only about eight hours'labour is required to take 2000 cuttings. Felkerand Clark (1981) reported the rooting of stemcuttings of a number of Prosopis species, andnoted a strong seasonality in rooting ability withcuttings rooting well in the spring but not in thelate fall. Klass et al. (1985) examined theenvironment for rooting cuttings in a growth .chamber, and found that optimum temperatureswere about 35°C and that irradiances greaterthan 500 gmol m2 s-1 were required. Theyobserved a photoperiod effect, with cuttingsrooting best when stockplants and cuttingsreceived a daylength of 12 h and 18 h,respectively Souza (1986) examined the rootingof stem cuttings under plastic bags, under mistof varying frequencies and with fog systems, andfound that almost no cuttings rooted under fog,whereas high rooting percentages (around 70%)were obtained under mist.

Following studies by Felker and Clark (1981)and Klass, Wright and Felker (1987), it becameclear that one of the most important factorsinfluencing rooting was the growth environmentof the stockplants. If shoots of the P albastockplants were growing rapidly (50 cm lengthsin three weeks) and free from insect damage,the cuttings rooted without hormones orfungicides. Even with high light intensities andair temperatures, there was occasionally adecline in the rooting percentage of Prosopis

184

cuttings. Souza and Felker (1986) examined theinfluence of nutrient application regimes oncarbohydrate contents, N contents andpercentage rooting of Prosopis alba cuttings, andfound optimal rooting was obtained whenstockplants received 500 mg 1-1 N twice a week.Under intensive management, Prosopis albastockplants produced about 50 two-node cuttingsper month.

In contrast to the ease with which cuttingsrooted when taken from intensively managedjuvenile stockplants grown in a glasshouse, it hasbeen difficult to root cuttings from maturefield-grown trees. Typical success rates havebeen less than 1%. The development ofpropagation techniques which enable materialfrom mature trees to be rooted successfully is,therefore_ a high priority

In 1988, a company named Plantclone Inc wasformed in Kingsville, Texas, to produce P albaclone B2V50 commercially by mist propagation,for sale as ornamentals. Commercially viablemist propagation systems were developed, using80 stockplants in 20 litre pots', that routinelyproduced about 1000 rooted cuttings per monthfrom June through October. This periodcoincided with air temperatures >32°C and high(>1000 gE m-2S-1) light intensities. Propagationof clone B2V50 was discontinued in 1991because of severe damage of outplanted treesby temperatures of -12°C.

Prosopis spp. are an important source offuelwood and fodder in many less-developedcountries, and clearly offer great potential fordomestication by vegetative propagation andclonal selection, if appropriate propagationtechniques can be developed. Sandys-Winch andHarris (1991) reported a technique for rootingcuttings that did not require electricity Thistechnique 'required placing large (1-2 cmdiameter and 30 cm long) cuttings withoutleaves in an enclosed environment. After 22weeks of twice-daily watering, these cuttings had97% rooting success. Similarly the low-technologynon-mist system described elsewhere in thisvolume (see papers by Leakey Newton & Dick,pp72-83; Mesén, Boshier & Cornelius,pp249-255; Ladipo et al., pp239-248), whichdoes not require an electrical or piped watersupply has also been used successfully for thepropagation of Prosopis juliflora (see Leakey etal., pp72-83, and references therein). Wojtusik,Boyd and Felker (1993) recently developed asolar-powered mist system that yielded 90%rooting success on location in Haiti. A watersupply with a-minimum pressure of 1.5 MPa wasrequired for the mist nozzles. This system used asolar panel to charge a 12V car battery whichoperated a solenoid valve when the moistureevaporated from a stainless steel 'leaf' under themist system.

Air layering

Air layering techniques for mature Prosopis spp.have been developed in the Rajasthan desert(Solanki, Kackar & Jindal 1986). A branch about70 cm long is girdled where it is about 1 cm indiameter. The bark is removed from a sectionabout 2 cm long, atuoin powder is applied, andmoist peat moss is wrapped tightly around thegirdled branch with clear plastic. Roots appearthrough the plastic after six to eight weeks.Some of the leaves are shipped from thesevered branch prior to potting. In the USA,20-30% success has been achieved using thistechnique on mature_ trees. The technique is,therefore, useful for obtaining the first asexualpropagule from mature field-grown trees.However, it is not appropriate for masspropagation because of the low multiplicationrate.

Grafting

Wojtusik and Felker (1993) recently examinedfour graft types for Prosopis spp. and found thatcleft and whip grafts were equally successful,while side-veneer grafting was less useful, andT-bud grafts failed completely Graft compatibilitybetween North American (P glandulosa andP articulata), South American (P alba andP chilensis) and tropical species (P juliflora) wasnearly 100%. The opportunity to use grafting tocapture mature elite genetic material for furthermultiplication was demonstrated by its successfulapplication in Texas on P alba rootstocks, usingscions from five fast-growing, erect, non-browsedand thornless P juliflora trees in a five-year-oldHaitian field trial (Wojtusik et al. 1993).

A significant rootstock effect on biomassproduction was noted in graft compatibility trials(Wojtusik & Felker 1993). Scions of all specieshad greater biomass production when grafted onto P alba or P chilensis rootstocks than whengrafted on to themselves.

USEFUL GENETIC CHARACTERISTICS OFPROSOPIS SPP.

The 44 species of Prosopis native to North .andSouth America, Africa, and Asia have greatvariability with regard to environmental tolerancesand useful economic characters. Burkart's (1976)monograph provides the most comprehensivedescription of variation in morphological traits.Prosopis species grow from sea level to 4000 melevation (P ferox in Argentina), and from theequator to 48°S latitude in Argentina. Somespecies are prostrate shrubs (P strombulifera),while others can grow into trees 20 m tall(P juliflora and P alba). Some species havehorny and thornless individual trees. ThornlessP alba trees were much more susceptible todamage from goat browse than P juliflora from

Peru in Haitian field trials with 70 open-pollinatedfamilies (Lee, Russell & Felker 1991). These trialsalso demonstrated great variation in the numberof mainstems produced and in the presence oferect form.

In a progeny trial in the California ImperialValley over a 100-fold range in the biomassaccumulation of 55 families of different Prosopisspp. was observed after two years' growth(Felker et al. 1983). P alba and P chilensisdisplayed higher growth rates than theCalifornian or Arizonan native Prosopis species.The slowest-growing species was P tamarugofrom the Chilean salt deserts. The largestindividual trees in this trial were subsequentlycloned by rooting stem cuttings.

There is also considerable range in tolerance ofProsopis spp. to temperature. Most Prosopis spp.tolerate air temperatures of over 40°C, if water isavailable. However, the species vary greatly intheir tolerance to absolute minimum temperatures(Felker et al. 1981). Prosopis juliflora from theCaribbean or Peru is a truly tropical specieswhich is completely killed by temperatures of—4°C. These temperatures are tolerated withoutdamage by P alba and P chilensis, although theyare killed by extended periods of —12°C(P Felker, unpublished data). In contrast, Texasnative P glandulosa var. glandulosa tolerates 5 hof —18°C without damage. However, P glandulosaappears to require sustained- 30°C temperaturesfor budbreak and temperatures of 35°C foroptimal growth.

There is great variation in the resistance ofProsopis spp. to attack by psyllid insects. Theyoung tender growth of P glandulosa var.torreyana from the California desert is very -susceptible to psyllid damage, while the SouthAmerican species P alba and P chilensis aremuch more resistant (P Felker, unpublished data).It should be noted that this is not the samepsyllid that damages Leucaena leucocephala,which is 100% resistant to the psyllids that attackProsopis spp. (P Felker, unpublished).Considerable variation in psyllid resistance hasalso been noted among the algarrobos inplantations in the Chilean salt deserts. In thiscase, resistant individuals could have beengrafted on to susceptible trees.

When Prosopis spp. native to the hot (42-48°C)California desert were planted in a cooler(32-38°C), but higher-humidity environment inTexas near the Gulf of Mexico, all 100 trees diedfrom attack by stem cankers within three years(P Felker, unpublished data). Prosopis velutinafrom Arizona and P flexuosa from the interiordeserts of Argentina were similarly affected butto a lesser extent. In contrast, the Texas nativeP glandulosa var. glandulosa and the Argentinespecies P alba and P chilensis appeared to becanker-resistant.

185

The variation in Prosopis pod production andchemical characteristics is large. The mean podproduction per mother tree from a five-year-oldCalifornia trial ranged from 7.1 kg tree-1 to0.0 kg tree-1 (Felker et al. 1984). Perhaps evenmore strildng was that, within the family with thegreatest mean production, the pod productionranged from 3.2 to 12.2 kg tree-1. Pod sugarand protein contents in a California field trialranged from 10% to 40% and from 10% to 17%respectively (Oduol et al. 1986). In spite of highanalytical values for pod sugar content of thetrees, some trees have a bitter after-taste whileothers are relatively mild and sweet. While mostProsopis pods are about 15 cm long, some inBrazil are nearly 50 cm long. Thus, there is greatopportunity for selection for pod size, sugar andprotein content, and for lack of astringency

There is also considerable variation in Prosopisspp. response to soil factors. Rhodes and Felker(1987) exposed 100 seedlings of nineopen-pollinated families to high salinity in agreenhouse environment, and found individuals ofP alba, P flexuosa, P chilensis and P juhflora thatcould grow in 3.2% NaC1, the same salinity asseawater. Individual trees that continued to growin this salinity have been repotted and could beasexually propagated for use as salt-tolerantrootstock. When some of these clones wereexamined in saline but waterlogged fieldconditions in southern Paldstan, they grew poorly(P Felker, unpublished data). The very thornyP articulata and P juliflora appear to be theProsopis species with the greatest tolerance ofwaterlogging. Singh, Abrol and Cheema (1989)found that P juliflora was able to grow in sodicsoils with a pH as high as 10.4, if the soils wereamended with zinc, gypsum and barnyardmanure. Without these amendments, P juliflorawas able to grow in soils with a pH of 9.0.

In a greenhouse trial examining nitrogen fixationin 13 Prosopis species, Felker and Clark (1981)found ten-fold differences in the rates of Nfixation for these species. The N fixation rateswere highly correlated with biomass productionin two-year-old field-grown trees. It is unclearwhether the greater N fixation caused greatergrowth, or whether greater growth provided morephotosynthate to make more N fixation possible.

Attributes of Prosopis species that are generallyoverlooked are the beautiful orange/red colour ofthe wood and its technical qualities for finefurniture and flooring manufacture (Tortorelli1956; Weldon 1986). It should not be surprising,given the beauty and excellent technicalproperties of other tropical legumes such asDalberghia sissoo (Indian rosewood) andPterocarpus spp. (cocobolo). The wood ofProsopis spp. is somewhat harder than speciessuch as Quercus (oak), Prunus (cherry), andJuglans (walnut), but its real advantage is that its

186

volumetric shrinkage (3-4%) is among the lowestof any species in the world. When solid furnitureis moved from high- to low-humidityenvironments, the wood contracts or expandsdepending on the air humidity Having one ofthe world's lowest values for volumetricshrinkage, Prosopis has much less of a tendencyto expand and contract than other finehardwoods. Prosopis alba from Argentina(Tortorelli 1956) and P glandulosa (Weldon 1986)from Texas also have low values for volumetricshrinkage.

The practical implication of these technicalqualities is that Prosopis timber should commanda price of about $400-800 m-3 (700 kg) whichis nearly five to ten times greater than its valuefor firewood. Thus, where economic developmentissues are important, the development andmanagement of Prosopis spp. for fine furniturewood could be considered. Value-addedmanufacture of fine furniture or flooring wouldfurther increase the value of this resource.

USE OF CLONAL MATERIAL IN THEGENETIC IMPROVEMENT PROCESS

To capture the first asexual propagule fromproven field trees, it is necessary to use eithergrafting or air-layering techniques, or pollardingto produce juvenile material. Grafting producesmany more individuals with less effort than airlayering, although many more individuals can beproduced from stem cuttings if the tree resproutscopiously after pollarding. After clonal stockplantshave been obtained, the resulting shoots caneither be used to provide rooted cuttings fortransplanting to the field, additional stockplants,material for clonal seed orchards, or scions forgrafting on to existing trees.

Where fuelwood plantations are desirable, cloneswhich produce high biomass will obviously bemost useful. When rooted cuttings of Prosopisalba were used to establish plantations forfuelwood, productivities of 21 Mg ha-1 drybiomass were obtained in the third growingseason (Felker et al. 1989). There are manyother applications where clonal material may alsobe of value. For example, the mature trees inthe 1000 ha Prosopis alba/flexuosa/chilensisplantations in northern Chile appear to varygreatly in pod colour, size and production, andin resistance to stem cankers and psyllids (PFelker, unpublished).

Grafting could be used to circumvent thenumerous problems that prevent successful treeestablishment in semi-arid regions.Well-established seedlings arising from naturalregeneration could be cut off above the browseline and superior, thornless material could begrafted on to them. This would reduce therequirements for nurseries, the transportation ofseedlings to the field, and the costs of labour

and maintenance. The hybridisafion ofgeographically distant trees could be greatlyfacilitated by grafting the desirable parents on tothe same tree or at least grafting all possibleparents into the same field plot. The need tosurvey dates of pollen shed, and the difficultiesin transporting live pollen to conduct the crossescould then be avoided. It would also be usefulto explore the influence of grafting on processessuch as flowering, dwarfing, date of budburst,and cold hardiness, that are known to beinfluenced by the rootstock in other tree species.

Some of the most intriguing rootstocks would bethose found by Singh et al. (1989) to tolerate pHvalues of 10.4 and those of Rhodes and Felker(1987) that grew in 3.2% NaCl. Our casualobservations suggest that P articulate, which isextremely spiny may be one of the most tolerantProsopis spp. to conditions which are bothsaturated and saline. Unfortunately the pods ofP articulate have some of the lowest sugarcontents of all Prosopis species, and the treeshave such long and profuse thorns that theywould be extremely undesirable in almost allcircumstances. The P juliflora in India thattolerates pH 10.4 conditions is not as spiny as Particulate, but the pods are quite bitter and notnearly of the same value as other P julfflora or Palba. Thornless clones of Prosopis alba and Pjuliflora with abundant high-sugar pods could begrafted on to rootstock which, other than itstolerance to severe edaphic conditions, would becompletely useless.

Salt-tolerant clones could be of immense value inareas such as southern Pakistan, where irrigationmismanagement has destroyed hundreds ofthousands of hectares of land that is now toosaline and too waterlogged to grow arable crops.Foresters in Gujarat, India, have begun tomanage Prosopis spp. as one of the few plantscapable of growing in the saline groundwaterresulting from seepage from the ocean. R Ahmad(unpublished) at the University of Karachi,Pakistan, has irrigated Prosopis spp. with highlysaline water along the coasts of Pakistan, toprovide windbreaks for controlling sand dunemovements on to roads and villages.

In some instances, rooted cutting technology maybe viewed as too technologically advanced, inwhich case it would be desirable to use seedseven if the full genetic potential of clones werenot achieved. There has been a considerabledemand for Prosopis seed that would producethornless trees, and it would be advantageous touse superior thornless clones in a seed orchardto produce genetically improved seed. Naturallyoccurring thornless trees produce seed thatresults in thorny and thornless trees due tooutcrossing with wild thorny trees. Thus, there isan opportunity to combine thornless clones in anisolated seed orchard to produce seed which

produces thornless plants, that also possessimproved growth rates, improved pod production,greater salt tolerance, etc. Because Prosopis isself-incompatible, a seed orchard of a singleclone would probably result in infertile seed.Thus, multiple thornless clones would berequired in the seed orchard. Heterosis shouldoccur in these seed orchards resulting in greatergrowth rates in the progeny than either of theparents. Mating schemes for clonal seed orchardshave been described in detail. Before anylong-term investment in clonal seed orchards, themating designs should be reviewed carefully

This author has frequently observed large plantsof Prosopis spp. being used around homes forshade trees in south-western USA, Haiti, Mexico,Chile and Argentina. In the United States, podproduction in yards is highly undesirable andtrees with minimal pod production would bepreferred. In contrast, in developing countries,where abundant forage for animals would be anadded advantage to shade, use of thornlessProsopis clones with a propensity for producinglong sweet pods would be most desirable. Theseclones could be produced by rooted cuttingsand grown in 20 litre polyethylene bags typicalof nursery stock, or existing wild trees could begrafted.

CONCLUSIONS

A suite of asexual propagation technologies isnow available to capture the qualities ofoutstanding elite genotypes and to multiply theseclones rapidly for use in genetic improvementprogrammes, to produce clonal seed orchards, orfor direct use in plantations. Grafting techniquesnow offer the possibility of combining trees witha resistance to adverse edaphic conditions, suchas high salinity with trees with other usefulcharacteristics, such as thornlessness or podproduction. Imaginative use of these techniquesoffers great potential for reforesting arid lands.

REFERENCESBatchelor, C.A., Yao, D., Koehler, M.J. and Harris,P.J.C. 1989. In vitro propagation of Prosopis species (Pchilensis, P cineraria, and P julillora). Anna les des SciencesForestieres, 46, 110-112.

Burkart, A. 1976. A monograph of the genus Prosopis(Legurninosae subfam. Mimosoideae). Journal of the ArnoldArboretum, 57, 217-247; 450-525.

Felker, P. & Clark, P.F. 1981. Rooting of mesquite(Prosopis) cuttings. Journal of Range Management, 34, 466-468.

Felker, P., Connell, G.H., Clark, P.R., Osborn, J.F. &Nash, P. 1983. Biomass production of Prosopis species(mesquite), Leucaena, and other leguminous trees grownunder heat/drought stress. Forest Science, 29, 592-606.

Felker, P., Clark, P.R., Nash, P., Osborn, J.F. &Cannell, G.H. 1981. Screening Prosopis (mesquite) for coldtolerance. Forest Science, 28, 556-562.

Felker, P., Clark, P.R., Osborn, J.F. & Cannell, G.H.1984. Prosopis pod production - a comparison of North

187

American, South American, Hawaiian, and germplasm inyoung plantations. Economic Botany, 38, 36-51.

Felker, P., Smith, D., Wiesman, C. & Bingham, R.L.1989. Biomass production of Prosopis alba clones at twonon-irrigated field sites in semiarid south Texas. ForestEcology and Management, 29, 135-150.

Goyal, Y. & Arya, H. C. 1981. Differentiation in cultures ofProsopis cineraria 'Linn. Current Science, 50, 468-469.

Green, W. 1989. Stabilization of mesquite (Prosopis alba clone82V50) during the multiplication stage of tissue culture andmultiplication of prickly pear nopal cactus (Opuntia spp.) usingtissue culture. MSc thesis, Texas A&I University

Green, B., Tabone, T. & Felker, P. 1990. A comparison ofamide and ureide nitrogen sources in tissue culture of treelegume Prosopis alba clone B2V50. Plant, Cell, Tissue andOrgan Culture, 21, 83-86.

Hunziker, J.H., Saidman, B.O., Naranjo, C.A., Palacios,R.A., Poggio, L. & Burghardt, A.D. 1986. Hybridizationand genetic variation of Argentine species of Prosopis ForestEcology and Management, 16, 301-316.

Jordan, M. & Balboa, 0. 1985. In vitro regeneration ofProsopis tamarugo Phil. and Prosopis chilensis (Mol.) Stuntzfrom nodal sections. Gartenbauwissenschaft, 50, 138-142.

Klass, S., Wright, J. & Felker, P. 1987. Influence of auxins,thiamine, and fungal drenches on the rooting of Prosopis albaclone B2V50 cuttings. Journal of Horticultural Science, 60,97-100.

Klass, S., Bingham, R.L., Finkner-Templemen, L. &Felker, P. 1985. Optimizing the environment for rootingcuttings of highly productive clones of Prosopis alba(mesquite/algarrobo). Journal of Horticultural Science, 62,275-284.

Lee, S.G., Russell, E. & Felker, P. 1991. Discovery ofthornless, non-browsed, erect tropical Prosopis in Haitianprogeny trials. Forest Ecology and Management, 46, 189-200.

Nandwani, D. & Ramawat, K. G. 1991. Callus culture andplantlets formation from nodal explants of Prosopis juliflora(Swartz) DC. Indian Journal of Experimental Biology 29,523-527.

Oduol, P.A., Felker, P., McKinley, C.R. & Meier, C.E.1986. Variation among selected Prosopis families for podsugar and pod protein contents. Forest Ecology andManagement, 16, 423-433.

Ramawat, K.G. & Nandwani, D. 1991. Propagation ofProsopis species: problem, perseverance and perspectives.Annals of Arid Zone, 30, 247-258.

188

Rhodes, D. & Felker, P. 1987. Mass screening Prosopis(mesquite) seedlings for growth at seawater salinity ForestEcology and Management, 24, 169-176.

Sandys-Winch, D.C. & Harris, P.J.C. 1991. A simplemethod for the vegetative propagation of Prosopis juliflora.Nitrogen Fixing Tree Reports, 9, 117-118.

Simpson, B.B. 1977. Breeding systems of dominant perennialplants of two disjunct warm desert ecosystems. Oecologia, 27,203-226.

Singh, G., Abrol, I.P. & Cheema, S.S. 1989. Effects ofgypsum application on mesquite (Prosopis juliflora) and soilproperties in an abandoned sodic soil. Forest Ecology andManagement, 29, 1-14.

Solanki, K.R., Kackar, N.L. & Jindal, S.K. 1986. Airlayering in Prosopis cineraria (L.) MacBride. The IndianForester, 112, 202-207.

Souza, D.S. 1986. Vegetative propagation of Prosopis albacuttings as ihfluenced by the stock plant, mineral nutrition, andmisting procedures. MSc thesis, Texas A&I University

Souza, D.S. & Felker, P. 1986. The influence of stock plantfertilization on tissue concentrations of N, p and carbohydratesand the rooting of Prosopis alba cuttings. Forest Ecology andManagement, 16, 181-190.

Tabone, T.J. 1986. Techniques in the clonal multiplication ofProsopis alba clone B2V50 through axillary bud culture. MScthesis, Texas A&I University

Tabone, T.J., Felker, P., Bingham, R.L., Reyes, I. &Loughery, S. 1986. Techniques in the shoot multiplication ofthe legiuninous tree Prosopis alba clone B2V50. ForestEcology and Management, 16, 191-200.

Tortorelli, L. 1956. Maderas y bosques Argentinos. BuenosAires: Acme Agency Press.

Weldon, D. 1986. Exceptional physical properties of Texasmesquite wood. Forest Ecology and Management, 16, 149-153.

Wojtusik, T. & Felker, P. 1993. Inter-specific graftincompatibility in Prosopis. Forest Ecology and Management,59, 329-340.

Wojtusik, T., Boyd, M.T. & Felker, P. 1993. Developmentof solar powered mist system for vegetative propagation ofProsopis in underdeveloped countries. Forest Ecology andManagement. In press.

Wojtusik, T., Felker, P., Russell, E.J. & Benge, M.D.1993. Cloning of erect, thornless, non-browsed nitrogen fixingtrees of Haiti: principal fuelwood species (Prosopis juliflora).Agroforestry Systems, 21, 293-300.

Domestication of lesser-known tropical treespecies: the Australian experience

T H Booth1 & J W Turnbull 2

1 Australian Tree Seed Centre, Commonwealth Scientific and Industrial Research Organisation, Division ofForestry, PO Box 4008, Canberra, ACT 2600, Australia2Australian Centre for International Agricultural Research, PO Box 1571, Canberra, ACT 2601, Australia

ABSTRACTThe CHRO Australian Tree Seed Centre has played a major role in domesticating tree species for over 25 years. This paperdescribes recent work carried out by the Centre s staff and other researchers in Australia, Zimbabwe, Kenya, Pakistan, Indonesia,Thailand and China. The work, which is being supported by the Australian Centre for International Agricultural Reseamh, hasassessed the potential of lesser-known Australian species for multipurpose use. This paper provides a brief description of themajor phases in the domestication process, including species selection, seed collection and distribution, nursery practice, speciesand provenance trials, species/site matching, tree improvement, tree management and utilisation. Progress with the domestication ofAcacia auriculiformis is described as a detailed case study

INTRODUCTION

Libby (1973) outlined strategies for thedomestication of forest trees in which herecognised six levels of tree improvement. Henoted that the utilisation of most tree species stillfollows a pattern established by neolithicsocieties. Wild seeds are used to establishplantations and these are the source of seed forexpanding the plantation resource, with little orno attempt to improve seed quality Only a smallnumber of Australian species have advancedbeyond this lowest rung of the domesticationladder, although in some of these species (suchas Eucalyptus grandis) the domestication processis well advanced (Table 1). There are dramatic

Table I. Increasing productivity of Eucalyptus plantations inBrazil due to silvicultural treatments (source: Campinhos 1991)

Period

1966-70

1971-75

1976-80

1981-85

1986-90

Silvicultural practicesMean productivity

(m3 ha-1 yr-1)

Low genetic quality 17.4seeds and fertilizeruse

Genetically pure 22.3seeds (unimproved)and fertilizer usage

As in previous 35.0stage, but fromselected stand

Improved seeds 44.9and vegetativepropagation

As in previous 60.0stage and additionalselection

improvements in productivity to be gained, evenin the early stages of tree improvement (seeTable 1). For example, an economic evaluation ofthe tree introduction programme in Chinasupported by the Australian Centre forInternational Agricultural Research (ACIAR)indicated internal rates of return on investment inthe range 27-45%, which were largely due tothe correct selection of provenances (McKenneyet al. 1991).

This paper outlines only the first stages in thedomestication process, and in particular thoseleading to the selection of the most suitableprovenances for use in a specific area. Thestages considered are: species selection, seedcollection and distribution, nursery practice,species and provenance trials, species/sitematching, tree improvement, tree managementand utilisation. The latter stages of domesticationdescribed by Libby (1973) are an iterativeprocess of tree improvement based aroundclones. Each stage uses a set of 'clonal' seedorchards, clone-holding orchards, 'seedling' seedorchards, clone evaluation plantings, and generalevaluation plantings. Clonal approaches are notdiscussed further here (see papers in thisvolume by Foster & Bertolucci, pp103-111;Felker, pp183-188; Leakey Newton & Dick,pp72-83; Ladipo et al,, pp239-248; Mesén,Boshier & Cornelius, pp249-255; Milimo, Dick &Munro, pp210-219).

The paper draws examples from two main areasof work. The first is an eight-year programme ofcollaborative research into the domestication of'lesser-known' Australian trees, such as Acacia,

189

Casuarina, Melaleuca, Grevillea and Sesbaniaspecies. The programme was co-ordinated byACIAR and involved researchers in national

• institutes in Zimbabwe, Kenya, Pakistan, Indonesia,Thailand and China, as well as Australia. Thesecond is the 25-year experience of theAustralian Tree Seed Centre (ATSC) based atCSIRO's Division of Forestry The main objectiveof the Centre is to support the collection,distribution and improved use of the geneticresources of Australian trees. The Centreresponds to over 2000 enquiries every year anddispatches about 15 000 seedlots from over 600species to researchers in 100 countries. A majorpart of the Centre's funds are provided by theAustralian International Development AssistanceBureau (AIDAB).

GENERAL METHODS

Species selection

The first stage of any domestication process is todecide which species will be evaluated. Thisdecision will depend on the needs which are tobe met and the number of species which aprogramme can examine. In April 1983, ACIARorganised a meeting of foresters and botanists toidentify 'lesser-known' Australian tree and shrubspecies with potential for multipurpose use.rrurnbull (1986) summarised the selection criteriaas follows:

• plants capable of providing products andservices in addition to fuelwood;

• adaptable plants that are easily establishedand maintained;

• plants capable of growing in extremeenvironments, including arid and humidtropical zones, infertile soils, heavy clays,saline, highly alkaline or waterlogged orexposed coastal situations.

Other desirable characteristics noted were anability to fix atmospheric nitrogen, a capacity forrapid growth, an ability to coppice, and goodburning properties. Using these criteria, 108species were selected from 170 initiallysuggested. As this was to be a large programmeof research, the selection criteria were broadand numerous species were chosen. If resourcesallowed only a small number of species to beexamined, then the selection criteria would haveto be more specific.

When the species have been chosen, existinginformation about them can be collated. For theACIAR programme, a book was producedsummarising the characteristics of 100 of thespecies (TUrnbull 1986). It included the botanicalname, common names, family main attributes,botanical features, geographical distribution,physiography (ie where it is found in thelandscape) and soils, vegetation type, utiisation,

190

silvicultural features, pests and diseases,limitations, related species, research contacts andreferences. The two-page description alsoincluded a map, showing the species distributionand a photograph of a typical specimen. Scantinformation was available for many of the'lesser-known' species. However, the collation ofinformation served to highlight gaps inImowledge which could be filled during thecourse of the programme. (Though a book maynot be produced by smaller projects, a similarformat could be followed.)

Seed collection and distribution

The availability of high-quality andwell-documented seed is a prerequisite for anyresearch underpinning a domesticationprogramme. Securing the seed may involvedetailed planning over several years and asubstantial investment of resources. Tree seedcentres play an important role in this activityTheir basic role is to collect, document anddistribute authenticated seed.

The objective of first seed collections is tosample some of the genetic variation within andbetween populations. For each species, seedshould be collected from at least twoprovenances in contrasting environments, but thesampling strategy may vary according to thevariability of the environment in which thespecies occurs and the extent of its naturaldistribution. If possible, seed samples should becollected from several widely spaced (>100 mapart) trees at each site. However, if the seedcrop is small, it may be necessary to bulk theseedlots. The techniques used in the ACIARcollections have been described by Searle (1989).

It is important to record information about thestand of trees and the environmental conditionsat the sampling site. The ATSC uses a standardseed collection data sheet which includesessential entries such as species, seedlotnumber, provenance name (to be used in a database), location name, latitude, longitude andaltitude. If possible, information on factors suchas aspect, slope, soil texture, pH, geology andvegetation associations are also added, as wellas mensurational data such as height, diameter,habit and crown dimensions. The data sheetindicates whether botanical voucher specimens,photographs and bulk or individual tree sampleswere taken. Space is also provided for seedviability assessments made in the laboratory

Nursery practice

It is pointless expending great effort to collectseed if they cannot be stored effectively andthen used to raise seedlings of a size suitablefor planting. Doran (1986) and Ryan et al. (1987)have summarised some of the techniques usedin the ACIAR programme. They describe

pre-sowing treatments, optimum temperature,moisture and light conditions for germination,choice of seedling containers, soils, use offertilizers and insecticides, soil sterilisation,inoculation with symbiotic organisms, time ofsowing, transplanting, shade, shelter, watering,weed control, root pruning and hardening off.They also consider tree establishment, includingsite preparation, spacing, time of planting,transport of plants to the field, planting method,fertilization, weed control and pest protection.

Relatively few Australian trees have 'recalcitrant'seeds, but storage is a problem with manytropical species. For example, difficulties may beexperienced with storing many dipterocarpspecies for more than a few weeks. Chin (1990)provides a general review of recalcitrant seedswhich cannot be successfully stored under colddry conditions (see also Tompsett, pp61-71).

Species and provenance trials

Arboreta containing a few individuals of manyspecies have demonstration value, but are nosubstitute for well-designed replicated trials.Boland and Ttrnbull (1989) summarised some ofthe design decisions behind the ACIAR speciestrials. The project initially used a randomisedcomplete block design, with square plots of 24or 36 trees and with three or four replications.The design was largely determined by theobjective of the project, which was to assess thegrowth and survival of about 25 lesser-knownAustralian species at each site over a period ofsix to ten years. The large plot sizes providedsome buffering between plots and allowed forthinning, which was required in some cases atthree years. Later in the project some trials wereestablished using incomplete block designs withfive to ten tree line plots. Such trials have ashort life (2-3 years), but are comparativelyinexpensive to set up. They are well suited toadverse environments, where planting may needto be repeated over several years to study theeffects of erratic climatic conditions.

Training in statistical design and analysis was animportant output from the ACIAR project. A bookentitled Design and analysis of field trials inforestry is in preparation, and Matheson (1990a,b) has provided a brief introduction to thesubject. The Australian Tree Seed Centre hasproduced a more general guide to nurserypractice and the establishment ofspecies/provenance trials (Carter 1987), whichhas been widely distributed and translated intoseveral languages.

As it was desirable to establish the species trialsover a range of relevant climatic and soilconditions, the initial ACIAR trials were located inAustralia, China, Kenya, Thailand and Zimbabwe,on sites located in areas where there was aperceived need for the species being tested.

Accessibility and security were major concerns.Assessment procedures were standardised as faras possible. The MFIDAT microcomputer database system developed by the United StatesAgency for International Development (USADD) inits Forestry and Fuelwood Research andDevelopment Project (F/FRED) and by the CSIRODivision of Forestry provides a Useful frameworkfor storing data from trials.

Species identified in the AC1AR trials as havinggood potential were further tested in range-wideprovenance trials (Gwaze 1992). Some of theseprovenance trials used individual progenies, sothat there was the potential to convert the trialinto a seed production area.

Species/site matching

Establishing, maintaining and assessing trials isan expensive operation. It is important to makethe most of information from these trials, so thecapability of other sites can be assessed. Recentdevelopments in interpolating climatic data havegreatly improved the ability to assess thepotential of sites in different regions, countries orthroughout the whole world. New methods topredict where and how well different species andprovenances may grow in different environmentshave been summarised by Booth (1991).

Tree improvement

The application of the results of species andprovenance trials is dependent on the availabilityof seed. Some native species may come fromareas which are not readily accessible. In thecase of exotics, seed may be available from thecountry of origin, but ultimately it is desirablethat a seed source is developed in the countrywhere the seed will be used. Species such asAcacia crassicarpa, Casuarina junghuhniana,Eucalyptus grandis and E. urophylla have beenidentified in species/provenance trials as havingpotential for wider planting, and seed orchardshave been established for these species. In somecases, the seed orchards have been developedas provenance/progeny trials, from which poorlyperforming progenies can be culled. Thisprocedure can accelerate the production ofhigh-quality seed, but it must be done carefullyand represents only the first step in treeimprovement.

Matheson (1990a, b) has provided a briefintroduction to tree improvement methods. He notesthat a worthwhile breeding stiategy requires:

• a breeding population;

• methods of selecting superior material;

• a method of regenerating and upgrading thebreeding population (by open or controlledpollination); and

• a method of transferring material from thebreeding population to the planting site (a

191

seed orchard, nursery cuttings or a tissueculture laboratory).

He outlines some alternative approaches,including simple mass selection methods or morecomplex recurrent selection strategies. He alsoprovides seven examples of strategies being usedfor different species in different parts of the world.

It is important to develop breeding plans at anearly stage in the domestication process, so thatstaff and financial resources can be allocatedefficiently Research to understand thereproductive biology of the species, includinglevels of inbreeding, is an integral part of anytree breeding programme. Breeding plans forAcacia mearnsli and Eucalyptus globulus havebeen prepared by Raymond (1987, 1988) andACIAR has supported research on thereproductive biology of both species. Vegetativepropagation can also be used for the masspropagation of the products of a treeimprovement programme. ACIAR has supportedresearch on cuttings and micropropagation oftropical acacias, and some practical techniqueshave been developed (Carron & Aken 1992).

The genetic engineering of tree species to addcharacteristics, such as partial resistance to insectdefoliation or herbicides, may be an increasinglyimportant part of tree improvement in the future.However, the first requirement is to ensure thattransformed trees are reproductively sterile. Thiswill prevent the 'escape' of foreign genes intonative forests.

Tree management

To realise the full potential of any species,research into tree management, includingnutrition, weed control, spacing, coppicing andthinning, should proceed in parallel with treebreeding (see Table 1). Where appropriate, thenutrition research should include work onsymbiotic organisms, which may dramaticallyaffect productivity particularly on poor sites(Reddell, Rosbrook & Ryan 1989; Dart,Umali-Garcia & Almendras 1991). ACIAR hassupported research on ectomycorrhizal,Rhizobium and Frankia associations aimed atselecting effective symbionts and developinglow-cost appropriate technologies.

Utilisation

All the work described in the preceding sectionsaims to produce effective management systems,but they are useless if the trees do not providethe product(s) required by the grower. Whetherthe desired products include sawn timber, pulp,fuelwood, fodder, tannin, honey or some otheroutput, the appropriate parts of the trees mustbe evaluated to ensure that a satisfactory productcan be produced (see, for examples, resourceevaluation and utilisation papers in Boland 1989and Turnbull 1991).

192

DOMESTICATION CASE STUDY: ACACIAAUltICULIFORMISThe domestication of Acacia auriculiformis, afast-growing nitrogen-fixing tree from Australia,Papua New Guinea and Indonesia, began about50 years ago. The original seed introductions tomost countries were haphazard, in the sense thatno records were kept of seed origin or geneticquality Initial introductions were for ornamentalpurposes and were undoubtedly collected mainlyfrom planted trees, with easy accessibility beingthe chief selection criterion.

A. auriculiformis was introduced into Thailand in1935 and is now planted widely; in India it hasbecome a major plantation species -since itsintroduction into West Bengal in 1946; and inChina since .1961 over 50 000 ha of plantationsand 3000 km of roadside plantings have beenestablished. It has been planted on a lesserscale elsewhere in Asia, Africa and SouthAmerica. The early plantings of A. auriculiformisdemonstrated its wide adaptability in the humidand subhumid tropical lowlands. It generallyformed a crooked, multistemmed and heavilybranched tree, which could be used for fuelwoodand small items of furniture, or for shelter andamenity purposes.

In 1982, the Food and Agriculture Organisationwas stimulated, by the recognition of A. mangiumin Malaysia as a fast-growing species capable ofproviding timber and pulpwood, to assess thepotential of other tropical acacias for woodproduction (Turnbull et al. 1983). In 1984, aplanning workshop of the International Union ofForestry Research Organizations (IUFRO)recognised the potential for increasing theproductivity of A. auriculiformis in Asia, andrecommended a programme of tree improvementand silvicultural research (IUFRO 1984). Therecommendation was implemented in aprogramme of co-ordinated research supportedby the United States Agency for InternationalDevelopment (USAID), through itsForestry/Fuelwood Research and DevelopmentProject (F/FRED) in Asia, and the ACIAR forestryresearch programme in Asia and East Africa.

Pulping studies showed that the wood ofA. auriculiformis was suitable for chemicalpulping and comparable with some of thecommercially accepted high-quality Eucalyptus(Logan 1987). The major drawback to itsutilisation for this purpose was its crooked stemform and multiple-leader habit. The naturalstands of this species occur primarily in sparselypopulated regions with poor road access, soexploration of the genetic resource was difficultand expensive. However, when it wasaccomplished, tall straight trees were found inPapua New Guinea and northern Australia, and itappeared likely that the introduction of thisgenetic material into a domestication programme

would extend the utilisation potential to polesand industrial wood.

Seed collected in a systematic way throughoutthe natural distribution of A. auriculiformis wascritical to the domestication process, andprovided the material for subsequent research.The collections were undertaken by CS1RO'sAustralian 'Free Seed Centre in co-operation withthe Papua New Guinea Department of Forests;between 1987 and 1990, about 3000 seedlotswere distributed to researchers (Gunn & Midgley1991). The field collections provided informationon the natural distribution and ecology of thespecies (Boland et al. 1990), and the seeds wereused for genetic studies, provenance trials,physiological screening, and the establishment ofseed orchards.

Provenance variation of the species wasconfirmed in glasshouse trials (Pinyopusarerk,Williams & Boland 1991) and through isozymestudies (Wickneswari & Norwati 1991), and thestriking degree of genetic diversity has beendemonstrated in field trials. In China, after twoyears, the fastest-growing provenance, which wasfrom Papua New Guinea, was over 5 m tall. Itwas 20% taller and 42% greater in stemdiameter than the local source of A. auriculiformis(Yang Minquan, Bai Jaiyu & Zeng Yutian 1989).The single Queensland provenance (SpringvaleHolding) included in this trial had the best stemform, with a high percentage of straightsingle-stemmed trees. Subsequent and morecomprehensive trials in China and Australia haveconfirmed the overall superiority in stem form ofthe Queensland provenances. Other internationaltrials have yet to be assessed and reported.

The best growth rates of A. auriculiformis areattained on deep, slightly acidic soils in thehigher rainfall areas of the humid tropics. Recenttrials have indicated that at least someprovenances have considerable potential forgrowing on degraded lands. In Zimbabwe, therehas been satisfactory growth on a site with lowrainfall (600 mrn yr-1) (Gwaze 1992). In addition,relatively high levels of alkalinity have beentolerated (McKinnell & Harisetijono 1991), andthere have been some promising results on siteswith a low-to-moderate salinity level whereseasonal waterlogging is a factor (Marcar et al.1991). Preliminary physiological studies showmajor differences between provenances in wateruse efficiency and this knowledge may providethe basis for selecting more drought-tolerantprovenances or clones (K C Woo, personalcommunication).

Information on the reproductive biology andvegetative propagation of A. auriculiformis is aprerequisite to tree improvement through seedorchards, hybridisation and the multiplication ofselected clones. Research in Australia, Malaysiaand Thailand has provided insights into flowering,

seed production, inbreeding, stem cuttings andtissue culture. The results are reported in ACIARworkshop proceedings (11.frnbull 1991; Carron &Aken 1992).

Adequate nutrition is essential for optimising theearly establishment and growth of trees,especially on degraded lands. It is particularlyimportant that the gains from tree breeding arenot squandered by inadequate culturaltechniques. Research on the role of symbioticmicro-organisms and the application of inorganicfertilizers in the nutrition of A. auriculiformis hasbeen conducted in Australia and the Philippines.Effective strains of R.12izobium have been isolatedwhich have given large growth responses in thefield (Dart et al. 1991), and mycorrhizas havebeen found to be beneficial in enhancingestablishment and growth (De la Cruz et al.1988).

This programme of research to domesticateA. auriculiformis stimulated the production of acomprehensive annotated bibliography on thespecies (Pinyopusarerk 1990), and results arebeing published as they become available. Theresearch will provide a firm basis on which toimprove the species' productivity and utilisation.Seed orchards have already been established inAustralia (Harwood et al. 1991) to provide seedsfor planting in Asia and elsewhere, until localseed production areas are operational.

CONCLUSIONS

Australian species have provided some of themost successful examples of domesticated trees.There is no doubt that many species with greatpotential for domestication around the worldawait discovery and development. Throughtraining and co-operative research programmes,the Australian experience can help realise thepotential of this 'bag of uncut cliamonds'. Theeconomic and social benefits of internationalco-ordinated approaches, as illustrated by theA. auriculiformis programme, to realise thepotential of fast-growing, adaptable and usefultree species in tropical developing countries arelikely to be very substantial. It is hoped that theinternational collaborative model for researchdeveloped between F/FRED, ACIAR, CSIRO andnational research agencies in Asia and Africa willbe applied to other potentially important treespecies.

ACKNOWLEDGEMENTS

We are grateful to our colleagues, Dr CHarwood and Dr J Doran, who made valuablecomments on a previous draft.

REFERENCES

Boland, D.J., ed. 1989. Trees for the tropics. (ACIARMonograph, no. 10.) Canberra: ACIAR.

193

Boland, D.J. & Turnbull, J.W. 1989. Australian tree speciesfor fuelwood and agroforestry in China, Kenya, Thailand andZimbabwe. In: Trees for the tropics, edited by D.J. Boland,13-20. (ACIAR Monograph, no. 10.) Canberra: ACIAR.

Boland, D.J., Pinyopusarerk, K., McDonald, M.W.,Jovanovic, T. & Booth, T.H. 1990. The habitat of Acaciaauriculiforrnis and probable factors associated with itsdistribution. Journal of Tropical Forest Science, 3, 159-180.

Booth, T.H. 1991. Where in the world? New climatic analysismethods to assist species and provenance selection for trials.Unasylva, 42, 51-57.

Campinhos, E., Jr. 1991. Plantation of fast-growing speciesfor tropical areas. Revue Forestiere Frangaise Hors Série, 5,111-120.

Carron, L.T. & Aken, K.M., eds. 1992. Breedingtechnologies for tropical acacias. (ACIAR Proceedings, no. 37.)Canberra: ACIAR.

Carter, E.J. 1987. From-seed to trial establishment. (DFRUser Series, no. 2.) Canberra: CSIRO Division of Forestry

Chin, H.F. 1990. Storage of recalcitrant seeds: past, presentand future. In: Tropical tree seed research, edited by J.WrIlirribull, 89-92. (ACIAR Proceedings, no. 28.) Canberra: ACIAR.

De la Cruz, R.E., Manalo, M.Q., Aggangan, N.S. &Taxnbalo, J.D. 1988. Growth of three legume trees inoculatedwith VA mycorrhizal fungi and rhizobium. Plant and Soil, 108,111-115.

Dart, P., Umali-Garcia, M. & Almendras, A. 1991. Roleof symbiotic associations in nutrition of tropical acacias. In:Advances in tropical Acacia research, edited by J.W TUrnbull,13-19. (ACIAR Proceedings, no. 35.) Canberra: ACIAR.

Doran, J.C. 1986. Seed, nursery practice and establishment.In: Multipurpose Australian trees and shrubs, edited by J.WTkirnbull, 59-80. (ACIAR Monograph, no. 1.) Canberra: ACIAR.

Gunn, B.V. & Midgley, S.J. 1991. Exploring and assessingthe genetic resources of four selected tropical acacias. In:Advances in tropical Acacia research, edited by J.W Turnbull,57-63. (ACIAR Proceedings, no. 35.) Canberra: ACIAR.

Gwaze, D. 1992. Species/provenance trials in Zimbabwe.ACIAR Forestry Newsletter, 13, 1-2.

Harwood, C.E., Matheson, A.C., Gororo, N. & Haines,M.W. 1991. Seed orchards of Acacia auriculiformis at MelvilleIsland, Northern Territory Australia. In: Advances in tropicalAcacia research, edited by J.W `Iltnbull, 87-91. (ACIARProceedings, no. 35.) Canberra: ACIAR.

International Union of Forestry Research Organizations.1984. Increasing productivity of multipurpose tree species: ablueprint for action. In: IUFRO Planning Workshop for Asia,edited by K. Shea & L. Carlson. Washington, DC: UnitedStates Department of Agriculture.

Libby, W.J. 1973. Domestication strategies for forest trees.Canadian Journal of Forest Research, 3, 265-276.

Logan, A.F. 1987. Australian acacias for pulpwood. In:Australian acacias in developing countries, edited by J.WTUrnbull, 89-94. (ACIAR Proceedings, no. 16.) Canberra:ACIAR.

Mc Kenney, D.W., Davis, J.S., Turnbull, J.W. & Searle,S.D. 1991. The impact of Australian tree species research inChina. (ACIAR Economic Assessment Series, no. 12.)Canberra: ACIAR.

194

McKinnell, F.H. & Harisetijono. 1991. Testing Acaciaspecies on alkaline soils in West Timor. In: Advances intropical Acacia research, edited by J.W Turnbull, 183-188.(ACIAR Proceedings, no. 35.) Canberra: ACIAR.

Marcar, N.E., Hussain, R.W:, Arunin, S. & Beetson, T.1991. Trials of Australian and other Acacia species onsalt-affected land in Pakistan, Thailand and Australia. In:Advances in tropical Acacia research, edited by J. W Turnbull,229-232. (ACIAR Proceedings no. 35.) Canberra: ACIAR.

Matheson, A.C. 1990a. Designing experiments for MPTgenotype evaluations. In: Tree improvement of multipurposespecies, edited by N. Glover & N. Adams, 55-66.(Multipurpose '11-ee Species Network Series, vol. 2.) Arlington,Virginia: Winrock International.

Matheson, A.C. 1990b. Breeding stategies for MPTs. In: Treeimprovement of multipurpose species, edited by N. Glover & N.Adams, 67-99. (Multipurpose Tree Species Network Series, vol.2.) Arlington, Virginia: Winrock International

Pinyopusarerk, K. 1990. Acacia auriculiformis: an annotatedbibliography Arlington, Virginia: Winrock International/Instituteof Agricultural Development/Australian Centre for InternationalAgricultural Research.

Pinyopusarerk, K., Williams, E.R. & Boland, D.J. 1991.Geographic variation in seedling morphology of Acaciaauriculiformis. Australian Journal of Botany 39, 247-60.

Raymond, C.A. 1987. Acacia mearnsii - a breeding plan forChina. Canberra: CSIRO Division of Forestry

Raymond, C.A. 1988. Eucalyptus globulus - a breeding planfor China. Canberra: CSIRO Division of Forestry

Reddell, P., Rosbrook, P.A. & Ryan, P.A. 1989. Managingnitrogen fixation in Casuarina species to increase productivityIn: Trees for the tropics, edited by D.J. Boland, 209-214.(ACIAR Monograph, no. 10.) Canberra: ACIAR

Ryan, P.A., Podberscek, M., Raddatz, C.G. & Taylor,G.W. 1987. Acacia species trials in southeast Queensland,Australia. In: Australian acacias in developing countries, editedby J.W Ttrnbull, 81-85. (ACIAR Proceedings, no. 16.)Canberra: ACIAR.

Searle, S.D. 1989. Seed collections of lesser-Imown trees andshrubs in Queensland, Australia. In: Trees for the tropics,edited by D.J. Boland, 2734. (ACIAR Monograph, no. 10.)Canberra: ACIAR.

Turnbull, J.W., ed. 1986. Multipurpose Australian trees andshrubs. (ACIAR Monograph, no. 1.) Canberra: ACIAR.

Turnbull, J.W., ed. 1991. Advances in tropical Acaciaresearch. (ACIAR Proceedings, no. 35.) Canberra: ACIAR.

Turnbull, J.W., Skelton, D.J., Subgyono, M. &Hardiyanto, E.B. 1983. Seed collection of tropical acacias inIndonesia, Papua New Guinea and Australia. Forest GeneticResources Information, 12, 2-15.

Wickneswari, R. & Norwati, M. 1991. Genetic structure ofnatural populations of Acacia auriculiformis in Australia andPapua New Guinea. In: Advances in tropical Acacia research,edited by J. W Turnbull, 94-95. (ACIAR Proceedings, no. 35.)Canberra: ACIAR.

Yang Minquan, Bai Jaiyu & Zeng Yutian. 1989. TropicalAcacia trials on Hainan Island, People's Republic of China. In:Trees for the tropics, edited by D.J. Boland, 89-96. (ACIARMonograph, no. 10.) Canberra: ACIAR.

Domestication of lesser-known species of thegenus Leucaena

J L Brewbaker & C T SorenssonDepartment of Horticulture, University of Hawaii, 3190 Mai le Way, Honolulu, Hawaii 96822, USA

ABSTRACT

The genus Leucaena (Leguminosae: Mimosoideae) is known worldwide by the cultivated and fully domesticated speciesL. leucocephala (Larn.) de Wit. It is a premier forage and fuelwood multipurpose (MPT) tree, but its acceptance is limited byfactors such as its intolerance of highly acid soils and susceptibility to psyllid insects. In the genus Leucaena Benth. there are 15other species, which offer solutions to these limitations. Most of these species are partly domesticated, and some have beenimportant ethnobotanically TWo of these lesser-known species, L. diversifolia and L. pallicla and their interspecific hybrids withL. leucocephala, have given superior yields in international yield trials and appear to deserve wide evaluation. The domesticationstatus and future prospects for these lesser-known species are reviewed.

Domestication of the lesser-known Leucaena species began in Latin America for fuelwood, food and forage, and has expandedgreatly in the past three decades through genetic improvement. Germplasm collection is well advanced, with all 16 speciesrepresented by 1875 seed collections in the three major germplasm banks (University of Hawaii, Australia's CommonwealthScientific and Industrial Research Organisation, and the UK's Oxford Forestry Institute). Most of these collections have beenevaluated during the past two decades in long-term trials, and all but one of the Leucaena species have been hybridisedexperimentally in Hawaii. Of the 232 hybrid combinations tested, over half (141) produced viable seed, leading us to the viewthat the entire genus can serve as a germplasm resource for future genetic improvement. Several species hold direct promise fordomestication as multipurpose trees for the tropics and subtropics. Eight activities that are seen to have a major influence ondomestication of leucaenas are demonstration, genetic improvement, germplasm protection, yield evaluation, quality evaluation,clonal propagation, communication and large-scale seed production. An emerging international network (LEUCNET) and thejournal Leucaena Research Reports (founded in 1980) enhance rapid exchange of knowledge and germplasm and will playsignificant roles in the domestication process in the future.

INTRODUCTION

The genus Leucaena (Leguminosae:Mirnosoideae) is a richly polymorphic complexof 16 species of leguminous trees and shrubsendemic from Texas in North America tonorthern Peru in South America (Brewbaker1987a, b; Brewbaker & Hutton 1979; Brewbaker& Sorensson 1990; Hughes 1989). The genus isof ethnobotanical importance as a human food inLatin America, as well as in parts of SE Asia,and the Mexican state of Oaxaca was namedafter it (after the Aztec word 'huaxin'). It becameknown internationally however, following theapparently casual transfer of one self-pollinatedvariety of the species L. leucocephala (Lam.) deWit subsp. leucocephala from Mexico to thePhilippines in the 16th century This 'common'variety (ipil-ipil, kubabul, koa haole, lamtoro)became one of the most versatile and usefulnitrogen-fixing trees in tropical agroforestry It iscurrently being replaced by even more vigorousarboreal types known as 'Salvador' or 'Hawaiiangiants' (L. leucocephala subsp. glabrata;Brewbaker 1987b).

Domestication of a plant species normally

involves adoption by growers for specific uses,and L. leucocephala is an example of a specieswhere this process is well advanced. Ultimatelythis process results in the careful collection,protection, characterisation, evaluation andincrease of germplasm of the species (Smith etal. 1992). The adoption and use of alesser-known species as a resource for geneticimprovement of trees such as L. leucocephala,however, lead to many of the same results.Important examples include species such as thewheat relative Triticum tauschii (Aegilopssquarrosa), the barley relative Hordeumspontaneum, and the sugarcane relativeSaccharum spontaneum. In the genus Leucaena,opportunities exist for the direct domestication ofa few species. However, as reviewed here, theopportunities for use of Leucaena species areeven greater through hybridisation and genetictransformation. These opportunities for both directand indirect use have prompted the collection,evaluation and hybridisation of all lesser-knownspecies in the genus. This paper considers thestatus of domestication of these species and theprocesses that may play significant roles in theirfurther domestication.

195

COLLECTION AND EVALUATION OFLEUCAENA GERMPLASM

There are three major world collections ofLeucaena germplasm, held respectively by theUniversity of Hawaii (UH), the Oxford ForestryInstitute (OH) and the Commonwealth Scientificand Industrial Research Organisation of Australia(CSIRO). Together, these include about 1875accessions, including duplicates, and at least halfhave been increased by seed production. Table 1gives details for the 16 Leucaena spp.Domestication status (Dom St) is indexed on a 1-5scale (1 high, 5 low), and information issummarised on botany forage biomass and othertraits. Botanical descriptions of all taxa are given inAppendix 1. Passport data on these accessions areavailable from the respective institutions.

All 967 accessions in the UH collection (Table 1)have been grown and evaluated. They represent15 species, and slightly over half are of L.leucocephala. Most of these accessions werecollected during botanical expeditions into LatinAmerica by the senior author and 'colleagues.Major expeditions occurred in 1967, in 1978(with A J Oakes, Germplasm Resources Center,USDA), in 1985 (under support of theInternational Bureau of Plant GermplasmResources (IBPGR) with S Zarate, R Van DenBeldt, C Sorensson and N Glover), and in 1988(with C Hughes and C Sorensson). The UHcollection also includes trees obtained fromCSIRO (42 lines) and OH (23). Cultivar K8 wasreleased in 1968 and an accession collected in1978; L. leucocephala K636 (PI 443740) is beingsubmitted for release in 1993.

CSIRO's collection of 815 accessions (Table 1)

Table 1 Domestication status and descriptive data for 16 Leucaena species

196

includes many from R Reid's botanical expeditionsbetween 1979 and 1981 (adding 183 newaccessions, mostly of lesser-known species) and in1985. TWo superior lines from these collections areL. diversifolia CPI46568 (a Guatemalan diploid, =K749) and L. pallida CPI 84581 (= K748). Mostindividual accessions in the UR and CSIROcollections are composed of seeds bulked fromseveral trees, although about 25% of the L.diversifolia (tetraploid) and L. leucocephalaaccessions were from individual elite trees (bothtaxa are self-pollinating). The CSIRO collection alsoincludes trees obtained from UH (23 lines) andother institutions.

OH has conducted Latin American expeditions forwoody legumes since the mid-1980s, largely underC Hughes (Hughes 1991a). The OH collections areprovenances based on seed composited from treesharvested separately at each site. The 93provenances (Table 1) represent 991 half-sibfamilies, collected in adequate quantity forextensive field evaluations and selection. Initialprovenance evaluations (Stewart, Dunsdon &Hughes 1991) identified two taxa that haveattracted interest in Central America, L.salvadorensis and L. shannonii subsp. magnifica.

The first Leucaena evaluation in Hawaii wasplanted in 1963 at the Waimanalo ResearchStation (sea level, 21.2°N 157.5°W) and the 86accessions (numbered K 1—K86) included fourspecies (Gonzalez, Brewbaker & Hamill 1967).Subsequent Waimanalo trials were planted in1966 (K87—K112), 1971 (K113—K338), 1978(K339—K600), 1979 (K 1—K316), 1980 (K317—K678),1983 (K682—K695), 1985 (K742—K893) and 1990(K903—K996). Seeds were freely distributed by

tRated on a 1-5 scale (1 = fully domesticated)2UH, University of Hawaii; CSIRO, Commonwealth Scientific and Industrial Research Organisation; OFI, Oxford Forestry Fomstry Institute3Numbers in parentheses are of half-sib families

U1-I and the Nitrogen Fixing Ree Association, andseveral species were entered into internationalLeucaena psyllid trials (Glover 1988). Most ofthese trials were designed with small plots of14-20 trees planted at 1 m x 1.5 m spacing,and maintained for at least six years. About halfof tlie 15 species were also tested after 1980 athigh-elevation sites on Haleakala, Maui (850 m)and Mealani, Hawaii (900 m). Prior to 1986, 12species had been field-tested extensively inHawaii, leading us to identify them as validspecies; most of the 40 other published epithetsin the genus were identified as synonyms(Brewbaker 1987a). Subsequent field expeditionshave led to recognition of four additional taxa,namely L. cuspidata, L. multicapitula, L.salvadorensis and the as-yet unpublishedpolyploid L. sp. 'glossy' (see Appendix 1).

The CSIRO collections were initiated in the1950s, but the entire collection was first plantedout in 1978 at the Lansdown Research Stationnear Townsville, Queensland (sea level, 19.2°N,146.5°W), and characterised by Bray (1984).Subsequent trials were established at five sites inQueensland (Bray et al. 1988) and weremanaged for the identification of superior foragetypes. Initial field evaluations of many OHprovenances were made in collaboration with twoHonduran forestry agencies at Comayagiia,Honduras (500 m, 14.3°N, 87.4°W) (Hawldns &Ochoa 1991; Stewart et al. 1991), and L.salvadorensis has recently been grown for seedproduction and evaluation in several CentralAmerican countries.

Many germplasm collections held internationallyare dominated by the cultivar 'common' ofL. leucocephala (subsp. leucocephala) thatappears to lack any significant genetic variability(W Sun, University of Hawaii, unpublished).Collections that include significant resources oflesser-known Leucaena spp. include those ofCentro Internacional de Agricultura 'Tropical inColombia, the International Center for Researchin Agroforestry in Kenya, the International Institutefor Tropical Agriculture in Nigeria, theInternational Livestock Center for Africa inEthiopia, the Taiwan Forest Research Institute, andthe Universidad Nuevo Leon in NE Mexico.Additional lines are maintained at a stateresearch centre in the Yucatan, Mexico, theNational University of Mexico in Mexico City theEscuela Agricultura Panamericana in Honduras,the Sta. Ana Research Station in El Salvador andthe University of the Philippines at Los Bafios. Allof these field trials have been observed by thesenior author.

DOMESTICATION STATUS OF LEUCAENA

SPECIES

The 16 Leucaena species are classified in fivegroups in Table 1 on the basis of their current

domestication status. Full domestication (Class 1)characterises only L. leucocephala, which iswidely planted and utilised throughout thetropics. Species placed in Class 2 include L.esculenta and L. pallida that serve as humanfood in Mexico (Zarate 1982), and L. diversifoliathat is now planted in many places in the tropicsand subtropics. Most species are viewed asintermediate in domestication (Table 1), whilefour species are considered undomesticated(Class 5). All leucaenas appear to provideattractive browse forage and fuelwood, and theyare normally known by name to indigenouspeoples. All are nitrogen-fixing (Bradyrhizobiumspp.), mycorrhizal (vesicular-arbuscular), and ofpotential importance as companion trees in thereforestation and stabilisation of degraded landsthroughout the tropics.

The current status of domestication in Leucaenaspp. (Table 1) does not necessarily reflect ourperception of the potential for full domestication,as can be seen for individual species inAppendix 1. The most promising are L.diversifolia and L. pallida. The tetraploid form ofL. diversifolia is a self-pollinator that grows into acolumnar tree at cooler sites where soils can bemore acidic than those preferred by L.leucocephala (Hutton 1990; Bray & Sorensson1992). It is being planted in the highlands of EAfrica and S Asia. The self-sterile tetraploid, L.pallida, and its hybrids have shown superiorforage performance when under heavy psyllidpressure. Most diploid species are rated lesshighly for domestication potential.

INTERSPECIFIC HYBRIDISATION

Interspecific hybridisation of leucaenas in regionsof endemism has been precluded by theecological and geographical separation thatcharacterises most species, and possiblyreinforced by floral preferences of pollinatinginsects and differences in the timing and lengthsof flowering seasons (Sorensson 1993). Hybridswarms are unknown in situ and the few hybridsfound appear to reflect agricultural movement ofL. leucocephala, the suspected pollen parent ofmost natural hybrids. Thus, domestication oflesser-known species cannot depend on naturallyoccurring hybrids.

Interspecific hybridisation among Leucaenaspecies became a major thrust of research inHawaii soon after it was started by Gonzalez etal. (1967), and continued under Pan andBrewbaker (1988); Sorensson (1993) intercrossed15 species in practically all combinations (232 of240 possible matings of a 16 x 16 matrix). Over75% of these crosses have been partially or fullyseed-fertile, and 99 hybrids have been grownfrom the 141 combinations producing viableseed. Several hybrids are promising for directproduction, but many more appear valuable in

197

breeding (Brewbaker & Sorensson 1990). Theseresults lead us to the view that the entire genusprovides a genetic base for future improvementresearch, and suggest that all species warrantfurther collection, protection and evaluation.

Most Leucaena breeding is based at present ontetraploid species (Table 1) and theirinterspecific hybrids, often with impressive growthand adaptability (Brewbaker & Sorensson 1990).Leucaena pallida has been the major source forpsyllid resistance and low seediness (owing to itsself-sterility), and tetraploid L. diversifolia for coldtolerance. Triploid hybrids between tetraploid anddiploid taxa have been of particular interest, andwere studied by Hutton (1981, 1990) as agenetic source of acid soil tolerance for L.leucocephala. Many triploids (Sorensson 1993)were fully seedless, a trait of great interest inimproving acceptability for reforestation andwood yields (Brewbaker & Sorensson 1990).Many hybrids are interesting for geneintrogression into L. leucocephala, a process thathas been accelerated by the use of unreducedgametes from diploid parents (Sorensson 1988).Colchicine-induced polyploids also hold promisefor the conversion of superior diploids to thetetraploid level, facilitating gene transfer and thesynthesis of new amphidiploids.

GENETIC EXPLOITATION OFLESSER-KNOWN SPECIES

Leucaena is a richly polymorphic genus thatdisplays extensive variability for nearly allimportant traits under study (Brewbaker 1987b).However, much of this genetic polymorphismresides in lesser-known species. Full exploitationof the genus is thus dependent on the successfulmanagement of these taxa. Important traits oflesser-known taxa include the following.

• Psyllid tolerance. Well over 50% of theLeucaena species have tolerance orresistance to psyllids (Homoptera:Heteropsylla cubana) (Table 1) (Sorensson &Brewbaker 1987). Leucaena pallida has beenused to incorporate resistance intoL. leucocephala.

• Cold tolerance. Most species are of highlandor subtropical origin (Table 1) and exhibitbetter tolerance of cool weather thanlowland-adapted L. leucocephala. Species thathave been used for this purpose includeL. diversifolia (Bray & Sorensson 1992),L. pallida, and L. pulverulenta (Gonzalez etal. 1967). Other species of promise includefrost-tolerant taxa (next section), the tetraploidL. sp. 'glossy' and three diploids: L.cuspidata, L. esculenta and L. greggil.

• Frost tolerance. Leucaena retusa is well knownfor frost resistance and L. pulverulenta forpartial resistance. Ongoing trials in New

198

Mexico and Texas, USA, have shown that L.greggli is also frost-tolerant, as L. cuspidatamay be.

• Acid soil tolerance. A highly sought trait inleucaenas internationally is tolerance of soilacidity (Hutton 1981, 1990; Oakes & Foy1984). Diploid L. diversifolia provided usefultolerance genes in Hutton's programme, andother species that looked promising to himincluded L. esculenta, L. pallida and L.shannonii.

• Mimosine content and forage quality Abouthalf of the species contain less of the toxicamino-acid, mimosine, than L. leucocephala,some with about 75% less. Species known tobe very low in mimosine are L. diversifolia,L. esculenta and L. pulverulenta (Bray 1983);others that are quite low include L. collinsiiand L. pallida (Brewbaker 1987b). Mimosineis fully degraded by a bacterium Synergistesjonesii in ruminants, and therefore isconsidered a problem only for non-ruminants,as are the dihydroxypyridine breakdownproducts. Nutritional quality as measured byintake and digestibility is generally high butvaries widely among the species (Austin etal. 1991).

• Wood density and lumber quality Leucaenashave fairly dense heartwood, with specificgravities ranging from 0.45 to 0.85(Brewbaker 1987b). The leadtrees' L.pulverulenta and L. retusa have high wooddensity (the former was widely used inrailway construction). Leucaena salvadorensis('sepia') is valued in Central America aspoles for its high density and dark-colouredheartwood (Hughes 1988), as are some L.collinsii populations.

• Drought tolerance. Most species toleratedrought very well, although none has beenshown conclusively to tolerate drought betterthan L. leucocephala. Following one of theworst extended droughts on record, theCHRO germplasm collection at the LansdownStation in N Queensland was observed bythe authors, and some lines of L. lanceolatalooked superior in drought tolerance. Thisspecies colonises some very arid Mexicancoastal soils. Other species occurring in aridregions include L. collinsii, L. greggii, L.macrophylla, L. retusa, L. salvadorensis and L.trichodes.

• Genetic diversity of tree form. The genusLeucaena is distinguished by unique,heritable intraspecific diversity in tree form.Many species range in habit from forestideotypes (tall, columnar, clear bole) toshrubs or spreading small trees, and all arehighly tolerant of coppicing, providinggrowers with great flexibility in sustainable

agricultural systems (Brewbaker & Shelton1993). The variation is well known inL. leucocephala, as it distinguishes 'common'cultivars from 'giant' or 'Salvador' types.Similar variation occurs in L. diversifolia,L. lanceolata and L. pallida. Leucaenatrichodes (small tree) and L. multicapitula(arboreal) are related taxa that also fit thisscenario.

• Waterlogging tolerance. Species such asL. diversifolia and L. multicapitula occur inhigh-rainfall areas and need more thoroughevaluation for waterlogging tolerance. Thediploid L. diversifolia, such as thosenaturalised in Indonesia, should also betested.

DOMESTICATION PROCESSES

Eight activities that appear especially importantin domesticating the lesser-known species ofLeucaena and similar multipurpose trees aredescribed below

i. Demonstration. Demonstration is a sine quanon to promoting interest in lesser-known treespecies. After three decades of research onLeucaena, Brewbaker (1987b) concluded thatmost people have no real understanding oftheir growth capability Under most favourableconditions, replicated yields achieved byleucaenas include 100 tonnes of total freshbiomass ha-1 yr-1 and 50 tonnes ediblebiomass ha-1 yr 1. These extraordinaryvalues are not easy to grasp, and theaverage grower must be shown to believe.Such demonstrations are often under-financedand mismanaged, and are most effectivewhen treated with care similar to that givento a high-value fruit tree.

Genetic improvement. Superior butunselected germplasm will rarely meetgrowers' demands in the 21st centurySuperior Fi species hybrids are known thatgreatly outyield L. leucocephala var.Cunningham and K8, two great cultivars ofthe 1970s, when grown for forage underpsyllid pressure. Major advances areoccurring with simple recurrent selection inbroad-based germplasm. Although breedingis often a slow, process, it must beemphasised that the generation time of mostleucaenas is about 18 months, and largepopulations can be screened at the densitiescommon for forage production (75 000-150000 per hectare). Without geneticimprovement it is unlikely that any of theself-sterile species can be advanced indomestication status.

Germplasm protection. Almost all naturalpopulations of leucaenas in Latin America areunder continuing pressures from man and

grazing animals. Some in situ germplasmprotection occurs for this genus in nationalparks and preserves, notably in Costa Rica,Honduras and Panama, but more is needed.The option of developing ex situ germplasmconservation deserves consideration.

iv. Yield evaluation. Most research trials haveinvolved only L. leucocephala, and yieldevaluations of other species and provenancesin diverse agroecosystems are thus of highpriority Acid soil tolerance provides a usefulexample of the need for thoroughcharacterisation of lesser-known species insoils that vary in pH and in concentrationsand forms of aluminium, manganese, calciumand phosphorus. Progress in pest resistancebreeding is only likely to occur whenevaluations are ,made under heavy anduniform infestations. Much can be learnedeven from two-year trials as small as 0.05 ha(adequate for a fully replicated design with20 entries, three replications, and rows 5 mlong). Seedling selection may be effective forsome traits (Sorensson 1993).

v Quality evaluation. Domestication leads toplants tailored to specific uses andmanagement; this will not occur in Leucaenawithout careful evaluations of quality Qualitycomponents needing study are those offorage (eg intake, digestibility tannins, lignins,saponins, mimosine, protein) and of wood (egdensity fibre length, wood colour, ease ofpulping). Quality evaluation and breedingresearch must be closely linked, to ensurethat yield improvement does not occur at theexpense of quality

vi. Vegetative propagation of specific genotypes.It is routine to fmd superior trees amongbreeding populations of outcrossed Leucaenaspp. Rapid genetic gains comparable tothose in Pinus and Eucalyptus can bepredicted using quick and reliable clonalpropagation methods. With such methods,cloning of seedless triploid or aneuploidhybrids could emerge as a predominantforce in domestication in the genus(Brewbaker & Sorensson 1990). Propagationin vitro holds promise, but explants havebeen difficult to sterilise and shoots grownfrom callus often fail to root or to survivetransplanting. Some grafting techniques areused regularly for propagating seedlessaneuploids in Indonesian coffee orchards, butit is probable that grafting will not meetcommercial-scale propagation needs. Studiesin Taiwan, India and Hawaii indicate thatcertain Leucaena genotypes can be rootedquickly and reliably from semi-woodycuttings, but genetic selection for this traitmay be essential.

199

vii. Communication. The distances involved amongLeucaena workers, many of them in developingcountries, often prevent personal exchange ofinformation, Effective written communication isessential. One journal has been devoted since1980 to the genus, namely Leucaena ResearchReports (LRR). The 13 volumes (1440 pp) havebeen sent to about 1500 associates of theNitrogen Fixing 11-ee Association (NFTA: Paia,Maui, Hawaii). The LRR has been a favouredoutlet for reports on efforts to domesticatelesser-known taxa, although many importanttrials are unreported to the internationalcommunity

viii. Large-scale seed production. Seed increase ofthe self-pollinating polyploid leucaenaspresents no major obstacles. However, seedproduction from self-sterile parents ofseedless hybrids and synthetics requiresfurther reproductive and developmentalstudies, especially where parents must becloned and flowering dates differ. Onset offlowering may be controllable by managingirrigation water, coppicing, or growthhormones. The production of seeds fromdifferent hybrids in the same orchard ispossible, reducing costs to seedsmen(Sorensson 1993).

CONCLUSIONS

Domestication of leucaenas is viewed here as aprocess that must involve demonstration, geneticimprovement, germplasm protection, yieldevaluation, quality evaluation, clonal propagation,communication and seed production of releasedcultivars. An international Leucaena research anddevelopment network (LEUCNET) has beenproposed by the Universities of Hawaii andQueensland, with an initial planning workshop tobe held in 1994. The network would seek toco-ordinate diverse teams of internationalscientists and streamline the activities essential todomestication.

The genus Leucaena exemplifies the diversedomestication processes affecting tropical trees.Prior to the 1500s, domestication was based onfood use by indigenous Americans, notably of L.leucocephala (Gomez-Pompa 1987) and thelarge-seeded L. esculenta (Zarate 1982). Duringthe 16th century a single genetically uniform lineof L. leucocephala (cv Common) was brought toSE Asia and dispersed worldwide. Domesticationof giant forms of L. leucocephala in the late1900s greatly expanded the acreage underLeucaena, and spurred interest in agroforestrysystems such as alley farming. The internationalmovement of the psyllid since 1984 has broughtspecial attention to other species of the genus.Interest in domesticating lesser-known species ofLeucaena started with Dutch efforts in the 1920sto use L. diversifolia and L. pulverulenta as shade

200

in highland Indonesian plantations. Progress indomestication has expanded with the collection,characterisation, and yield evaluation of manyspecies and interspecific hybrids. All studiesconfirm the superiority of the genus as a woodylegume for forage and green manure use andfully justify future research to domesticate theselesser-known species (Brewbaker 1987b).

ACKNOWLEDGEMENTS

We thank a number of critics for generouslysuggesting alterations to the manuscript, notablyC Hughes and M Shelton.

REFERENCES

Austin, M.T., Williams, M.J., Hammond, A.C. &Chaxnbliss, C.G. 1991. Cattle preference ratings for eight

Leucaena spp. in Florida. Leucaena Research Reports, 12,109-110.

Bray, R.A. 1983. Sources of variation for mimosine content.

In: Proceedings of Australian Plant Breeding Conference,

Adelaide, Australia. 278-279. Adelaide: Standing Committee onAgriculture.

Bray, R.A. 1984. Breeding Leucaena. In: Forage legumes for

energy-efficient animal production, edited by R. Barnes,317-322. Palmerston North: US Department of Agriculture/ARS.

Bray, R.A., Cooksley, D.G., Hall, T.J. & Ratcliff, D. 1988.

Performance of fourteen Leucaena lines at five sites inQueensland. Australian Journal of Experimental Agriculture, 28,69-76.

Bray, R.A. & Sorensson, C.T. 1992. Leucaena diversifolia -

fast-growing highland NFT species. (NFT Highlights, no. 92-05.)Paia: Nitrogen Fixing 'nee Association.

Brewbaker, J.L. 1987a. Guide to the systematics of thegenus Leucaena (Mimosoideae: Leguminosae). Leucaena

Research Reports, 7, 6-20.

Brewbaker, J.L. 1987b. Leucaena: a multipurpose tree genus

for tropical agroforestry In: Agroforestry: a decade of

development, edited by HA. Steppler & PK.R. Nair, 289-323.Nairobi: International Centre for Research in Agroforestry

Brewbaker, J.L. & Hutton, E.M. 1979. Leucaena -versatile

tropical tree legume. In: New agricultural crops, edited byG. A. Ritchie, 207-259. Washington, DC: American Assodiation

for the Advancement of Science.

Brewbaker, J.L. & Shelton, H.M. 1993. Leucaena forage

and wood production and use. Paia: Nitrogen Fixing neeAssociation.

Brewbaker, J.L. & Sorensson, C.T. 1989. Leucaena

leucocephala (Lam.) de Wit. In: Plant resources of south-east

Asia, 172-175. Wageningen: Pudoc/PROSEA.

Brewbaker, J.L. & Sorensson,*C.T. 1990. New tree crops

from interspecific Leucaena hybrids. In: Advances in new

crops, edited by J. janick, 283-289. Portland, Oregon: TimberPress.

Glover, N. 1988. Evaluation of Leucaena species for psyllid

resistance. LeuCaena Research Reports, 9, 5-18.

Gomez-Pornpa, A. 1987. On Mayan silviculture. Mexican

Studies, 3, 1-17.

Gonzalez, V., Brewbaker, J.L. & Hamill, D.E. 1967.Leucaena cytogenetics in relation to the breeding of lowmimosine lines. Crop Science, 7, 140-143.

Hawkins, T.H. & Ochoa, 0. 1991. The effects of seedpretreatment on the germination of seventeen Leucaena taxa.Leucaena Research Reports, 12, 19-22.

Hughes, C.E. 1988. Leucaena salvadorensis: status,

distribution, and genetic conservation. Leucaena Research

Reports, 9, 99-105.

Hughes, C.E. 1989. New opportunities in Leucaena genetic

development. In: Breeding tropical trees, edited by G.L.

Gibson, AR Griffin & A.G. Matheson, 218-226. Oxford:Oxford Forestry Institute/Arlington, Virginia: WinrockInternational.

Hughes, C.E. 1991a. Leucaena improvement thzough use of

lesser known species. Leucaena Research Reports, 12, 98-100.

Hughes, C.E. 1991b. 'Ago new subspecies of Leucaena

(Leguminosae - Mimosoideae) from Guatemala. Kew Bulletin,46, 547-557.

Hutton, E.M. 1981. Natural crossing and acid tolerance in

some Leucaena species. Leucaena Research Reports, 2, 2-4.

Hutton, E.M. 1990. Field selection of acid-soil tolerant

Leucaena from L. leucocephala x L. diversifolia crosses in atropical oxisol. Tropical Agriculture, 67, 2-8.

Oakes, A.J. & Foy, C.D. 1984. Acid soil tolerance of

Leucaena species in greenhouse trials. Journal of Plant

Nutrition, 7, 1759-1774.

Pan, F.J. & Brewbaker, J.L. 1988. Cytological studies in the

genus Leucaena Benth. Cytologia, 53, 393-399.

Robustiano, A. & Sorensson, C.T. 1987. Leucaena

multicapitula: a true giant from Panama, and its soil

environment. Leucaena Research Reports, 8, 54-55.

Smith, N.J.H., Williams, J.T., Plucknett, D.L. & Talbot,J.P. 1992. Tropical forests and their crops. Ithaca, New York:

Comstock Publishing Association.

Sorensson, C.T. 1988. Utilizing unreduced gametes for

production of novel hybrids of Leucaena species. Leucaena

Research Reports, 9, 75-76.

Sorensson, C.T. 1989. Status and mechanism of

self-incompatibility in Leucaena species. Plant Cell

Incompatibility Newsletter 21, 77-85. Cortland, New York:

University of Cortland.

Sorensson, C.T. 1993. Production and characterization of

interspecific hybrids of the tropical tree Leucaena

(Leguminosae: Mimosoideae). Doctoral thesis, University of

Sorensson, C.T. & Brewbaker, J.L. 1987. Psyllid resistance

of Leucaena species and hybrids. Leucaena Research Reports,

7, 29-31.

Stewart, J.L., Dunsdon, A.J. & Hughes, C.E. 1991. Early

results from a Leucaena species provenance trial in Honduras.

Leucaena Research Reports, 12, 101-104.

Zarate, S.P. 1982. Las especias de Leucaena Bentham de

Oaxaca con notas sobre la sistematica del genero para Mexico.

BA thesis, Universidad Nacional de Mexico.

Zarate, S.P. 1984. Taxonomic revision of the genus Leucaena

Benth. from Mexico. Bulletin of the International Group for the

Study of Mimosoideae, 12, 24-34.

APPENDIX

Botanical description of the 16 Leucaenaspecies

The 16 recognised Leucaena species arediscussed below in alphabetical order with abrief description of their botany distribution, usesand potential for domestication (see also Table 1).

1 Leucaena collinsii Britton and Rose(COLL), in North American Flora, 23 (2), 126.

1928. Type: Collins & Doyle 157, 11ixtlaGutierrez, Chiapas, Mexico. 1907. Fieldvariants include a southern population inGuatemala, subsp. zacapana, and a northernpopulation with larger trees and flowers (to13 m) in Chiapas, Mexico, subsp. collinsii(Hughes 1991b). This species is a hardyspreading tree of dry hilly midlands with adistinctive shape like Ulmus spp. Leaves arelarge, to 20 cm, with 8-15 pairs of pinnaeand small (3-7 mm) leaflets, flowering headsare small, to 18 mm, and white, pods 16-20mm x 12-18 cm, seeds 6-8 mm. It isself-sterile but produces heavy pod setunder most conditions, often with seedlesspods that ripen very slowly, about 300 daysin Waimanalo. It is often found dominatingthe calcareous soils at low to mid-elevations(eg 300-800 m in the Grijalva and MotaguaRiver basins). L. collinsii is of some localimportance for shade and is lopped for fuel.The wood is dense and used for smallconstruction, but it is not widely fed toanimals. Its high psyllid and droughttolerance make this species valuable forgene transfer, while domestication per semay offer little.

2 Leucaena cuspidata Standley (CUSP), inContributions from the United States NationalHerbarium, 20, 189, 1919. Type: USNI-1Purpus 5183, San Luis Potosi, Mexico. Little isknown of this apparently rare andendangered species. Morphologically itresembles L. greggii, a small multistemmedshrub (to 5 m height), native tohigh-elevation areas that may receive frost,yellow coloration in the inflorescence, longpeduncles (to 3.5 cm), strikingly long stipules(to 4 mm), long and narrow woody pods,squarish seeds and small coriaceous leaflets.Its inflorescence is smaller and lighter yellowthan that of L. greggii, however, and itsleaflets are cuspidately tipped, hence itsname. Like L. greggii, it occurs in NuevoLeon, where the two species are separatedgeographically by the large San Luis Potosivalley which lies west of the Sierra MadreOriental. L. cuspidata also extends south intoHidalgo, also at high elevations (about 2000m), and these populations may bemorphologically distinct (Zarate 1982, 1984).It is summer-flowering and autumn-seeding. Itappears to have no indigenous use.

3 Leucaena diversifolia (Schlecht.) Benth.(DIV2, DIV4), in Journal of Botany, 4, 417,1842. Type: Shiede, Jalapa, Veracruz, Mexico.This species is widespread, and occurs astwo distinct subspecies diversifolia (2n=104)and trichandra (2n=2) (Bray & Sorensson1992). The type specimen was of subspeciesdiversifolia. It occurs as highly variable

201

shrubs or trees, 5-18 m in height, with adistinctive dark bark. Leaflets are very small(3-6 mm) on leaves (8-18 cm) with 15-30pairs of pinnae. The flowers areconspicuously reddish, small (8-12 mm) indiploid L. diversifolia, larger (to 15 mm) intetraploid L. diversifolia. Pods are reddish,fairly small (1-1.5 cm x 12-18 cm), withsmall dark seeds (to 5 mm). The diploidsubspecies is self-incompatible andmorphologically diverse, while the tetraploidsubspecies is self-compatible, with limitedoutcrossing. Several published taxa areconsidered synonyms (Brewbaker 1987a).The diploids range from Oaxaca in Mexicosouth to Nicaragua between 800 m and 2000m, while the tetraploid is largely found inVeracruz, Mexico. The diploids appear lessrestricted to calcareous sites, and have beenused by Hutton (1981, 1990) in breedingacid tolerance. Leucaena diversifolia trees arelopped for fuelwood and occasionally grownor protected for shade (coffee plantations)and in fencerows. Yield trials under coolertemperatures in Queensland, Hawaii andNepal show this species and its hybrids tobe much superior to L. leucocephala (Bray &Sorensson 1992).

4 Leucaena esculenta (Moc. and Sesse)Bentham (ESCU), in Transactions of theLinnaean Society of London, 30, 442, 1875, asAcacia esculenta Moc. and Sesse, in DeCandolle, 2, 470, 1825; no type specimenexisted of this taxon, but type locality wasgiven as Mexico'. Subspecific variationincludes a rounded-branch variant at lowerelevations in W Mexico (subsp. matudae). Nosynonyms occur, although L. pallida has beenincorrectly referred to. it. Leucaena esculentais an attractive tree that is easilydistinguished from all other species. It islarge, to 20 m in height, with basal diameterto 60 cm and a distinctively thick, corky,whitish bark marked by horizontal scars onolder trees. Branches are sharply angulate.Leaves are very large (to 40 cm) with largenumbers (to >12 000) of tiny leaflets (3-4mm), on 30-50 pairs of pinnae. Flowers arelarge and white, while pods are reddish andamong the largest of the genus (to 2 cm x25 cm) and seeds are large and squarish(7-9 mm). flowering occurs in the autumn.This species is distributed widely in themidlands of Mexico. The large pods aremarketed at physiological maturity andnormally eaten green, and are ofethnobotanical importance (Zarate 1982),although the mimosine and dihydroxypyridinecontent will preclude frequent cultivation forthis use. Trees are occasionally planted onfarms or used in agroforestry cropping

202

systems with interplanted crops.Establishment is rapid and yields are high inforage trials, but IVDMD digestibility valuesfor this species are the lowest amongleucaenas (Austin et al. 1991). The species isvirtually immune to the leucaena psyllid(Table 1). Gum exudation from some hybridshas been of interest because the gumresembles gum arabic (Brewbaker &Sorensson 1990). It is a most interestingspecies as a genetic resource.

5 Leucaena sp. 'glossy' (GLOS) (syns. L.cuspidata subsp. compactiflora, 'confertiflora).This small tree, common to San PedroChapulco and other mountain villages inPuebla and Oaxaca, Mexico, was treated asL. cuspidata subsp. compactiflora var.adenostricta by Zarate (1984), but should betreated under a new epithet. It was shownby Sorensson (1989) to be a self-compatiblepolyploid with 112 somatic chromosomes(confirmed by R A Bray unpublished) andoccurs south of the volcanic axis, whereas L.cuspidata occurs north of it. Theinflorescence is reddish, small (about 1.5 cmdiameter) and open, containing only about75 florets (comparable to L. pulverulenta).Leaves are glossy hence the temporaryname L. sp. 'glossy'. Leaflets are small androunded, and the leaf gland is subcylindrical.Zarate (1984) noted that it was occasionallycultivated and that its pods were sold inregional markets by villagers.

6 Leucaena greggii S. Watson (GREG), inProceedings of the American Academy of Artsand Sciences, New Series XV, Whole SeriesXXIII (May 1887—May 1888). Series XVIII.Contributions to American Botany 249-272.Type USNH 2636386. A small tree of themidlands of NE Mexico extending fromLinares, Nuevo Leon, into the hills westwardand northwards to Coahuila. Leaflets arefinely divided, about the size of L.leucocephala, but waxy and coriaceous.Leaves have conspicuous subcylindricalglands at the base of each pinna, as in L.retusa. Flower heads are golden in colour, onlong peduncles to 6 cm long. Pods arenarrow long and woody and seeds aresquarish, as in L. retusa. It withstandsoccasional frosts and extended dry seasons.Roadside trees are regularly lopped forcut-and-carry fodder for goats, andprovenance evaluations for use inreforestation are ongoing in NW Mexico.

7 Leucaena lanceolata S. Watson (LANC), inProceedings of the American Academy of Artsand Sciences, 21, 427, 1886. Type: E. Palmer6, Batopilas, Chihuahua, Mexico, 1885.Evident subspecific variation includes awidespread shrubby phenotype and two

populations of arboreal variants in Oaxaca,one that has been treated as subsp. sousae(eg K952). This species has the widestdistribution of any species in the genus, fromSonora to Chiapas, Mexico, and istaxonomically complex with a number ofsynonyms. It commonly occurs as a shrub indry often calcareous lowlands, normallygrowing to 7 m, although the arboreal formreaches 16 m. Leaves range from 10 to 25cm, with only 3-6 pairs of pinnae and 6-7pairs of large (2-4 cm) leaflets. Flower headsare large (to 25 mm), white and heavilyscented. Pods are 16-25 mm wide and15-30 cm long, with seeds like those of L.leucocephala (a probable amphiploidderivative of L. lanceolata). It is used forforage and fuelwood but seems to beplanted. Some accessions show unusualdrought tolerance in Australia, and highgrowth rates. This species could be of greatinterest for drought-prone areas.

8 Leucaena leucocephala (Lam.) de Wit(LEUC), a taxon fully described in theliterature (Brewbaker & Sorensson 1989;Smith et al. 1992).

9 Leucaena macrophylla Benth. (MACR), inBotany of the Voyage of HMS Sulphur. .90, 1844. Type: Hinds 1841, Acapulco,Mexico. Hughes (1991a) identified twosubspecies, macrophylla and nelsonii. This isa small tree (/1 8 m) of dry midlands in WMexico and has very large leaflets (30-70mm) in leaves ranging to 25 cm, with only2-3 pairs of pinnae. The white flower headsare small (9-14 mm) and dense, and podsare 1.5 cm x 15-20 cm, with seeds of 7-8mm. Flowering is late and synchronous withL. esculenta. Leucaena macrophylla occursfrom coast to dry midlands and from Oaxacanorth to Sinaloa in W Mexico. Pods areseasonally marketed as 'guaje verde' in amanner similar to that of L. esculenta, andtrees are protected but not commonly planted.

10 Leucaena multicapitula Schery (MULT)(syn. L. multicapitulata), in Annals of theMissouri Botanical Garden, 37, 303-304, 1950.Type: P White 135, Canal-zone, Miraflores,Panama. This species is related to L.macrophylla and L. trichodes, all with largeleaflets, small inflorescences, and few seedsper pod. Leucaena multicapitula isdistinguished by its tall arboreal habit (to20 m), by a distinct geographical distributionin Panama and Costa Rica, and bysemi-paniculate inflorescences, broadcupulate petiolar glands and 16-grain polyadpollen (Sorensson 1993). The inflorescencehas relatively few florets (about 50-70) andis quite fragrant. It is autumn-flowering andwinter-seeding. Pods are short (about 12 cm)

and wide (to 3 cm). This fine tree is foundin high-rainfall areas, and has much promisefor domestication (Robustiano & Sorensson1987).

11 Leucaena pallida Britton & Rose (PALL),in North American Flora, 23, 126-127, 1928.Type: Rose 2359, Huejuquilla, Jalisco, Mexico,1897. Several taxa are synonymised with thisspecies, including four designated by Brittonand Rose and the L. esculenta subsp.paniculata of Zarate (1984). Shrubby andarboreal forms segregate in this species, andare found in dry highlands in Mexico. It wasshown by Pan and Brewbaker (1988) to be aprobable amphidiploid hybrid of L. esculentaand L. diversifolia, and closely resembles thediploid hybrid of those species (Sorensson1993). Leaves range from 15 cm to 40 cm,with 10-20 pairs of pinnae. Flower heads arelarge (to 20 mm), pinldsh and with manyflorets. Pods are variably long (10-18 cm)and wide (12-16 mm). It is sporadic in thecentral midlands of Mexico as a backyard orfencerow tree coppiced for fuel, fodder orfood. It is of great importance for thegenetic improvement of other species inpsyllid tolerance and rapid seedling growth.

12 Leucaena pulverulenta (Schlecht.)Bentham (PULV), in Journal of Botany 4,417, 1842. Basionyrn: Acacia pulverulentaSchlecht., Linnaea, 12, 571. Type: Schiede,Berlandier 288, San Antonio, Veracruz,Mexico. This distinctive tall tree has nosynonyms. It grows to 18 m with basaldiameters reported >50 cm. Bark and leavesare greyish, leaves ranging from 15 cm to 25cm with 10-20 pairs of pinnae and 15-30pairs of small (3-6 mm) leaflets. Flowerheads are small and open, white, with few(50-75) florets. L. pulverulenta is distributedon dry often calcareous soils, from Veracruznorth into S Texas, where frost often reducesthe tree height. It is known as 'tepeguaje' or'leadtree', and is widely retained andprotected through its range as a shade andfuelwood tree, but rarely planted. Trees areoften found in fencerows and are routinelylopped for fuel. Fast-growing hybrids (2n=80)with L. leucocephala were produced in Hawaii(Gonzalez et al. 1967) and even named inIndonesia, from which sterile segregants wereused as coffee shade. This species deservesgreat interest as a domesticate.

13 Leucaena retusa Bentham, ex Gray(RETU), in Plantae WrightianaeTexano-Neo-Medicanae, 1, 64, 1852. Type: C.Wright 171, Rio Nueces, Texas, 1849. This isa distinctive, yellow-flowered, subtemperateshrub. It grows to 5 m in height, and leavesrange up to 15 cm, having only 3-4 pairsof pinnae and 4-8 pairs of large (15-30 mm)

203

reticulated leaflets. Blooming heads areattractively golden, ranging to 3 cm, with150-200 florets, while the pods are narrow10 mm, and long (20-25 cm). It occurs inSW Texas and NE Mexico (Coahuila,Chihuahua) and favours limestone soils. It isslow to nodulate, possibly with uniquerhizobial affinities. It is the most cold-tolerantspecies of the genus, surviving snow andheavy frosts and ranging to 2000 melevation. Leucaena retusa is known as'littleleaf leadtree', and is grown as anornamental. It is often browsed by cattle andgoats. It is probably the best genetic sourceof frost tolerance genes; it has high droughtand heat tolerance and high-density woodthat encourage further domestication.

14 Leucaena salvadorensis Stand ley exBritton & Rose (SALV), in North AmericanFlora, 23, 125, 1928; Type: Calderon 2031,Jocoro, Morazan, Salvador, 1924. Hughes(1988) rediscovered and developed majorcollections of this species. It is an erect andtall tree (that can exceed 18 m) with leavesranging from 16 cm to 24 cm, with 4-7 pairsof pinnae and 15-25 pairs of long thinleaflets (4 mm x 20 mm). The white flowerheads are 12-22 mm in diameter, and thepods are large and heavily thickened, 18-30mm x 15-22 cm. Pod ripening is extremelyslow (as in L. collinsii). It is sparselydistributed from Salvador into Nicaragua indry foothills, and is known by the commonname 'sepia' in reference to the dense, darkwood. It makes straight poles valued inhouse construction, and is lopped forfuelwood. Foliage is relatively sparse butbrowsed by animals. The species has beenpromoted by OFT for greater domestic use inCentral America.

15 Leucaena shannonii Donn. Smith (SHAN),in Botanical Gazette, 57, 419, 1914. Type:Shannon 5032, Cuscutlan, Cojutepeque,

204

Salvador, 1892. No synonyms. Hughes(1991b) formally described and named twosubspecies, shanonnii (shrubby to 8 m) andmagnifica (arboreal, to 14 m), and correctedthe specific epithet from the collector's'shannonf to conform to taxonomic traditions.The two subspecies mimic the 'common' and'giant' forms of several other species. Leavesrange from 15 cm to 25 cm, with 4-5 pairsof pinnae and large leaflets (18-25 mm).Flower heads are heavily scented, white,10-15 mm in diameter, and pods are small(1 cm x 12 cm) and usually velutinous. Thisis a shrub or small tree of dry oftencalcareous lowlands from SE Mexico toNicaragua, with subsp. magnifica as a tree inGuatemala. It serves throughout its range asa fencerow fuelwood or shade tree and aslopped fodder. It is native in drought-proneareas, and the vigour of subsp. magnfficamakes it attractive for breeding.

16 Leucaena trichodes (Jacq.) Bentham(TRIC), in Journal of Botany, 4, 417, 1842.Type: Vienna Herb. as 'Mimosa trichodesJacq.' 32049, Brazil. At least five epithets areconsidered synonyms (Brewbaker 1987a).This small tree resembles L. macrophylla andL. multicapitula (Table 1), with which it iscross-fertile, but differs in several importantfeatures, including pollen (Sorensson 1993).The leaflets are very large (20-50 mm) andhave 2-4 pairs of pinnae. The flower headsare white, aromatic and very small (7-12mm), and pods are 2 cm x 10-15 cm,glabrous, with small seeds (6-7 mm).Leucaena trichodes colonises dry coastalareas to 300 m elevation from northern Peruthrough Ecuador and Colombia to Venezuela.Throughout this range it is known as a hardyfuelwood tree, but is rarely planted. Like somany Leucaena spp., this one is ofquestionable value as a domesticate exceptas a possible genetic reservoir.

Domestication of Sesbania sesban for agroforestrysystems in eastern and southern Africa

F Owino, P A Oduol & F EseguInternational Centre for Research in Agroforestry, PO Box 30677, Nairobi, Kenya

ABSTRACTSesbania sesban has potential for soil improvement and pole production when grown in short-rotation fallows or as scattered trees

in cropland. Examples of the domestication of S. sesban by farmers are presented, together with research results from field trials

by the International Centre for Research in Agroforestry (ICRAF). Results from two field trials in western Kenya indicated that

variation in growth rate and crown form among and within provenances is considerable. Genetic gains in height growth based on

progeny performance at two-location trials are reported to range from 14%, when the mean of the best ten families is used, to

61%, when the mean of the best five trees in the trial is used, for the Maseno location. Growth and reproductive phenology also

displayed significant variation, offering further possibilities for genetic improvement.

Above- and below-ground characteristics of S. sesban which constitute an ideotype for a specific agroforestry technology are

discussed, together with techniques for the grading of superior trees. Parallel research on tree genotype x Rhizobium strain

specificities and on the rooting pattern of S. sesban is described.

INTRODUCTION

With the current emphasis on integrated landmanagement practices such as social forestryand agroforestry many hitherto 'lesser-known' (or'Cinderella') tree and shrub species arereceiving increased research attention. In the pastten years, information/data inventories, treeimprovement and management research havebeen initiated in many parts of the world, withan emphasis on the multipurpose tree and shrubspecies that have the potential to provide acombination of products and services (Burley &von Carlowitz 1984; Burley 1984). Some prioritymultipurpose tree species for agroforestrydevelopment in humid and subhumid parts oftropical Africa that have been identified arecurrently the focus of research at theInternational Centre for Research in Agroforestry(ICRAF) (Maghembe 1989; Owino 1991).

One such species is Sesbania sesban, which hasbeen shown to enhance soil fertility (Rao, Gill &Abrol 1989; Onim et al. 1990) when grown inshort-rotation fallows. At the same time, thisspecies provides the farmer with fodder andpoles. Indeed, with the elimination of its prevalentlarge stem knots leading to weak poles, andproblems with pests and nematodes, S. sesbancould be an excellent species for agroforestrydevelopment in the humid and subhumid tropicsof Africa. It is with this background that S. sesbanhas been identified as a priority species forICRAF's tree improvement research programme.

ECOLOGY AND NATURAL VARIATION

Sesbania sesban (Leguminosae) is common innatural vegetation and in cultivated fields incentral, eastern and southern Africa (Lewis 1988).It is a fast-growing and often heavily branchedtree, which attains 3-8 m height at maturityActively growing parts of the stem arecharacteristically pinkish. It closely resemblesS. bispinosa in gross morphology but for thepinkish stem colour. Some forms resemble theslender and sparse-foliaged S. macrantha.

Within natural vegetation, S. sesban is commonlyfound on sites with permanent or seasonalswamps. Over a long period of time, farmers inwestern Kenya have favoured S. sesban trees withhigher growth rates and good (knot-free andstraight) stems in the fallows and cultivated fields(authors' personal observation). Although theextent of genetic improvement through farmers'practices has not been analysed, this mayrepresent an example of tree domestication byfarmers.

On good sites, S. sesban grows about 3 m inheight per annum for the first two years, afterwhich the growth rate declines rapidly Terminaldieback occurs at about four years of age, sothe species is only briefly perennial. Plants flowerand fruit at an age of two months in the nurseryand three months in the field. flowering startsshortly after the onset of the rains. In areas withtwo rainy seasons in a year, it flowers and setsfruit twice, coinciding with the rainy periods(Figure 1).

205

206

JFM AMJJ A

Months

Field trials have revealed significant differences ingrowth rate among local landraces. An analysis ofheight growth of 22 half-sib families at two sites(Kisii and Maseno, Kenya) indicated that the Kisiilandrace (K7 series) showed superiority at bothsites, whereas the Kakamega and Kisumulandraces (K1 and K5, respectively) showedintermediate performance (Figure 2). Greatvariation in crown form and branching habit wasalso recorded among the half-sib families. TheKisii landrace was characterised by strong apicalgrowth and limited lateral branching, whereas theKakamega landrace was much more highlybranched (Figures 3 & 4). Furthermore, earlyresults from progeny trials indicated a stronggenetic basis to the observed variation (narrowsense heritabilities of height for Maseno, Kisiiand combined sites were 0.4, 0.64 and 0.52,respectively). This pronounced variation in growthrate and crown form offers an opportunity forselecting the most desirable genotypes forfarmers by further genetic improvement research.

S OND J FM A MJ J AS OND

Months

Figure 1. Flowering and fruiting calendar of Sesbania sesban in western Kenya, 1990. Data collected by authors during plus-treeselection in 1990; observations based on plus-trees and their neighbours (n>5 per region)

POTENTIAL OF S. SESBAN INAGROFORESTRY SYSTEMS

S. sesban has greatest potential when managedin short-rotation fallows. On favourable sites, itcan provide valuable poles and improve soilfertility within a period of about six months(Oduol, Akunda & Wambugu 1993).Furthermore, it is a valuable source of greenmanure (Evans & Rotar 1987). For example, inone experiment at Maseno in Kenya, S. sesbanadded 448 kg N ha -1 yr -1 to the soil whenused as a source of green manure (Onim et al.1990). Natural regeneration in cultivated fieldsis prevalent and the species rarely becomes adifficult weed to eradicate as farmers haverelatively little trouble in removing mature treesduring field preparation.

The appropriate period of tree growth infallows or on cropland will depend on thespecific combination of products and servicesrequired. For the combination of poles,fuelwood and soil fertility improvement, one to

*cp 10

9

8

7

6

5

4

3

2

Kisii SE = 0.26

Maseno SE = 0.23

CO (O 0 CO ts (3) •Cr CM LC) CD CO CO CO LO •Cl•CON COCOOJCNICNICOCOCOCD0001-1—CMCN.—.-1—..—NN- N. is- N. N- LO LO st

YYYYYYYYYNCYYYYYYYYYY

Sesbania sesban Kisii landrace

Figure 4. (below) Diagrammatic representation of a typicalProgeny codes Sesbariia sesban tree from the Kakamega landrace

K1 KakamegaK4 SiayaK5 KisumuK7 Kisii

0,1

H11,1,

' 1) 1111o\I1:1'111,;Id [if 111;1(11; I y

111—":.": •

two years of tree growth would be ideal onfavourable sites (Dutt & Urmilla-jamwal 1989;Weerankoon 1989). For the combination offuelwood and soil fertility improvement, asix-month rotation (followed by clearing andreplanting) might be more appropriate (Yamoah& Getahun 1989; Rao et al. 1989).

TREE IMPROVEMENT RESEARCHINITIATIVES FOR S. SESBAN

Given the great potential of S. sesban foragroforestry in humid and subhumid parts ofAfrica, ICRAF initiated a germplasm improvementresearch programme for the species in 1990. Thefirst phase of this research was the evaluation ofa wide range of germplasm accessions collectedby the International Livestock Centre for Africa(ILCA) planted at one site (Maseno) in westernKenya. Results from this field evaluation indicatedthat a landrace from Kenya (Kakamegaprovenance) was superior in growth rate and

Figure 2. (left) Mean heights of 22 Sesbania sesban half-sibfamilies at Kisii and Maseno, Kenya, after one year

F7gure 3. (below left) Diagrammatic representation of a typicalSesbania sesban tree from the Kisii landrace

Sesbania sesban Kakamega landrace

1 /

fir

207

total above-ground biomass production. TheKakarnega provenance also proved superior intwo other provenance trials in Burundi andRwanda (unpublished data). It was, therefore,decided to select phenotypically superior treesfor evaluation and breeding research from withinthe Kakamega provenance.

The next step involved the definition of the idealideotype towards which selection and breedingresearch should concentrate. The ideotype wasdefined in relation to the agroforestry technologieswith the greatest potential for the region. The threemost promising technologies within which S. sesbancould play important roles are (i) short-rotation treefallows, (ii) relay follows with trees grown togetherwith crops in hedges, and (iii) scattered trees oncropland. For these technologies it is consideredthat S. sesban should have:

• high growth rate in terms of height;

• straight stems with few large knots;

• narrow crowns to minimise competition withcompanion crops;

• sparse foliage to facilitate light penetration;

• branching architecture which maximisespole quality (ie small evenly spacedbranches);

• narrow and deep rooting patterns tominimise competition with companion crops;

• freedom from pests and diseases (includingroot diseases and nematodes);

• copious nodulation and high rates of fixationof atmospheric nitrogen.

Most of the above tree characteristics can becombined in a single phenotype, as exeMplifiedby an eight-month-old tree of the Kisii landraceat the Kakamega site in Kenya (Figure 5).

Sesbania sesban

208

4 3

A

The practical challenge for S. sesbanimprovement research is, therefore, to reduceprimary and secondary branching and increaseapical growth, while at the same time selectingfor a deep and narrow rooting pattern (asdepicted in Figure 3) and confirmed potential forN fixation. Consequently initial selections ofsuperior phenotypes have been made on thebasis of height growth, stem diameter, branchdensity crown diameter, foliage density stemstraightness, branching architecture, and freedomfrom pests and diseases. This initial massselection began in Kenya in mid-1990. To date,38 phenotypically plus-trees have been selectedand their progeny are under field-testing ashalf-sib families. The target is to extend therange of search into Uganda and to assemble aninitial breeding population of some 100genetically plus-trees.

In collaboration with staff at the Department ofSoil Science, University of Nairobi, a parallelresearch initiative has been studying treegenotype x Rhizobium strain interactions. Morerecently rooting pattern studies at four sites inKenya involving 22 half-sib families have beencompleted. Observed trends from thesebelow-ground tree characteristics will bereported elsewhere.

ESTIMATES OF GENETIC GAINS FROMFIRST-GENERATION SELECTION

Data from one-year-old progeny trials of S.sesban at two sites in western Kenya were usedto estimate genetic parameters and genetic gains(lhble 1). Variance components for family andwithin-family were estimated via expected meansquares in an analysis of variance. Narrow-senseheritabilities (h2) were calculated using intra-class

2

FYgure 5. Diagrammatic representation of contrasting ideotypes of Sesbania sesban

correlations (Falconer 1981) as follows:

a2bt =

b + o6ist

where t = intra-class correlation; a2b =between-family variance; and sa2w = within-familyvariance.

The intraclass correlation (t) is an estimate of thecorrelation of family values and can be used toestimate narrow-sense heritability (h2). Forhalf-sibs, t = 1/4 h2. Estimated values of h2 forheight growth from Maseno, Kisii and combinedprogeny trials were 0.4, 0.64 and 0.52,respectively

Table 1. Genetic gains' (%) calculated from Sesbania sesban

progeny trials for height growth at Maseno and Kisii sites in

Kenya

Percentage genetic gain

Selection method/intensity Maseno2 Kisii2 Maseno/Kisii3

Mean of best five trees in 61 127 174

the trial irrespective offamily

Mean of best ten trees in the 54 116 165

trial irrespective of family

Mean of best five families 20 49

Mean of best ten families 14 34

TFormula for genetic gain used: selection differential x

narrow sense heritability (h2)2Based on progeny test results at separate sites3Based on combined progeny test results at two sites

CONCLUSIONS

Results from progeny trials with S. sesban haveconfirmed large differences between half-sibfamilies in stem growth rate and quality withnarrow-sense heritability estimates for heightgrowth ranging from 0.4 to 0.64. Substantial gainscan be made through selection and breeding forthese traits. Research is on-going to improveunderstanding of the nature and amount ofgenetic variation in the species.

REFERENCESBurley J. 1984. Global needs and problems of collection,

storage and distribution of multipurpose tree germplasrn. In:

Multipurpose tree gennplasm: proceedings of a planning

workshop to discuss international co-operation, edited by

J. Burley & P von Carlowitz. Nairobi: International Council for

Reseamh in Agroforestry

Buz ley J. & von Cazlowitz, P., eds. 1984. Multipurpose tree

germplasm: proceedings of a planning workshop to discuss

international co-operation. Nairobi: International Council for

Research in Agroforestry

Dutt, A.K. & Urrnilla-Jamwal. 1989. Woodfuel properties of

Sesbania bispinosa and Sesbania sesban. In: Perennial

Sesbania species in agroforestry systems, edited by B. Macklin

& DO. Evans, 73-77. Waimanoloi: Nitrogen Fixing Ree

Association.

Evans, DA:). & Rotar, P.P. 1987. Sesbania in agriculture.

(Westview Tropical Agriculture Series, no. 8.) Boulder,

Colorado: Westview Press.

Falconer, D.S. 1981. Introduction to quantitative genetics. New

York: Ronald.

Lewis, G.P. 1988. Sesbania Adans in the Flora Zambesiaca

region. Kirlda, 13, 11-51.

Maghembe, J. 1989. Preliminary performance of some

multipurpose trees hi the unimodel rainfall ecozone of

southern Africa. In: Trees for development in sub-Saharan

Africa, edited by J.N. Wolf, 95-97. Stockholm: International

Foundation for Science.

Oduol, RA., Akunda E.N. & Warnbugu, P.N. 1993.32 Nodulation survey of leguminous and non-leguminous trees22 growing at ICRAF's field station. In: Proceedings of the 3rd

African Association for Biological Nitrogen Fixation, Dakar

Senegal, 1988. In press.

Onim, J.M., Mathuva, M., Otieno K. & Fitzhurg, H.A.1990. Soil fertility changes and response of maize and beans

to green manures of Leucaena, Sesbania and pigeon pea.

Agroforestry Systems, 12, 197-215.

Owino, F. 1991. Improving multipurpose tree and shrub

species for agroforestry systems. Agroforestry Systems, 16, 7-13.

Rao, D.L.N., Gill, G.S. & Abrol, J.P. 1989. Regional

experience with perennial Sesbania in India. In: Perennial

Sesbania species in agroforestry systems, edited by B. Macklin

& DO. Evans, 189-198. Waimanolo: Nitrogen Fixing Tree

Association.

Weerankoon, W.L. 1989. Sesbania in indigenous farming

systems in Sri Lanka. In: Perennial Sesbania species in

agroforestry systems, edited by B. Macklin & no. Evans,

181-187. Waimanolo: Nitrogen Fixing 'free Association.

Yarnoah, C. & Getahun, A. 1989. Alley cropping and crop

yield enhancement with Sesbania species. In: Perennial

Sesbania species in agroforestry systems, edited by B. Macklin

& DO. Evans, 109-122. Waimanolo: Nitrogen Fixing Tree

Association.

209

Domestication of trees in semi-arid East Africa:the current situation

P B Milimol , j McP Dick2 & R C Munro 2

1Kenya Forestry Research Institute, PO Box 20412, Nairobi, Kenya2Institute of Terrestrial Ecology*, Bush Estate, Penicuik, Midlothian EH26 OQB, UK*A component of the Edinburgh Centre for Tropical Forests

ABSTRACTIn East Africa, pastoralists in the dry savannahs rely heavily on trees and shrubs for browse, fodder, shade, fencing and firewood,etc. This review focuses on the use of non-industrial trees and shrubs, and discusses the role of domesticated indigenous trees inagroforestry systems.

To date, much attention has been paid to matching species to different ecoclimatic zones, especially with regard to droughttolerance. More recently specific studies have been initiated on provenance variation, for example in Acacia species, andlow-technology vegetative propagation techniques have been developed for a number of species. Multipurpose tree species offernumerous opportunities for genetic selection and domestication, such as improvements in survival, growth rate and yield, as wellas various quality attributes associated with different forest products.

Mycorrhizal inoculation has been found to enhance greatly the survival of trees planted in degraded areas, which have lowmycorrhizal inoculum potential. Inoculated trees have been used to restore the soil inoculum and have been shown to enhancethe growtli of interplanted agricultural crops. Studies in progress in Kenya are screening for the effectiveness of indigenousrhizobial strains associated with N-fixing tree species. There are opportunities to exploit tree/crop symbiotic associations further inagroforestry systems, using trees selected both for their own attributes and for soil-improving qualities.

INTRODUCTION

Africa is faced with the serious problem of notbeing able to feed its population or supply itwith wood fuel. Production of more than onecrop plant by each farmer is a common featureof traditional farming systems in Africa (Sanginga,Mulongoy & Swift 1992), and plant species thatprovide more than one product are favoured forplanting (Konuche & Milimo 1989). Multiplecropping systems usually involve the use ofmultipurpose trees and shrubs, includinglegumes, in intimate association with seasonaland perennial food crops and livestock. Thesuccess of these systems lies in a high overallbiomass yield, good nutritional value of theedible products, and a wide diversity of products.This diversity also reduces the chances ofcomplete crop failure (Torquebiau 1992).

In East Africa, pastoralists in the dry savannahsrely heavily on trees and shrubs for browse,fodder, shade, living fences and windbreaks.These species are also important as sources ofwood (firewood, poles, posts) and non-woodproducts (gums, resins, fruits, food, medicines)and for maintaining soil fertility (Table 1). Certaindecorative woods, especially the black ones(Diospyros spp., Dalbergia melanoxylon, etc), arealso commercially very important for carvings forthe tourist trade.

210

This review focuses on the non-industrial treesand shrubs in semi-arid East Africa which couldbe selected for use in agroforestry systems. Themultipurpose nature of trees and shrubs inagroforestry poses difficulties for tree improvers,such as whether single Or multipurposecharacteristics should be selected when breedingor selecting improved lines of multipurpose trees(see Owino, Oduol & Esegu, pp205-209; Simons,MacQueen & Stewart, pp91-102). This papercovers progress in three stages of thedomestication process:

i. species selection;

propagation and intraspecific geneticselection, and

the incorporation of these selections intosustainable management systems. -

In East Africa, a start has been made with theseprocesses by screening species for adaptation todifferent climatic/edaphic conditions (Table 2), theevaluation of provenance variation and thecreation of a small number of clones. In addition,investigations are in progress to examine the roleof symbiotic soil micro-organisms, includinginodulating trees with vesicular—arbuscularmycorrhizal fungi (VAM) and nitrogen-fixingbacteria (rhizobia).

Table 1. Tree and shrub species listed by their common uses in East Africa

Poles and posts Fuelwood Erosion control Carving

Acacia albida Acacia nilotica Acacia albida Acacia albida

A nilotica A polycantha A elatior A polycantha

A tortilis A. saligna A senegal A tortilis

Azadirachta indica A seyal A tortilis Balanites aegyptiaca

Balanites aegyptiaca A tortilis Albizia coriaria Commiphora spp.

Ca &his glauca Azadirachta indica A lebbeck Croton megalocarpus

Cassia siamea Balanites aegyptiaca Cajanus cajan Delonix elata

C. spectabilis B. orbiculans Cordia sinensis Diospyros scabra

Casuarina equisetifolia Callitris glauca C. ovalis Dobera glabra

Cordia sinensis Cassia siamea Croton megalocarpus Erythrina spp.

C. ovalis Casuarina equisetifolia Combretum micranthum Hyphaene ventricosa

Croton megalocarpus Combretum micranthum Dalbergia melanoxylon 7brminalia brovvnii

Delonix elata Cordia sinensis Euphorbia tirucalli Ziziphus mauritiana

Eucalyptus camaldulensis Croton megalocarpus Lannea alata

E. terencornis Diospyros scabra Lantana carnara

Hyphaene ventricosa Eucalyptus camaldulensis Leucaena leucocephala

Leucaena leucocephala E. tereticornis Prosopis chilensis

Melia azedarach Leucaena leucocephala P juliflora

Melia volkensil Melia azedarach Salvadora persica

Prosopis chilensis Parkinsonia aculeata Tamarix aphylla

P juliflora Prosopis chilensis Ziziphus spp.

Salvadora persica Salvadora persica Ziziphus mauritiana

Tamarindus indica Syzygium cumini

Terminalia brownii Ximenia americana

Windbreaks

Acacia albida

A. nflotica

A. polyacantha

A. tortilis

Berchemia discolor

Cassia siamea

C. spectabilis

Croton megalocarpus

DeloniX elata

Diospyros scabra

Euphorbia tirucalli

Ficus sycomorus

Leucaena leucocephala

Schinus molle

Tamarindus indica

Ornament

Acacia albida

A. nilotica

A. polycantha

A. tortilis

Adansonia digitata

Adenium obesum

Berchemia discolor

Cassia siamea

C. spectabilis

Croton megalocarpus

Delonix elata

D. regia

Diospyros scabra

Erythrina abyssinica

Ficus sycomorus

Melia azedarach

M volkensil

Panax fruiticosa

Phyllanthus roseoptia

Schinus molle

7brminalia spinosa

SPECIES SELECTION

Until recently when it became government policyin Kenya to promote the replanting of indigenousrather than exotic tree species, most of the workon the selection of trees of semi-arid lands inKenya was with exotic fast-growing species. Forexample, a trial with Australian Acacias on twosemi-arid Kenyan sites (Marimanti andLanchiathurio) at 18 months indicated that Acaciaholosericea, A. cowleana, A. plectocarpa, A.leptocarpa and A. brassii were promising (Chege

Hedges

Bixa orellana

Cassia edulis

C. grandiflora

Caesalpinia decapetala

C. pulcherrirna

Chrysophyllum albidurn

Duranta repens

Dovyalis caffra

Opuntia dellenii

Pithecellobium dulce

Human food

Adansonia digitata

Annona cluysophylla

Azanza garckeana

Balanites glabra

Bequaertiodendron natalense

Berchemia discolor

Borassus aethiopum

Canthium gueinzli

Carissa edulis

Cordia sinensis

C. somaliensis

Cordylogyne africana

Diospyros mespfliformis

Ficus glumosa

urceolanS

F vaflis-choudae

Flacourtia indica

Grewia ectasicarpa

Hirtella zansibarica

Hyphaene coriacea

Lannea alata

Human food

Manilkara butugi

Mimusops fruticosa

Phoenix reclinata

Rhus natalensis

R. vulgaris

Sclerocarya birrea

S. caffra

Sorindeia obtusifoliolata

Strychnos decussata

Syzygium guineense

Vangueria apiculata

V tomentosa

Vitex payos

Ximenia americana

X. caffra

Ziziphus abyssinica

Z. mucronata

& Stewart 1991). In other trials in theEmbu/Meruilsiolo project, species showingpromise were Acacia albida, A. cyanophylla, A.nilotica, A. polycantha, A. victoriae, Atnplexnummularia, Atriplex semi-baccata, Balanitesaegyptiaca, Melia volkensii, Prosopis juliflora,Termihalia spihosa and Ziziphus mauritiana(Armstrong & Lugadiru 1986). Subsequently theperformance of exotics was compared withindigenous species but seldom with a widerange of provenances. Comparative results of 58exotic and indigenous tree and shrub species at

211

Table 2. nee and shrub species listed accon:ling to siterequirements

11, 150-350 mm; 2, 300-350 mm; 3, 450-900 mm; 4,

600-1100 mm21 , depressions/river banks; 2, plains; 3, hills/slopes; 0, notspecific31 , clays; 2, loam, sand, groundwater alluvial. 3, rocks, gravel,hardpans4 1 , dependent on groundwater; 2, tolerates seasonalinundation; 3, requires moist, well-drained soils; 0, not specific

212

two years of age at Loruk (Kenya) are reportedby Kimondo (1991). From these results, theAustralian Acacias generally were superior inheight growth, but had poor survival with onlythree out of the 15 species achieving greaterthan 80% survival (Table 3), compared With nineout of 12 of the indigenous species. The

indigenous Sesbania sesban was the tallest, buthad only 25% survival. Of the other indigenousspecies tested, the height and survival of Acaciaseyal and Melia volkensii were acceptable(Kimondo 1991).

Similarly, trials with exotic (25) and indigenous(16) species have also formed part of theobjectives of the Baringo fuel and fodder projectbased near Lake Baringo in Kenya Mortalitiesless than 50% occurred with only Acacia tortilisand the exotics Prosopis pallida, P chilensis,Acacia anew-a, Cassia sturtii, Gleditsia tnacanthos,Prosopis tamarugo, Acacia albida, Simmondsiachinensis, Azadirachta indica and Atnplexcanescens.

In order to facilitate domestication, it isnecessary to determine ..the tree and shrubcharacteristics desired by a particular communityand to identify genetically outstanding naturalpopulations and individual8 for propagation byseed or vegetative techniques. In semi-aridareas, there is considerable variation in soiltypes, rainfall (quantity frequency duration of dry

Table 3. Mean height and survival of Australian Acacia spp.and 12 other species indigenous to Kenya at two years ofage at Loruk, Kenya (500 mm mean annual rainfall) (source:Kimondo 1991)

Ecoclimateand

rainfallzonel

Geomoology

- Soil/waterSoils3 regirn. es4

Acacia albida 1-4 1 1,2 1,2A gerrardii 3-4 1,2 1 3A mellifera 2-3 0 1,2 3A. nilotica 1-4 0 1 3A polyacantha 4 1,0 1,2 3A. senegal 1-4 1 3 2,3A. seyal 2-4 2 3 3A. tortilis 1-4 1,0 1,3 1,2,3Adansonia digitata 2-4 1,2 1 3Azadirachta indica 2-4 2 1 1,3Balanites aegyptiaca 1-4 0 1,3 3,0Berchemia discolor 3-4 1,0 3 3Boscia angustifolia 2-3 0 1,3 3B. coriacea 2-3 0 1,3 3Boswellia hildebrandtii 2-3 2,0 1 3Carissa edulis 3-4 1,2 1 3Combretum molle 3-4 2,0 1,3 3,0Commiphora africana 2-3 2 1,3 3Conocarpus lencifolius 3-4 1 1 1,3Cordia sinensis 1-3 1,2 1,3 1,3Croton megalocarpus 4 2 1 3Dalbergia

melanoxylon2-4 2 1 3

Delonix elata 1-3 2,1 3 3D. regia 4 0 1 3Diospyros

mespiliformis3 0 1,3 3

D. scabra 1-3 1 1,3 3Dobera glabra 2-3 0 1,2 2,3Dodonaea viscosa 4 3,0 2,3 3Euphorbia tirucalli 2-4 0 2,3 0Grewia villosa 2-3 0 2 3Hyphaene ventricosa 1-3 1 2 1Melia azedarach 3-4 0 2 3M. volkensii 3-4 1 2 3M. stenopetala 3-4 1 2 1Parkinsonia aculeata 2-3 0 1,2,3 0Piliostigma. thonningii 4 0 2,3 3Pithecellobium dulce 4 2 2 3Populus ilicifolia 2-4 1 1,2 1Salvadora persica 1-3 1,0 2 0Syzygium cumini 2-4 1 1,2 1

CountryCSIROseedlot

Meanheight Survival

Species of origin number (m) (%)

Acacia holosericea Australia 14660 2.8 85A. holosericea 14632 2.8 93A famesiana 2.6 55A. torulosa 14888 2.6 33A. plectocarpa 17207 2.5 74A. brassii 15480 2.4 15A. difficilis 14623 2.4 51A ampliceps 14668 2.4 74A. ampliceps 14631 2.3 55A shirleyi 14622 2.2 26A. cowleana 14885 2.1 35A. cowleana 14683 2.1 32A. julifera 14656 2.1 44A raddiana 1.7 82A trachycarpa 15767 1.6 18

A. nilotica India - 2.8 78Tamarindus indica Kenya - 2.9 97A. nilotica - 1.9 82A. nubica - 1,0 23A. tortili - 1.6 99Cordia sinensis - 1.6 91Dalbergia

melanoxylon _ 1.5 97krminalia brownli _ 1.3 95A. mellifera - 1.1 98Croton megalocarpus - 1.1 94Berchemia discolor - 1.1 74Sesbania sesban - 3.2 25Balanites aegyptiaca - 0.9 68

season, reliability etc), proximity to water table,pH, salinity and topography. Thus, there is alikelihood that wild tree populations will haveadapted to their environments on a local scale.In addition to this small-scale variation, there aresuperimposed regional gradients of increasingaridity Species trials in Kenya at Ho la, Ramogi,Baringo, Kibweze, Lodwar, Embu and Meru havescreened more than 500 different exotic andindigenous tree and shrub species for theiradaptation to different ecological zones. In EastAfrica, therefore, it is clear that there has beenconsiderable genetic testing at the species level.This testing now needs to be extended to astudy of the genetic variation betweenprovenances, as well as the likely variation withindifferent populations.

INTRASPECIFIC GENETIC SELECTION

International provenance trials with dry-zoneAcacia spp. are currently in progress,co-ordinated by the Oxford Forestry Institute(Fagg & Barnes 1990). Four African Acaciaspecies (A. torah's, A. nilotica, A. senegal and A.albida [now called Faidherbia albida]) with widegeographical distributions, which include EastAfrica, were identified as having particularpotential as producers of a range of productsand as being suitable for planting on arid,degraded sites. As a prerequisite to thesuccessful establishment of provenance trials, astudy was made in 1987-90 of:

• morphological and phenological variation ofthe tree species and of their rhizobialsymbionts throughout their natural range;

• the breeding systems of the species.

In addition, a sampling strategy was developedand utilised to provide the basis for geneticexploration, evaluation and conservation as abasis for subsequent domestication (Fagg &Barnes 1990).

In the second phase of this project (1990-93), A.

eriloba was added to the project and the aboveresearch was extended and seed collectionswere made. These collections were fordistribution* to collaborators in the establishmentof international provenance trials. In addition tosuperior performance in terms of growth, thesetrials are intended to test other characteristics,such as nutritional value, fodder production, andgum . quality and productivity (Fagg et al. 1990).

By 1991, over 30 provenances had beendeposited in the UK Forestry Commission'sseedbank at Alice Holt, Surrey, England. Fora number of these provenances, seedcollections were made from individuallysampled trees for more detailed studies ofgenetic variation and breeding systems. Leafcollections were also made for studies onperoxidase activity and nitrogen fixation

assessments. Wood samples were collected fordendrochronology studies, and gum sampleswere taken for chemical analysis. In a parallelstudy funded by the European Commission,further seed collections of Faidherbia albidahave been made from East Africa for acollaborative study between institutions in theUK, France and Senegal on its reproductivebiology and its genetic variability in agro-forestry characteristics and in its isoenzyrnepatterns (Lockhart, Fagg & Barnes 1990).

In another study, 30 families within threeprovenances of Melia volkensii, in which seedswere collected along a sharp climatic gradient,were established on three sites in Kenya(Embu, Bura and Malindi) in 1990. Preliminaryresults have indicated pronounced variation atthe family level, but not at the provenancelevel, in growth, drought tolerance, rootarchitecture, branch density and size, leafphenology and resistance to spider miteattacks. These results have importantimplications for the selection of genotypes fordry and wet season fodder production, theprovision of shade for livestock, and forintercropping with agricultural crops (P Miimo,unpublished data).

In other areas of semi-arid Africa, studies are inprogress on species also indigenous to EastAfrica. In two different experiments involvingalbida, provenances raised from West Africanseed had generally smaller shoots comparedwith those from South Africa, under moist soilconditions (Wanyanja, Mills & Gwaze 1991;Vandenbeldt 1991). In the drier West Africanplots, all provenances of F albida from SouthAfrica had 100% mortality at the end of thesecond year, while survival of both provenancesin South Africa was the same. In South Africanplots, local provenances outperformed the WestAfrican ones in height growth. The South Africanprovenances had greater shoot dry biomasscompared with those from the West Africanprovenances, which allocated 20% more drymatter to the roots. These results indicate that,in this instance, local provenances were betteradapted to the prevailing conditions thanintroduced ones.

A detailed study of variation between 75provenances of Sesbania sesban has been madeby the International Centre for Research inAgroforestry (see Owino et al., pp205-209). Afurther component of the study with this speciesis to select individual and contrastingphenotypes for their apparent suitability forproducing poles, fodder and biomass. These willbe propagated vegetatively to assess theirpotential for maximising productivity of aparticular product by selecting single-purposeclones of this multipurpose species (P Oduol,personal communication; see also Owino et al.,pp205-209).

213

Variations between and within species in theircapacity to resist drought are likely to be amajor characteristic influencing the survival andgrowth of trees in the semi-arid lands of EastAfrica. Although stable performance of aprovenance over a wide range of environmentsis generally regarded as desirable, past attemptsto identify and select trees for semi-arid areashave resulted in a wide range of responsesamong species and provenances to differentenvironments, as would be expected. This hasbeen demonstrated in progeny/provenance testsof Melia volkensii and IVI. azedarach establishedon three sites with 400 mm, 950 mm and 2000mm annual rainfall, in northern Queensland,Australia. Results from these studies showmarked variation within and between species inleaf phenology survival and above-ground drymass production (P Milimo, unpublished data).Thus, given the range of soil types and wateravailability in these conditions, genotype/environment interactions are almost certain toprevent the selection of a widely acceptableprovenance.

The detection of early indicators of good fieldperformance is a valuable step in treeimprovement. In hot dry sites, high water useefficiency and the absence of photo-inhibition,may be determined by examining morphologicaland physiological characteristics. Thisexamination, however, is likely to be moresuccessful at the clonal rather than theprovenance level. Success would depend on thecapacity to use juvenile tissues to predict theresponses of mature trees to drought conditions.However, because mature trees are more likelyto be able to utilise water deep underground,drought tolerance is probably much moreimportant in seedlings and young plants.

Most of the indigenous trees of semi-arid EastAfrica could be selected for a range bf genetictraits. Thus, selection in multipurpose tree speciescould include not only the typical, forestry-relatedcharacteristics such as growth rate and stemform, but also selection for resistance to drought,salinity alkalinity; fodder and pod yields andquality (nutritive value, digestibility freedom fromalkaloids, thornlessness, etc); yields and quality ofgums, resins and pharmaceuticals (see Waterman,pp42-48); firewood *and charcoal quality (highwood density presence of oxalates (Prior & Cutler1992), etc); as well as for nitrogen-fixing ability inthe leguminous species. The problems associatedwith selection of multipurpose species areconsidered by Simons et al. (pp91-102): see alsoFelker (pp 183-188) for studies on Prosopis spp.

PROPAGATION BY SEED

Although propagation by seed is the mainmethod of plant multiplication in E Africa,availability of quality seed is a major constraint

214

affecting planting (Shakacite 1987; Shehaghilo1987; Wate 1987; Milimo 1987). Problems thatprevent the use of seed can be broadlyclassified as irregular or infrequent flowering andseed inviability and/or dormancy

Fruit-bearing species like Ziziphus mauritiana,Diospyros scabra, Balanites aegyptiaca andDobera glabra have well-defmed seed collectionperiods, although their viability declines rapidlywith time (Zumer-Linder 1983). However,seedlings of many of the popularwood-producing indigenous tree species cannotbe propagated easily by seed because theyexhibit dormancy Amongst these are Meliavolkensii (Mllimo & Hellum 1989a, b), Terminaliaspp. (Specht & Schaefer 1990) and Hyphaenacoriacea (Zumer-Linder 1983). The techniquesthat have .been used to break dormancy are notpractical for large-scale seed production, andtherefore more research on this aspect isneeded. Insect damage is a major constraint onseed availability in some genera, eg Balanitesand Acacia.

VEGETATIVE PROPAGATIONVegetative propagation is becoming increasinglyimportant in forestry and agroforestry for themultiplication of limited seed material and for theproduction of genetically uniform stock forplanting. The value of vegetative propagation tomultiply selected planting material and to capturegenetic potential has long been known (Libby1973; Zobel & Talbert 1984; Leakey 1987).Vegetative propagation gives the tree improverthe ability to multiply test, select from and utilisethe large genetic diversity present in most treespecies. In this way selected, highly productivebut unrelated clones can be used commerciallyfor reforestation and agroforestry (Leakey 1991).

Studies to determine the best propagationenvironment for semi-arid species have indicatedthat a non-mist propagation system is generallymore effective than conventional mist propagation(Dick, East & Leakey 1991). In particular, thebetter rooting of cuttings in non-mist propagatorsseems to be related to a lower susceptibility torotting and consequent mortality (Figures 1 & 2).The low-technology non-mist propagator currentlyused in Kenya has been described by Leakey etal. (1990) (see also Ladipo et al., pp239-248).This design does not require electricity or pipedwater, and is therefore particularly suitable forrural areas in the tropics. Several groups workingin Kenya have successfully utilised thistechnology to root a variety of species (Table 4).

Trials have started in Kenya to determineprotocols for the larger-scale propagation ofsemi-arid zone species. Initial work has shownthat Acacia tortilis and Prosopis juliflora both rootwell from pollarded material. Three months afterpollarding P julillora trees 1 m above ground

100

80

60

40

20

00

100

80

E. 60-0a)0 40

20

0

1 2 3

Weeks

Figure 1. Effect of three propagation environments: (i) alow-technology non-mist propagator, (ii) an open mist system,and (iii) a mist system enclosed in polythene, on thepercentage of Prosopis juliflora cuttings forming adventitiousroots

0 1 2 3

Weeks

0 Non-misto Open mist• Enclosed mist

4 5 6

0 Non-mistoOpen mist• Enclosed mist

4 5

Figure 2. Percentage mortality of Prosopis juliflora cuttings inthree propagation environments

level (to avoid grazing by goats), between 50and 250 cuttings were harvested and transportedthe same day to the propagators in Nairobi, fourhours' drive away l'wo-node cuttings(approximately 5 cm length) were taken fromeach shoot after discarding the very soft 10 cmat the tip. The cuttings were dipped in acommercial rooting powder 'Seradix 2' (May &Baker Ltd, active ingredient 0.8% indole-3-butyricacid, IBA), and placed in the non-mist

propagators. The number of cuttings rooted wasassessed five weeks later when 43% and 96%rooting had occurred in A. tortilis and P juliflora,respectiveiy. Three harvests were collected fromone tree of P juliflora pollarded in December1988; cuttings were collected in February Augustand November 1989, with rooting success of43%, 56% and 83%, respectively The differencesin rooting success are probably attributable tothe stage of growth of the resprouted materialwhen collected, rather than to season. Older,more lignified tissue was collected in the firsttwo harvests and is less likely to root thanyounger material.

It is clear that the system developed in Kenya hasgreat potential for practical implementation insemi-arid areas with species which coppice. Theability to root tropical tees from leafy stem cuttingsdepends upon many facto's, such as thephysiological state of the stockplant, the propagationenvlionment and the treatment applied to the cuttingprior to propagation, eg the leaf area and theconcentration of applied auxins (Leakey 1985). Agood understanding of these factors is necessary tosustain output from vegetative propagation systems(Leakey Newton & Dick, pp72-83).

The state of the leaf on the cutting is animportant factor in determining the rootingpotential of a cutting. For example, leaves oncuttings of Terminalia spinosa trimmed to 30 cm2produced roots in a shorter period than those

6 trimmed to only 7.5 cm2 (Newton, Muthoka &Dick 1992), whereas leafless cuttings did notroot. The mean length of the longest rootincreased with increasing leaf area (0, 7.5, 15and 30 cm2), which is almost certainly importantfor subsequent cutting survival and establishmentin the nursery Large-leaved cuttings weredroughted, displaying the lowest foliar relativewater content. In response to this drought, thesecuttings closed their stomata and thus had thelowest stomatal conductance, and displayed thehighest rate of leaf shedding (Newton et al.1992). These self-regulated changes in leaf arearesulted in cuttings with optimal leaf area forrooting in the environment tested.

Stockplant management is also critical forsustained, successful rooting. The light

Table 4. Dry zone species propagated by vegetative stem cuttings at ITE, Edinburgh, National Museums of Kenya and Baringofuel and fodder project

215

environment to which stockplants are subjectedalso influences rooting potential of cuttings. InAcacia tortilis grown under different light qualitiesat constant irradiance (150 gmol IT1-2 CI) incontrolled environment conditions, stockplantsgrew tallest in light with a red/far-red ratio (R:FR)of 0.61, rather than at 1.09, 3.14 or 4.48 R:FR(Dick & East 1992). Cuttings from shootsharvested from plants grown at low R:F'R (0.61and 1.09) rooted better (77-79%) than thosefrom R:FR 3.14 (63%) or R:FR 4.48 (22%). Thisresponse is similar to that of other tropical trees(see Leakey et al., pp72-83).

TREE ESTABLISHMENT AND SUSTAINABLEMANAGEMENT

The value of domestication in capturing a rangeof genetic potential is lost if, when the selectedmaterial of multipurpose trees is planted out inthe field, they either do not establish or growpoorly Such problems have often occurred whenreforesting drylands where soil moisture andnutrients are limiting (Wood 1989; Armstrong &Lugadiru 1986; Chege &. Stewart 1991; Chirchir& Stewart 1989), or when the seedlings havebeen inadequately protected from grazinganimals. The change in environmental conditionsexperienced by the plants during transplantingmay be severe, and may result in severedroughting of the plants, which can be at leastpartly alleviated by irrigation (Newton, Munro &Wambugu 1993). Although the application ofwater-retaining polymers has also beenrecommended to improve tree establishment byalleviating the effects of drought (see Callaghanet al. 1989), initial field results have not beenencouraging (Newton et al. 1993; Wilson et al.1991).

Many tropical soils are degraded, contain lowconcentrations of plant nutrients, and have highphosphorus-fixing capacity (Sanchez & Salinas1981). Attempts to solve this problem usuallyinvolve application of chemical fertilizers or theaddition of organic matter. Since chemicalfertilizers are unaffordable by most subsistencefarmers, except through credit schemes whichcan plunge the fanner into debt, this reviewexamines the role of soil microflora in treeestablishment.

The role of VA mycorrhizas

Symbiotic soil micro-organisms possess thepotential to assist trees and crops to accessgreater quantities of nutrients (Torquebiau 1992)and therefore survive better in difficult habitats(Wilson et al. 1991; Sprent, pp176-182; Mason &Wilson, pp165-175). Vesicular—arbuscular (VA)mycorrhizal fungi perform a range of functions,including enhancement of mineral nutrient uptake(particularly phosphorus) by the plant rootsystem (Harley & Smith 1983). Under tropical

216

conditions, little is known about the occurrenceand distribution of VA mycorrhizal fungi, but theyare considered essential for the growth of manygrain crops, forage legumes, pasture plants, foresttrees and horticultural plants (Janos 1987). Indegraded arid areas, there may be a need toenhance the mycorrhizal and rhizobial symbiosesif, as is likely the processes of degradation havehad detrimental effects on the soil inoculum.

In Kenya, observations of spore numbers andtypes were made around naturally occurringAcacia tortilis trees at Olorgasailie, in the RiftValley Samples were taken in April 1989, at theend of the wet season, from the top 15 cm layerof soil along transects from the boles of A. tortilistrees outwards to bare ground. The trees were ofvariable size and unknown age, and transectlengths were varied according to tree size so asto accommodate variation in the areas of canopyand rooting zone influence.

Following extraction from the collected soilsamples, the numbers of live and dead sporeswere counted and types were distinguished.rIWelve types were found, including Acaulosporascrobiculata, Entrophospora sp., Scutellospora sp.,Glomus geosporum, six other Glomus species,and two unknown types (J Wilson, unpublisheddata). This is a reasonably diverse population.Total numbers of spores were 5-6 times greaterclose to the trees than at a distance (Table 5).Live spores, which probably represent animportant component of the soil inoculum indegraded sites (see Mason & Wilson,pp165-175), outnumbered dead spores in eachsample, and the ratio of live/dead spores was1.6-2.7 close to the trees, while it was lower(1.2-1.3) away from the trees. The presence ofhigher spore numbers with a greater proportionof viable spores close to trees may be a directeffect of their root systems or an indirect effectof soil temperature, moisture, chemistry or otherfactors.

Table 5. Distribution of live and dead VA mycorrhizal sporesin soil samples taken close to and on bare ground at adistance from naturally occurring Acacia tortilis trees atOlorgasailie, Kenya (results are numbers per 50 g fresh massof soil)

Live

Tree A

Dead Total Live

'Free B

Dead Total

In Kenya two trials have recently shown thebenefits of nursery inoculation on tree survival insemi-arid lands (Wilson et al. 1991). Inoculationof four tree species was examined: Acaciatortilis, Prosopis juliflora, Termihalia brownii andTerminalia prunioides (Wilson et al. 1991).Pre-germinated tree seedlings were inoculated

with root and adhering soil inoculum ofendomycorrhizal fungi cultured on maize andcowpea. The leguminous trees were alsoinoculated with rhizobia. After growth in anursery they were planted in two successive wetseasons at two sites, Marirnanti and Olorgasailie,both semi-arid but with annual rainfalls of -847 mm and 476 mm, and soil pH of 6.5 and8.1, respectively. The percentage of seedlingssurviving at each site 30 weeks after planting isshown in Table 6.

Table 6. Pementage survival of inoculated (+S) anduninoculated (-S) plants 30 weeks after outplanting, andimprovement in survival (% gain) as a result of inoculationwith VA mycorrhizal and with rhizobia for leguminous treespecies. Experiment at two semi-arid sites in Kenya (databased on 90 plants per treatment of Acacia and Prosopis and45 plants for the Tbrminalia spp. in each experiment)

* Leguminous tree

Planting date

Inoculation generally improved plant survival,although effects varied with both season andspecies. Substantial effects of inoculation on treegrowth were also observed in the nursery priorto outplanting. Observations on theseexperiments and others indicate that benefit frommycorrhizal inoculation is most marked whenroots are more than 70% mycorrhizal at planting(Wilson et al. 1991) and when environmentalconditions are most stressful.

Irrespective of the cause, the presence of agreater quantity of inoculum in the soil closeto trees may lead to higher rates ofmycorrhizal infection in crops grown amongtrees than in those grown on their own,provided that the types of inoculum arecompatible. The dynamics of VA mycorrhizalpopulations in agroforestry systems areunknown, although the cycle of soil andvegetation disturbance is likely to have adverseeffects. Agricultural crops can benefit

substantially from inoculation with VA mycorrhizalfungi (Howeler, Sieverding & Saif 1987). Whileinoculation of agricultural crops is resliicted bythe large quantities of inoculum required,inoculation of nursery-grown trees intended foragroforestry systems may be worthwhile,particularly in stressful environments. Because ofthe wide host ranges of these fungi, and thediversity of fungal species occurring with trees,the tree roots may sustain inocula which are alsoof value to the intercrop (Wilson et al. 1991).Inoculation of trees, therefore, may . have twobenefits:

i. improvement of tree survival and growth, and

maintenance of populations of mycorrhizalspecies suitable for intercrops in the periodbetween harvesting one intercrop and sowingthe next.

Past experience would indicate that, if anadequate heterogeneous VA mycorrhizalpopulation is present, crops should have noproblem in finding sufficient compatible fungi forgrowth (Sieverding 1991). However, moreextensive observations are needed to confirmthis effect of trees and to examine the effects onintercrops.

Rhizobia and nitrogen fixation

A large programme of research is in progress onnitrogen-fixing trees at the Kenya ForestryResearch Institute (KEFRI). The work entailsisolation, characterisation and screening for theeffectiveness of indigenous rhizobial strains, foundin association with selected N-fixing trees. Thisprogramme is also developing appropriateinoculation techniques for tree seedlings, forenhanced nodulation, nitrogen fixation and treeestablishment (D Odee, personal communication).As already mentioned, microsym.biont inoculation(VA mycorrhizas and rhizobia) has improved treeestablishment on degraded sites (Table 6). Incollaboration with the Department of BiologicalSciences, University of Dundee, KEFRI is alsoengaged in programmes to map and testindigenous tree rhizobial flora across the variousecoclimatic zones of Kenya; and to test inoculantsfor Acacia spp. of East and southern Africa.Further studies are planned to determine thecompatibility of Kenyan rhizobial strains withAustralian Acacia spp., when grown in Kenya(J Sutherland, personal communication).

CONCLUSIONS

The domestication of trees for the production ofwood and other products in semi-arid areas ofEast Africa is at an early stage. A wide range ofexotic and indigenous species have been testedfor their suitability In general, indigenous treeshave shown better survival but slower growththan exotics, especially the Australian acacias

217

(see Booth & Turnbull, pp189-194). To whatextent this better survival of indigenous speciesmay be related to their symbiotic relationshipswith soil microflora is not known, but it seemsclear that inoculating trees with root symbiontscan greatly enhance tree establishment. Followingon from species selection, a start has beenmade on provenance selection in the Africanacacias and the development of vegetativepropagation techniques so that the potential ofclonal approaches can be evaluated.

ACK1VOWLEDGEMENTS

We wish to thank the Kenya Forest ResearchInstitute and the National Museums of Kenya forprovision of facilities and assistance atOlorgasallie, the EML project for assistance atMarimanti, and the staff of the AMSAL project fortheir contribution to the experimental work. TheUK Overseas Development Administration isgratefully acknowledged for the provision offunds (R4319 and R4730). The Australian Centrefor International Agricultural Research is thankedfor providing funds to allow the senior author toattend this Conference.

REFERENCES

Armstrong, G.D. & Lugadiru, J.K. 1986. Embu-Meru-IsioloProgramme, Kenya. Forestry Project, Silviculture] ResearchProgramme 1983-86. Surbiton: Land Resources DevelopmentCentre.

Callaghan, T.V., Lindley, D.K., Ali, 0.M., Abd el Nour,H. & Bacon, P.J. 1989. The effect -of water-absorbingsynthetic polymers on the stomatal conductance, growth andsurvival of transplanted Eucalyptus microtheca seedlings in theSudan. journal of Applied Ecology, 26, 663-672.

Chege, F. & Stewart, M. 1991. Early results from trials ofAustralian acacias in semi-arid Kenya. In: Advances in tropicalAcacia research, edited by J.W TUrnbull, 201-205. Canberra:ACIAR.

Chirchir, R.K. & Stewart, M. 1989. Hill afforestation: casehistory of Kuani Hill, Meru, Kenya. Embu: Embu-Meru-IsioloArid and Semi-arid Lands Programme.

Dick, J.McP. & East, K. 1992. The influence of light qualityon growth and rooting of Acacia tortilis, a semi-arid treespecies of East Africa. Nitrogen Fixing Tree Research Reports,10, 97-101.

Dick, J.McP., East, K. & Leakey, R.R.B. 1991. Influenceof tropical environment and cutting length on the rooting ofProsopis juliflora (Swartz) DC. Nitrogen Fixing Tree ResearchReports, 9, 114-116.

Fagg, C.W. & Barnes, R.D. 1990. African acacias: study andacquisition of genetic resources. Oxford Forestry InstituteAnnual Report 1990, 26-27.

Fagg, C.W, Barnes, R.D., Gourlay, I.D. & Lockhart,L.A. 1990. African acacias: study and acquisition of geneticresources - phase 2. Oxford Forestry Institute Annual Report1990, 27-28.

Harley, J.L. & Smith, S.E. 1983. Mycorrhizal symbiosis.London: Academic Press.

Howe ler, R.H., Sieverding, E. & Saif, S. 1987. Practicalaspects of mycorrhizal teclmology in some tropical crops andpastures. Plant and Soil, 100, 249-283.

218

Janos, D.P. 1987. VA mycorrhizas in humid tropicalecosystems. In: Ecophysiology of VA mycorrhizal plants, editedby GR. Safir, 107-134. Boca Raton, Florida: CRC Press,

Kimondo, J.M. 1991. Early growth of lesser-knownAustralian acacias at Loruk, Kenya. In: Advances in tropicalAcacia research, edited by J.W Tlirnbull, 206-208. Canberra:ACIAR.

Konuche, P.K.A. & Milimo, P.B. 1989. Future needs ofASAL forestry research in Kenya. In: Proceedings of theWorkshop on Setting National Forestry Research Priorities inKenya, 1989, 30-32. Nairobi: Kenya Forestry ResearchInstitute.

Leakey, R.R.B. 1985. The capacity for vegetativepropagation in trees. In: Attributes of trees as crop plants,edited by J.E. Jackson & M.G.R. Carmell, 110-113. AbbotsRipton, Huntingdon: Institute of Terrestrial Ecology

Leakey, R.R.B. 1987. Clonal forestry in the tropics - areview of developments, strategies and opportunities.Commonwealth Forestry Review 66, 61-75.

Leakey, R.R.B. 1991. Towards a strategy for clonal forestry:some guidelines based on experience with tropical trees. In:Tree breeding and improvement, edited by J.E. Jackson, 27-42.Ti-ing: Royal Forestry Society of England, Wales and NorthernIreland.

Leakey, R.R.B., Mesén, J.F., Tchoundjeu, Z., Longman,K.A., Dick, J.McP., Newton, A., Matin, A., Grace, J.,Munro, R.C. & Muthoka, P.N. 1990. Low technologytechniques for vegetative propagation of tropical trees.Commonwealth Forestry Review 69, 247-257.

Libby, W.J. 1973. Domestication strategies for forest trees.Canadian Journal of Forest Research, 3,..265-277.

Lockhart, L.A., Fagg, C.W. & Barnes, R.D. 1990.Evaluation of the genetic resources and study of reproductivebiology of Faidherbia albida, with a view to improving itseffectiveness in agroforestry systems. Oxford Forestry InstituteAnnual Report 1990, 29-30.

Milimo, P.B. 1987. Tree seed problems constraining thedevelopment of dryland afforestation systems in Kenya. In:Proceedings of the International Symposium on Forest SeedProblems in Africa, edited by S.K Kamra & AR. Ay ling,170-176. Uppsala: Sveriges Lantbruksuniversitet, Institutionenfor Skoglig och Vaxtfysiologi.

Milimo, P.B. & Helium, A.K. 1989a. Dormancy in seeds ofMelia volkensii. East African Agriculture and Forestry journal,54 (3), 111-122.

Milimo, P.B. & Hellum, A.K. 1989b. Studies of thestructure and development of seeds of Melia volkensii Gurke.East African Agriculture and Forestry Journal, SS (3), 27-36.

Newton, A.C., Munro, R.C. & Wambugu, T.W. 1993. Thewater status of Terminalia brownii Fresen seedlings duringestablishment on semi-arid sites in Kenya. journal of AridEnvironments. In press.

Newton, A.C., Muthoka, P.N. & Dick, J.McP. 1992. Theinfluence of leaf area on the rooting physiology of leafy stemcuttings of lbrminalia spinosa Engl. Trees - Structure andFunction, 6, 210-215.

Prior, J. & Cutler, D. 1992. Trees to fuel Africa's fires. NewScientist, 135 (1836), 35-39.

Sanchez, RA. & Salinas, J.G. 1981. Low-input technologyfor managing oxisols and ultisols in tropical America.Advances in Agronomy, 34, 279-406.

Sanginga, N., Mulongoy, K. & Swift, J.J. 1992.Contributions of soil organisms to the stability andproductivity of cropping systems in the tropics. Agriculture,Ecosystems and Environment, 41, 135-152.

Shakacite, 0. 1987. Seed problems in Zambia. In:Proceedings of the Internadonal Sympoum on Forest SeedProblems in Africa, edited by S.K Kamra & AR Ay ling,263-272. Uppsala: Sveriges Lantbruksuniversitet, Institutionenf6r Skoglig och Vaxtfysiologi.

Shehaghilo, I.M. 1987. Some studies on the collection andextraction of Cupressus lusitanica seed at Lushoto, Tanzania. In:Proceedings of the International Symposium on Forest Seed

Problems in Africa, edited by S.K Kamra & AR. Ay ling,280-286. Uppsala: Sveriges Lantbruksuniversitet, Institutionenfor Skoglig och Vaxtfysiologi.

Sieverding, E. 1991. Vesicular-arbuscular mycorrhizamanagement in tropical ecosystems. Eschborn: TechnicalCo-operation, Federal Republic of Germany

Specht, C.E. & Schaefer, C. 1990. Interim recommendationsfor the presowing treatment of Terminalia brownii andTerminalia spinosa. (Technical Note, no. 9.) Nairobi: KenyaForestry Research Institute.

Torquebiau, E. 1992. Are tropical agroforestry home gardenssustainable? Agriculture, Ecosystems and Environment, 41,189-207.

Vandenbeldt, R.J. 1991. Rooting systems of western andsouthern African Faidherbia albida (Del.) A. Chey (syn. Acaciaalbida Del.) - a comparative analysis with biogeographicimplications. Agroforestry Systems, 14, 233-244.

Wanyanja, J.M., Mills, W.R. & Gwaze, D.P. 1991. Geneticvariation in Acacia albida and its agroforestry potential inZimbabwe. (Paper presented at a Confetence on AgroforestryResearch and Development in the Miombo Ecozone ofSouthern Africa, Lilongwe, Malawi, 1991.) (Unpublished.)

Wate, P.A. 1987. Seed problems in Mozarnbique. In:Proceedings of the International Symposium on Forest SeedProblems in Africa, edited by S.K Kamra & AR Ay ling,378-379. Uppsala: Sveriges Lantbruksuniversitet, Institutionenfor Skoglig och Vaxtfysiologi.

Wilson, J., Munro, R.C., Ingleby, K., Mason, P.A.,Jefwa, R., Muthoka, P.N., Dick, J.McP. & Leakey,R.R.B. 1991. Agroforestry in semi-arid lands of Kenya - roleof mycorrhizal inocula tion and water retaining polymer. ForestEcology and Management, 45, 153-163.

Wood, P.J. 1989. Report on forestry research in Embu, Meruand Isiolo Project area. (ODA consultancy report.)(Unpublished.)

Zobel, B. & Talbert, J.A. 1984. Applied forest treeimprovement. New York: Wiley

Zumer-Linder, M. 1983. Re-assessment of the Turkanaforestry project. (Report prepared for the Kenya ForestryDepartment, Forestry Research Organization, KAR1, and theNorwegian Agency for International Development.)(Unpublished.)

219

Domestication potential of indigenous fruit trees ofthe miombo woodlands of southern Africa

A Maghembel , F Kwesiga 2 , M Ngulube 3 , H Prins' &F M Malaya411nternational Centre for Research in Agroforestry, SADCC/ICRAF Agroforestry Project, Makoka AgriculturalResearch Station, PO Box 134, Zomba, Malawi2International Centre for Research in Agroforestry, Zambia/ICRAF Agroforestry Project, PO Box 510046,Chipata, Zambia3Forestry Research Institute of Malawi, PO Box 270, Zomba, Malawi4Division of Forest Research, PO Box 22099, Kitwe, Zambia

ABSTRACT

This paper presents results of various studies on the identification, propagation and growth of fruit trees indigenous or naturalisedto the miombo ecozone of southern Africa. In 1989, ethnobotanical surveys were conducted in eastern Zambia, western Tanzaniaand Malawi to identify the range, uses and management of undomesticated trees in farmersfields,and around their homesteads.During these surveys, over 50 indigenous tree species were identified by farmers as providing edible fruits. Those mentioned inmore than one country included Adansonia digitata, Annona senegalensis, Azanza garckeana, Flacourtia indica, Strychnos spinosa,Uapaca kirkiana, Sclerocarya caffra, Tamarindus indica, Grewia bicolor Parinari curatellifolia, Diospyros mespiliformis, Syzygiumcordatum, Rothmannia englerana, Ximenia caffra, Vangueria infausta, Ziziphus mauritiana, Trichilia emetica, Par Ida filicoidea, Bauhiniapetersiana, Canthium crassum and Bride lia micrantha.

Considerable information is available on the nutritional attributes of many of these species but little is known about propagationand field performance following artificial regeneration. A series of studies are in progress in Malawi and Zambia to address thisgap in lmowledge as a 'first step towards the conservation and domestication of this valuable resource.

INTRODUCTION

More than 50 indigenous or naturalised treespecies of the miombo ecosystem of southernAfrica bear edible fruits which are importantsources of vital nutrients (Food and AgricultureOrganisation 1983; Fan,shawe 1972; Saka &Msonthi 1993). A few of these, such asAdansonia digitata, Bauhinia petersiana, Grewiaflavescens and Tamarihdus indica, serve as foodreserves during seasonal food shortages and intimes of famine (see also Campbell 1987).Studies on the nutritional value of indigenousfruits show many to be rich in sugars, essentialvitamins, minerals, vegetable oils and proteins(Saka, Msonthi & Sarnbo 1989; Wehmeyer 1966;Hans, Parker & Kumar 1978; see also Table 1).Thus, Annona senegalensis and Trichilia emeticafruits are important protein supplements, while-Strychnos spinosa, Azanza garckeana and Parinaricuratellifolia have on average high crude fat(31%), crude fibre (45%) and total carbohydrates(88%) (Saka & Msonthi 1993).

In spite of their importance, however, indigenousfruit trees of the miombo ecozone have not beensubject to agricultural or forest policies(Campbell 1987). There has been limited interestin their conservation, propagation andimprovement. Farmers are interested in

220

indigenous fruit trees only to the extent that theycan extract edible fruits and medicines. Beyondthat, the only other effort made is to select themore valuable fruit trees before clearfelling themiombo forest prior to cultivation.

Domestication, or even the planting of wild fruittree seedlings, is not common. For example, inthe Condo area of Zimbabwe, only nine out of102 households had planted a wild fruit tree(Campbell 1987). In Zambia, most indigenousfruit-bearing tree species are left standing duringland clearing, with a view to continue providinga food supplement to the diet in the form offruit. Wild fruit tree cultivation is not practisedtraditionally In eastern Zambia, none of the 112farmers interviewed had ever planted anindigenous fruit tree (Kwesiga & Chisumpa 1992).

It is clear, therefore, that there is need for aconcerted effort towards understanding whyfarmers do not domesticate indigenous fruit trees,and to encourage this process by developingstrategies for domestication of this valuableresource through tree improvement programmes.Failure to do so may lead to loss of geneticdiversity of the indigenous fruit trees in themiombo ecozone. This paper presents results ofethnobotanical surveys, nursery and fieldperformance in parts of the miombo ecozone.

Table I. Nutritional value of some indigenous fruits (source: Hans et aL 1978)

DETAILS OF RESEARCH PROGRAMMES

The International Centre for Research inAgroforestry (ICRAF), through the southern AfricaAFRENA (Agroforestry Research Networks forAfrica) and the collaborating national institutionsin Malawi, Tanzania and Zambia, undertookethnobotanical studies in 1989 with the followingobjectives:

• to explore what uses the farmers actuallymake of indigenous trees;

• to incorporate indigenous 1mowledge of themanagement and conservation of themiombo woodland through new or improvedagroforestry systems;

• to provide a basis for research onidentification, selection, propagation and treeimprovement strategies of the mostcommonly preferred indigenous trees.

In addition, propagation studies and field plantingof selected indigenous fruit trees were initiatedby both the Forestry Research Institute of Malawi(FRIIVI) and the Southern Africa DevelopmentCoordination Conference (SADCC)/InternationalCentre for Research in Agroforestry (ICRAF)programme on the improvement of multipurposetrees (MPTs) at Makoka, Zomba, in Malawi. Theobjectives of these studies were to:

• develop or simplify propagation techniquesfor native fruit trees;

• raise suitable planting stock of native fruittrees;

• evaluate growth and fruit yield from these

6

5

4

3

2

fruit trees after artificial regeneration in thefield.

In Zambia, the Department of Forestry startedresearch on the domestication of wild trees asfar back as the 1930s, but the emphasis thenwas on timber-producing species. It is onlyrecently that fruit trees have been included inthe research programme. Since 1976, researchwork has been conducted on the domesticationof various indigenous timber and fruit treespecies at Chati Forest Research Station, 40 kmwest of Kitwe. The results so far show that fruittrees in general are growing faster than mosttimber species (Figure 1).

FruitO Parinari curatellifolia

Strychnos cocculoidesTimberO Pterocarpus angolensis• Afzelia quanzensis

1 2 3 4 5 6 7 8 9 10Age (yrs)

Figure I. Comparison between growth of indigenous fruittrees and timber trees in a field nial at Chati Forest Research

Station, Zambia. 'flee species were allocated randomly to plots

of approximately 28 m x 28 m, and were direct sown at the

onset of the rainy season

221

Table 2. Common indigenous fruit trees of the miomboecozone of southern Africa (source: Kwesiga & Chisumpa1992; Maghembe & Seyani 1991; Karachi et al. 1991;Campbell 1987)

Species

Adansonia digitata

Adenia senensis

Annona cherimolia

Annona senegalensis

Azanza garckeana

Bauhinia petersiana

Berchemia discolor

Borassus aethiopum

Bridelia cathartica

Bridelia micrantha

Canthium crassum

Canthium burtii

Canthium huillense

Diospyros mespiEformis

Ficus vallis—choudae

Ficus sycomorus

Flacourtia indica

Friesodielsia obovata

Garcinia smeathmanni

Grewia biscolor

Hexalobus englerane

Hexalobus monopetalus

Lannea discolor

Lannea edulis

Manilkara mochisia

Parinari curatellifolia

Parkia filicoidea

Piliostigma thonningii

Ricinodendron

rautanenii

Rothmannia englerana

Rothmannia fischeri

Sclerocarya caffra

Sclerocarya birrea

Strychnos cocculoides

S. pungens

S. innocua

S. madagascariensis

S. spinosa

Syzygium

cordatum/guineense

Syzygium guineense

ssp. afromontantim

Tamarindus indica

Uapaca kirkiana

Uapaca nitida

Vangueriopsis

lancillora

Vangueria infausta

Vitex doniana

Vitex iringensis

Vitex. mombassae/payos

Ximenia americana

Ximenia calfra

222

EasternZambia

Wholeof Western

Malawi Tanzania

X

X

Communalareas

of Zimbabwe

X

-

x

X

ETHNOBOTANICAL SURVEYS

The results of ethnobotanical surveys in Malawi(Maghembe & Seyani 1991), the Tabora region inTanzania (Karachi et al. 1991), the easternprovince in Zambia (Kwesiga & Chisumpa 1992)and in the Condo area of Zimbabwe (Campbell1987) have shown that there are over 50 knownindigenous trees with edible fruit in the miomboecozone (Table 2). However, these fruit trees arestill largely in the wild because no systematicefforts have been made to domesticate themthrough selection and planting, and very limitedscientific effort (Mwarnba 1989) has beenchannelled towards understanding theirreproductive biology growth rates and geneticvariation. This neglect has resulted in the lack ofa genetic conservation strategy for these species.

As a result of burning and clearfelling foragricultural expansion to meet the needs of thefast-growing population of southern Africa,genetic erosion of native fruit trees has probablyoccurred in the miombo region. A narrow rangeof exotic fruit trees (mangoes, guavas, pawpaws,avocado, etc) has been promoted by extensionagencies, resulting in further neglect ofdomestication, utilisation and marketing ofmiombo tree species.

GERMINATION OF FRESHLY COLLECTEDSEED

Germination studies were initiated with freshlycollected seed (Maghembe & Prins 1993) atthe SADCC/ICRAF' nursery at Makoka inZomba, Malawi, using 17 indigenous fruit treespecies (Table 3). Seeds were subjected toeight treatments under conditions that can berepeated at village level (see Table 4) involvingno soil sterilisation and no chemical treatments.Seeds of most species were collected inAugust and December 1989 from an areabetween Namadzi (15°34'S latitude, and 36°13'Elongitude) and Mangochi (14°34'S latitude,35°16'E longitude) in Malawi. The seeds werestored at ambient temperature and sown withinthree weeks of collection. For each treatment,30 seeds were used in a randomised completedesign, replicated three times. Although theInstitute for Seed Testing Association (ISTA)recommends 100 seeds per replicate, it wasnot possible in this case.

Germination tests were carried out in transparentplastic containers (20 cm x 10 cm x 7.5 cm)filled with sieved river sand and kept moist witha sprinkler. During experimentation, temperaturesranged from 15.5°C to a maximum of 37°C.Germination was recorded daily and germinantswere removed and transplanted into polythenetubes. Observations continued for 45 days. Aftergermination studies, seedlings were grown atboth sites (FRIM and SADCC/ICRAF) in a soil

mixture of forest soil and river sand in a 50:50 ratio,thereby ensuring that local micruflora and rootsymbionts were present. The seedlings werewatered twice a day and root-pnmed weekly

At least one of the simplest treatments, such asdirect sowing (T1), soaking (T2) and cleaningand soaking (T3), resulted in over 80%germination in all species, except Azanzagarckeana, Canthium foetidurn, Flacourtia indica,Parinari curatellifolia and Vangueria infausta(Table 4). For Azanza garckeana and Canthiumfoetidum, high germination rates (100% and 93%,respectively) were achieved by chipping (T7).Flacourtia indica and Parinari curatellifoliagerminated poorly in all treatments.

Table 3. Some information on occurrence, distribution, tree form and fruits of the tree species tested at Makoka, Malawi

Species

Azanza garckeana

Br/delia carthartica

Bridelia micrantha

Canthium foetidurn

Diospyros mespiliformis Ebenaceae

Flacourtia indica

Parinari curatellifolia

Parkia filicoidea

Piliostigma thonningii

Strychnos spinosa

Syzygium cordatum

Tamarindus indica

Tbrininalia catappa

Uapaca kirkiana

Vangueria infausta

Ziziphus abyssinica

Ziziphus mauritiana

It must be emphasised that conclusions drawnfrom these studies are based on the use ofseeds stored for not longer than three weeks atambient temperature. With Uapaca kiridana,

completely different results can be experiencedwhen using older seeds. In nature, the fruits ofUapaca kirkiana ripen at the beginning of therainy season, facilitating germination of the young

seedlings. Seeds stored for long periods in coldroom conditions show poor germination, andstudies in Malawi (Ngulube & Kananji 1989)found that viability of carefully stored seeds of

Uapaca kirkiana was lost within a month ofcollection.

Family Characteristics

Malvaceae In all types of woodland, sea level to 1700 m. Bushy shrub to medium-sized tree,3-10 m height. Fruits Feb-Sep, spherical, 2-4 cm. Sweet mucilage comes outwhen chewed

Euphorbiaceae Stream banks, riverine fringe thicket and littoral scrub. Multistemmed shrub ortree, 4-6 m. Fruits Apr-Sep, 1 cm

Euphorbiaceae Riverine forest, relic forest and open woodland. Small- to medium-sized tree, 7-15m. nulls Jan-Mar, ellipsoidal, 0.5-1 cm. Taste like blackcurrants

Rubiaceae Medium altitudes. Open woodland. Small tree, 3-5 m. Fruits Jul-Oct, 1 cm. Ediblebut slightly astringent

Low to medium altitudes. Woodlands, along rivers, rarely on rocky hill slopes.Medium-sized tree, 10-15 m, with dense, rounded crown giving excellent shade.Fruits Apr-Sep, ovoid, fleshy 2.5 cm. Eaten fresh and preserved

Flacourtiaceae Woodland, from sea level to 1600 m. Thorny shrub or small tree, 3-5 m. FruitsApr-Sep, berry-like, fleshy up to 2.5 cm. Sweet/acid

Chrysobalanaceae Open deciduous woodlands on sandy soils, 500-1200 m. Large tree, up to 21 m,with a heavy, rounded crown. Evergreen. Fruits Oct-Jan, oval, 5 cm, with astringentpulp and three edible seeds

Leguminosae Riparian forests, 500-1200 m. Large tree, up to 25 m, with spreading,umbrella-shaped crown. Fruits Nov-Feb; pods (15-30 cm) contain sweet, orangefloury pulp

Leguminosae Woodland and wooded grassland, medium to low altitude, 300-1200 m. Small- tomedium-sized tree, 3-5 m height. Fruits Jun-Sep; dry pods (pulp) and seeds asfamine food (10-25 cm)

Loganiaeae From sea level to 1500 m. Open woodland and riverine fringes. Shrub ormedium-sized tree, 1-9 m height. Fruits Mar-Aug, spherical, with woody shell,edible pulp, 8-15 cm

Myrtaceae Medium to high altitude. Riverine thickets and forests; always near water or alongwatercourses. Medium-sized tree, 8-15 m in height. Fruits Nov-Mar, ovoid, 1.5 cmlong, fleshy slightly sweet/acid

Leguminosae Low-altitude woodland, wooded grassland and bush along rivers. Medium to largeevergreen tree (20-24 m) with a dense crown. Fruits Jun-Jul; pods with stickyedible pulp, acid •

Combretaceae Introduced from Middle East and now naturalised. Low altitudes. Medium to largeevergreen tree (8-15 m). Fruits Jul-Sep; stringy edible pulp with edible nut, 10-15cm long

Euphorbiaceae Medium altitude, 500-1300 m. Open woodland, dominant on gravelly soil. Mediumto small tree (5-10 m), with open rounded crown. Fruits Oct-Dec, spherical,2-3.5 cm

Rubiaceae Medium altitude. Wooded grasslands, among rocks and sand dunes. Smalldeciduous tree (3-7 m) with open crown. Fruits Feb-Apr (Aug), spherical,2-3.5 cm, sweet, edible pulp

Rhamnaceae Medium to low altitudes. Open woodland, wooded grassland and along riverbanks. Shrub or small tree, up to 7 m. Fruits Jun-Sep, spherical, 1-3 cm, edible,slightly acid

Rhamnaceae Medium to low altitudes. Small tree, 7-12 m. Open woodland, wooded grassland.Fruits Jun-Sep, edible, slightly acid, also used for distilling spirit, 1-3 cm

223

nble 4. Mean cumulative germination (%) of freshly collected seed studies at SADCC/1CRAF nursery Zomba, Malawi

Species

Azar2za garckeana

Bridelia cathartica

Bride lia micrantha

Canthium foetidum

Diospyros mespiliformis

Flacourtia indica

Parinari curatellifolia

Parkia filicoidea

Piliostigma thonningil

Strychnos spinosa

Syzygium cordatum

Tamarindus indica

Tbrminalia catappa

Uapaca ldrldana

Vangueria infausta

Ziziphus

Ziziphus mauritiana

T1 T2 T3

Strychnos spinosa is a species known to bedifficult to germinate (Ngulube & Kananji 1989).Its seed coat may contain a germination inhibitorwhich is only deactivated by proper washing.Such a strategy would allow germination onlyafter heavy rains when enough moisture isavailable to secure suitable growing conditions.

GERMINATION OF SEED SUBJECT TOCOLD STORAGE

Seeds of 13 indigenous fruit tree species werecollected locally in 1988 by the FRIM (Table 5).The seeds were kept under cold storage(4-10°C) for six months before they were tested

224

Treatments1

T4 T5 T6

Table 5. Seed sources of the indigenous fruit trees tested at Naungti, Machinga

Leastsignificant

F difference

T7 T8 probability (P<0.05)

*NS, not significant

1T1, Direct sowing of seed after removing the fruit pulp; T2, Soaking seeds in cold water for 24 hours after removing the fruitpulp; T3, Cleaning seeds thoroughly ie removal of all mesocarp carefully and soaking at room temperature for 24 hours;T4, Cleaing seeds and placing in warm water at 65°C, and left to cool for 24 hours; T5, Cleaning seeds and placing in hotwater at 100°C and leaving to cool for 24 hours; T6, Cleaning seeds and placing in hot water at 100°C, allowing to boil for oneminute then leaving to cool for 24 hours; T7, Scarification, ie making a hole in the seed coat, through abrading or nicking;T8, Complete removal of the seed coat

for germination. Six pre-treatments, including acontrol, were applied to the seeds, includingmechanical hand scarification, soaking in coldwater and soaking in boiling water (see Table6). After pre-treatment, the seeds were nurserysown into black polythene tubes (10 cm x 15cm) filled with woodland soil. Treatments werearranged in blocks each containing 100 seeds.The plants were watered twice a day (morningand evening) for 30 days during whichgermination counts were made daily

Untreated seed of Adansonia digitata, Tamarindusindica and Ximenia caffra germinated successfullyduring the 30-day experimental period. The

17 40 43 23 0 0 100 63 0.001 17

93 75 73 0 0 90 - 0.001 25

0 93 60 - 0 - 73 0.001 20

10 47 43 13 0 0 93 87 0.001 39

80 60 60 0 3 87 0.001 22

20 0 7 - - 3 13 3 0.026 19

0 - 17 13 0.290 NS*

100 77 3 13 97 0.001 42

50 90 77 10 - - 0.001 31

80 - - 0 0 0 0.001 13

0 93 27 - - - 27 0.007 44

100 93 - - 87 0.444 NS

70 97 17 - - 0.008 37

43 93 - - - 87 100 0.021 34

- - - 40 -

100 80 87 20 0 - - 0.001 36

100 87 100 0 0 - 0.001 16

Species Locality Latitude Longitude Elevation (m)

Adansonia digitata Shire valley Nsanje 16°43' 35°00' 200

Azanza garckeana Liwonde, Machinga 15°05' 35°25' 700

Strychnos spinosa Liwonde, Machinga 15°05' 35°25' 700

Tamarindus indica Shire valley Nsanje 16°43' 35°00' 200

Ziziphus mauritiana Liwonde, Machinga 15°05' 35°25' 700

Uapaca kirkiana Malosa, Machinga 15°10' 35°25' 900

Uapaca nitida Chongoni, Dedza 14°15' 30°15' 1300

results also indicate that mechanical scarificationis effective with Adansonia digitata, Uapacakirkiana, Ximenia caffra and Ziziphus mauritiana(Table 6). Cold and hot water treatments wereonly effective with Adansonia digitata and Ximeniacaffra, but only IL digitata responded to hotwater. Germination with the rest of the specieswas very poor: Flacourtia indica, Parinaricuratellifolia, Parkia filicoidea, Strychnos spinosaand S. innocua did not germinate at all,regardless of pre-treatment. Boiling water did notenhance germination in any species, andprevented it in most instances. Observations onthe development of Uapaca kirkiana seedrevealed that the mature seed contains primordialleaves rather than fully developed cotyledons. Itis possible that the cotyledons are killed whenthe seed is exposed to heat. Seed qualityassessments indicated that around 90% and 12%of the seeds of Flacourtia indica and Uapacakirkiana, respectively were empty whichobviously affected the overall germinationsuccess, regardless of pre-treatment procedures.

The study has shown that, although it is possibleto collect, germinate and store some indigenousfruit trees, there are difficulties with most species.These germination problems are possibly due towrong timing of seed collection periods, poorprocessing and storage procedures, and lack ofproper pre-treatment methods. It is apparent thatinformation on the reproductive biology of mostMalawian and indeed southern Africanindigenous fruit trees is lacking. Such informationis very important for the development of suitablepropagation techniques for successfuldomestication programmes. It is, therefore,recommended that detailed biological studies

7'able 6. Germination response (%) of indigenous fruit tree seeds following pre-treatment at FRIM, Zomba

1Immersed for 30 s then sown2Immersed in boiling water and allowed to cool for 12 h

(including vegetative propagation) be undertakento address the above aspects on selectedspecies commonly found in traditional fanningsystems.

FIELD PLANTINGS AT MAKOKA ADMNAUNGU, MACHINGA

The fruit trees at Makoka were assessed in June1992 (ie alter 27 months' growth), whereas thoseat Machinga (PRIM) were assessed after 18months. Survival, height, root collar diameter,crown diameter and fresh mass of fruits(Makoka) were assessed. Data were based on allsurviving trees at Makoka; at Machinga, 25 treesper plot were assessed. All data were subjectedto analysis of variance using GENSTAT 5 (Payneet al. 1987).

At the SADCC/ICRAF site at Makoka, seedlingswere planted out in February 1990, except forVangueria infausta which was planted in January1989. The experiment was arranged in arandomised complete block design (RCBD) with19 accessions replicated three times. Each treeaccession constituted a line plot of 2.25 m x14 m, with seven trees planted on the ridge 2 mapart. The plots were kept free of weeds, and nofertilizers or herbicides were applied.

All fruit tree species reported survival above90%, except Cordyla africana at 85% (Table 7).Differences in height were quite prominent. Thefastest-growing species was Ziziphus mauritiana,which achieved a mean height of more than4 m, followed by Cordyla africana (3.62 m),Bride lia micrantha (3.43 m), B. cathartica (3.31 m)and Parkia filicoidea (3.23 m). Both Ziziphusspecies showed good growth and heavy

225

Table 7. Performance of indigenous fruit trees at 27 months after planting at Makoka, Malawi

2Calculated as the fresh mass of fruits per tree

226

branching with thorns, suggesting their potentialfor live fencing. In addition, the species providefodder from their leaves, and their flowers arerich in pollen and nectar and are relished bybees.

Azanza garckeana, Bridelia cathartica, Ziziphusabyssinica, Z. mauritiana, Ficus vallis-choudae andVangueria infausta all produced fruits within twoyears after planting. For Azanza garckeana, thisresult was rather unexpected: this tree grows toover 10 m in height at maturity yet the youngtrees produced fruits at 1.6 m tall. This earlyfruiting provides an opportunity for farmers togrow the tree and harvest fruits continuously asthe tree matures. Terminalia catappa flowered butproduced no fruits, perhaps a result of the severedrought gripping southern Africa at the time.

Seedlings of seven indigenous fruit tree species(Table 7) were tested at Naungu forestryresearch experimental site in Machinga (15°24'Slatitude and 35°21'E longitude, 650 m asp. Thesite receives a mean annual rainfall between500 mm and 800 mm. Seedlings were planted inan RCBD at 2 m x 2 m in January 1991.However, because of variation in numbers ofseedlings available, the size of blocks varied forthe seven species.

Tamarindus ihdica and Ziziphus mauritiana showedexcellent survival (>90%), followed by Adansonia

digitata and Azanza garckeana (>70%) (Table 8).The survival of Strychnos spinosa, Uapaca larkianaand U nitida was generally poor (Table 8).Seedlings of the three species were relativelysmall at planting time, which could explain thepoor survival. Alternatively they may haverequired mycorrhizal inoculation (see Lapeyrie &Hogberg, pp158-164).

In terms of height, the fastest-growing specieswas ZiziPhus mauritiana, which achieved morethan 2 m (Table 9), followed by Azanzagarckeana (1.4 m), but the rest were relativelyslow-growinp (<1.0 m). Similar growth rates of<1.0 m yF were recorded for Strychnoscocculoides and Parinari curatellifolia in a fieldtrial at the Chati Forest Research Station inZambia. Branching height was generally less than20 cm above-ground in all the species (Table 8),the highest being Azanza garckeana and Uapacanitida (19 cm), and the lowest being Uapacakirkiana and Ziziphus mauritiana (6 cm and 5 cmrespectively). Z. mauritiana had the highestnumber of branches (25), followed by Adansoniadigitata and Tamarindus indica (11); the otherspecies had fewer than ten branches.

Flowering and fruiting was observed in Azanzagarckeana and Ziziphus mauritiana 18 monthsafter planting. It was, however, observed that lessthan 40% of the trees planted produced fruits.

Table 8. Species and source of the indigenous fruit trees seed studied at the Forestry Research Institute of Malawi

Furthermore, fruit size and taste varied betweentrees. Early fruiting is an important characteristicas it may encourage farmers to plant indigenousfruit trees in anticipation of early benefits. Thefruiting results of Ziziphus mauritiana in thepresent study also provide an opportunity ofselecting individual trees for early fruitproduction, as well as quality Early fruiting wasalso recorded 'in Strychnos cocculoides (Wildorange) and Carissa edulis, which fruited withinfive years of growth at the Chati Forest ResearchStation in Zambia, thus comparing favourablywith exotic Citrus species.

PROSPECTS FOR DOMESTICATION OFFRUIT TREES OF THE MIOMBO REGION

Even though the indigenous fruit trees are acheap source of food, medicine and fodder,there are some problems which discouragetheir domestication. They are perceived to grow

Table 9. Survival and growth of 18 months old indigenous fruit trees at Naungu site, Machinga

slower and to take longer to produce fruitscompared with exotic species, and often have alow ratio between edible and non-edible parts,as in the case of Vangueria infausta and Uapacakirkiana. These problems could be overcomethrough selection and improvement programmes.Indigenous fruits often have little market value.For example, out of 30 indigenous fruit treespecies in Malawi, only five were sold onroadside markets (Maghembe & Seyani 1991),while in eastern Zambia only 20% of the farmershave ever sold indigenous fruits (Kwesiga &Chisumpa 1992). Even if they sold some fruitson the roadside, farmers complained of lowprices. Indigenous fruits with a market potentialin the region include Adansonia digitata, Uapacakirkiana, Annona senegalensis, Vangueria infausta,Ziziphus mauritiana, Azanza garckeana andStrychnos spinosa, depending on location withinthe miombo. Marketing will remain the single

227

most important constraint on the domesticationand genetic improvement of indigenous fruittrees in the miombo ecozone.

In addition, there is lack of appropriatetechnologies for preservation and processing ofthe fruit. Preparation of some indigenous fruitsbefore consumption is very labour-intensive.For example, in order to consume Tamarihdusindica, some form of processing is required toimprove its taste and palatability In theLuangwa valley of Zambia, farmers soak Tindica fruits in water and add ash fromColophospermum mopane for two days beforemixing with sorghum flour to make porridge.Others which require processing are Trichinaemetica and Parinari curatellifolia. In the caseof Ricihodendron rautenenii, the seeds have tobe crushed to remove kernels which in turnhave to be pounded before the oil is extracted.

Farmers in northern Malawi and most parts ofZambia do not perceive deforestation as a bigproblem. The majority of these farmers have notfound wayS of propagating the indigenous fruittrees. On the other hand, nearly every householdin the miombo ecozone has several exotic fruittrees, the most widespread being Mangiferaindica. Most adults have ignored indigenous fruittrees as they consider them only of interest tochildren. However, during drought and famine,these trees assume significance for the wholefamily Other constraints, such as the insecurity ofland tenure, technical knowledge in handling

228

Ethnobotanical

Farmers

Fruit quality Studies on variation in fruitassessment quality characteristics

Farmers

Priority species

Priority cultivars/varieties

Fruit processing Horticultural/genetic control ** Economic justification of ++requirement of dominant characteristics improvement programme

Farmers

Selection and breeding research

propagules, and termite and livestock damage toseedlings, are all common concerns of farmers.Irregular fruiting times and the short periodwhen preferred fruits are available add to thediscouragement of domestication. Even if thenutritional value of indigenous fruit trees is wellknown, very little knowledge is available tofarmers about propagating this germplasm; thereis no institutional framework for the promotion ofplanting materials, and therefore they are notavailable through the traditional channels ofresearch and extension. Without overcoming mostof these constraints, it will take some time beforethe domestication process is initiated.

The upsurge of interest in commercialisation ofindigenous fruit tree products in the region isvery recent. In Zambia, several years ofexperimentation with Uapaca kirkiana by theNational Council for Scientific Research (NSCR)has resulted in a new brand of wine, 'MASUKIJ'.Following such successful ventures, furtherresearch has been initiated on natural variation infruits of U kirkiana with a view to grading fruitsaccording to size at harvest and to provideadvice in selecting orchard sites and plantingstock (Mwamba 1989). Research at NSCR inZambia is underway and aimed at selection, treeimprovement and domestication of severalindigenous fruit trees, including Anisophylleapomifera, Parinari curatellifolia, Uapaca kirkanaand Ricinodendron rautenenii, which havepotential for beverages, conserves, wine and oil,respectively

Market surveys

Conservation sheds

Economic appraisal/ Growth and fruitmarket projection production requirements

Tree managementresearch

Phenology studies

Reproductivebiology studies

Geneticvariation studies

ngure 2. Diagramatic representation of the strategy developed by ICRAF for domestication of indigenous fruit trees

In Malawi, Ziziphus mauritiana is traditionally usedin the brewing of a potent liquor, 'KACHASU';more recently, Z. mauritiana fruits have beenintegrated with fruits of Uapaca kirkiana in thebrewing of a new popular brand of'MULUNGUZI' wine. In Tanzania, Trichilia emeticais already domesticated for the production ofvegetable oil and soap.

These examples further indicate thatcommercialisation is the key incentive for theselection, improvement and eventualdomestication of indigenous fruit trees. Forexample, in southern Malawi, there is a veryhigh demand for planting material of Uapacakirkiana, as well as for conserving miombowoodlands with U kirkiana trees: indigenous fruittrees are known to have a market value,because the Mulunguzi winery in Blantyre iswilling to purchase good-quality U kirkiana fruits.Consequently sources of the fruit are beingbrought under individual farmers' ownership,stretching claims to those areas where theselocal fruit trees are encountered. Thisdevelopment has implications for land tenure andthe allocation of land resources betweenagriculture and forest conservation.

Indigenous fruit trees are a key component ofagroforestry systems of the miombo ecozone,given their potential to contribute to the nutritionand cash economy of small-scale farmers(Aiyelaagbe 1992). Individual species of fruit treecut across national boundaries and arewidespread throughout the region (Table 2).Therefore, international agricultural researchcentres, and ICRAF in particular, should take thelead with the national institutions to developresearch into domestication of this valuableresource.

ICRAF is now in the process of developing astrategy for the improvement of indigenous fruittrees (Figure 2) through the multipurpose treeimprovement and management programme. Initialwork in the miombo ecozone is to prioritisethose indigenous fruit trees that the nationalinstitutes and farmers wish to see improved. Thecriterion being followed is marketability Alreadystudies are underway to determine the marketpotential of indigenous fruit trees of the miombo.

Deforestation has almost certainly eroded muchof the genetic resource of indigenous fruit trees.It is important that clear policies are adopted forconservation, and that selection, multiplicationand improvement of this germplasm are initiatedsoon. The different interested agencies, such asuniversities in the region, ICRAF, forestry andagricultural research institutes andnon-governmental organisations, should create a

forum to discuss and co-operate on conservationand domestication stiategies for indigenous fruittrees, if their genetic potential is to be fullydeveloped in the future.

REFERENCES

Aiyelaagbe, 1.0.0. 1992. Fruit crops in agroforestry: theiruse management and potential. (Final project report of studycommissioned by ICRAF/G12.) Nairobi: International Centrefor Research in Agroforestry

Campbell, B.M. 1987. The use of wild fruits in Zimbabwe.Economic Botany 41, 375-385.

Fanshawe, D.B. 1972. Useful trees of Zambia for theagriculturist. Lusaka: Government Printer.

Food and Agriculture Organisation. 1983. Food and fruitbearing forest species: examples from eastern Africa. (FAOForestry Paper, no. 44/1.) Rome: FAO.

Hans, A.S., Parker, E.J. & Kumar, V. 1978. Potentialities ofZambian forest fruit trees. Lusaka: National Council forScientific Research.

Karachi, M., Ruffo, C.K., Lema, N.K. & Minae, S. 1991.Use of multipurpose trees in western Tanzania: results of anethnobotanical survey (AFRENA Report, no. 40.) Nairobi:International Centre for Research in Agroforestry

Kwesiga, F. & Chisumpa, S.M. 1992. Multipurpose trees ofeastern Zambia: an ethnobotanical survey of their use infarming systems. (AFRENA Report, no. 49.) Nairobi:International Centre for Research in Agroforestry

Maghembe, J.A. & Prins, H. 1993. Germination studies onseed of fruit trees indigenous to Malawi. Forest Ecology andManagement. In press.

Maghembe, J.A. & Seyani, J.H. 1991. Multipurpose treesused by smallholder farmers in Malawi: results of anethnobotanical survey (AFRENA Report, no. 42.) Nairobi:International Centre for Research in Agroforestry

Maghembe, J.A., Prins, H., Sambo, E.Y. & Ngulube, M.1991. Established methods for indigenous fruit trees in Makoka,Malawi. (Paper presented at the Regional Conference onAgroforestry Research and Development in the MiomboEcozone of Southern Africa, Lilongwe, Malawi, 1991.)

Mwamba, C.K. 1989. Natural variation in fruits of Uapacakirkiana in Zambia. Forest Ecology and Management, 26,299-303.

Ngulube, M.R. & Kananji, B. 1989. Seed problems ofedible indigenous fruits in Malawi. Zomba: Forestry ResearchInstitute of Malawi.

Payne, R.W, Lane, P.W., Ainsley, A.E., Bricknell, K.E.,Digby, P.G.N., Harding, S.A., Leech, P.K., Simpson,H.R., Todd, A.D., Verrier, P.J., White, R.P., Gower, J.C.,Tunnicliffe Wilson, G. & Paterson, L.J. 1987. Genstat 5:reference manual. Oxford: Clarendon Press.

Saka, J.D.K. & Msonthi, J.D. 1993. Nutritional value ofedible fruits of indigenous wild trees in Malawi. ForestEcology and Management. In press.

Saka, J.D.K., Msonthi, J.D. & Sambo, E.Y. 1989.Identification, chemical analysis, propagation and processing ofindigenous fruits of Malawi. Lilongwe: Ministry of Agriculture.

Wehmeyer, A.S. 1966. The nutrient composition of someedible wild fruits in the '17ansvaal. South African MedicalJournal, 40, 1102-1104.

229

Amazonian fruits and nuts: potential fordomestication in various agroecosystems

c R Clement1 & H Villachica21Centro de Pesquisas de Ciências Agronômicas, Instituto Nacional de Pesquisas da Amazeinia, Cx Postal

478, 69.011 Manaus, Amazonas, Brazil21nstituto Nacional de Investigacián Agraria y Agroindustrial, Apartado 248, Lima 100, Peru

ABSTRACTNew crops and new agroecosystems are required to move tropical agricultural development towards sustainability whileconserving sufficient tropical forest to avoid serious erosion of tropical biodiversity and to slow global climate change. InAmazonia, numerous fruit and nut species have been identified as having sufficient potential to merit research attention; some ofthese are already being planted commercially It is useful to classify these species by their degree of domestication and by theirpopulation biology because these factors can suggest the most suitable agroecosystem for a species' continued development. Atleast 12 fruit species were semi- or fully domesticated by native Americans before European contact, one of which was probablya staple. Several dozen were managed in swidden agroforestry or in the indigenous forest management practices that followswiddens. A hundred or more were collected from the wild without apparent management. The population biology of most ofthese species requires scarcity as a strategy to limit pest and disease damage, but a few commonly occur at moderate to highdensities. One domesticate, two managed species and several wild species can occur as monospecific populations, covering tens,hundreds or thousands of hectares. Because of extreme biotic pressures, conventional agroecosystems (ie monocultures) are onlysuitable where sufficient capital and infrastructure are available to sustain them, unless the species is pre-adapted to monocultureor research can identify production techniques that allow monoculture at low capital levels. Numerous agroforestry practices, fromsimple to highly heterogeneous systems, are currently the subject of intensive research and many development projects, butrequire new tree components. The transformation of 'natural' forests into multipurpose forest management systems, based upon theharvest of sustainably extracted fruits, nuts, and- other forest products, is starting to be evaluated. The latter option offers thepotential for sustainable economic yields, while maintaining forest biomass and considerable biodiversity To develop sustainablyAmazonia must evolve into a mosaic of human settlements, conventional agroecosystems, agroforests, managed forests for multipleproducts (only one of which will be timber) and preservation areas. International assistance to support and expand this researchand development effort is essential, and will benefit not only Amazonians but all humanity

INTRODUCTION

During this century agricultural developmentinspired by the rich societies of the northernhemisphere has reduced the number of cropsplanted, their genetic diversity and thecomplexity of the agroecosystems used toproduce them. Concurrently yields of manycrops have increased dramatically owing toimproved agrotechnology and increased capitalinputs. Nonetheless, the current preoccupationwith sustainability suggests that, with thesetechnologies, temperate agriculture has becomemore unstable, rather than less so. This is alsotrue for tropical agriculture because thebiophysical and economic factors that maketemperate agriculture unstable are more intensein the tropics. Students of modern tropicalagriculture now recognize that: (i) diversity canand should contribute to stability andsustainability although this is not always true; and(ii) there is a need for the conservation ofsufficient tropical forest to avoid serious erosionof tropical biodiversity (Smith et al. 1992).Nevertheless, the erosion of crop geneticdiversity is reducing our ability to develop new

230

crop varieties in response to current and futureagricultural crises. In parallel with the increasingawareness of the need for genetic diversityrecognition is dawning of the need to conservecultural diversity as well, as the techniques ofdiversified agriculture are still used by manynon-Western cultures and these cultures are thede facto curators of the majority of the world'scrop genetic diversity

In short, it is recognised that agriculturalsustainability in the tropics requires 'new' cropsand 'new' agroecosystems, to run in parallel withincreased conservation. 'New' crops arenecessary because markets for the currentlyrecommended tropical cash crops are at or nearsaturation. 'New' agroecosystems are necessarybecause of the poor adaptation of the northernhemisphere conventional agricultural model inmany parts of the humid tropics. This approachmay help to slow global climate change,because many of the 'new' agroecosystems willbe based upon tree crops, and may also help toreduce the instabilities of agricultural productionin many areas. This combination of new cropsand new agroecosystems, with conservation, is a

tall order for the research community and thedevelopment agencies. Recognition of thechallenge is increasing, however, and positivesteps are being taken in many countries.

In Amazonia, research on 'new' crops has beenunder way for the past 50 years, althoughcontinuity and comprehensiveness have beenproblems in all countries of the region. Duringthe past two decades, agroforestry has alsoreceived considerable attention. Additionallyethnobiological research has found that manynative American practices extend agroforestryinto multipurpose forest management systems .(eg Posey 1985). Recent research suggests thatat least 12% of the Amazonian forest originatedfrom these systems (Balée 1989). Someresearchers even suggest that a primevalArnazonian forest does not exist and that all theforest was modified by human intervention (Clay& Clement 1993)1.

Among the potential 'new' crops, many fruit andnut species have sufficient merit to warrantresearch attention. Some of these are alreadyplanted commercially both in monoculture and inagroforestry It is useful to classify 'new' crops bytheir degree of domestication and by theirpopulation biology because these classificationscan suggest the most suitable agroecosystem fora species' continued development. This paperwill focus upon a few species, their potential, thecurrent research efforts to utilise them, and howthey fit into different types of agroecosystemssuitable for use in Amazonia.

Some definitions

Domestication is human-induced change in thegenetics of a plant population to conform tohuman desires and agroecosystems, culminatingin the plant's loss of its ability to survive innatural ecosystems (Harlan 1975). A fulldomesticate, therefore, requires humanintervention to survive. A semi-domesticate issignificantly modified from the wild state, but cansurvive if abandoned. A managed species is onethat receives some degree of care but is notcultivated in Harlan's sense, ie no agroecosystemis prepared for it. A managed species may beprotected during clearance of the forest forswiddens, or its volunteer seedlings may beencouraged along paths and around dwellings, orselected seed may be sown outside of cultivatedareas. As Rindos (1984) points out, evenmanaged species are subject to genetic changesand thus start to become domesticated.Cultivation is the preparation of anagroecosystem for the growth of either wild ordomesticated plants (Harlan 1975).

THE DOMESTICATION CONTINUUM INAMAZONIA

In Amazonia, it is possible to find indigenous

fruit crops and potential crops in all stages ofthe domestication process, from wild to fullydomesticated.

Wild and oligarchic species

As indicated by Prance (pp7-15), extractivismhas been practised in Amazonia for millenniaand may initiate the domestication of theextracted species. A hundred or more Amazonianfruits and nuts are currently collected from thewild without apparent management, although theymay have been managed in the past. Thepopulation biology of most of these speciesapparently requires scarcity (ie a low populationdensity) as a strategy to limit pest and diseasedamage in nature, but a few commonly occur atmoderate to high densities in oligarchic forests.

Caryodendron orinocensis (inchi) is distributedthroughout north-western Amazonia, both in thelowlands and in the Andean foothills. A closelyrelated species, C. amazonicum, which mayactually be an ecotype of C. orinocensis, occursin north-western Brazil. The dove egg-sized nutof C. orinocensis is reputed to taste like the nutof Corylus avellana (hazel) (Recklin 1982),although one of its common names is 'treepeanut'. There are several small plantations inColombia and Ecuador, where some researchhas been initiated. Some small germplasmcollections exist and are partially evaluated(J B Martinez, University of Nario, personalcommunication, 1985). In addition, in thetransitional ecotone between the ColombianLlanos and the humid tropic forest, theCorporación Araracuara systematically collectedgermplasm and established this in several smallgermplasm collections in San José de Guaviare,Colombia (E Duran, Corp. Araracuara, personalcommunication, 1990). However, caterpillars -havebecome an extremely serious pest, causingcomplete defoliation several times a year,reducing growth and eliminating yield. This pesthas not been reported in other areas (eg seeRecklin 1982), but highlights the susceptibility ofsome wild species in low-input monocultureagroecosystems in Amazonia.

One domesticate, two managed species andseveral wild species can occur in Amazonia asnatural or anthropogenic monospecificpopulations, covering tens, hundreds orthousands of hectares. Peters et al. (1989a) haveused the term 'oligarchic' to describe this groupof species (see Prance, pp7-15) and discuss fourpalms, Jessenia bataua (managed), Euterpeoleracea, Mauritia flexuosa and Orbignya phalerata(all wild), and two dicots, Grias peruviana(Lecythidaceae) and Myrciaria dubia (Myrtaceae)(both wild). Some castanhais (small areas of highabundance of managed or formerly managedtrees) in Amazonia have a high enough densityto be classified as oligarchic, as is the case with

231

the domesticated Poraqueiba sericea(Icacinaceae), planted in low-inputnear-monocultures in Tamshiyacu, Peru (Padochet al. 1987).

In both Brazil and Peru, research has beenundertaken on the shrub M. dubia (cagari orcamu-camu). This species is indigenous to thefloodplains, where it can spend three to fourmonths partly submerged. Its red/purple-sldnnedfruits contain a translucent white pulp and oneor two seeds. Both the pulp and the sldn containhigh concentrations of vitamin C, up to 4 g in100 g fresh pulp with skin, with higherconcentrations in the sldn than in the pulp(J Andrade, National Research Institute forAmazonia (11\TPA), personal communication, 1990).In natural stands, Peters (1990) estimated yieldsof 9-12 Mt ha-1 yr-1. These yields require noagronomic or capital input, being fertilized andhaving pests controlled by the yearly floods(Peters et al. 1989a).

The Peruvian Agricultural and Agro-IndustrialResearch Institute (INIAA) has a small germplasmcollection of the shrub-type M. dubia located atIquitos with 40+ accessions (Mendoza, Picon &Gonzales 1989), while the Brazilian-NationalResearch Institute for Amazonia (INPA) maintainsa much smaller one (4 accessions) at Manaus(W B Chavez INPA, personal communication,1990). INIAA researchers recently collected atree-type Myrciaria, very similar in most respectsto M dubia, but with lower fruit acidity It has notyet been described taxonomically

Managed species

Several dozen Amazonian fruit and nut specieswere managed in swidden agroforestry or in theindigenous forest management practices thatfollow swiddens. Bertholletia excelsa(Lecythidaceae) (Brazil nut) is probablyindigenous in south-eastern Amazonia, but wasdistributed by native Americans to most cornersof the basin (Moh & Prance 1990). Until recentlythis forest emergent was managed sustainably inmost areas where it occurs, but some of theprime B. excelsa areas in south-eastern Arnazoniaare now being subjected to severe deforestationfor pasture, charcoal production, and other landuses (Kitamura & Willer 1984).

Recently Posey and collaborators at the MuseuParaense Emilio Goeldi, in Be lém, Brazil, haveshown that Amazonia's famous castanhais aremostly, due to native American agroforestry andforest management practices (eg Posey 1985).B. excelsa seedlings were transplanted into newswiddens and then managed during thesucceeding swidden fallow thus creating thecastanhais. In south-eastern Amazonia andsouthern Peru, these castanhais contain 50-200trees each, at densities of 5-20 trees ha-1, andoccur scattered over areas of hundreds to

232

thousands of hectares. At one time, Brazil nutswere second only to rubber (Hevea brasfliensis,Euphorbiaceae) as an export product fromAmazonia (Mori & Prance 1990) and retain theirimportance in some areas, eg Madre de Dios,Peru (Sanchez 1973) and Acre, Brazil (Mori &Prance 1990).

Since the mid-1970s, the Agricultural ResearchCenter of the Humid Tropics (CPATIJ-EMBRAPA),in Be lém, has researched B. excelsa (Miller 1981;Mori & Prance 1990), CPATU has a small clonalgermplasm collection, A small seedling progenyand clonal germplasm collection also exists inAcre state, at the local EMBRAPA station. A fewgrafted clones are being planted commerciallyon 3000 ha about 200 km east of Manaus(C H Miller & A Kato, CPATU, personalcommunication, 1990). INIAA is evaluating(1988-92) individual tree yield in castanhais nearPuerto Maldonado and Iquitos to identifyhigh-yielding trees with non-alternating harveststo form a clonal germplasm collection at Iquitos.

Theobroma grandiflorum (Sterculiaceae)(cupuassu) is a T cacao relative that isenormously popular for juices, ice-creams andother confectionery products in eastern andcentral Amazonia (Clement & Venturieri 1990).Like B. excelsa, T grandiflorum is indigenous toeastern Amazonia and was either collected fromthe wild or managed in agroforestry systems.Today it is cultivated around Be lém and Manaus,where hundreds of hectares are planted annuallyto supply the strongly increasing demand for itspulp.

Unlike T cacao, T grandiflorum is exploitedprincipally for its aril, the pulp around the seed.This has a strong flavour and aroma, which mostconsumers prefer diluted (eg in juice orice-cream). The pulp (35-45% of fresh fruitmass) is frozen for export to other parts ofBrazil, the United States and Europe. The seed(15-20% of fresh fruit mass) is rich (50% drymass) in unsaturated fats (45-50% unsaturation),but has a different fatty acid profile from T cacaoand so makes only low-quality chocolate(Clement & Venturieri 1990). The fat is beingtested for use in sldn creams and other beautyproducts in Manaus.

There is a seedling germplasm collection of100+ progenies of T grandiflorum at INPA,Manaus, a clonal collection of more than 30genotypes at the Amazonian AgroforestryResearch Center (CPAA-EMBRAPA), Manaus, anda further collection at CPATU, Be lém (20+clones). Most of the current planting, however,uses unselected or mass-selected germplasmfrom established plantations (Venturieri, Alves &Nogueira 1985). CPAA and CPATU currently haveactive tree improvement programmes.

Domesticated species

At least a dozen Amazonian fruit species weresemi- or fully domesticated by native Americansbefore European contact. One of these, Bactrisgasipaes (pejibaye or peach palm), was probablya staple (Clement 1988, 1992) and is native tosouth-western Amazonia. It is a thorny,multistemmed palm that produces bunches ofstarchy fruit that are popular throughout thelowland humid tropics, from Costa Rica to BrazilWhile it may have entered the domesticationprocess in several places (Mora Urpi 1984), themost derived landraces occur in north-westernAmazonia (Mora Urpi & Clement 1988). Clementand Mora Urpi (1987) have discussed itspotential, and Clement and Arkcoll (1989) reviewthe current state of knowledge.

Although B. gasipaes was domesticated as astaple starch crop, it is now widely planted forits heart of palm. There are about 5000 haplanted to high-density monocultures, mostly inCosta Rica, although Brazil, Colombia, Ecuadorand Peru are rapidly planting large areas.Current world trade in palm hearts exceedsUS$50 million, mostly extracted from wildpopulations of Euterpe oleracea in the AmazonRiver estuary (Mora Urpi et al. 1991). Severaltropical American countries have activegermplasm banks of B. gasipaes (Clement 1991),the largest being located in Costa Rica. In1983-84, the US Agency for InternationalDevelopment (AID) financed a pan-Amazoniangermplasm collection (Clement & Coradin 1988).Although this germplasm is conserved in allactive banks, it is most actively used forimprovement in Costa Rica, and to a lesserdegree in Brazil and Peru.

The Costa Rican programme is concentrating onimproving B. gasipaes for palm heart, fruit forhuman consumption and fruit for animal feed,with the first by far the most important. Manycontrolled hybridisations were made and severalprogeny trials installed in Costa Rica G MoraUrpi, University of Costa Rica, personalcommunication, 1991). INIAA is using massselection for fruit yield and quality and takingthe resultant progeny to field trials; initialselections for palm heart have also been made.In addition to the Costa Rican priorities, theBrazilian programme, located at INPA, includesfruit for vegetable oil (Clement & Arkcoll 1991).

The state of São Paulo, Brazil, has recentlystarted a programme for palm heart (Bovi,Godoy & Saes 1988).

Another domesticate is Eugenia stipitata(Myrtaceae) (araza). This small shrub yields alarge (50-800 g) berry with a canary-yellow,highly perfumed rind, and a canary-yellow acid,

pleasantly flavoured mesocarp. The fruit's flavouris ideal for juices and flavourings that requirelittle or no cooking (ice-creams, creams, etc), but

its flavour is lost on cooking (Clement 1990a).The perfumed rind requires investigation as asource of essential oils. Pinedo, Ramirez andBlasco (1981) suggest that the domesticated formoriginated in Peru, between Iquitos and thelower Ucayali River basin. Chavez Flores andClement (1984) note that the domesticate is thesubspecies sororia and that the wild form issubspecies stipitata, possibly native to Acre state,Brazil Recent germplasm collections made byINIAA found both subspecies in Peru, adjacent toAcre state (Bello 1991).

INIAA maintains an active germplasm bank ofE. stipitata at Iquitos with 20+ accessions (Bello1991) and 1NPA maintains a much smaller one (5accessions) at Manaus (Clement 1990a). INIAAhas established field trials near Iquitos toevaluate mass-selected progenies from plantsyielding the equivalent of 11-15 Mt ha-I. INIAAand 1NPA have used phenotypic selections inexperimentation and for distribution to farmers inthe Iquitos and Manaus areas, respectively

AGROECOSYSTEMS

Given our defmitions of domesticated, managedand wild, it is logical to expect that eachcategory is best adapted to the agroecosystemin which it is found or in one with a higherlevel of inputs. Thus, domesticates grow best inmonocultures and in specific locations inagroforestry systems, although pest and diseaseincidence may increase as crop diversitydecreases. Managed species grow best in theirtraditional agroecosystem but are easilyintroduced into monocultures, again withincreased risk of pest and disease attack. Whileit is relatively easy to take a species from anypart of the domestication continuum into ahigher-input system, eg monoculture, the reverseis not true. A domesticate, adapted to high-inputsystems, cannot be grown successfully in sy-Stemswith significantly lower inputs, eg a multipurposeforest management system. For example, Bactrisgasipaes does not yield in closed second-growthforest (Clement 1990b). Hence, continueddomestication of these crops should be designedwith the appropriate agroecosystem in mind. .Agronomic research in various agroecosystemson the six species discussed above is reviewedhere.

Monoculture

Because of extreme biophysical pressures in .Amazonia, monocultures are only suitable wheresufficient capital and infrastructure are availableto sustain them (Villachica et al. 1990), unlessthe species is pre-adapted to monoculture, egoligarchic species such as M. dubia. Nonetheless,considerable research has been undertaken withmonocultures, both as a final objective and tocompare with agroforestry systems.

233

Table I. Export-quality palm heart yields (kg ha-1) of Bactrisgasipaes from the first harvest at several densities andnutrient availabilities on Amazonian oxisols (nutrientavailabilities given as g plant of N, P205 and K20) (source:Gomes et al. 1988; Gomes & Arkcoll 1988)

234

Iquitos Manaus

Table 2. Yields (Mt ha-1) of the first three harvests (year 3-5)of Eugenia stipitata at 3 m x 4 m, with and withoutfertilization, on a yellow oxisol (Manaus) and on a sandyentisol (Iquitos) (nutrient availabilities given as g plant-1 of N,P205 and K20) (source: Villachica 1993; Alfaia et al. 1987)

Monoculture yields of B. gasipaes for palm heartof up to 1.6 Mt ha-1 have been estimated withunselected germplasm at several densities onlow-fertility oxisols near Manaus (Gomes &Arkcoll 1988) and in Ronclônia state (Gomes,Menezes & Viana Filho 1988), the latter at threenutrient availabilities (Table 1). These yieldscompare favourably with those obtained in CostaRica on more fertile soils and with a morefavourable rainfall distribution. Similar work existsin Peru, but data have not yet been published.

Monoculture yields of B. gasipaes for fruit fromplants of different ages have been reported tovary between 3 and 30 Mt ha-1 yr 1 (Clement -&Arkcoll 1989), although most of these are basedupon estimates rather than measurements. Szott,Palm and Sanchez (1991) suggest yields of up to20 Mt ha-1 when the plantation reaches fulldevelopment on an ultisol at Yurimaguas, Peru.Picon and Ramirez (1991) report yields of 2.3 Mtha-1 for the first harvest, obtained at three tofour years after transplanting on to a sandyinceptisol at Iquitos. Both experiments usedannual crop interplantings to simulate farmerestablishment practices.

B. gasipaes is susceptible to many pests anddiseases, although none have proved to beeconomically severe except locally Fruitfall is aserious problem in fertilized monocultures. It isprobably a nutritional (micro-nutrient) problem in

which fruits from fertilized plants are susceptibleto Heteroptera damage and fruitfall (Couturier,Clement & Viana Filho 1993). Fruitfall has notbeen reported from unfertilized plots or fromagroforestry systems. A coleopteran has recentlybecome a significant pest in the Chocó region ofColombia, causing severe losses (A Velasco F,Palmito del Pacifico, personal communication, 1991).

E. stipitata monocultures have proven to be veryproductive, both in Brazil and in Peru (Table 2).In Manaus, maximum yields were obtained in thefourth harvest (sixth year), estimated at 35 Mt

- 1ha with the maximum fertilization at a densityof 1667 plants ha-1 (2 m x 3 m). Yields in Peruare similar and still increasing; it is expected thatmaximum yields will be attained at nine or tenyears. Fruitflies limit yields, however, althoughthey can be controlled at economic levels withbaited traps and good plantation sanitation.

A moderately fertilized, high-density (1667 plants-ha 1), ten-year-old monoculture of M. dubia at

INIAA in Pucallpa has yielded 7-8 Mt ha-1.Growth was slow for the first six years, whichstimulated research to use grafted plantings.Grafted plants often flower and produce fruitsearlier than seedlings (2-3 versus 4-5 years), AtINPA, 300 plants were established on the dryuplands (oxisols) where the species does notnormally occur. Only five plants (2%) presentedgood yields, extrapolated to 7 Mt ha.-1 yr 1(W B Chaves E INPA, personal communication,1990). INIAA has started to encourage smallcommercial plantings in the floodplain nearIquitos and on poorly drained soils nearPucallpa, on the lower Ucayali River,

Experimental yields of B. excelsa, T grandiflorumand C. orinocensis have not been reported todate. The fertilization and spacing requirementsof T grandiflorum are being studied in Brazil, butthe two nut species are not receiving the sameattention. C. orinocensis is limited by a defoliatorin monocultures. T grandfflorum is limited byCrinipellis perniciosa (witch's broom) inAmazonia, which is more severe in monoculturethan in agroforestry systems (Clement &Venturieri 1990). Neither B. excelsa nor M dubiahave had severe pest problems reported to date.

Agroforestry

Many agroforestry practices, from simple tohighly heterogeneous systems, are now thesubject of intensive research and manydevelopment projects in Amazonia. Many of thefruit and nut crops listed here have greatpotential for agroforestry systems and are nowbeing incorporated in both experimental anddevelopmental plots.

Clement (1986, 1989) has reviewed the earlymore empirical studies and discussed theagroforestry potential of B. gasipaes. More

recently Szott et al. (1991) discussed resultsobtained at Yurimaguas, while Picon and Ramirez(1991) presented results of a B. gasipaes x E.stipitata trial near Iquitos. Iit the latter trial,Manihot esculenta yielded close to the regionalmean in the first two years, but yielded poorly inyear three because of competition from B.gasipaes and declining soil fertility (Table 3). Itis, however, too early to determine whether thepolycrop systems yield better than themonocultures.

Table 3. Establishment phase yields (Mt ha.-1) of Bactris

gasipaes and Eugenia stipitata in monoculture (625 plants-1

ha 1) and polyculture (312 plants ha of each species) at

Iquitos, with Manihot esculenta used as an intercrop (source:

Picon & Ramirez 1991)

Promising results were obtained from B. gasipaesinterplanted with Artocarpus altilis (Moraceae)and A. integrifolia at INPA (Arkcoll 1982).B. gasipaes as a shade for T cacao gave suchpromising results that the Brazilian CacaoCommission (CEPLAC) recommends its use asproductive shade in Amazonia (P T Alvim,CEPLAC, personal communication, 1989).B. gasipaes is intercropped with T grandiflorumand many other species in Brazil, Peru andelsewhere in tropical America, as well as beinga common component of most Amazonianswiddens (Clement 1986, 1989).

The managed species, B. excelsa andT grandiflorum, are traditionally planted inswiddens, although monocultures have recentlybecome fashionable. The CPATU station, inBelém, has several experiments with B. excelsaas shade for T cacao and other small-staturedperennials. Because T grandiflorum has a growthhabit similar to that of T cacao, currentrecommendations are that it should also beplanted in partial shade (G A Venturieri,University Federal do Para, personalcommunication, 1990).

C. orinocensis is occasionally planted in swiddensor private gardens (Recklin 1982). Given itssusceptibility to pests, however, its furtherdevelopment should be encouraged in,agroforestry systems or multipurpose forest

management areas. M dubia is an ecologicalweed in its natural environment, which is why itcan occur at monoculture densities and hasadapted so well to agroecosystems. Peters (1992)has recommended sustained yield managementof natural populations of M. dubia in Peru, andthis species is now being researched by INIAA.

Multipurpose forest management

The transformation of 'natural' forests intotraditional multipurpose forest managementsystems, based upon the harvest of sustainablyextracted fruits, nuts, and other forest products, isstarting to be evaluated (Lamb 1991; Nepstad &Schwartzman 1992). This practice offers thepotential for sustainable economic yields, whilemaintaining forest biomass and considerablebiodiversity (Myers 1984).

Peters, Gentry and Mendelsohn (1989b) suggestthat this type of management in a Peruvian forestnear Iquitos could sustainably yield sufficientincome to be attractive to farmer-foresters, buttransportation and marketing are limiting factorsthere. Anderson (1990) states that this type ofmanagement of an oligarchic forest of Euterpeoleracea is sustainable near Belém. Anderson,May and Balick (1991) suggest that oligarchicforests of Orbygnia phalerata can be sustainablymanaged in south-eastern Amazonia, althoughthese forests are being degraded rapidly today

To make multipurpose forest management viablein Amazonia or elsewhere, new markets must befound for traditional products and developed fornew ones (Clay & Clement 1993). Often,economic densities of species with markets mustbe created in the forest to facilitate harvesting(Clement 1993). Of the species discussed in thispaper, only the managed species, B. excelsa andT grandiflorum, and C. orihocensis (wild nut) aresuitable for inclusion in forest management plots,because they are still adapted to forestconditions, and especially to competition for light.

SUSTAINABLE DEVELOPMENT BASED UPONTREES

To develop sustainably Amazonia must evolveinto a mosaic of human settlements, conventionalagroecosystems, agroforests, managed forests formultiple products (only one of which will betimber) and preservation areas. Sustainability canbe attained technologically as shown by Sanchezet al. (1982) at Yurimaguas and by manyBrazilian and Peruvian researchers (Villachica etal. 1990), but this alternative is only viable if thetechnology is available to farmers and they havethe capital to pay for it.

Unfortunately the vast majority of Amazonianfarmers do not have either the capital or theaccess to the technology that would allow themto farm sustainably in conventional high-input,low-diversity agroecosystems. For these farmers,

235

agroforestry holds great promise because it is apractice with which they are already familiar.Furthermore, it husbands scarce resources, bothbiologically and in terms of family labour andcapital. The establishment phase of agroforestrysystems requires special attention, so thatfarmers can generate income while their treecrops are maturing (Villachica et al. 1990).However, the number of crops that currently haveextensive markets and that can be grown inagroforestry systems is small. Thus, to enhancethe viability of expanded agroforestry productionin Amazonia, it will be essential to search fornew crops and to develop new markets for them(Clay & Clement 1993).

Multipurpose forest management systems arealready practised by many smallfarmers/extractivists in Amazonia and elsewherein the wet tropics (eg Gómez-Pompa & Kaus1990; Groube 1989) and potentially offer themost sustainable form of land use. Research inthis area has barely started, however, so thatalmost everything remains to be done to developthis desirable blend of production andconservation. International assistance to supportand expand research and development into thisarea is essential, and will benefit not onlyAmazonians but all humanity

NOTE

1Although this suggestion has not yet beenproved, it is based upon a synthesis of thefollowing sources: large human populations inpre-history (Denevan 1992); occupation of allmajor natural ecosystems in Amazonia, both inpre-history (Denevan 1992) and currently; theexistence of sophisticated resource managementsystems to support these populations inpre-history and in the present (Posey & Balée1989); agroecosystems that extend our westernideas of such systems and include all of theenvironment manipulated by human groups(Alcorn 1989), including both indigenous (Alcorn1989; Balée 1989; Moran 1989) and folk groups(Moran 1989; Parker 1989). Balée's (1989)estimate of 12% of Amazonia beinganthropogenic forests includes only those withthe most obvious changes caused by humanintervention. Less obvious changes are morelikely in most of the region but more difficult todetect.

REFERENCES

Alcorn, J.B. 1989. Process as resource: the traditionalagricultural ideology of Bora and Huastec resourcemanagement and its implications for research. In: Resourcemanagement in Amazonia: indigenous and folk strategies,edited by D.A. Posey & W Balée, 63-77. (Advances inEconomic Botany 7.) New York: New York Botanical Garden.

Alfaia, S.S., Chavez Flores, W.B., Ferreira, S.A.N. &Clement, C.R. 1987. Efeito de espacamento e adubacdomineral no araca-boi (Eugenia stipitata McVaugh). I. Producao

236

de frutos. Anais Congresso Brasileiro de Fruticultura, 9,119-124.

Anderson, A.B. 1990. Extraction and forest management byrural inhabitants of the Amazon estuary In: Alternatives todeforestation: steps toward sustainable use of the Amazonrainforest, edited by A.B. Anderson, 65-85. New York:Columbia University Press.

Anderson, A.B., May, RH. & Balick, M.J. 1991 Thesubsidy from nature - palm forests, peasantry anddevelopment on an Amazon frontier New York: ColumbiaUniversity Press.

Arkcoll, D.B. 1982. Considerac6es sobre a produção dealimentos por arvores e florestas. Acta Amazonica, 12, 247-249.

Balée, W. 1989. The culture of Amazonian forests. In:Resource management in Amazonia: indigenous and folkstrategies, edited by D.A. Posey & W Balée, 1-21. (Advancesin Economic Botany 7.) New York: New York Botanical Garden.

Bello, S. 1991. Informe de las expediciones de recolecciOn degermoplasma de araza (Eugenia stipitata), guarana (Paulliniacupana) y pifia (Ananas comosus) en el ambito de losdepartamentos de Loreto y Ucayali durante 1989. (InformeTécnico no. 14, Programa de Investigación en CultivosTropicales.) Lima: Instituto Nacional de Investigacion Agraria yAgroindustrial.

Bovi, M.L.A., Godoy G. Jr & Saes, L.A. 1988. Pesquisascorn os generos Euterpe e Bactris no Instituto Agron6nmicode Campinas. In: Anais do 10 Encontro Nacional dePesquisadores em Palmito, 1-44. Curitiba: Centro Nacional dePesquisa de Florestas/Empresa Brasileira de PesquisaAgropecuaria.

Chavez Flores, W.B. & Clement, C.R. 1984.Consideragoes sobre a araca-boi (Eugenia sti)Jitata McVaugh)na Amaz6nia brasileira. Anais Congresso Brasileiro deFruticultura, 7, 167-177.

Clay, J.W. & Clement, C.R. 1993. Introduction -conservation and development. In: Selected species andstrategies to enhance income generation from Amazonianforests, edited by J.W Clay & C.R. Clement. (FAOForestry Paper.) Rome: Food and AgricultureOrganisation. In press.

Clement, C.R. 1986. The Pejibaye palm (Bactris gasipaesH.B.K.) as an agroforestry component. Agroforestry Systems, 4,205-219.

Clement, C.R. 1988. Domestication of the pejibaye palm(Bactris gasipaes): past and present. In: The palm - tree oflife, edited by M.J. Balick, 155-174. (Advances in EconomicBotany, 6.) New York: New York Botanical Garden.

Clement, C.R. 1989. The potential use of the pejibaye palmin agroforestry systems. Agroforestry Systems, 7, 201-212.

Clement, C.R. 1990a. Araza. In: Fruits of tropical andsubtropical origin: composition, properties, uses, edited byS. Nagy PE. Shaw & W Wardowski, 260-265. Lake Alfred,Florida: Florida Science Source.

Clement, C.R. 1990b. Regeneracdo natural de pupunha(Bactris gasipaes). Acta Amazonica, 20, 399-403.

Clement, C.R. 1991. Amazonian fruits: a neglected,threatened and potentially rich resource requires urgentattention, Diversity, 7, 56-59.

Clement, C.R. 1992. Domesticated palms. Principes, 36,70-78.

Clement, C.R. 1993. Creating income generating forests inAmazonia. In: Selected species and strategies to enhanceincome generation from Amazonian forests, edited by J.W Clay& C.R. Clement. (FAO Forestry Paper.) Rome: Food andAgriculture Organisation. In press.

Clement, C.R. & Arkcoll, D.B. 1989. The pejibaye palm:

economic potential and research priorities. In: New crops for

food and industry, edited by G. Wickens, N. Hag & P Day

304-322. London: Chapman and Hall.

Clement, C.R. & Arkcoll, D.B. 1991. The pejibaye (Bactris

gasipaes H.B.K, Palmae) as an oil crop: potential and

breeding strategy Oleagineux, 46, 293-299.

Clement, C.R. & Coradin, L., eds. 1988. Final report

(revised): peach palm (Bactris gasipaes H.B.K) germplasm

bank (US-Agency for International Development project

report.) Manaus: Instituto Nacional de Pesquisas da

AmazOnia/Centro Nacional de Recursos Genéticos.

Clement, C.R. & Mora Urpi, J. 1987. The pejibaye (Bactris

gasipaes H.B.K. , Arecaceae): multi-use potential for the lowland

humid tropics. Journal of Economic Botany, 41, 302-311.

Clement, C.R. & Venturieri, G.A. 1990. Bacuri and

cupuassu. In: Fruits of tropical and subtropical origin:

composition, properties, uses, edited by S. Nagy PE. Shaw &

W Wardowski, 178-192. Lake Alfred, Florida: Florida Science

Source.

Couturier, G., Clement, C.R. & Viana Filho, P. 1993.

Leptoglossus lonchoides Allen (Heteroptera, Coreidae), agente

de la caida de los jovenes frutos de Bactris gasipaes

(Palmae) en las plantaciones de la region de Manaus.

Turrialba, 41. In press.

Denevan, W.M. 1992. Native American populations in 1492:

recent research and a revised hemispheric estimate. In: The

native population of the Americas in 1492, edited by WM.

Denevan, xvii-xxix. Madison, Wisconsin: University of

Wisconsin Press.

Gomes, J.B.M. & Arkcoll, D.B. 1988. Estudos iniciais

sobre a produgao de palmito de pupunha. In: Anais do 10

Encontro Nacional de Pesquisadores em Palmito, 271-278.

Curitiba: Centro Nacional de Pesquisa de Florestas/Empresa

Brasileira de Pesquisa Agropecuária.

Gomes, J.B.M., Menezes, J.M.T. & Viana Filho, P. 1988.

Efeito de niveis de adubagdo e espagamento na produgao de

palmito de pupunheira (Bactris gasiPaes H.B.K.) em solo de

baixa fertilidade na região de Ouro Preto D'Oeste, RondOnia.

In: Anais do I o Encontro Nacional de Pesquisadores em

Palmito, 261-266. Curitiba: Centro Nacional de Pesquisa de

Florestas/Empresa Brasileira de Pesquisa Agropecuaria.

Gómez-Pompa, A. & Kaus, A. 1990. Traditional

management of hopical forests in Mexico, In: Alternatives to

deforestation: steps toward sustainable use of the Amazon

rainforest, edited by AB. Anderson, 45-64. New York:

Columbia University Press.

Groube, L. 1989. The taming of the rain forests: a model for

Late Pleistocene forest exploitation in New Guinea. In:

Foraging and farming - the evolution of plant exploitation,

edited by DR. Harris & G.C. Hillman, 292-304. London: Unwin

Hyman.

Harlan, J.R. 1975. Crops and man. Madison, Wisconsin:

American Society of Agronomy/Crop Science Society of

America.

Kitamura, P.C. & Muller, C.EI. 1984. Castanhais nativos de

Maraba, Para: fatores de depredagdo e bases pam sua

preservagao. (Documentos 30.) Belem: Empresa Brasileira de

Pesquisa Agropecuaria/Centro de Pesquisas Agropecuárias do

TrOpico Omido.

Lamb, D. 1991. Combining traditional and commercial uses

of rain forest. Nature and Resources, 27, 3-11.

Mendoza, 0., Picon, C. & Gonzales, J. 1989. Informe de

la expediciOn de recolección de germoplasma de camu-carnu

(Myrciaria dubia) en la Amazonia Peruana. (Informe Técnico,

no. 11, Programa de InvestigaciOn en Cultivos Tropicals.)

Lima- Instituto Nacional de InvestigaciOn Agraria y

Agmindustrial.

Mora Urpi, J. 1984. El pejibaye (Bactris gasipaes H.B.K):

origen, biologia floral y manejo agronOmico. In: Palmeras

poco utilizadas de America Tropical, 118-160. 'ffirrialba: Food

and Agriculture Organisation/Centro Agronómico Tropical de

InvestigaciOn y Ensehanza.

Mora Urpi, J. & Clement, C.R. 1988. Races and

populations of peach palm found in the Amazon basin. In:

Final report (revised): peach palm (Bactris gasipaes H.B.K)

germplasm bank edited by C.R. Clement & L. Coradin,

78-94. (US Agency for International Development project

report.) Manaus: Instituto Nacional de Pesquisas da

AmazOnia/Centro Nacional de Recursos Genéticos.

Mora Urpi, J., Bonilla, A., Clement, C.R. & Johnson,

D.V. 1991. Mercado internacional de palmito y futuro de la

explotacion salvaje vs cultivado. Boletin Pejibaye, 3, 6-27.

Moran, E.F. 1989. Models of native and folk adaptation in

the Amazon. In: Resource management in Amazonia: indigenous

and folk strategies. edited by D.A. Posey & W Balée, 22-29.

(Advances in Economic Botany 7.) New York: New York

Botanical Garden.

Mori, S.A. & Prance, G.T. 1990. Taxonomy, ecology and

economic botany of the Brazil nut (Bertholletia excelsa Humb.

& Bonpl., Lecythidaceae). In: New directions in the study of

plants and people, edited by G.T Prance & M.J. Balick,

130-150. (Advances in Economic Botany 8.) New York: New

York Botanical Garden.

C.H. 1981. Castanha-do-Brasil - estudos agron6micos.

(Documentos 1.) Belem, Para, Brazil: Empresa Brasileira de

Pesquisa Agropecuaria/Centro de Pesquisas AgropecuArias do

Tropico Umido.

Myers, N. 1984. The primary source - tropical forests and

our future. New York: Norton.

Nepstad, D.C. & Schwartzman, S., eds. 1992. Non-timber

products from tropical forests - evaluation of a conservation

and development strategy (Advances in Economic Botany 9.)

New York: New York Botanical Garden.

Padoch, C., Inuma, J.C., Jong, W. & Unruh, J. 1987.

Market-oriented agroforestry at Tamshiyacu. In: Swidden-fallow

agroforestry in the Peruvian Amazon, edited by WM. Denevan

& C. Padoch, 90-96. (Advances in Economic Botany 5.) New

York: New York Botanical Garden.

Parker, E.P. 1989. A neglected human resource in Amazonia:

the Amazon caboclo. In: Resource management in Amazonia:

indigenous and folk strategies, edited by D.A.Posey & W

Balee, 249-259. (Advances in Economic Botany 7.) New York:

New York Botanical Garden.

Peters, C.M. 1990. Population ecology and management of

forest fruit trees in Peruvian Amazonia. In: Alternatives to

deforestation: steps toward sustainable use of the Amazon

rainforest, edited by AB. Anderson, 86-98. New York:

Columbia University Press,

Peters, C.M. 1992. The ecology and economics of

oligarchic forests. In: Non-timber products from tropical

forests - evaluation of a conservation and development

strategy, edited by D.C. Nepstad & S. Schwartzman, 15-22.

(Advances in Economic Botany 9.) New York: New York

Botanical Garden.

Peters, C.M., Balick, M.J., Kahn, F. & Anderson, A.B.

1989a. Oligarchic forests of economic plants in Amazonia:

utilization and conservation of an important tropical resource.

Conservation Biology 3, 341-349.

Peters, C.M., Gentry, A.R. & Mendelsohn, R.O. 1989b.

Valuation of an Amazonian rainforest. Nature, London, 339,

655-656.

237

Picon, C. & Rarnirez, F. 1991. Cultivo intercalado de araza(Eugenia stipitata) y pijuayo (Bactris gasipaes). (Informe deAvance, Prograrna de Investigación en Cultivos Tropicales.)Lima: Instituto Nacional de Investigación Agraria yAgroindustrial.

Pinedo, M., Ramirez, F. & Blasco, M. 1981. Notaspreliminares sobre el araza (Eugenia stipitata), frutal nativo dela Amazonia Peruana. Lima: Ministerio de Agricultura yAgropecuaria/Instituto Nacional de InvestigaciónAgricola/Instituto InterAmericano de Cooperación Agricola

Posey, D.A. 1985. Indigenous management of tropical forestecosystems: the case of the Kayapo Indians of the BrazilianAmazon. Agroforestry Systems, 3, 139-158.

Posey, D.A. & Balée, W. eds. 1989. Resource management inAmazonia: indigenous and folk strategies. (Advances inEconomic Botany 7.) New York: New York Botanical Garden.

Beck lin, J. 1982. The orinoconut - a promising tree crop forthe tropics. International Tree Crops Journal, 2, 105-119.

Rindos, D. 1984. The origins of agriculture: an evolutionaryperspective. New York: Academic Press.

Sånchez, J.S. 1973. ExplotaciOn y comercializaciOn de lacastalia en Madre de Dios. (Informe no. 30.) Lima: Ministeriode Agricultura, Dirección General de Florestas y Caza.

238

Sanchez, P.A., Bandy, D.E., Villachica, J.H. &Nicholaides, J.J. 1982. Amazon basin soils: management forcontinuous crop production. Science, 216, 821-827.

Smith, N.J.H., Williams, J.T., Plucknett, D.L. & Talbot,J.P. 1992. Tropical forests and their crops. Ithaca, New York:Cornell University Press.

Szott, L., Palm, C.A. & Sanchez, P.A. 1991. Agroforestryin acid soils of the humid tropics. Advances in Agronomy, 45,275-301.

Venturieri, G.A., Alves, M.L.B. & Nogueira, M.D. 1985.0 cultivo do cupuaguzeiro. lnformativo Sociedade Brasileira deFruticultura, 4, 15-17.

Villachica, H. 1993. Crop diversification in Bolivia, Colombiaand Peru: potential to enhance agricultural production. In:Agricultural alternatives to coca production. Washington, DC:Office of Technology Assessment, Congress of the UnitedStates of America. In press.

Villachica, H., Silva, J.E., Peres, J.R. & Rocha, C.M.C.1990. Sustainable agricultural systems in the humid tropics ofSouth America. In: Sustainable agricultural systems, edited byCA. Edwards, 391-437. Ankeny Iowa: Soil and WaterConservation Society

Genetic improvement of West African tree species:past and present

D 0 Ladipol , S P K Britwum2 , Z Tchoundjeu3 , 0 Oni 4

& R R B Leakey51International Centre for Research in Agroforestry, do International Institute of Tropical Agriculture,

PMB 5320, Ibadan, Nigeria2Forestry Research Institute of Ghana, University Post Box 63, Kumasi, Ghana3Office National de Dévéloppement des Forêts, Forest Management and Regeneration Project, BP 163,

Mbalmayo, Cameroon4Forestry Research Institute of Nigeria, PIV1B 5054, Ibadan, Nigeria5Institute of Terrestrial Ecology*, Bush Estate, Penicuik, Midlothian EH26 OQB, UK

*A component of the Edinburgh Centre for Tropical Forests

ABSTRACTInternal and external demand for indigenous tropical hardwoods is increasing in many West African countries, while natural forests

have become depleted as a result of over-exploitation. Selective logging has probably eroded the gene pools of some

tiaditionally well-known species. Owing to a lack of information and various biological problems, insufficient seed is available for

the reforestation programmes which are required to reduce the deforestation pressures on the surviving areas of natural forest.

The identification of these biological problems, and the realisation of their imminent environmental and commercial consequences,

resulted in enhanced research efforts in various West African countries in the late 1960s. Studies on seed management,

provenance evaluation, vegetative propagation and genetic selection for improved yield and quality are some of the areas of

reseamh initiated for the domestication of a few important indigenous hardwood species. These developments by the individual

countries of West and Central Africa have been supported by the UK Overseas Development Administration, the International

nopical Timber Organisation, the Danish International Development Agency and the World Bank.

Provenance trials and gene banks have been established in Ghana, Cameroon, Nigeria and in COte d'Ivoire, representing much of

the geographical range of certain species, with clonal testing also initiated in Nigeria, C6te d'Ivoire, and currently Cameroon.

Clonal evaluations have resulted in better understanding of genetic variability, and for 7'riplochiton scleroxylon (obeche) an overall

gain in stem volume of over 30% has been achieved. Studies on the early growth of obeche clones have included the

relationship between branching frequency and apical dominance, as well as measurements of photosynthesis, as aids to early

selection of desirable genotypes. Developments from these research efforts have enhanced reforestation, particularly in Côte

d'Ivoire and Nigeria.

At least in the short term, substantial benefits could be achieved in West Africa by the establishment of clonal plantations of

indigenous trees, exploiting variation present in wild populations. Renewed efforts in this area could help to ensure a continuing

timber supply in the future.

INTRODUCTION

In most West African countries, the demand for

indigenous tropical hardwoods for export anddomestic consumption is increasing, while thenatural forest is being depleted as a result of

Table 1. Balance of supply and demand for major wood products in Nigeria (millions of m3 roundwood equivalent) (source:

Agricultural Development in Nigeria, 1983-2000, Forestry Sector Review)

over-exploitation (Table 1). Population pressure on

the forests and demand for land for other uses

are major issues affecting the forest industry

(Table 2). In addition to these problems, selective

exploitation of the remaining forest resources has

239

Table 2. Area of tropical moist forests in some West Africancountries (source: FAO/UNEP 1981)

probably reduced the diversity of the availablegene pool of most of the important timberspecies, such as Milicia (syn. Chlorophora)excelsa, Entandrophragma angolense, Terminaliaspecies and Triplochiton scleroxylon, amongothers. However, the information on this aspect isscant.

Hardwood timber supplies have traditionallydepended on natural forest managementtechniques for regeneration (Kio 1983), until thismethod was found to be insufficiently productive(Food and Agriculture Organisation 1989). Thedependence on natural regeneration, coupledwith poor control of logging in the ForestReserves, has led to over-exploitation of theforest in humid W Africa, except where theterrain does not allow easy access, or where theareas are traditionally protected (fetish groves).Such areas are a valuable resource ofgermplasm, but they are being threatened bydeforestation, which is continuing to have seriousenvironmental and economic consequences.Reviewing the forest situation in Côte d'Ivoire,Caulfield (1982) reported that this country hadlost more than 70% of its rainforests since theturn of the century A continuing annual rate ofdeforestation of around 10% was expected tomake Côte d'Ivoire a net importer of timber inthe 1990s (Bourke 1987). Similarly Nigeria andGhana are expected to be unable to meet theinternal demand for timber in a very short time.Because of this, Nigeria and Ghana bannedfurther exports of certain timbers in 1976 and1978, respectively in order to help meetdomestic wood requirements (Kio 1983). Partly asa result of its lower population and poor accessto the eastern part of the country Cameroon stillhas a significant timber resource, but at theInternational Tropical Timber Organization (1110)meeting at Yaoundé in May 1992, it was revealedthat Cameroon loses 200 000 hectares of itsmoist forests annually

With the high rates of deforestation, it becameclear in the late 1960s that action was needed torestock the moist forests and to initiate intensiveplantation forestry At this time, mostre-afforestation schemes in West African countries

240

depended on exotic fast-growing hardwoodssuch as Tectona grandis and Gmelina arborea tomeet the future hardwood needs of thepopulation. Eucalyptus and Pinus species weregrown for pulp. Seeds of these exotic specieswere available from trials located at various sitesin these countries, so allowing theimplementation of large-scale reforestationprogrammes for the production of lumber,industrial cellulose and poles. Attempts to plantindigenous species were also made, wheneverseeds were available. However, the tendencywas to plant exotics rather than indigenousspecies, as the silviculture of the former wasbetter known. In the case of Tnplochitonscleroxylon, the lack of planting was due to thepoor and irregular seeding habit and thesusceptibility of the seeds to insect attack(Britwum 1973b; Jones 1974; Howland & Bowen1977). In others (iroko and mahogany), therewere severe pest problems. In general, however,exotics were preferred because of their fastergrowth and the expectation of greater economicreturns.

Despite early interest in indigenous hardwoodsgoing back to colonial times (eg 1930-50s), itwas not until the late 1960s and 1970s that itwas realised that research on regenerationmethods was needed, and that tree improvementcould generate improved economic returns. Treeimprovement programmes were, therefore,initiated at this time in Ghana, Côte d'Ivoire andNigeria. The present paper presents what hasbeen accomplished over the past 25 years underthese and other more recent programmes withindigenous species.

COLLABORATIVE SCHEMES FORRESEARCH AND DEVELOPMENT

The history of tree improvement in West Africacannot be complete without mention of variousoverseas governments, institutes and agencies.The example of T scleroxylon will beemphasised as this was recognised as thespecies for which there were major problemsand great commercial interest. The first actionwas in Ghana, at the Forest Products ResearchInstitute at Kumasi, where clonal seed orchards ofT scleroxylon and Terminalia ivorensis wereestablished in 1972 and 1973 and techniques ofrooting juvenile cuttings were pioneered (Britwum1970b), with external funding by the UKOverseas Development Administration (ODA).Subsequently in 1971, the West Africanhardwoods improvement project (WAHIP) wasinitiated for the most promising indigenoushardwood tree species of West Africa at theFederal Department of Forest Research (later theForest Research Institute of Nigeria [FRIN]) atIbadan. WAHIP continues to the present day butoverseas funding ceased in 1977. This project

was based on the use of vegetative propagationand clonal selection, again with external fundingfrom ODA (Howland & Bowen 1977). In 1974,WARP was supplemented by a project to studythe physiology of rooting in T scleroxylon at theUK Institute of Tree Biology (later the Institute ofTerresti ial Ecology MED in Edinburgh. Researchon this and other West African hardwoods at rrEhas continued up to the present. Together, theseprojects have developed techniques inT scleroxylon to store seeds, root stem cuttings,select superior clones, induce flowering andestablish breeding programmes, as part of anoverall package to domesticate this commerciallyimportant timber tree (Longman & Leakey 1993).

As a result of this research, commercial plantingof clonal T scleroxylon was initiated by SafaSplintsin the 1980s, using a mixture of selectedclones provided by FRIN. A somewhat similarprogramme of vegetative propagation and clonalselection started in Côte d'Ivoire in the 1970s atCentre Technique Forestier Tropical (CTI-'1'), withfunds from the French Government (Verhaegen1992). The practical application of thesetechniques, particularly with Terminalia ivorensis,was initiated as part of a World Bank/Commonwealth Development Corporation projectbased in the Société de Dévéloppement desForêts (SODEFOR), in collaboration with C11-1.Later in 1987 with funds from French Aid andCooperation Fund, a similar industrial-scaleproject was initiated by SODEFOR at Téné(Verhaegen et al. 1992). Currently work continuesat the Institut de Dévéloppement des Forêts(IDEFOR, previously C'll-T) with clonal selectionof Triplochiton scleroxylon.

In the early 1980s, an attempt was made toinitiate a 'regional programme for improvement oftropical hardwoods for West and Central Africa'.Detailed proposals were prepared with fundingfrom the European Commission (Leakey &Grison 1985) but the programme, which spannedSierra Leone, Liberia, Côte d'Ivoire, Ghana,Nigeria, Cameroon and Congo, never materialised.

In the late 1960s and early 1970s, C11-1 in theCongo was also developing a clonal approach toindigenous hardwoods (in parallel with theirwell-known programme on Eucalyptus species atPointe Noire [Delwaulle, Laplace & Quillet 1983]),but with Terminalia superba instead ofTriplochiton scleroxylon (Martin & Quil let 1974;Koyo 1983, 1985). Similarly early studies inGabon by CTI-T concerned another hardwood,Aucoumea klaineana (Grison & Hamel 1978).

In addition to timber tree species, geneticimprovement studies were in progress in Nigeriain the 1970s and 1980s with indigenous fruittrees (see Okafor & Lamb, pp34-41). Furtherwork to domesticate fruit trees is currently inprogress in:

• Cameroon with Ricinodendron heudelottii andlrvingia gabonensis at the Forest ResearchStation of Agronomic Research Institute atKurnba (P Shiembo, unpublished) and withDacryodes ednlis (J Kengue, unpublished) atthe Agronomic Research Institute atNkolbisson; and

• Côte d'Ivoire with Coula edulis (L Bonnélin,unpublished) in `raj National Park, incollaboration with Wageningen University

Currently, much of the research done withT scleroxylon over the past 20 years is beingapplied in Cameroon by the Office National deDévéloppement des Forêts (ONADEF formerlythe Office National de Regénération des Forêts(ONAREF)) in collaboration with ITE and withexternal funding from ODA (see Lawson,pp112-123). This project is part of the Cameroonand UK governments' response to the TropicalForest Action Plan. This project has built on theearlier activities of the ONAREF Parc deBouturage (Leakey 1985), which was part of theCameroon/World Bank forestry project.

Current studies are also in progress in Ghana,with ri-ro funds, to select and propagatevegetatively clones of Milicia (syn. Chlorophora)spp. resistant to the gall-forming psyllidPhytolyma lata (Cobbinah 1990). Further work onthe vegetative propagation of M excelsa is inprogress at FORIG, Kumasi (D Ofori, unpublished).

Studies have also been made with clones ofGmelina arborea in Cameroon (Bibani 1983) andin Côte d'Ivoire (Kadio 1990), while otherapproaches to tree improvement are in progresswith exotic hardwoods in Ghana, eg Tectonagrandis provenance trials in collaboration withDanish International Development Agency; Cordiaalliodora and Cedrela odorata provenance trialsin collaboration with Oxford Forestry Institute;and Gliricidia sepium with the InternationalCentre for Livestock in Africa.

DEVELOPMENTS FROM RESEARCHACTIVITIES, ESPECIALLY WITHTRIPLOCHITON SCLEROXYLON

Substantial information has been acquired overthe past 20-25 years of research on variousaspects of tree improvement in West Africa. Toestablish plantations, and to make better use ofavailable land, there is a need to domesticatecommercially important species. Genetic variationwithin a tree species offers the opportunity forselection of superior high-yielding populationsand individuals. However, very few tropical treeshave been improved genetically because of thedifficulties of making improvements by breedingin tree species which have such a long lifecycle, which do not flower regularly and do notbecome sexually mature for many years, andwhich are not self-fertile (Burley 1985).

241

Seed storage

It is now known that seed of T scleroxylon isavailable each year in small quantities and thatmast years occur every four to seven years.Further, with adequate seed harvesting,processing and storage as developed byHowland and Bowen (1977), viability can bemaintained for six years (Ladipo 1984). Regularpower failures, however, result in considerablelosses of stored germplasm. Seed of a numberof other species (eg Mansonia altissima andEntandrophragma angolense) have also beenstored successfully (Gyimah 1984; Damptey 1968).

Provenance trials

Provenance trials provide information ongeographical variation within species, whichenables the best seed sources to be identifiedand selected. This was the first strategyemployed in tree improvement in West Africa(Britwum 1973a, 1978).

Triplochiton scleroxylon

A major provenance trial established in Nigeriaunder WAHIP included investigation based onseed collected from the following six countries:Sierra Leone, Liberia, Côte d'Ivoire, Ghana,Nigeria and Cameroon, forming the West Africanmoist forest region and spanning most of thenatural range of this species. Results revealedsubstantial variation in growth, with the Nigerianand C6te d'Ivoire provenances showing thegreatest promise (Howland & Bowen 1977).

Working on some Nigerian provenances, Adedire(1986) also reported substantial early variationbetween the seecllots evaluated. The Oturkpoprovenance performed better in height anddiameter than the Obubra and Ubiajaprovenances.

Terminalia superba

In 1969, C11-'1; Cote d'Ivoire, started provenancetrials of T superba using 16 provenances fromSierra Leone (1), Côte d'Ivoire (3), Ghana (3),Cameroon (3), Gabon (3), Congo (1) and CentralAfrican Republic (2). Results of assessment threeyears after establishment showed significantdifferences between provenances in heightgrowth and stem form (Delaunay 1978a). Thebest provenance was from Ghana (Abofour) witha mean height of 9.4 m, almost double that ofDahomey (Quedo) provenance with a height of5.3 m. The Ghana-Abofour provenance wasfollowed by two provenances from Côte d'Ivoireand two from Cameroon. The provenances fromCongo and Gabon were poor and that ofDahomey was the worst. Studies usingelectrophoresis have examined the polymorphismof enzymatic systems differentiating 'fraké' inC6te d'Ivoire from limba' in Congo (Vigneron1984). The former were more polymorphous.

242

Terminalia ivorensis

This species has been planted extensively inmany countries in West Africa, especially underthe taungya system in high forest areas.Provenance trials exist in Ghana, Côte d'Ivoireand Cameroon.

In 1973, Ghana and C6te d'Ivoire exchangedT ivorensis seeds for provenance studies. Theprovenances tested in Ghana included five fromGhana, three from Côte d'Ivoire and one fromCameroon. Results of assessment for heightgrowth in Ghana showed a highly significantdifference between the Cameroon provenanceand the provenances from Ghana and Côted'Ivoire (Britwum 1978; Delaunay 1978b). TheCameroon seed source was the poorest on allthe three sites tested. There were no significantdifferences between the Ghana and the Côted'Ivoire provenances. Some of the trees in thisstudy started to flower in 1976. There was noobserved provenance variation in flowering.

Tree breeding

Plus-tree selection and seed orchards

To aid genetic improvement, it is important thatsuperior genotypes are identified and centralised.In Nigeria (FRIN), budwood from 105 plus-treesof Triplochiton scleroxylon were collected andpropagated by grafting before 1976 (Howland &Bowen 1977). These were preserved in fieldplantings at three sites and in the nursery Manyof the identified and labelled plus-trees infarmland were subsequently felled. Thus, thegrafted material provides an importantgermplasm resource, although relatively few ofthese grafts have subsequently provided seeds.These nursery plants at least are not subjectedto the risks of fire or logging.

A similar programme of selecting provisionalplus-trees of T scleroxylon and T ivorensis inGhana, in the natural high forest, was started in1968 (Jones 1968, 1970; Britwum 1970a). Thephenotypic characteristics on which selection wasbased, as in Nigeria, included height growth andtree form. Most of these plus-trees, other thanthose in clonal seed orchards, have subsequentlybeen lost through logging.

Clonal seed orchards

Apart from the nursery stock described above,all other propagules derived from 30 matureplus-trees were in established experiments andhave now been converted into clonal seedorchards (Howland & Bowen 1977) at field sitesin southern Nigeria.

Clonal seed orchards of T scleroxylon and7brminalia ivorensis were established in Ghana in1972 and 1973, using scion material collected fromprovisional plus-trees and budded on to seedlingroot stock. r1Wenty clones of each species were

used to establish the two orchards in SouthFonnangsu Forest Reserve (Britwum 1973a).

Flowering and fruiting in both the Ghanaian andNigerian 7'riplochiton scleroxylon seed orchardshave generally been poor. Some of the cloneshave never fruited, but clone 4 in Ghana hasflowered almost every year since 1974. InNigeria, two trees of S93 in Sapoba haveflowered consistently and much more frequentlythan the other clones.

Vegetative propagation

Cuttings of T scleroxylon were rooted originallyto circumvent seed problems. Subsequently, theextent of clonal variation indicated the potentialof selecting clones for their superiority in growthand form. Research on the rooting of single-nodeleafy cuttings has reflected these changes in thereason for clonal propagation.

In the early 1970s, the objective was to achievea practical and reliable method of bulking upplanting stock (Britwum 1970b; Okoro 1974;Howland 1975) using mist and non-mistpropagators. Ladipo (1985) further demonstratedthat rooted cuttings could be as good asseedlings for commercial field use and that noproblems of plagiotropism occurred. From 1974,greater emphasis was placed on understandingthe sources of variation in rooting ability and theachievement of physiological understanding, suchthat stockplant management could ensurecost-effective and sustainable methods of rootingto retain juvenility during mass propagation(Leakey 1983; Leakey Dick & Newton 1992). Thiswork to acquire physiological understanding ofthe rooting process continues and is presented ingreater detail by Leakey Newton and Dick(pp72-83).

More recently since 1986, progress has beentowards scaling up the operation and applyingthe existing knowledge in a forest regenerationproject in Cameroon (Lawson, pp112-123) andC6te d'Ivoire (Verhaegen et al. 1992). Currently,in Cameroon, propagation in the ONADEF/ODAforest management and regeneration project'sParc de Bouturage is producing about 20 000rooted cuttings for planting about 30 ha eachyear (Leakey Njoya & Bockett 1992; Lawson,pp112-123). This project is utilisinglow-technology non-mist propagators (Figure 1),which have been adapted from an earlier designused in Nigeria. These propagators are cheapand simple and so are suitable for ruraldevelopment projects in the tropics (Leakey et al.1990).

In Côte d'Ivoire, cuttings are rooted under mistpropagators with a capacity of 250 000 cuttingsper year. For research purposes, thesepropagators are enclosed with shade cloth (40%shade) and misted twice a day (Verhaegen et al.

1992). On average, the rooting success is 68%.In the period December 1990 to March 1991,40 000 cuttings had been set in thesepropagators, and mean rooting success hadincreased to 80%.

The developments with T scleroxylon have nowbeen transferred to other species (Table 3), withintensive projects on Khaya ivorensis and Lovoatrichilioides (Tchoundjeu 1989; see also Newtonet al., pp256-266). In addition, detailed studieson in vitro micropropagation have beencompleted with Khaya ivorensis and Naucleadiderrichii (Mathias 1988).

Clonal selection

Selection for improvement is one element of thedomestication process currently employed forTfiplochiton scleroxylon and some other nativehardwood species in West Africa. It has beenused at FRIN and by HEin Nigeria (Howland &Bowen 1977; Leakey & Ladipo 1986), atC11-1/IDEFOR in Côte d'Ivoire (Verhaegen et al.1992), and has recently started at ONADEF inCameroon (Leakey Njoya & Bockett 1992). Manyspecies have problems of seed availability andvegetative propagation is capable of

25 cm

25 cm Wat-e—r1

./_-

1 • 'tk..)..)

Rootingmedium Water

ravel tableStones

SandPolythene

Figure I. Diagram of the construction of an ITE non-mistpropagator (source: Leakey et al. 1990)

o0

243

7'able 3. Indigenous timber tree species of moist forests of W Africa vegetatively propagated using intermittent mist and simplelow-technology propagators

244

circumventing them. With propagation methodsalready well developed, it is then possible toincrease the productivity of the plantations oragroforestry systems by selecting elite clonesand propagating them vegetatively to providelarge quantities of genetically superior plantingstock.

In Nigeria, clonal trials representing the fullgeographical range of T scleroxylon have beenestablished at five sites. The first phase of thestudy examined the variation in a wide range ofgrowth characteristics among large numbers ofclones of different seed origins in an attempt toidentify criteria for clonal selection (Ladipo et al.1983). Selection of the ten tallest clones at 18months old from this randomly chosen clonalpopulation of 98 clones originating from 15seedlots showed a potential height gain of 16.5%over the mean of all plants. At nearly five years,a selection of 33% of the clones with bothabove-average mean stem volume and meanstem scores resulted in potential overall gains involume of 30.5%. These gains could besubstantially improved, however, by the selectiOnof the best ten clones, resulting in about 80%improvement in yield (Leakey & Ladipo 1986).For a large-scale clonal reforestation programme,it is necessary to have a strategy of clonalselection and deployment that will maximiseproductivity while minimising the risk of utilisinga narrow genetic base and also minimising therisks of ecological problems. Consequently anapproach has been developed for T scleroxylonto introduce a large number of unrelated clones(Leakey 1991; see also Foster & Bertolucci,pp103-111), and to ensure the conservation andwise use of germplasm.

To achieve this objective by testing largenumbers of clones in field trials is notpracticable, so efforts have been made toidentify desirable characteristics at an early age.

This approach is geared towards screening for'superior' seedlings in the nursery as apreliminary procedure to reduce the numbers ofclones to be tested in the field. Havingexamined the possibilities of identifying selectioncriteria, it became apparent that branchingfrequency was important to both yield and form,with low branching frequency being preferred.Consequently after standardising environmentaland morphological variables, Ladipo (1981)examined clonal variation in apical dominance(the process of determining branchingfrequency). He identified conclitions under whicha nursery study of sprouting followingdecapitation could be used as a 'predictive test'to select potentially superior clones (Ladipo,Leakey & Grace 1991a, b). Using this test, astrong relationship (r=0.76; P=0.001) was foundbetween the bud activity at weeks 3-4 in youngdecapitated plants of T scleroxylon and thebranchiness of four-year-old trees of the sameclones in plantations (Ladipo et al. 1991c). It,therefore, seems that, at least for treesconforming to Rauh's model of branchingarchitecture (Ha 116, Oldeman & Tomlinson 1978),there is potential for screening seedlings in thenursery for those worthy of inclusion in clonalfield trials.

The predictive test described above is currentlybeing utilised in the Parc de Bouturage of theONADEF/ODA forest management andregeneration project at Mbalmayo in Cameroon.Recently 14 000 seedlings of T scleroxylon of tenseed origins have been subjected to the test toidentify those for subsequent cloning and fieldplantings (see Lawson, pp112-123).

In C6te d'Ivoire, a different approach has beentaken to clonal selection in T scleroxylon. In thisinstance, a four-year-old plantation of 191 ha ongood soil at Mopri was used to select plus-trees(Verhaegen et al. 1992). The seeds for this

plantation were of local origin. It was felt that atfour years of age these trees, which had a meanheight of 11 m and basal area of 10 m2 ha-4,had expressed their potential and yet wereyoung enough to be multiplied vegetatively Asthe first stage, the best 35 trees ha-1 wereselected (5% selection intensity) on the basis ofvisual appearance. In the second stage, these 35trees were assessed for their form and vigourrelative to their neighbours (Boutin 1983, 1991).Relative vigour (E) was assessed as stemcircumference at 1.5 m of the selected tree (Cs)and the mean circumference of its eight nearestneighbours (MCv):

(C52—MCv 2)

E8 cv

where cv is the coefficient of variation. On thebasis of this analysis, 308 trees were selected, ofwhich 154 were selected by a second visualinspection as plus-trees. This represents an 0.3%selection intensity

The plus-trees were then coppiced at 0.4-0.5 min 1984 and cuttings collected one month later;116 clones were established. Stockplants of theSeclones have subsequently been planted at Téné,and clonal trials covering 23.8 ha established in1987-90 at Mopri, Téné and Sangoué. Theseclones have performed better at Mopri than atTéné (Kadio, Legaré & Bohousson 1991). Aftertwo to five years, the mean performance of theclones was 14% better than a control plot ofunselected seedlings (Verhaegen et al. 1992).

Ladipo et al (1984) reported considerable clonalvariation between clones of T scleroxylon in netphotosynthesis, stomatal resistance and mesophyllresistance. A good relationship was found(r=0.81; P=0.05) between gas exchangeparameters and yield. To date, this relationshiphas not been used in selection programmes, andfurther work is needed to take advantage ofsuch physiological selection criteria.

RESEARCH ON WILD FRUIT OR FOODTREES

The basic principles of vegetative propagationand clonal selection of indigenous hardwoods are

Table 4. Some indigenous fruit tree species vegetativelY propagated using simple horticultural methods in West Africa

equally applicable to the domestication of wildfruit tree species (see Okafor & Lamb, pp34 41)(see Table 4).

In addition to Okafor's (1975, 1981) extensivestudies with Irvingia gabonensis, Treculia africana,and other species, there are currently otherstudies in Nigeria by the International Centre forResearch in Agroforestry (ICRAF) (multipurposetree species project), in Cameroon by theAgronomic Research Institute, and in Côted'Ivoire by Wageningen University Furtherobjectives of these projects are to determine theecological, biological and physiological factorsinfluencing the domestication process, to selectsuperior trees for mass production and to testthese cultivars under agroforestry situations.Techniques of marcotting have been developedin Nigeria (ICRAF/International Institute of TropicalAgriculture/Oregon State University) for Treculiaafricana, Irvingia gabonensis and Dacryodesedulis, while the rooting of stem cuttings hasbeen used in Cameroon for Ricinodendronheudelotii and Irvingia gabonensis. This approachis particularly aimed at improving the compoundfarm system of agroforestry

Currently there are proposals to add studies onthe vegetative propagation and clonal selection offruit trees to the ODA/ONADEF project inCameroon, perhaps including the domesticationof Cola nitida, Cola lepidota, Xylopia spp., Colapachycarpa, Tetrapleura tetraptera, etc, and theirgenetic improvement by selection at ICRAF(Nigeria/Cameroon). This aspect of the project isimportant as the demand for some local fruitproducts in the urban areas of West Africa isexpanding, while some products like the cola nuthave considerable importance in regional trade.

GENETIC RESISTANCE TO PESTS

Not surprisingly there are pest and diseaseproblems associated with cultivating indigenoustree species. Recently however, attempts havebeen made to search for pest-resistant genotypes(Ladipo 1986; Cobbinah 1990). In Nigeria, workon Terminalia ivorensis leaf gall commenced in1986, while at the Forestry Research Institute ofGhana (FORIG) work is in progress with ITTOfunds to screen Iroko half-sib progenies of Milicia

245

excelsa for resistance to Phytolyma lata galls.Evidence has been obtained that someM excelsa trees are probably resistant to attackby this psyllid (Cobbinah 1990). This ITTOproject aims to quantify possible sources ofresistance and to test them using two bioassaytechniques (insect bioassay and the standardchoice bioassay). For these tests, bothsusceptible and resistant trees will be clonedusing stem cuttings. Initial results have shown thatsome individuals are less frequently attacked andthat insects in the .galls formed do not developto maturity Similar studies on genetic resistanceto Hypsipyla shoot-borers in Central Americanmahogany (Newton et al., pp256-266) could beapplied to the W African members of theMeliaceae, such as Khaya spp.

REPRODUCTIVE BIOLOGY AND BREEDINGThere have been limited studies on the breedingsystems of tropical hardwoods of West Africa.Howland and Bowen (1977) reported thatTnPlochiton scleroxylon is self-sterile and ispollinated by many insect species from severalfamilies. Pollen management and storage ofT scleroxylon have been investigated (LeakeyFerguson & Lonyinan 1981; Oni, Fasehun &Ladipo 1988) and deep-frozen pollen has beensuccessfully used for cross-pollination. Attemptsto induce precocious flowering in T scleroxylon,while not successful in all clones, diddemonstrate that flowering can be initiated inyoung plants (18 months old) by inducing rootdormancy in plants with active shoots (Leakey etal. 1981). Subsequently cross-pollinations resultedin the development of normal seedlings and, inone instance, a second-generation progeny wasalso created as a result of precocious flowerinduction. It thus appears that there is potentialto add early breeding techniques to thedomestication procedures of this species and soobtain full-sib progeny on a short generationcycle (1.5-2 years). Studies by Grison (1978) onAucoumea klaineana described the flowers andtheir pollination mechanisms, and reported somesuccessful controlled pollinations.

Recent studies on Terminalia ivorensis haveshown that the species is andromonoecious. Theflowers are either male or bisexual, with bothhermaphrodite and male flowers occurring onthe same spike. The distribution of male andhermaphrodite spikes on flowering twigs ofT ivorensis differed significantly between sites at1% probability The ratio of male/hermaphroditespikes (MF:HF) is 1.9:1 (Oni 1989). The flowersare mostly insect-pollinated. Results of controlledpollination carried out on the species indicatedself-compatibility The reproductive efficiency ofthe species is 21%. Pollen grains of T ivorensisare amenable to storage at —17°C or —30°C and

246

can be grown successfully on a mediumcontaining 0.1% boric acid + 20% sucrose (Oni1989). In Ghana, seeds have been obtained froma clonal seed orchard. The size of the fruitsvaried among the different clones.

Flowering and flower morphology of T superbahave also been studied in Ghana (see FORIGTechnical Note, no. 29). Flowering occurs at notearlier than ten years of age, and takes placebetween mid-October and early January Theflowers appear in a whorl of spikes, and last forone month. The whorl contains 19 flower spikes,of which nine are entirely composed ofhermaphroditic flowers, two of only male flowers,and the remaining eight contain hermaphroditicflowers at the base with staminate flowersarranged above them in an average ratio of29:47. The ovary is inferior and unilocular. Thehermaphroditic flowers are self-compatible, andpollination by the staminate flowers is restrictedif not inhibited.

GERMPLASM CONSERVATION

Deforestation poses a major threat to the geneticresources of indigenous tree species in WestAfrica. In Nigeria, 12 strict Natural Reserves havebeen established in various ecological zones toprovide ih situ conservation of these and manyother species. Similar action has been or isbeing taken in several other countries in theregion. With more efficient control andmanagement, Forest Reserves can play a majorrole in germplasm conservation. However, asalready mentioned, ex situ conservation by seed,pollen and live plants also has an importantcontribution to play especially when linked totree improvement programmes.

CONCLUSIONS

In conclusion, much progress has been made inWest Africa by local research organisations withand without international collaboration. What isneeded is greater regional collaboration,particularly in germplasm collection andexchange, and the widespread application of thisresearch, together with ongoing studies todeepen understanding and broaden the spectrumof species domesticated for the production oftimber and non-timber products. Thisdevelopment would have significant implicationsfor the timber and wood supply economy andenvironment of West Africa.

ACKNOWLEDGEMENTS

We thank various national governments and theinternational organisations mentioned for theircontributions to the research reported here. Thispaper is dedicated to our past and present theprogress that has been made over the years.

REFERENCESAdedire, M.O. 1986. Studies on early provenance variations

in the growth and development of Triplochiton scleroxylon K

Schum. MSc thesis, University of lbadan.

Bibani, M. 1983. Selection et amelioration genetique des

essences papetières a Edea: bouturage du Gmelina arborea(Linn). Yaounde: Institut de la Recherche Agronomique.

Bourke, G. 1987. Forests in the Côte d'Ivoire face extinction.

New Scientist, 114 (1564), 22.

Boutin, B. 1983. Samba (Triplochiton scleroxylon K. Schum).

Project de selection d'arbres '+' en plantation (parcelles 1979)du chantiér SODEFOR de Mopri. Abidjan: Centre Technique

Forestiér Tropical.

Boutin, B. 1991. Selection phénotypique des arbres '+' at

regles établies pour leur mobilisation par bouturage dans les

parcelles de samba (friplochiton scleroxylon K Schum.)plantées en 1979 sur le chantier SODEFOR de Mopri (COte

divoire). Abidjan: Centre Technique Forestiér Tropical.

Britwum, S.P.K. 1970a. Preliminary selection of provisional

plus-trees in tropical high forest. 7bchnical Newsletter of the

Forest Products Research Institute of Ghana, 4 (2).

Britwum, S.P.K. 1970b. Vegetative propagation of sometropical forest trees. Technical Newsletter of the Forest ProductsResearch Institute of Ghana, 4, (1), 10-15.

Britwum, S.P.K. 1973a. Status of forest improvement inGhana. In: Tropical provenance and progeny research and

international co-operation, edited by J. Burley & G. Nildes,588-589. Oxford: Commonwealth Forestry Institute.

Britwum, S.P.K. 1973b. Seed problem of indigenousplantation species in Ghana. In: Proceedings of the 1UFRO

International Seed Symposium, Norway, vol. II (7), 1-6.

Britwum, S.P.K. 1978. Species and provenance selection of

tropical hardwood species with special reference to WestAfrica. In: Proceedings of the Third World Consultation on

Forest Tree Breeding, vol. I, 147-153. Canberra: CSIRO.

Burley, J. 1985. Global needs and problems of the collection,

storage and distribution of multipurpose tree germplasm:science and practice of agroforestry. Nairobi: InternationalCentre for Research in Agroforestry

Caulfield, C. 1982. Tropical moist forests. London: Earthscan

Cobbinah, J.R. 1990. Biology seasonal activity and control of

Phytolyma lata (Homoptera: Psyllidae). In: Pests and diseases

of forest plantations, edited by C. Hutacharern,K.G. MacDicken, M.H. Ivory & K.S.S. Nair, 180-185. (RAPAPublication 1990/9.) Bangkok: Food and Agriculture

Organisation.

Damptey, H.B. 1968. Storage of seeds of Entandrophragma

angolense (We lw) C.D.C. Technical Newsletter of the ForestProducts Research Institute of Ghana, 2 (3/4).

Delaunay, J. 1978a. 7brminalia superba Engl. and Die Is.provenance trials six years after their establishment in the

Ivory Coast. In: Proceedings of IUFRO Workshop on the

Progress and Problems of Genetic Improvement of Tropical

Forest Trees, edited by D.G. Nikles, J. Burley & R.D. Barnes,

vol. 2, 879-885. Oxford: Commonwealth Forestry Institute.-

Delaunay, J. 1978b. Preliminary results from provenance-trials of 7brminalia ivorensis A. Chev, two and a half years

after their establishment in the Ivory Coast, In: Proceedings of1UFRO Workshop on the Progress and Problems of Genetic

Improvement of Tropical Forest Trees, edited by D.G. Nikles,

J. Burley & RD. Barnes, vol. 2, 874-878. Oxford:Commonwealth Forestry Institute.

Delwaulle, J.C., Laplace, Y. & Quillet, G. 1983.Production massive de boutures d'Eucalyptus en Republique

Populaire de Congo. Silvicultura, 8, 779-781.

Food and Agriculture Organisation. 1989. Management of

tropical moist forests in Africa. FAO Forestry Paper, 88,107-132.

Food and Agriculture Organisation/United NationsEnvironment Programme. 1981. Tropical forest resources

assessment project (in the framework of GEMS). Forestresources of tropical Africa. Part I. Regional synthesis. Rome:FAO.

Grison, F. 1978. Recherches sur la biologie floral del'Okoumé. In: Proceedings of the 3rd World Consultation onForest Tree Breeding, vol. 2, 915-923. Canberra: CSIRO.

Grison, F. & Hamel, 0. 1978. Essais sur le bouturage del'Okoumé. In: Proceedings of the 3rd World Consultation on

Forest Tree Breeding, vol. 2, 925-932. Canberra: CSIRO.

Gyimah, A. 1984. Storage of Mansonia altissima seeds.7bchnical Newsletter of the Forest Products Research Institute ofGhana, 4, 16-20.

Ha llé, P., Oldeman, R.O. & Tomlinson, P.B. 1978.Tropical trees and forests: an architectural analysis. New York:

Springer.

Howland, P. 1975. Vegetative propagation methods forTriplochiton scleroxylon K. Schum. In: Proceedings ofSymposium on Variation and Breeding in Triplochitonscleroxylon (K Schum), lbadan, Nigeria, 99-109. Ibadan:Federal Department of Forest Research.

Howland, P. & Bowen, M.R. 1977. Triplochiton scleroxylonK Schum. and other West African tropical hardwoods. (West

African Hardwoods Improvement Project, Research Report

1971-1977.) Ibadan: Forestry Research Institute of Nigeria.

Jones, N. 1968. Plant breeding: problems with forest treespecies in West Africa. Ghana Journal of Agricultural Science,1 (1), 21-28.

Jones, N. 1970. Progress in the improvement programme for7brminalia ivorensis and 7'riplochiton scleroxylon in Ghana.Ghana journal of Agricultural Science, 3 (1), 31-38. •

Jones, N. 1974. Records and comments regarding theflowering of Tnplochiton scleroxylon K. Schum. Commonwealth

Forestry Review, 53, 52-56.

Kadio, A. 1990. Amelioration génétique du Gmelina arborea

(Linn) en Côte dIvoire: bilan et perspectives. Abidjan: Centre

Technique Forestier Tropical.

Kadio, A., Legar, D. & Bohousson, D. 1991. Multiplication

et plantations clonales de Samba en Côte d'Ivoire. Revue

Forestière Frangaise Hors Série, 5.

ICio, RR.O. 1983. Management potential of the tropical high

forest with special reference to Nigeria. In: Tropical rainforest:

ecology and management, edited by S.L. Sutton, T C. Whitmore

& AC. Chadwick, 445-455. Oxford: Blackwell Scientific.

Koyo, J.P. 1983. Le bouturage au service des plantationsameliorees de 7brminalia superba en Republique Populaire duCongo. Silvicultura, 8, 331-333.

Koyo, J.P. 1985. Current status of genetic improvement work

with 7brminalia superba in the People's Republic of Congo. In:

Provenance and genetic improvement strategies in tropical

forest trees, edited by RD. Barnes & GL. Gibson, 547-552.Oxford: Commonwealth Foresty Institute/Harare: Forest

Research Centre.

Ladipo, D.O. 1981. Branching patterns of the tropical

hardwood 11-iplochiton scleroxylon K Schum with special

reference to the selection of superior clones at an early age.

PhD thesis, University of Edinburgh.

Ladipo, D.O. 1984. Seed problems in fuelwood plantations inNigeria. (Paper prepared for the International Symposium on

Seed Quality of Tropical and Sub-tropical Species, Bangkok,

Thailand.)

247

Ladipo, D.O. 1985. Comparative study of the growth ofseedlings, vegetative propagules, ortets and the effects oftopophysis on the growth of Triplochiton scleroxylon cuttings.(Research Paper (Forest series), no. 57.) lbadan: ForestResearch Institute of Nigeria.

Ladipo, D.O. 1986. Screening of indigenous forest speciesfor resistance to insect pests in Nigeria. In: Proceedings of the18th IUFRO World Congress, Lubljiana, Division 2, vol. 1,228-235. Vienna: IUFRO.

Ladipo, D.O., Leakey, R.R.B. & Grace, J. 1991a. Budactivity of decapitated, nursery-grown plants of Tnplochitonscleroxylon in Nigeria: effects of light, temperature andhumidity Forest Ecology and Management, 50, 287-298.

Ladipo, D.O., Leakey, R.R.B. & Grace, J. 1991b. Clonalvariation in apical dominance of Tnplochiton scleroxylon K.Schum. in response to decapitation. Silvae Genetica, 40,135-140.

Ladipo, D.O., Leakey, R.R.B. & Grace, J. 1991c. Clonalvariation in a four year old plantation of Triplochitonscleroxylon K. Schum. and its relations to the predictive testfor branching habit. Silvae Genetica, 40, 130-135.

Ladipo, D.O., Grace, J., Sandford, A. & Leakey, R.R.B.1984. Clonal variation in photosynthesis, respiration anddiffusion resistances in the tropical hardwood, TnPlochitonscleroxylon K. Schum. Photosynthetica, 18, 20-27.

Ladipo, D.O., Leakey, R.R.B., Longman, K.A. &Last, F.T. 1983. A study of variation in Tnplochitonscleroxylon K. Schum: some criteria for clonal selection.Silvicultura, 8, 333-336.

Leakey, R.R.B. 1983. Stockplant factors affecting rootinitiation in cuttings of Tnplochiton scleroxylon K. Schum., anindigenous hardwood of West Africa. Journal of HorticulturalScience, 58, 277-290.

Leakey, R.R.B. 1985. Tree improvement and seed productionunit. (Natural Environment Research Council contract report tothe World Bank.) Penicuik, Midlothian: Institute of TerrestrialEcology

Leakey, R.R.B. 1991. Towards a strategy for clonal forestry:some guidelines based on experience with tropical trees. In:Tree breeding & improvement, edited by J.E. Jackson, 27-42.Tring: Royal Forestry Society of England, Wales and NorthernIreland,

Leakey, R.R.B. & Grison, F. 1985. Regional programmes forthe improvement of tropical hardwoods: West and CentralAfrica. (Natural Environment Research Council contract reportto Commission of the European Communities.) Penicuik,Midlothian: Institute of Terrestrial Ecology/Nogent-sur-Marne:Centre Technique Forestier Tropical.

Leakey, R.R.B. & Ladipo, D.O. 1986. Selection forimprovement in vegetatively propagated tropical hardwoods.In: Improvement of vegetatively propagated plants, edited byR.K. Atkin & AL. Campbell, 230-241. London: Academic Press.

Leakey, R.R.B., Dick, J. McP. & Newton, A.C. 1992.Stockplant-derived variation in rooting ability: the source ofphysiological youth. In: Mass production technology forgenetically improved fast growing forest tree species, vol. 1,171-178. Nangis: AFOCEL.

248

Leakey, R.R.B., Ferguson, N.R. & Longman, IC.A. 1981.Precocious flowering and reproductive biology of Tnplochitonscleroxylon K. Schum. Commonwealth Forestry Review, 60,117-126.

Leakey, R.R.B., Njoya, C.Q. & Bockett, G. 1992.Vegetative propagation and clonal forestry, Parc de Bouturage.(Natural Environment Research Council contract report to UKOverseas Development Administration.) Penicuik, Midlothian:Institute of Terrestrial Ecology

Leakey, R.R.B., Mesén, J.F., Tchoundjeu, Z., Longman,K.A., Dick, J. McP., Newton, A.C., Matin, A., Grace, J.,Munro, R.C. & Muthoka, P.N. 1990. Low-technologytechniques for the vegetative propagation of tropical trees.Commonwealth Forestry Review, 69, 247-257.

Longman, K.A. & Leakey, R.R.B. 1993. La domesticationdu Samba. Anna les des Sciences Forestières. In press.

Martin, B. & Quillet, G. 1974. Bouturage des arbresforestiers au Congo. Bois et Forks des Tropiques, 154, 41-57;155, 15-33; 156, 39-61; 157, 21-40.

Mathias, P.J. 1988. Micropropagation of the tropicalhardwoods, Khaya ivorensis A. Chev and Nauclea diderrichii(de Wild. and Th. Duv) Merrill. PhD thesis, University ofNottingham.

Okafor, J.C. 1975. Varietal delimitation in lrvingia gabonensis(Irvingiaceae). Bulletin du Jardin Nationale Belgique, 45,211-221.

Okafor, J.C. 1981. Delimitation of a new variety of Treculiaafricana Decne. subsp. africana (Moraceae). Bulletin du JardinNationale Belgique, 51, 191-199.

Okoro, 0.0. 1974. Preliminary investigation of rooting of stemcuttings of Thplochiton scleroxylon. (Research Paper (Forestseries), no. 28.) Ibadan: Forest Research Institute of Nigeria.

Oni, 0. 1989. The reproductive biology of the West Africanhardwood Terminalia ivorensis A. Chev PhD thesis, University ofIbadan.

Oni, 0., Fasehun, F.E. & Ladipo, D.O. 1988. Effect ofstorage method, growth media, and source on pollen viabilityin Tnplochiton scleroxylon K. Schum. Nigeria Journal ofGenetics, 7, 7-12.

Tchoundjeu, Z. 1989. Vegetative propagation of tropicalhardwoods, Khaya ivorensis (A. Chev) and Lovoa trichilioides.PhD thesis, University of Edinburgh.

Verhaegen, D. 1992. La sortie varietale d'especes de boisd'oeuvre a moyenne revolution en Cote d'Ivoire. In: Massproduction technology for genetically improved fast growingforest tree species, vol. 1, 387-395. Nangis: AFOCEL.

Verhaegen, D., Kadio, A., Boutin, B., Delaunay, J. &Legal*, D. 1992. Le samba. Selection phénotypique d'arbres'+' et production industrielle de boutures en COte d'Ivoire.Bois et Foréts des Tropiques, 234 (4), 13-27.

Vigneron, P. 1984. Polymorphisme enzymatique et variabilitégénétique des provenances Ivoriennes et Congolaises de7brininalia. superba Engl. et Diels. Bois et Forêts des Tropiques,204 (2), 41-45.

Genetic improvement of trees in Central America,with particular reference to Costa Rica

F Mesén1 , D H Boshier2 & J P Cornelius11Centro Agronomico Tropical de Investigacion y Ensefianza, Tree Improvement Project, Turrialba 7170,

Costa Rica2Oxforcl Forestry Institute, Department of Plant Sciences, University of Oxford, South Parks Road, Oxford

OX1 3R8, UK

ABSTRACTCentral America has little or no forestry tradition other than widespread clearance and exploitation of its natural forests. Planting

of trees to provide for a shortfall of timber and other products is a recent event, and has only reached a significant scale in the

past decade. Tree improvement activities in Central America, starting in 1977 at the Centro Agronomico Tropical de Invesfigacion

y Ensefianza (CATIE) in Costa Rica and the National School of Forest Sciences, Honduras, therefore developed initially in a

vacuum, without the usual demand for improved seed, The Honduran project was curtailed in 1982 owing to the lack of finance.

Work started again in 1987 with the conservation and genetic improvement of Honduras forest resources project (CONSEFORH),

worldng with *a greater number of species, a wider remit, and greater emphasis on the conservation of native tree genetic

resources. In Costa Rica, the tree improvement project (TIP) has worked continually since 1977. Central to the approach of the

TIP has been the establishment of trials on farmers' land to ensure the performance of selected provenances/genotypes on the

type of land to be reforested. More recently a bilateral project funded by the Danish International Development Agency started in

Nicaragua, focusing on the improvement and conservation of various native species. No formal tree improvement programmes

exist in the other Central American countries, although seed stands have been established for a number of species.

Given population densities and the land tenure situation in Central America, the majority of reforestation is and will continue to

be by farmers with small and medium-sized holdings, using both timber and multipurpose trees. The great diversity of climatic

and edaphic conditions encountered in Central America, and the varied demands of farmers for trees, result in the need for work

on a greater number of species than is normal in tree improvement programmes. These include both native species, such as

Albizia guachapele, Alnus acuminate, Bombacopsis quinata, Cedrela odorata, Cordia alliodora, Gliricidia sepium, Pinus spp., Swietenia

macrophylla and Vochysia guatemalensis, and exotics, mainly Acacia mangium, Gmelina arborea and various Eucalyptus species..

With particular reference to activities in Costa Rica, this paper reviews past, present and planned tree improvement activities in

the region and the effect of some of the peculiarities of the region in developing strategies for tree improvement and

domestication.

INTRODUCTION

Few areas of the tropics of a similar size toCentral America show such wide variations oftopography climate, soil and vegetation.According to Holdridge's life zone classification,there are 17 major zones in Central America(Holclridge 1967). Vegetation ranges from verydry forest to rainforest, depending on the amountof rainfall, and with altitude from basal to paramo.Vavilov (1935, 1951) and Zohary (1970) identifiedCentral America as one of the principal areas ofthe world for genetic diversity of plant life. Thisbotanical wealth is shown as much in treespecies as in shrubs and herbs. With a totalnumber of more than 4000 tree species, theregion's tree genetic resources are enormous.

Various tree species now planted on a largescale in many tropical countries of the world areindigenous -6 Central America, eg Calliandracalothyrsus, Cupressus lusitanica, Glihcidia sepium,Leucaena spp., Pihus caribaea, P oocarpa andP tecunumanii. Others have been planted on a

smaller scale but have great potential for useand development, eg Alnus acumihata,Bombacopsis quihata, Cedrela odorata, Cordiaalliodora, Juglans olanchana, Parkinsonia aculeata.Furthermore, there are numerous species in theregion which may have potential for use in thefuture. The increasing levels of deforestationendanger the genetic resourCes of many of theregion's species. Particularly in the dry Pacificzone where deforestation has been extensive,species have suffered serious genetic erosionand in some cases are in danger of localisedand even regional extinction (Janzen 1986).

Central America has little or no forestry traditionother than widespread clearance and exploitationof its natural forests. The high population growthrate in the region has resulted in a continuingdemand for land for agriculture, ranching andforest products, leading in turn to an indiscriminatedestruction of the forest. Hartshorn (1982)estimated that 60-70 kha of forest are cut eachyear in Costa Rica; for every hectare of plantation

249

established, 80 ha of mature forests are cut. Thereis little indication of a change in this pattern, andthe majority of the natural forests outside protectedareas will be exhausted by the end of the century

The necessity for reforestation on a large scaleas part of the solution to this critical situationhas been evident for some time. Flores (1985)estimated that, for the period 1985-95, 900 khaneeded to be planted in the Central Americanregion to supply the internal demand for woodand wood-based products. The demand for seedhas, therefore, grown in recent years and willcontinue to increase, even if only a part of thearea needed is planted. As the number ofreforestation projects in the region increases,experience shows that inappropriate source ofseed, or propagative material, is a majortechnical cause of observed failures or lowyields. Much of the seed collection in the regionis carried out by farmers or other unqualifiedpeople, without adequate knowledge of the bestcollection techniques, or of the characteristicsthat the trees or stands should possess. As aresult, the collections are often made from badlyformed trees and stands. For this reason,established plantations often show poor growthand a lack of uniformity that makes managementdifficult and gives reforestation a poor image.These problems have been accentuated by thelack of knowledge about the importance of usingappropriate sources of seed and a lack ofinformation of the best sources among theregion's scientific personnel and farmers.

In the past ten years, reforestation has increaseddramatically in Costa Rica. This increase was dueinitially to the motivation of a number of .small-scale farmers and extension agents andtheir perceived need to meet the shortfall oftimber for local communities, within an affordableprice. Within this context, it was important toachieve visible results early on, and specieswere chosen which were easy to establish, suchas Bombacopsis quinata, Gmelina arborea andTectona grandis.

Capitalising on this initial success, planting hasincreased under new fiscal incentives and softloans, directed more towards small-scale farmers(typically 2-3 ha) organised into groups. Thesuccess has 'snowballed', with muCh largernumbers now being involved, and the resultantplantations have been of a much higher qualitythan previously As a result, with respect toforestry the situation in the region as a whole isnow very different from what it was 10-15 yearsago. Even though the species chosen have beentraditionally regarded as 'industrial species', theyare planted by farmers on a small scale for theirown use in a non-industrial context. Seed standsof Gmelina arborea were later set up using smallplots, and sale of seed has become an importantsource of extra income for farmers.

250

The availability of propagative material ofsuperior quality could have an important impacton tree planting in the region. Productionincreases in plantation forestry achieved in thepast 30 years in tropical and subtropicalcountries, through genetic improvementtechniques, have been well documented.However, the majority of these programmes, withthe exception of Leucaena leucocephala (seeBrewbaker & Sorensson, pp195-204), have beendeveloped by large businesses at a purelyindustrial level, and the potential gains from treeimprovement have not been made available tomedium- or small-scale farmers. Givenpopulation densities in Central America, themajority of reforestation will continue to be bysmall- and medium-scale landowners using bothtimber and multipurpose trees. It is evident thatthe farmers of the region plant trees for avariety of products: timber, wood for rural use(construction, carpentry work tools, posts),protection (live fences, windbreaks), shade (forlivestock, crops and the home), firewoodproduction, recuperation of soils, and morerecently to produce forage and feed livestockduring the dry season, from a variety of species.For this reason and the high degree ofenvironmental variation, tree improvementprogrammes are obliged to work with a greaternumber of species than is normal andconsequently accept a slower rate of advancewith any one species (see Table 1).

Past, present and future plans for treeimprovement within the region have thereforedeveloped within these constraints, taking intoaccount the above factors. The present paper,with particular reference to activities in CostaRica, and without any pretensions to cover allwork in the region comprehensively considersthree ongoing projects, with reference to:

• strategies for improvement and conservation;

• the state of progress;

• implementation; and

• future plans.

The three projects are:

i. CATIE tree improvement project;

CAllE tree crop production and managementproject (Madeleria); and

conservation and genetic improvement ofHonduras forest resources (CONSEFORH)project.

ONGOING PROJECTS FOR GENETICIMPROVEMENT OF TREES

CATIE tree improvement project

Planting trees in Costa Rica to provide for ashortfall of timber and other products is a recentevent and has only reached a significant scale in

'Table I. Tree improvement activities by species in the Central American region

*x, species propagated vegetatively

the past decade. Up until the beginning of 1983,the total area reforested in Costa Rica wasestimated at 4 kha (Hartshorn 1982).Reforestation has increased during the pastdecade because of a series of incentives givenby the government, to reach rates of 12-15 khaper year. However, plantation failures are high,mainly owing to poor quality of nursery stock,inadequate maintenance after planting, and poorseed quality

The CAPIE tree improvement project (TIP),financed by the UK Overseas DevelopmentAdministration (ODA), the Department forDevelopment Cooperation (DDC/Norway), theSwiss Office of Development Cooperation(COSUDE) and the US Agency for InternationalDevelopment (USAID), has been active in thegenetic improvement of traditional agroforestryand timber species since 1977. The TIP began,from 1977 onwards, with the establishment ofprovenance trials, mainly of Pinus caribaea,P oocarpa, Cordia alliodora, Gmelina arborea,Eucalyptus grandis, E. urophylla and Acaciamangium. They were established as a part ofinternational trials co-ordinated by organisationssuch as the International Union of ForestryResearch Organizations (IUFRO), the Food andAgriculture Organisation (FAO) of the UnitedNations, the Oxford Forestry Institute (OH), the

either by leafy cuttings or stakes

Commonwealth Scientific and Industrial ResearchOrganisation (CSIRO) and the 'llee Seed Centrein Denmark, funded by the Danish InternationalDevelopment Agency Species were selected onthe results from species trials and plots thatwere established in many parts of the country(Combe & Gewald 1979; Camacho 1981;Martinez 1981). The strategy was based on thefact that the most economical and rapid meansof obtaining gains in the productivity of trees isthrough the use of the most appropriate speciesand provenances (Zobel & Talbert 1984). Thetrials were planted on a variety of sitesthroughout the country covering the majorecological zones considered suitable forreforestation with the particular species. Theproject aimed also to identify any patterns ofgenotype/environment interaction for eachspecies and their effect on any recommendationsfor the establishment of particular provenances.

At the time the project started, there was littleawareness of, or demand for, the information ormaterial that the project could provide. Theproject, therefore, aimed to establish trials thatwould provide appropriate information, geneticmaterial and demonstration units for the mediumterm when the need became apparent.

The TIP's trials and seed orchard programmerelies heavily on the active participation of

251

smallholder farmers, both individually and inorganised groups. Virtually all trials areestablished on small farms under writtenagreements with the landowners. After finalthinning, the trials and seed orchards are usedby the individuals or groups in question fortimber and/or improved seed sources. Theestablishment of trials on farmers' land ensuresthe performance of selected provenance/genotypes on the type of land to be reforested,and permits the combination of the process ofgenetic testing, seed production and technologytransfer.

An important part of the project is thestrengthening of capabilities for tree improvementwork throughout the region. Short courses andcourses at postgraduate level are given toenhance capability within the region for treegenetic improvement.

The need to increase the work of the project toensure the uptake of its research results by theend users (ie tree planters of all types) hasbecome evident. One example came from thesuperior growth of Pihus tecunumanii to bothP caribaea and P oocarpa observed in TIP trials.The results were reported by the project as longago as 1986 (Boshier & Mesén 1986), but thespecies was not officially accepted by the CostaRican General Forestry Directorate as suitable forreforestation until 1991. This acceptance cameonly as a result of the. TIP establishing a numberof small demonstration plantations (1-2 ha) onfarmers' land. Firm recommendations on seedsources can now be made for Gmelina arborea,Acacia mangium, Pinus caribaea, P tecunumaniiand Eucalyptus urophylla (Mesén 1991).

While provenance research provides a soundbasis for avoiding catastrophic losses in forestryplantations, the TIP's goal is higher than simplyto avoid plantation failures. Even within the bestprovenances, trees show a great variability intraits of economic value. Therefore, a furtherstage in the TIP was the selection of individualsuperior trees within provenances, both for theestablishment of progeny tests and seedorchards in a traditional breeding programme,and for inclusion in a vegetative propagation andclonal selection programme. An additionalcomponent within the TIP's activities is thescreening of Cedrela odorata and Swietenia spp.families for genetic resistance to the shoot-borer(Hypsipyla grandella), and the eventual cloning ofpest-resistant genotypes (see Newton et al.,pp256-266).

Progeny tests

The natural ranges of all the native speciesincluded in the TIP progeny test programme arenot restricted to Costa Rica, but extend to manyother Central and South American countries. Inthis regard, international co-operation for

252

exploration, tree selection and exchange of seedbecomes a critical matter. A particularly fruitfullink was initiated in 1989 between the CentralAmerica and Mexico Coniferous ResourceCooperative (CAMCORE) and CA.1'1E, through aUSAJD-funded project which focused on theexploration and collection of seed from superiortrees of six threatened broadleaved species inGuatemala, Honduras, Colombia and Costa Rica.The species selected — Albizia guachapele, Alnusacuminata, Bombacopsis quinata, Cordia alliodora,Sterculia apetala and Vochysia guatemalensis —were chosen because all have populations indanger of extinction or genetic impoverishmentand, at the same .time, have exhibitedcommercial potential in CAIIE or CAMCOREtests (Mesén & Dvorak 1992). In addition, the TIPhas carried out selections within Costa Rica for anumber of promising non-native species,particularly Cupressus lusitanica, Eucalyptusdeglupta and Ginehha arborea. In total, 597phenotypically superior trees have been selectedand their seed established in provenance/progenytests in a variety of suitable sites in Costa Rica.

As for the provenance trials, the provenance/progeny tests are established under co-operativeagreements with farmers involved in reforestationunder the Costa Rican Forestry Directorateincentive scheme. Most trials are designed insuch a way that they can be converted intoseedling seed orchards after the evaluationperiod. Thus, in addition to the basic objectivesof genetic testing and providing material foradvanced selections, the trials are also used fordemonstration purposes and the production ofimproved seed, tested directly in the area wherereforestation is to be carried out.

Vegetative propagation and clonal selection

It is now recognised that vegetative propagationand clonal selection offer a means to enhancegreatly the yield and quality of forest products(Leakey 1987; Libby & Rauter 1984). Thetremendous improvements in productivity andform of clonal Eucalyptus in Brazil, for example,are well known by foresters. Most advances inclonal silviculture, however, have been developedby large companies and the techniques are notavailable to farmers in developing countries.Since 1989, the TIP has concentrated on thedevelopment of low-cost, appropriate technologycloning techniques, through a co-operation linkwith- the Institute of Terrestrial Ecology (114Scotland. The main obstacle for the use ofcloning in small-scale programmes, that of thesophisticated and expensive equipment needed,has been overcome by the use oflow-technology non-mist propagators, asdescribed by Leakey et al. (1990), Leakey andMesén (1991) and Mesén, Leakey and Newton(1992) (see also Leakey Newton & Dick,

pp72-83). Through the ITE/CAUE link, the TIPhas developed considerable expertise in the useof these techniques and in understanding thebasic physiological principles, both pre- andpost-severance, affecting the rooting ability ofleafy cuttings. All priority species included in theprogramme, some of which were formerlyconsidered to be difficult to root, are now beingcloned routinely They include Albizia guachapele,Bombacopsis quinata, Cordia alliodora, Cedrelaodorata, Gmelina arborea, Swietenia macrophyllaand Vochysia guatemalensis. The use of low-costand easily transferred cloning methods offersexcellent prospects for both the conservation andthe commercial use of genetically high-qualitygermplasm of valuable species.

Seed stands and the Made lam project

The focus of the Madelefia project is towards thedevelopment of techniques for the planting ofmultipurpose trees in different systems, includingagroforestry in small or relatively small farms inCentral America. The main aim was to developand strengthen the capabilities of CATIE andpublic and private organisations in the CentralAmerican region to allow them access, topromote and to disseminate on-farm,market-oriented tree crop technologies for theuse and benefit of small- and medium-scalefarmers and rural industries. This projectdeveloped from an initial attempt to addressfuelwood shortages on a regional basis, begun in1980, under the fuelwood and alternative energysources project based at CATIE. The projectscreened more than 100 species to determinetheir feasibility for fuelwood and latterlymultipurpose uses. Fourteen priority specieswere identified and a variety of promotional,extension and training activities carried out. Theexclusive focus on fuelwood was broadenedunder the project extension (USAID/RegionalOffice for Central America and Panama tree cropand management project) to include a mix ofspecies and management systems which can beutilised to provide a variety of products andeconomic benefits.

Traditional guides to the establishment of seedstands argue that areas of 2-5 ha arerecommended for seed stands (Matthews 1964).Generally in the Central American region, thereare few large plantations and the problems ofcontamination from surrounding stands of treesare reduced. Genetic constraints can, however,be avoided by the use of much smaller areas,and economies of scale may be very significantif a farmer can receive income from both thetimber and the seed from his trees. Thisapproach has worked in all the countries of theregion in conjunction with each country's nationalinstitutions, and national seed banks. The projectaims to establish a network of seed stands for

the multipurpose species it has identified aspromising and of high priority for the region.

CONSEFORH project

The genetic improvement programme inHonduras started in 1977 at the National Schoolof Forest Sciences, but was curtailed in 1982because of a lack of finance, leaving only a fewestablished trials (Gibson & Romero 1988). Workrecommenced in 1987 with the CONSEFORHproject, through an agreement between theHonduran Corporation for Forestry Developmentand ODA. CONSEFORH is worldng with a largenumber of species, both native and exotic, withparticular emphasis on the conservation,evaluation and genetic improvement of nativespecies. This work involves explorations todetermine the current distribution andconservation status of the species, seedcollections for the establishment of genetic trials,gene conservation banks and seed orchards,establishment of silvicultural trials, and researchon the properties and potential uses of thetimber. Work with exotics involves theestablishment of seed orchards and provenancetrials.

CONSEFORH is active in the four major foresttypes found in Honduras, namely dry forest,cloud forest, humid forest and coniferous forest.The number of forest species in these areaS isenormous, some of which (eg Pinus caribaea var.hondurensis) are planted on a large scale inmany tropical countries of the world, or mayhave great potential for future use anddevelopment. Up until the present, the efforts ofthe project have been concentrated in the dryforest, being the most severely disturbed forestassociation in Honduras and where there aremore than 300 tree species. Large-scaleecological restoration of the type currently beingimplemented in the dry forest of Costa Rica isnot an option in Honduras, because of the landtenure pattern and consequent pressure fromsubsistence farmers (Gibson 1993). The projecthas thus taken a varied approach, depending onthe importance and usage of the individualspecies. It involves collection and management ofgermplasm in both in situ and ex situconservation units, while for some species thepossibilities of conservation through use by localcommunities are also being explored(Gibson 1993).

Work began with priority species selected asendangered, commercially valuable and withcharacteristics making them acceptable tofarmers and others interested in tree planting(Gibson 1993). Extensive exploration was carriedout to define the distribution and status ofpopulations and to assess options forconservation of these species. Seed collectionswere then made, followed by the establishment

253

of breeding seedling orchards (BS0s) as a basisfor their conservation and improvement. Thespecies in this programme included Albiziaguachepele, A. niopoides, A. saman, Bombacopsisquihata, Cordia alliodora, Enterolobiumcyclocarpum, Gliricidia sepium, Hymanaeacoubard, Leucaena salvadorensis, Simaroubaglauca and Swietenia humilis (Gibson 1993). Workwith Pinus has also involved the establishment ofBSOs to ensure the conservation of endangeredprovenances of P caribaea.

.Most trials have been planted in one of theexperimental stations that the project maintains inHonduras, which range from 100 m to 1100 mabove sea level. The concentration of activities inthe experimental stations allows for bettermaintenance and protection of trials andconservation of the local flora and fauna, and atthe same time facilitates education anddemonstration activities. Many valuable speciesgrow naturally in these stations: for example,Albizia guachapele, Bursera simarouba,Enterolobium cyclocarpum, Guazuma ulmifolia,Pithecellobium dulce and Swietenia humilis arecommon in the dry forest station. Therefore, thestations themselves offer opportunities fordetailed growth and phenological studies and forthe development of appropriate techniques for insitu conservation. At present, a large proportionof the project's resources are concentrated in thisarea.

CONCLUSION

In Central America, the justification for treeimprovement has never been the straightforwardeconomic one appropriate to countries andcompanies with large reforestation programmes;the lack of adequate levels of reforestation isprecisely one of the problems which the regionfaces. At CATIE, at least, workers in treeimprovement have long been conscious that treeimprovement in Central America must lead,rather than follow reforestation, through theprovision of excellent-quality material whose fieldperformance will in itself stimulate tree plantingby farmers.

This concept must be reflected in the treeimprovement strategies adopted in the region.For this reason, at CATIE, the TIP is pursuingand advocating ambitious strategies of individualtree selection based, at the highest intensity onclonal selection and vegetative propagation.Obviously lower-intensity activities such asprovenance selection will remain cruciallyimportant. However, the best provenance containsbad as well as good genotypes, and in itselfeven the use of the best provenance is unlikelyto achieve the sort of 'quantum leap' in thequality of plantations that is necessary to changecompletely perceptions of and attitudes towardsplantation forestry and to stimulate reforestation

254

on a large scale. Clonal methods have thepotential to transform smallholder forestry aseffectively as they have transformed industrialforestry although different strategies of clonaldeployment may need to be developed (cfFoster & Bertolucci, pp103-111), and acomplementary programme of sexualimprovement should be maintained.

Domestication by vegetative propagation andclonal selection is just one example of animproved production system made possible bybut not solely dependent on, geneticimprovement. Workers in tree improvement in theregion should be looking for other possibilitiesoffered by genetic improvement to developimproved production systems which meet specificproblems or which offer new opportunities. Anobvious example is the work on selection forpest resistance in mahoganies mentioned above;other possibilities abound.

At a wider level, there is a need for an overallstrategy of management of Central Americanforest genetic resources, of which thedevelopment of genetically improved forestry andagroforestry production systems would constituteonly a part. The genetic base of many species isbeing rapidly eroded by deforestation, oftenbefore we know anything about their geneticstructure or even have complete knowledge oftheir natural distribution. For this reason, there isnow some initial work in progress in the regionon genetic conservation and baseline exploration,taxonomic and population genetic studies.However, there is little co-ordination betweenthese activities on a regional or even nationalbasis. An integrated forest genetics resourcemanagement strategy might be founded upon aregional network of resource management units,each consisting of a specific type of population.They might range from national parks orreserves, with the objective of in situ geneticconservation, through to clonal multiplicationgardens, dedicated exclusively to the productionof improved germplasm for particular productionsystems. Between these two extremes there couldbe a variety of other options, such as seedstands, natural relict stands, seed orchards, etc,each with its own clearly defined role in relationto the overall genetic management of the species.Through such a network, priorities for fundingand action would be made clearer, and nationaland regional co-operation on tree improvementand related activities could be greatly improved.

REFERENCESBoshier, D.H. & Mesén, J.F. 1986. Proyecto demejoramiento genético de arboles de CATIE: estado de avancey principales resultados. (Trabajo presentado en el PrimerCongreso Forestal Nacional.) San José, Costa Rica.

Carnacho, P. 1981. Ensayos de adaptabilidad y rendemientode especies forestales en Costa Rica. Cartago: ITCR-MAG.

Combe, J. & Gewald, N., eds. 1979. Gu la de campo de los

ensayos forestales del CAT1E en Turrialba, Costa Rica

TUnialba. Centro Agronomico Tropical de Investigacion yEnsehanza.

Flores, R.J. 1985. Diagndstico del sector forestal San Jose: EIJNED.

Gibson, G.L. 1993. Genetic conservation of nativehardwoods in the degraded dry forests of Honduras. In:Proceedings of the 111FRO Conference on Breeding TropicalTrees, Cartagena, Cali, Colombia, 1992. In press.

Gibson, G.L. & Romero, S.E. 1988. Proyecto conservaciOn

y mejoramiento genetico ODA-ESNACIFOR. Noticiero

Mejoramiento Genético y Semi Ilas Forestales para America

Central, 2, 3-7.

Hartshorn, G.S. 1982. Costa Rica: country environmental

profile; a field study San José: Tropical Science Centre/AID.

Holdridge, L. 1967. Life zone ecology San José: 'tropicalScience Centre.

Janzen, D.H. 1986. Guanacaste National Park: tropical

ecological and cultural restoration. San Jose: ITCR-FPN-PEA.

Leakey, R.R.B. 1987, Clonal forestry in the tropics - areview of developments, strategies and opportunities.Commonwealth Forestry Review 66, 61-75.

Leakey, R.R.B. & Mesén, J.F. 1991. Propagación vegetative

de especies forestales: enraizamiento de estacas suculentas.In: Manual sobre mejoramiento genetico con referencia

especial a America Central, edited by J. Cornelius, J.F. Mesén

& E. Corea, 113-133..Tlirrialba: Centro Agronomico 'tropicalde Invesfigacion y Ensehanza.

Leakey, R.R.B., Mesen, J.F., Tchoundjeu, Z., Longman,K.A., Dick, J. McP., Newton, A., Matin, A., Grace, J.,Munro, R.C. & Muthoka, P.N. 1990. Low-technologytechniques for the vegetative propagation of tropical trees.Commonwealth Forestry Review 69, 247-257.

Libby, W.J. & Ranter, R.M. 1984. Advantages of clonalforestry' Forestry Chronicle, 145-149.

Martinez, H. 1981. Evaluación de ensayos de especies

forestales en Costa Rica. MSc' thesis, Centro AgranomicoTropical de Invesfigacion y Ensefianza-UCR.

Matthews, J.D. 1964. Seed production and seed certification.

Unasylva, 18, 104-118.

Mesén, J.F. 1991. Resultados de ensayos de procedencias en

Costa Rica. (Serie Tecnica, Informe Técnico, no. 156.)Turrialba: Centro Agronomico Tropical de Investigacion yEnseilanza.

Mesén, J.F. & Dvorak, W.S. 1992. Seed collections and

vegetative propagation of threatened multipurpose species in

Central America. (Networking/Workshop Meeting of the Boardon Science and Technology for International Development,National Research Council.) 'Riffle Boa. Centro AgronomicoTropical de Investigacion y Ensehanza.

Mesén, J.F., Leakey, R.R.B. & Newton, A.C. 1992. Hacia

el desanollo de técnicas de silviculture clonal para elpequefio finquero. El Chasqui (Costa Rica), 28, 6-18.

Vavilov, N.I. 1935. Theoretical basis of plant breeding.

Moscow (In Russian.)

Vavilov, N.I. 1951. Phytogeographic basis of plant breeding.The origin, variation, immunity and breeding of cultivatedplants. Chronica Botanica, 13, 1-366.

Zobel, B. & Talbert, J.A. 1984. Applied forest treeimprovement. New York: Wiley

Zohary, D. 1970. Centers of diversity and centers of origin.

In: Genetic resources in plants - their exploration and

conservation, edited by GELFrankel & E. Bennett, 33-42. (1BP

Handbook, no. 11.) Oxford: Blackwell Scientific.

255

Domestication of mahoganies

A C Newton', R R B Leakey1 , W Powell2, K Chalmers 2

R Waugh2 , Z Tchound6jeu3 , P J Mathias4 *, P G Alderson4,J F Mesén5 , P Baker & S Ramnarine711nstitute of Terrestrial Ecologyt, Bush Estate, Penicuik, Midlothian EH26 OQB, UK2Cell and Molecular Genetics Department, Scottish Crop Research Institute, Invergowrie,Dundee DD2 5DA, UK3Office National de Dévéloppement des Forêts, Forest Management and Regeneration Project, BP 163,Mbalmayo, Cameroon4Faculty of Agricultural and Food Sciences, Department of Agriculture and Horticulture, University ofNottingham, Loughborough, Leicestershire LE12 5RD, UK5Centro Agronomico Tropical de Investigacion y Enseiianza, Turrialba 7170, Costa Rica6International Institute of Biological Control, Gordon Street, Curepe, Trinidad and Tobago7Trinidadian Forestry Division, Balisier Street, Pleasantville, San Fernando, Trinidad and Tobago* Present address: Australian Nursery Industry development officer; NSW Agriculture, Locked Bag 11,WMclsor NSW 2756, Australiaf A component of the Edinburgh Centre for Tropical Forests

ABSTRACTDespite the economic importance of mahoganies, few attempts have been made at genetic improvement, partly because of thehigh incidence of pest attack when mahoganies are grown in areas where they are native. A suggested domestication strategy formahoganies is outlined, centred on the selection for pest resistance as part of a genetic improvement programme, the capture ofselected genotypes using vegetative propagation techniques, and the deployment of selected material in appropriate silviculturalsystems which optimise pest control. Current progress in developing such a strategy is described, including the assessment ofgenetic variation using field tests and molecular methods, and the development of vegetative propagation techniques with in vitroapproaches. The importance of conserving genetic resources of mahogany species is highlighted, and the prospects for the futuredevelopment of a domestication strategy are discussed. It is suggested that the development and implementation of such astrategy should be given high priority if a sustainable resource of mahogany is to be guaranteed in the 'future.

INTRODUCTION

Mahoganies are among the most economicallyimportant tropical timber species, accounting fora significant proportion of world trade in tropicalhardwood. Despite this fact, the mahoganiesremain largely undomesticated: very little isknown about the extent of genetic variation inwild populations, and very few attempts havebeen made at genetic improvement (see Palmer,pp16-24).

In the strict sense, the term 'mahogany' appliesto members of the genus Swietenia (Meliaceae),which comprises three species, all native to theneotropics (see Styles 1981, for a detailedreview). The natural distribution of S. humilis isthe Pacific coast region of Central America,whereas S. mahagoni is found on a number ofCaribbean islands and mainland USA (southernFlorida). S. macrophylla, now the principalmahogany of commerce, occurs over a largegeographical area, from Mexico to the southernAmazon in Bolivia and Brazil (see Styles 1981).

256

In this paper, mahogany is also taken to includethe closely related genus Khaya (Africanmahogany), which bears a number ofmorphological and ecological similarities toSwietenia. A number of points are illustrated byreference to other economically important generain the same family such as Cedrela and Lovoa.About seven species of Khaya are recognised byStyles (1981), including K anthotheca, Kgrandifoliola, K ivorensis, K madagascariensis, Knyasica and K. senegalensis. Most of the speciesare native to tropical Africa; K ivorensis is nativeto coastal rainforests of West Africa, whereas Ksenegalensis occurs in the drier northern parts ofthe same region.

The most important product obtained frommahogany is timber, which is principally usedfor furniture and veneers; it is easily worked andstrong for its weight (Lamb 1966). Often,mahogany species are also favoured for use inagroforestry systems (eg in Central America andparts of Indonesia), where they may provideshade for crops and fuelwood. Other products

derived from Meliaceae include oil (derived fromthe seed of Carapa spp.; see Prance, pp7-15)and biological insecticides (such as neem,obtained from Azadirachta indica). Medicinalproducts, such as treatments for whoopingcough, rheumatism and lumbago, are derivedfrom Khaya spp. (Abbiw 1990). Products such asthese could potentially be derived from othermeliaceous species by appropriate selectionprogrammes.

The main factor which has limited the cultivationof mahoganies is attack by shoot-boring moths(Hypsipyla spp.), which are widespreadthroughout the tropics. The moth larvae destroythe terminal bud of the young tree, which thenfrequently branches or forks, reducing theeconomic value of the timber considerably. Thispest has resulted in the failure of many attemptsat reforestation with mahoganies in countrieswhere they are native, including Puerto Rico,Guatemala, Peru and Cuba in the case ofneotropical species (see Newton et al. 1993 fordetails). Similarly planting of naya spp. hasbeen almost completely abandoned in bothGhana and Nigeria because of shoot-borerattacks (Wagner, Atuahene & Cobbinah 1991).For this reason, selection for pest-resistantgenotypes may form a critical part of thedomestication strategy for mahoganies.

As few successful examples exist of mahoganycultivation in plantations, most timber continues tobe derived from the exploitation of naturalforests. This work is largely undertaken in anon-sustainable way Domestication of mahoganyis crucial for the development of an alternativeresource, to guarantee the supply of high-qualitytimber into the future. In this paper, we considerthree stages in the domestication process:

i. the assessment and selection of geneticvariation;

the capture of selected genotypes by theuse of propagation techniques; and

the deployment of genetically improvedmaterial in silvicultural or agroforestrysystems to realise the full genetic potential.

THE ASSESSMENT AND SELECTION OFGENETIC VARIATION

Assessments of genetic variation have traditionallybeen made by comparing the growth of materialfrom different geographical origins in provenanceand progeny tests. However, recently developedmolecular techniques enable the extent ofgenetic differentiation between genera, speciesand populations to be quantified directlyPreliminary results from both these approachesare described below, together with aconsideration of selection for pest resistance andgenetic conservation.

Provenance and progeny tests

Very few genetic tests have been establishedwith either New or Old World mahoganies (seePalmer, pp16-24). For example, the NationalResearch Council (1991) reported that there areno active tree improvement activities withSwietenia species. The most extensive provenancetests of Swietenia which have been established todate are those of the Institute of Tropical Forestryin Puerto Rico (Geary Barres & Ybarra-Coronado1973; see also Boone & Chudnoff 1970),although no data have apparently been publisheddescribing the variation observed (but seeGlogiewicz 1986). However, the broad ecologicaland geographical ranges of Swietenia species,coupled with their ability to hybridise, suggestthat a high degree of genetic diversity may existwithin the genus (Newton, Leakey & Mesén1993; see also Liu 1970).

Even less is known about the extent of geneticvariation in Khaya spp. than in their neotropicalrelatives. A number of workers have outlined theearly stages of genetic improvement programmeswith Khaya spp. Betancourt, Marquetti and Garcia(1972) described the possibility of hybridisationbetween K nyasica and K senegalensis, andnoted that a programme for selection ofresistance to stem cankers was initiated in Cuba.A preliminary programme of plus-tree selectionwas undertaken in Ghana, inclucling six trees ofK anthotheca and four of K ivorensis (Britwum1970). However, there are apparently nopublished data describing results from progenyor provenance tests of Khaya spp. in any area,although Chapuis (1990) gave brief details of thebreeding programme with both Khaya andSwietenia spp. in Cuba.

The only species of the Meliaceae which hasbeen investigated in any detail with respect togenetic variation is Cedrela odorata (Spanishcedar). A series of international provenance trialswere co-ordinated by the Oxford ForestryInstitute, UK, in the 1960s and 1970s (Chaplin1980; see also Burley & Lamb 1971). In 1967,seed.lots of 14 provenances were distributed to21 collaborating countries throughout the tropics,for use in trials Provenance differences in meanheight growth by up to a factor of six weresubsequently recorded (see papers in Burley &Nikles 1973; Nikles, Burley & Barnes 1978). Ingeneral, the most promising provenances interms of height growth were those from CostaRica and Belize (Chaplin 1980). These resultsindicate the extent of genetic variation whichcould potentially be recorded in other species ofMeliaceae, were they to be investigated (Newton,Leakey & Mesén 1993). The pattern of geneticvariation within the genus Cedrela is obscured,however, because some of the species (such asC. angustifolia) are poorly defined taxonomically(see Styles 1981) and susceptible to hybridisation.

257

O 60

5

250

•

100

Figure 1. Preliminary results from field trials of Swietenia macrophylla, illustrating the extent of genetic variation

i. progeny test, including 23 half-sib progenies from a range of sites in Costa Rica, Honduras and `frinidad,growth at Bajo Chino, CATIE, Costa Rica

n. progeny test, including 16 half-sib progenies from a range of sites in T1-inidad, after 17 months' growth at

iii. provenance test, including five provenances from the Central American/Caribbean region, after 14 months'Sur, CATIE, Costa Rica

258

80

70

50

40

30

300

200

150

140

120

80

.1. too 0) ('1 (0 co o cv C0 NI. 1"... O)0) 0 CV Nr t (D N.Lo Loin 1.0 0 CO CO co r N. I,- N- COa) co co 00 co 03

•

cv CV CV Ci C (Nj cv CV oj CV CV CV CNJ CV CV cvCi csi cq CV CV CV CV cy c‘j CV (N Cl (V CV CV CV CV CV CV cv csi c‘j

Progeny number

1 3 4 5 6 7 8 9 10 11 12 1316 17 18 19

Progeny number

2023 2025 2027 2028 2029

Provenance number

A number of small-scale provenance/progenytrials of neotropical mahogany species haverecently been established in Central Americaand the Caribbean, in an attempt to initiate aprogramme of genetic selection andconservation. For example, two progeny tests anda provenance test of Swietenia macrophylla havebeen established in Costa Rica and Thnidad aspart of a collaborative link between the Instituteof Terrestrial Ecology (ITE), the CentroAgronomico Tropical de Investigacion yEnsefianza (CATIE) and the International Instituteof Biological Control (Newton 1990; 'Newton,Mesén & Leakey 1992; Newton, Leakey & Mesén1993). In addition, the conservation and geneticimprovement of Honduras forest resources(CONSEFORH) project (see Mesén, Boshier &Cornelius, pp249-255) has established twoprogeny tests of S. hurnilis, which are probablythe first for this species. No results of these trialshave been published so far.

Preliminary results from the S. macrophylla trialsin Costa Rica and Trinidad indicate a significantdegree of genetic variation in rate of heightgrowth. In a progeny test at CATIE, Costa Rica,half-sib progenies differed by a factor of two inmean height after 15 months' growth (Figure li).A similar degree of variation (by a factor of 1.5)was recorded in a progeny test in Thnidad after17 months (Figure 1 ii), but five provenancestested at CATlE were less markedly different(Figure I iii).

Molecular techniques

Traditionally genetic resources have beencharacterised on the basis of morphological andagronomic traits. The effectiveness of thisapproach for estimating genetic diversityhowever, has been questioned by several authors(Gottlieb 1977; Brown 1979). The subsequentdevelopment of isozyme and other biochemicalmarkers represented a significant improvement.However, the effectiveness of such biochemicalmarkers is limited by the number ofpolymorphic loci detected.

With the advent of molecular techniques,DNA-based procedures for detecting geneticvariation have been proposed. They includerestriction fragment length polymorphisms(RFLPs) which have the potential to detect almostunlimited amounts of variation. Althoughchloroplast DNA (Palmer et al. 1988) and nuclearRFLPs (Debener, Salamini & Gebhardt 1990)have been used for taxonomic studies, the

in height growth

after 15 months'

Moruga, Ihnidad

growth at Florencia

-

0 -

Selection for pest resistance

As attack by shoot-borers (Hypsipyla spp.) is themain factor restricting the cultivation ofmahoganies in plantations, selection for pestresistance could be considered to be a key aimof a domestication strategy Pest resistance mayarise through three main mechanisms (Grijpma1976):

i. non-preference, when the insect is notattracted to or is actively repelled fromovipositing or feeding on the tree;

ii. antibiosis, in which the insect is ldlled,injured or prevented from completing its lifecycle after feeding on the tree; and

iii. tolerance, in which the tree recovers fromattack to an acceptable level.

There is evidence for all three mechanismswithin the Meliaceae family as a whole. Withrespect to non-preference, some mahoganyspecies are clearly less susceptible to attackthan others, such as S. mahagoni compared withS. macrophylla (Whitmore & Hinojosa 1977). Suchdifferences in susceptibility may reflect variationin the production of chemical attractants, althoughdifferences in growth rate may also be influential(Grijpma 1976). Antibiosis is demonstrated byspecies such as Toona ciliata, a native of SE Asiaand Australasia, which produces water-solublecompounds toxic to Hypsipyla grandella, thenative shoot-borer of the Americas (Grijpma &Roberts -1975). Some mahoganies produce resins,which may also hinder shoot-borer attack(Wilkins 1972; Lamb 1968; Whitmore 1978). Theability of individual trees to tolerate attack bystrong apical growth has also been observed inboth Cedrela spp. (Chaplin 1980; Grijpma 1976;Vega 1976) and Svvietenia spp. (A C Newton,personal observation).

However, little information is available on theintraspecific variation in these mechanisms ofpest resistance. To investigate this aspect, thegenetic tests established by the ITE/CATIE linkand CONSEFORH project (see above) have beenintensively assessed for the incidence of pestattack. Preliminary results, from combinedprovenance/progeny tests of C. odorata in CostaRica, have indicated intraspecific variation indifferent forms of resistance. Apart frompronounced differences in growth rate, differentfamilies displayed three-fold variation insusceptibility to attack (Newton, Leakey & Mesén1993). In addition, some individuals were able totolerate attack by vigorous growth of a newdominant lateral shoot, although the genetic basisof this characteristic has not yet been examinedin detail. These preliminary results suggest thatselection for pest resistance may be anachievable objective in mahoganies, althoughfurther research on this aspect is clearly required.

260

Genetic conservation

Concern has recently been voiced about theconservation status of neotropical mahoganies(Newton, Leakey & Mesén 1993; Rodan, Newton& Verissimo 1992), as reflected in the listing oftwo species (S. humilis and S. mahagoni) onAppendix II of the Convention on InternationalTrade in Endangered Species (criEs).Aproposal to include S. macrophylla on this listingwas made in 1992 by the governments of theUSA and Costa Rica, but was eventuallywithdrawn prior to consideration by thecommittee (Rodan et al. 1992). It is possible thatthis proposal will be renewed in the future. S.macrophylla is considered by some to beendangered or vulnerable in a number ofcountries (US CITES proposal 1992), althoughothers have suggested that large stocks still exist(Anon 1992). In fact, little detailed informationexists on the extent of remaining populations.

Many of the Old World mahogany species areperhaps in an even more precarious state, andare considered to be vulnerable or endangeredin many parts of their range (WorldConservation Monitoring Centre, Cambridge, UK,unpublished information). Germplasm collectionand exploration of Khaya spp. have beenaccorded high priority by the Food andAgriculture Organisation (1989), and there havebeen suggestions that this genus should also belisted on Appendix II of CITES (Flora and FaunaPreservation Society UK, personal communication).

The concerns about genetic conservation arisefrom the fact that the vast majority of mahoganytimber is harvested from natural stands. Selectivelogging, involving removal of the mosteconomically desirable phenotypes, may result inthe genetic depletion of the forest stand and areduction in its future economic value. Svvieteniamahagoni, which has been logged intensivelyover the past 400 years, is perhaps the moststriking example of genetic erosion in tropicalforestry: most individuals which remain are highlybranched or forked (Styles 1981). The sameprocesses are undoubtedly acting on S.macrophylla and other mahogany speciescurrently being harvested, although the extent ofany genetic erosion which may be occurring isdifficult to assess quantitatively

CAPTURE OF GENETIC VARIATION

Selected genotypes may be captured for use incultivation by seed and vegetative propagationtechniques. The requirements for the storage ofmahogany seed are described elsewhere(Tornpsett, pp61-71) and are therefore notdiscussed further in this paper. Instead, theprogress made in developing practical protocolsfor the vegetative propagation of mahoganies isdescribed, including both propagation by leafycuttings and in vitro techniques.

Vegetative propagation by rooting of leafycuttings

A number of Meliaceae species, including mostmahoganies, have now been successfullypropagated by rooting leafy cuttings (Leakey Last& Longman 1982; Newton, Leakey & Mesén1993). Successful results have been obtainedwith a number of different propagation systems,including traditional mist propagators (Howard,Verkade & DeFilippis 1988; Tchoundjeu 1989),and also low-technology non-mist propagators(Leakey et al. 1990; see also Leakey, Newton &Dick, pp72-83; Mesén et al., pp249-255).However, if mahoganies are to be propagated ona commercial scale, detailed information isrequired on the appropriate treatments whichshould be applied to both the stockplants andthe cuttings to obtain consistently high rootingsuccess. Such information is gained primarilythrough specific experimental programmes withindividual species.

The most extensive propagation studies to datehave been with Kliaya ivorensis under mist(Tchoundjeu 1989) and with Lovoa trichilioides(African walnut) in non-mist propagators(Tchoundjeu 1989). These examples areconsequently described here in some detail.Initial experiments were designed to determinethe optimal conditions for rooting single-node,leafy cuttings from hedged juvenile stockplants. Itwas hypothesised that, for a species for whichlittle is known about the conditions for rooting,the most important factors to test and optimisewere auxin concentration, leaf area, cutting lengthand node position (Tchoundjeu & Leakey 1993).The basic methods and mist propagation systemused were as previously described by Leakey etal. (1982) for the West African hardwoodTriplochiton scleroxylon.

By comparison with T scleroxylon, the highestrooting percentages of K ivorensis were obtainedwith a considerably higher applied auxinconcentration (200 )..ig IBA per cutting) but asmaller leaf area (10 cm2). Subsequently Asanga(1989) determined that the optimal leaf areaunder the conditions tested was about 30 cm2.As with many other species, long cuttings (39mm) rooted better than short ones (19 mm),especially if associated with a supra-optimal leafarea. Unlike T scleroxylon and some otherlight-demanding species, the cuttings from basal

'nodes rooted better than those from apicalnodes. These basal node cuttings had higher N,P, K, soluble carbohydrate and starch contentsthan those from apical nodes. One otherobservation from this study was that cuttingsdeveloped a one-sided root system if the cuttingbase was made by an oblique cut as opposedto a square cut (Tchoundjeu 1989).

Like many other members of the Meliaceae,plants of K ivorensis grow by recurrent flushing;

there are, therefore, alternating periods ofterminal bud activity and dormancy Higherrooting percentages were obtained when cuttingswere taken from dormant shoots than fromflushing shoots, although the latter had higherconcentrations of soluble carbohydratesthroughout the period of propagation.

In a more detailed study of the rooting of K

ivorensis cuttings, an attempt was made toinvestigate the relationships between rooting andthe carbohydrate dynamics of the cuttings(Tchoundjeu 1989). This study included anexamination of the effects of stockplant irradianceand nutrient applications on the dynamics ofreducing sugar and starch contents of both theleaf and stem portions of cuttings in thepropagator. Results showed that rooting neverseemed to be limited by the storedcarbohydrate reserves of the cuttings. Leakey etal. (pp72-83) present evidence derived mostlyfrom light-demanding, pioneer species thatrooting generally tends to becarbohydrate-driven. It, therefore, appears that,for K ivorensis, and perhaps other relativelyshade-tolerant species, rooting ability may not belimited by either carbohydrate reserves or theproduction of current assimilates. This conclusionwas clearly demonstrated in a further study ofthe effects of stockplant nutrition on rooting. Inthis case, there was no effect of nutrientapplication on stockplant growth or rooting,although there were very considerable effects onthe conversion of stored starch to sugars. Cuttingmortalities were, however, greatest in cuttingsfrom stockplants receiving the highest rate ofnitrogen application (Tchoundjeu 1989).

Additional experiments were undertaken inCameroon, investigating the factors whichinfluence the rooting of Lovoa trichilioides, usinga non-mist propagator as described by Leakeyet al. (1990). In the early experiments, thepropagators were not as air-tight as in laterexperiments, and the rooting percentages werefrequently less than 50%. Nevertheless, bycomparison with K. ivorensis, the auxinrequirements of L. trichilioides cuttings wererelatively low (Tchoundjeu 1989), with an optimalconcentration in one experiment of 50 jig percutting, while, in another, untreated controlsrooted as well as treated cuttings. The highestrooting percentages were achieved with leaflamina areas of 200 cm2, about ten times that ofK ivorensis. These large-leaved cuttings alsoproduced the most roots and had the lowestcutting mortalities. In a number of experiments, itwas found that, as in T scleroxylon but incontrast to K ivorensis, a higher proportion ofcuttings rooted from apical nodes of the topshoot, while those from basal nodes had thegreatest mortality rates (Tchoundjeu 1989).Higher rooting percentages were obtained with

261

cuttings from basal shoots, these having thegreatest leaf and stem nitrogen concentrations,and high foliar carbohydrate contents.

In an attempt to examine the effect of cuttingsize (stem length and diameter), cuttings of threesize categories were collected from similarpositions within shoots. In this case, higherrooting percentages were obtained with long thincuttings (38 mm x 4 mm) than, long thickcuttings (45 mm x 8 mm); short thin cuttings (15mm x 4 mrn) were intermediate. To examine theeffects of cutting origin on rooting, cuttings werecollected from hedged stockplants producingone, two, three or four shoots per plant. In thisinstance, the mean percentage rooting of allcuttings harvested per plant was similar in allfour treatments (Tchoundjeu 1989). However, therelative rooting percentage of cuttings from thedifferent shoots was strongly influenced by thenumber of shoots per plant and their position onthe plant.

As in K ivorensis, the effects of stockplantmanagement treatments, such as nutrientapplication and shading (irradiance and lightquality), were not conclusive. It seems that, unlikelight-demanding species such as T scleroxylonand Eucalyptus grandis (Leakey & Storeton-West1992; Hoad & Leakey 1992), rooting in relativelyshade-tolerant hardwoods is not predeterminedby the stockplant's light environment and theinteractions of light with nutrients. Further studiesare, therefore, required to examine thedifferences between these two groups of treesand determine the reasons for these differencesin rooting physiology However, despite this lackof conformity with other well-studied tropicalhardwoods, it is clear that both these species ofthe Meliaceae are relatively easy to root as stemcuttings under either mist or non-mistpropagation systems.

Few detailed vegetative propagation experimentshave been undertaken with the neotropicalspecies of the Meliaceae. In a preliminaryinvestigation using non-mist propagators, thepercentage rooting of Swietenia macrophyllacuttings was found to be higher when a rootingmedidm with a high proportion of sand wasused; maximum rooting of over 60% wasachieved with 75:25 sand/gravel (Mesén, Leakey& Newton 1992; Newton, Leakey & Mesén 1993).The concentration of IBA applied to the base ofthe cuttings was found to have only a slighteffect on rooting. In these experiments, thecuttings were relatively slow to root (11 weeks),indicating that further research is needed ifpropagation protocols are to be improved. Ingeneral, Cedrela odorata appears relatively easyto root, displaying higher rooting percentages insand than gravel, and with relatively low(0.2-0.4%) concentrations of applied IBA(Maldonado, Salazar & Mesén 1992).

262

In vitro micropropagationA number of mahogany species have now beensuccessfully micropropagated using in vitrotechniques, including Cedrela odorata andSwietenia macrophylla (Lee & Rao 1988;Maruyama et al. 1989). One of the few specieswhich has been investigated in any detail,however, is Khaya ivorensis (Mathias 1988), and isdescribed here.

By comparison with another W African hardwoodspecies, Nauclea diderrichil, explants of Kivorensis were easy to steriise with commercialsterilant (5%, 10% and 20% for 10, 20 or 30minutes) (Mathias, Alderson & Leakey 1989).Those explants treated with 5-10% sterilant werefree from tissue browning and were viable. Themedium used in this study was that ofMurashige and Skoog, with a carbon source of20 g 1-1 galactose (Mathias 1988). Thestockplants were grown under tropical glasshouseconditions in Britain and explants cultured at25°C at a photon flux of 50-60 i.tmol rri-2

s- 1 for16 h each day The experimental programmeexamined the effects of pre-severance stockplanttreatments on culture initiation and the conditionsrequired for shoot proliferation, and is describedin full by Mathias (1988).

The environmental factors investigated in order toimprove the success of culture initiation were thephoton flux of photosynthetically active radiation(PAR), light quality (red/far-red ratio), daylength,day/night temperatures and stockplant nutrition. Inaddition, because K ivorensis grows by recurrentflushing, experiments tested the effects ofcollecting explants at different times during theflushing cycles, as well as at different times afterremoval of the terminal bud.

Explants collected from dormant shoots had thehighest bud activity in culture and the lowestmean callus score. When dormant shoots weredecapitated prior to collecting explants, greaterbud activities were found in explants collectedeither two to three or eight to nine days afterdecapitation than in those collected at othertimes. The application of fertilizers to stockplantshad some effects on shoot growth prior to thecollection of the explants, but little effect onculture initiation. In contrast, the stockplant lightenvironment did influence bud activity in culture,with the greatest activity occurrin in explantsfrom plants grown at 60 i.tmol s-1 at R/F'R of0.3, especially in the absence of appliednutrients (Mathias 1988). Analysis over a numberof different treatments, however, showed that thegreatest increases in explant activity wereachieved by increases in the redlight/photosynthetic photon flux ratio (ie theproportion of red light [660 nm] in the wholeband of photosynthetically active radiation[400-700 nm]).

Regarding the stimulation of shoot proliferation inK ivorensis, the cytokinins benzylaminopurine(BAP) and zeatin at 2, 5 and 10 mg 1-increased the mean number of axillary shootsformed per explant. Subsequently, a study of theeffects of auxin (naphthalene acetic acid[NAAD/cytokinin (BAP) ratio on bud activityshowed that the optimal combination was around1:100-1:200 in the first subculture, but that in thesecond subculture even greater bud activitiesoccurred, with an .optimum NAA/BAP ratio of 1:25(Mathias 1988). In another experiment, thetransfer of cultures which had previouslyproliferated and had their shoots harvested, tomedia containing gibberellic acid (GA3),stimulated further proliferation/elongation ofslioots. However, in all these studies there wasevidence that growth regulatory substancesaccumulated in the tissues, and consequently thatthey could reach inhibitory concentrations, ifapplied repeatedly through several subcultures.

In conclusion, it is clear that .K ivorensis (Africanmahogany) is amenable to micropropagation, andthat practical protocols could be developed withfurther study The major problem encountered inmicropropagation was the initiation of aproliferating culture, owing to variability in theexplants.

DEVELOPMENT OF APPROPRIATESILVICULTURAL SYSTEMS

In order to realise the full genetic gains obtainedthrough selection, the trees should beestablished in appropriate silvicultural systems.The choice of an appropriate system isdetermined partly by the physiological responsesof the individual species (see Fasehun & Grace,pp148-157). The photosynthetic responses ofSwietenia macrophylla, Cedrela odorata and Khayaivorensis have now been analysed in some detailunder controlled conditions (Kwesiga & Grace1986; Kwesiga, Grace & Sandford 1986; Ramos &Grace 1990). In general, Swietenia and Cedrelaspp. are highly light-demanding, and this factshould be taken into account in cultivation. Manyof the failures in mahogany cultivation insilvicultural systems, such as line enrichment, canbe attributed to inadequate intervention leadingto excessive shading (cf Palmer 1988).

A wide range of different silvicultural approacheshas been applied to the production ofmahoganies with the aim of controllingshoot-borer attack. Most have resulted in failure,but there are examples of trials whereshoot-borer damage has been at least partlycontrolled by cultural methods (Newton et al.1993). For example, in Puerto Rico, lineenrichment plantings resulted in as few as 11%of the trees being attacked (Weaver 1987;Weaver & Bauer 1986). Similarly Vega (1976)described a series of trials in Surinam, involving

establishment of Cedrela spp. in naturalregeneration, line enrichment and open plantationsystems. After two years, the proportion of plantsattacked was higher in plantations established inthe open (10-60%) than in enrichment plantings(4-40% attacked). In other enrichment trials, theproportion attacked was less than 10% after 22months. In trials established in Brazil, Yared andCarpanezzi (1981) reported that shoot-borerdamage of Swietenia macrophylla was virtuallyabsent in the line enrichment system employed.

The reasons for the success of these exampleshave not been investigated in detail, but mayinvolve a variety of processes, such as the effectsof shading on the growth rate of the trees and theproduction of terniinal shoots (Newton et al. 1993).In particular, it has been suggested that thepresence of other tree species may hinder locationof meliaceous trees by the adult moth (Grijpma1976; Morgan & Suratmo 1976). Very little preciseinformation is available to indicate whether thisprocess actually occurs, but it is conceivable thatlow densities of susceptible trees may prevent thebuild-up of moth populations (Weaver & Bauer1986). In addition, populations of natural predatorsof Hypsipyla could be maintained in systems suchas line enrichment, where much of the originalvegetation is left intact (see Gibson & Jones 1977).

It should be noted that the planting ofmahoganies in mixtures with crops ornon-susceptible tree species does not guaranteesuccessful shoot-borer control. For example,when a number of different silvicultural andagroforestry systems were tested in Colombia,no consistently successful method of shoot-borercontrol was identified (Vega 1987; Neyra &Martinez 1985). These results emphasise theimportance of viewing the silvicultural system asone aspect of an integrated domesticationstrategy Such a system might involveincorporation of pest-resistant genotypes into asilvicultural system optimising natural biologicalcontrol, such as a line enrichment system,thereby providing an integrated system of pestmanagement (Newton et al. 1993).

Mahoganies have been established successfullyin monocultures in a number of countries wherethey are not native, such as Swieteniamacrophylla in Indonesia and the S Pacific(Evans 1982). In such situations, mahoganies areoften (but not always) resistant to the nativeshoot-borers, and can be grown successfully athigh density offering the prospect of rapidgenetic gains in improvement programmes.Plantation establishment of exotic mahoganies islikely to increase in the future (Newton 1993),perhaps including the introduction of Khaya spp.into the neotropics (Betancourt et al. 1972), anapproach which has so far not been tested on alarge scale despite its obvious potential.

263

CONCLUSIONS

The development of a domestication strategy formahoganies offers the prospect of overcomingthe problems which have limited mahoganycultivation so far. Such a strategy should involveselection for pest resistance as one componentof a genetic improvement programme, andshould also involve the deployment of selectedgenotypes in appropriate silvicultural systems tooptimise pest control as well as growth.Techniques for capturing selected genotypes,including in vitro techniques and propagation byleafy cuttings, have been successfully developedfor a number of mahogany species, althoughfurther research is required to refine the precisetreatments required for sustained successfulrooting. Application of these propagationtechniques to genetic improvement should enablerapid progress to be made in generatingsuperior clonal populations for use inreforestation.

Apart from increasing efforts at geneticimprovement of mahoganies, attention should alsobe directed towards genetic conservation, both ofpopulations in situ and of selected genotypes exsitu. Increased exploration and testing of geneticresources of mahogany species are urgentlyrequired, particularly of the African species,which have hardly been investigated in thisregard. High rates of deforestation in bothpalaeotropical and neqtropical regions continue todeplete these genetic resources, and could limitthe potential for sustainable production ofmahogany in the future.

It is to be hoped that the prospects ofdeveloping a successful domestication strategyfor mahoganies will stimulate interest within thetimber industry and encourage attempts toregenerate a resource. The economic incentivesfor such an initiative certainly exist: the demandand value of the timber are likely to remain highfor the foreseeable future (Palmer, pp16-24).

ACKNOWLEDGEMENTS

The lTE/CATIE link project is financed by the UKOverseas Development Administration (ODA). P JMathias was supported by a NERC CASE award,and Z Tchoundjeu was supported by theEuropean Commission and ODA.

REFERENCESAbbiw, D.K. 1990. Useful plants of Ghana. London:Intermediate Technology Publications Ltd. and Royal BotanicGardens, Kew

Anon. 1992. New research shows Brazil has plentiful supplyof mahogany Marwood Incorporated News, 14, 1. (Newsletterproduced by Marwood Incorporated, Jeffersonville, Indiana.)

Asanga, A.C. 1989. Effects of leaf area on rooting successin leafy stem cuttings of the tropical hardwood, Khayaivorensis A. Chev (African Mahogany). MSc thesis, Universityof Edinburgh.

264

Betancourt, A., Mazquetti, J.R. & Garcia, J.R. 1972.Khaya nyasica and K senegalensis. Behaviour of the twospecies in Cuba. Baracoa, 2 (2), 32-52. (In Spanish.)

Boone, R.S. & Chudnoff, M. 1970. Variations in wooddensity of the mahoganies of Mexico and Central America.Turrialba, 20, 369-371. .

Britwum, S.P.K. 1970. Preliminary selection of provisionalplus trees in the tropical high forest. Forest Products ResearchInstitute of Ghana, Rchnical Newsletter, 4, 5-7.

Brown, A.D.K. 1979. Enzyme polymorphisms in plantpopulations. Theoretical Population Biology, 15, 1-42.

Burley, J. & Lamb, A.F.A. 1971. Status of the CFIinternational provenance trial of Cedrela odorata (including C.mexicana and C. tubillora). Commonwealth Forestry Review, 50,234-237. •

Burley, J. & Nikles, D.G., eds. 1973. Tropical provenanceand progeny research and international co-operation.(Proceedings of IUFRO Working Party Meeting at Nairobi,Kenya). Oxford: Commonwealth Forestry Institute.

Carlson, J.E., Tulsierant, L.K., Glaubitz, J.C., Luk,V.W.K., Kauffekle, C. & Ruthedge, R. 1991. Segregationof randomly amplified DNA markers in Ft progeny of conifers.Theoretical and Applied Genetics, 83, 194-200.

Chalmers, K.J., Waugh, R., Sprent, J.I., Simons, A.J. &Powell, W. 1992. Detection of genetic variation between andwithin populations of Gliricidia sepium and G maculata usingRAPD markers. Heredity, 69, 465-472.

Chaplin, G.E. 1980. Progress with provenance explorationand seed collection of Cedrela spp. In: 1 I th CommonwealthForestry Conference, Oxford, 1-17. Oxford: CommonwealthForestry Institute.

Chapuis, P. 1990. Twenty years of forestry research in Cuba.Bois et Foréts des Tropiques, 223, 43-50. (In French.)

Debener, T., Salamini, F. & Gebhardt, C. 1990.Phylogeny of wild and cultivated Solanum species based onnuclear restriction fragment length polymorphisms (RFLPs).Theoretical and Applied Genetics, 79, 360-368.

Evans, J. 1982. Plantation forestry in the tropics. Oxford:Clarendon Press.

Food and Agriculture Organisation. 1989. Report of panelof experts on forest gene resources. Rome: FAO.

Geary, T.F., Barres, H. & Ybarra-Coronado, R. 1973.Seed source variation in Puerto Rico and Virgin Islands grownmahoganies. (USDA Forestry Service Research Papers, no.ITF-17.) Rio Piedras: US Department of Agriculture.

Gibson, I.A.S. & Jones, T. 1977. Monoculture as the originof major forest pests and diseases. In: Origins of pest,parasite, disease and weed problems, edited by J.M. Cherrett& GR. Sagar, 139-161. Oxford: Blackwell Scientific.

Glogiewicz, J. S. 1986. Performance of Mexican, CentralAmerican and West Indian provenances of Swietenia grown inPuerto Rico. MSc thesis, State University of New York.

Gottlieb, L.D. 1977. Electrophoretic evidence and plantsystematics. Annals of the Missouri Botanical Gardens, 64,161-180.

Grijpma, P. 1976. Resistance of Meliaceae against the shootborer HypsiPyla with particular reference to 7bona ciliata M.J.Roem, var. australis (Fv.Muell.) C.DC. In: Tropical trees:variation, breeding and conservation, edited by J. Burley & B.T.Styles, 69-78. London: Linnean Society

Grijpma, P. & Roberts, S.C. 1975. Studies on theshootborer Hypsipyla grandella (Zeller) (Lep., Pyralidae)XXVIII. Biological and chemical screening for the basis ofresistance of Ibona ciliata M.J. Roem. var. australis (Fv.M.)C.DC. Turrialba, 25, 152-159.

Hoad, S.P. & Leakey, R.R.B. 1992. Morphological andphysiological factors induced by light quality and affecting

rooting in Eucalyptus grandis. In: Mass production technology

for genetically improved fast growing forest tree species, vol. 1,

51-58. Nangis- AFOCEL.

Howard, F.W., Verkade, S.D. & DeFilippis, J.D. 1988.Propagation of West Indies, Honduran and hybrid mahoganies

by cuttings, compared with seed propagation. Proceedings of

the Florida State Horticultural Society, 101, 296-298.

Kwesiga, F. & Grace, J. 1986. The role of the red/far-redratio in the response of tropical tree seedlings to shade.

Annals of Botany, 57, 283-290.

Kwesiga, F., Grace, J. & Sandford, A.P. 1986. Some

photosynthetic characteristics of tropical timber trees as

affected by the light regime during growth. Annals of Botany,

58, 23-32.

Lamb, A.F.A. 1968. Cedrela odorata L. (Fast growing timber

trees of the lowland tropics, no. 2.) Oxford: Commonwealth

Forestry Institute.

Lamb, F.B. 1966. Mahogany of tropical America: its ecology

and management. Ann Arbor, Michigan: University of Michigan

Press.

Leakey, R.R.B. & Storeton-West, R. 1992. The rooting

ability of Triplochiton scleroxylon K. Schum. cuttings: theinteractions between stockplant irradiance, light quality and

nutrients. Forest Ecology and Management, 49, 133-150.

Leakey, R.R.B., Last, F.T. & Longman, K.A. 1982.

Domestication of forest trees: a process to secure the

productivity and future diversity of tropical ecosystems.Commonwealth Forestry Review, 61, 33-42.

Leakey, R.R.B., Mesén, J.F., Tchoundjeu, Z., Longman,K.A., Dick, J.McP., Newton, A., Matin, A., Grace, J.,Munro, R.C. & Muthoka, P.N. 1990. Low-technology

techniques for the vegetative propagation of tropical trees.

Commonwealth Forestry Review, 69, 247-257.

Lee, S.K. & Rao, A.N. 1988. Plant let production of Swietenia

macrophylla King through tissue culture. Gardens Bulletin,

Singapore, 41, 11-18.

Liu, C.P. 1970. The genetic improvement of Honduras

mahogany I. Studies on natural variation and individual

selection. Quarterly journal of Chinese Forestry, 3(3), 41-56.(In Chinese.)

Maldonado, E.R.A., Salazar, R. & Mesén, J.F. 1992.Enraizamiento de estacas juveniles de Cedrela odorata L.

Silvoenergia, 51, 1-4.

Maruyama, E., Ishii, K., Satio, A. & Migita, K. 1989.

Screening of suitable sterilization of explants and proper

media for tissue culture of eleven tree species of

Peru-Amazon forest. Journal of Agricultural Science, 33, 252-261.

Mathias, P.J. 1988. Micropropagation of the tropical hardwoods,

Khaya ivorensis A. Chev and Nauclea diderrichii (de Wild and

Th. Dur) Merrill. PhD thesis, University of Nottingham.

Mathias, P.J., Alderson, P.G. & Leakey, R.R.B. 1989.

Bacterial contamination in cultures of tropical hardwood

species. Acta Horticulturae, 212, 43-48.

Mes6n, J.E, Leakey, R.R.B. & Newton, A.C. 1992. Hacia

el desarrollo de tecnicas de silvicultura clonal pare el

pequefio finquero. El Chasqui, 28, 6-18.

Morgan, F.D. & Suratmo, F.G. 1976. Host preferences of

Hypsipyla robusta (Moore) (Lepidoptera: Pyralidae) in West

Java. Australian Forestry, 39, 103-112.

National Research Council. 1991. Managing global genetic

resources: forest trees. Washington DC: National Academy Press.

Newton, A.C. 1990. Selection por resistencia al perforador

de las Meliaceas. Noticiero (5), Mejoramiento Genetico y

Semi llas Forestales para America Central, 4-7

Newton, A.C. 1993. Prospects for growing mahogany inplantations. Ln: Proceedings of the Second Pan American

Furniture Manufacturers' Symposium on Tropical Hardwoods,

Grand Rapids, 1992, 97-108. Grand Rapids, Michigan: Center

for Environmental Study.

Newton, A.C., Baker, P., Ramnarine, S., Mesén, J.F. &Leakey, R.R.B. 1993a. The mahogany shoot borer: prospects

for control. Forest Ecology and Management, 57, 301-328.

Newton, A.C., Leakey, R.R.B. & Mesén, J.F. 1993.

Genetic variation in mahoganies: its importance, utili7ation and

conservation. Biodiversity and Conservation, 2, 114-126.

Newton, A.C., Mesén, J.F. & Leakey, R.R.B. 1992.ITE/CATIE link Second interim report to the ODA. Edinburgh:

Institute of Terrestrial Ecology.

Neyra, M. & Martinez, H. 1985. Comportamiento de

Cedrela odorata bajo ties percentajes de mezcla con Cordia

alliodora y Tabebuia rosea en Ilimaco. (CONIF Serie Tecnica,

no. 11.) Bogota: CONE

Nikles, D.G., Burley, J. & Barnes, R.D., eds. 1978.

Progress and problems of genetic improvement of tropical

forest trees. Oxford: Commonwealth Forestry Institute.

Palmer, J.D., Jansen, R.J., Michaels, H.J., Chase, M.W.& Manhart, J.R. 1988. Chloroplast variation and plantphylogeny Annals of the Missouri Botanical Gardens, 75,1180-1206.

Palmer, J.R. 1988. Summary on enrichment plantings in the

neotropics. Unpublished report to Centre Technique Forestier

Tropical/International Tropical Timber Organisation.

Ramos, G. & Grace, J. 1990. The effects of shade on the

gas exchange of seedlings of four tropical trees from Mexico.

Functional Ecology, 4, 667-677.

Rodan, B.D., Newton, A.C. & Verissimo, A. 1992.

Mahogany conservation: status and policy initiatives.Environmental Conservation, 19, 331-338.

Roy, A., Frascaria, N., Mackay, J. & Bousquet, J. 1992.Segregating randomly amplified polymorphic DNA (RAPD) in

Betula alleghaniensis. Theoretical and Applied Genetics, 85, 173-180.

Styles, B.T. 1981. Swietenioideae. In: Meliaæae, edited by TD.Pennington, B.T Styles & DA.H. Taylor, 359-418. (Flora Neotropica

Monograph, no. 28.) New York: New York Botanical Garden.

Tchoundjeu, Z. 1989. Vegetative propagation of the tropical

hardwoods Khaya ivorensis A. Chev and Lovoa trichilioides

Harms. PhD thesis, University of Edinburgh.

Tchoundjeu, Z. & Leakey, R.R.B. 1993. Vegetative

propagation of African mahogany: effects of auxin, node

position, leaf area and cutting length. New Forests. In

press.

US CITES. 1992. Proposal to list the genus Swietenia on

Appendix ll of the Convention on International Trade in

Endangered Species (CITES). Washington DC: US Department

of the Interior, Fish and Wildlife Service.

Vega, C.L. 1976. Influencia de la silvicultura en elcomportamiento de Cedrela en Suriname. In: Studies on tile

shootborer Hypsipyla grandella (Zeller) Lep. Pyralidae, edited

by J.L. Whitmore, vol III, 26-49. (IICA Miscellaneous

publication, no. 101.) "farrialba: Centro Agronomico Tropical

de Investigacion y Ensefianza.

Vega, L.E. 1987. Crecimiento del cedro Cedrela odorata

manejado en fajas de restrojo y en el asocio inicial con

cultivos. (CONIF INFORMA, no. 10.) Bogota: CONT.'

Wagner, M.R., Atuahene, S.K.N. & Cobbinah, J.R. 1991.

Forest entomology in West tropical Africa: forest insects of

Ghana. Dordrecht: Kluwer.

Waugh, R. & Powell, W. 1992. Using RAPD markers for

crop improvement. Trends in Biology, 10, 186-191.

265

Weaver, P.L. 1987. Enrichment plantings in tropical America.In: Management of the forests of tropical America: prospectsand technologies, edited by J.C. Fig-ueroa, F.H. Wadsworth & S.Branham, 259-278. Rio Piedras: US Department of Agriculture.

Weaver, P.L. & Bauer, G.P. 1986. Growth, survival andshoot borer damage in mahogany plantings in the LuquilloForest in Puerto Rico. Turrialba, 36, 509-522.

Welsh, J. & McClelland, M. 1990. Fingerprinting genomesusing PCR with arbitrary primers. Nucleic Acids Research, 18,7213-7218.

Whitmore, J.L. 1978. Cedrela provenance trial in Puerto Ricoand St Croix: establishment phase. (USDA Forest Service ResearchNote, no. ERF 16.) Rio Piedras: US Department of Agriculture.

Whitmore, J.L. & Hinojosa, G. 1977. Mahogany (Swietenia)hybrids. (USDA Forest Service Research Paper, no. 1TF-23.)Rio Piedras: US Department of Agriculture.

266

Wilde, J., Waugh, R. & Powell, W. 1992. Geneticfmgerprinting of Theobroma clones using arbitrary primers.Nucleic Acids Research, 18, 7213-7218.

Wilkins, R.M. 1972. Suppression of the shoot borerHypsipyla grandella (Zeller) (Lepidoptera, Phyticidae) withcontrolled release insecticides. PhD thesis, University ofWashington.

Williams; J.G.K., Kubelik, A.R., Livak, K.J., Rafalski,J.A. & Tingey, S.V. 1990. DNA polymorphisms amplified byarbitrary primers are useful as genetic markers. Nucleic AcidsResearch, 18, 6531-6535.

Yared, J.A.G. & Carpanezzi, A. A. 1981. Conversao decapoeira alst da Amazônia em povoamento de producaomadereira: o me todo do 'Recru' e especies promissoras.Brasil Florestal, 45, 57-73.

Production of dipterocarp planting stock bycuttings in Indonesia

W T M Smits, A C de Fraiture & I YasmanTROPENBOS-Kalimantan, PO Box 319, Balikpapan 76103, Indonesia

ABSTRACT

Diptemcarps belong to one of the most important tropical timber tree families in the world, and, as a result, they have been

heavily exploited thmughout their range, especially over the past 30 years. When exploited too heavily logging may lead to an

excessive degree of canopy opening, or to longer-lasting damage from which the forest needs a very long time to recover. In

Indonesia, it is clear that a continued supply of diptemcarp timber is of the highest importance, with a growing local demand

and the importance of forest industries in the overall economy This paper describes some research work currently being

directed towards maintaining a supply of dipterocarp timber, and presents results of its application. The work described is that of

the TROPENBOS-Kalimantan project, where research is centred on a priority list of 20 dipterocarp species. In these species,

shortages of wildlings arise because the number of seedlings in the natural forests decreases sharply in the years after a mast

fruiting. Other means of providing planting stock are, therefore, necessary For this reason, effort was put into developing a system

for vegetative propagation of dipterocarps, involving the setting up of hedge orchards in which the stockplants can be induced to

produce vertical shoots suitable for propagation. A propagation system using aerated water is described, and the factors

important for successful rooting of dipterocarp cuttings are discussed. Various methods for ectomycorrhizal inoculation are also

described. The operational application of these techniques, and the scope for future genetic improvement of dipterocarps are

assessed. The research results reported suggest that in the future vegetative propagation of . dipterocarps may be applied to the

production of higlffy productive clones, providing a completely new economic outlook on the use of dipterocarps in plantation

forestry as well as in the enhchment of natural stands.

INTRODUCTION

Dipterocarps belong to one of the mostimportant tropical timber tree families in theworld. In the world markets, their timber makesup a very large part of the total timber trade.Their presence in the forests of south andsouth-east Asia is often characterised by theirdomination of the upper canopy: in some forests,dipterocarps may make up more than 80% ofthe trees in this layer. With their timber availablein large volumes of relatively homogeneousquality and with the fact that most of theirtimbers float enabling cheap transportation byriver, they have become a popular timber forindustrial uses.

As a consequence of their economically attractivecharacteristics and their common occurrence inthe forests of SE Asia, as well as the politicalsituation in a number of countries withdipterocarp forests, their exploitation has beentaken up with great vigour, especially over thepast 30 years. Countries which started timberexploitation earlier, such as the Philippines, nowhave virtually no natural forest left. In othercountries, the area of dipterocarp forest is alsodecreasing quickly The reasons for thediminishing areas of dipterocarp forest arenumerous and varied. Exploitation is beingpractised mostly under selective cutting systems

and should not necessarily lead to thedisappearance of the whole forest. Whenexploited too heavily however, such cutting leadsto a large degree of opening up of the uppercanopy Unscrupulous exploitation leads to morelong-lasting damage, and the forest will need a

very long time to recover from the loggingoperation. Furthermore, when the initial harvest isfollowed too soon by relogging, irreparabledamage can result; again, the forest will need tobe left alone for extremely long periods beforesimilar vegetation grows back. The opened uplogged-over forest is more easily encroached byshifting cultivators, while at the same time it

becomes more prone to the risks of fire. Withhuman encroachment, the fire risk may becomevery serious. In addition to the unintendedsquandering of their diminishing resources, manycountries also convert natural forest to otheruses, such as rubber and oil palm plantations,

agriculture and settlements.

The forest industries are of major importance tothe Indonesian economy In 1991, 2.5 million

people were directly involved in the forestrysector for their income and another 1.2 millionwere involved indirectly Including their families,

this makes a total of more than 14 millionpeople earning their livelihood from theIndonesian forests (MPI 1992). Indonesia has

267

invested heavily over the past years in varioustimber industries such as plywood mills (156units), sawmills (468), and furniture construction(340), all of which rely mostly upon dipterocarptimber for their functioning. The dipterocarptimber being exported generated US$3.981million for the country in 1991 (MPI 1992). It isclear that a continued supply of dipterocarptimber is of highest importance for Indonesia,with its growing local demand for timber.

The Ministry of Forestry in Indonesia has inrecent years taken many steps to encouragesustainable dipterocarp forest management. Mostof the forests are managed under a natural standmanagement system, called the Indonesianselective felling and planting system (TPTI). Thissystem is based upon the harvesting of a fewtrees of mature size (>50 cm dbh) and theconsequent tending of trees of commercialspecies in the diameter range of 20-50 cm dbh.The system also involves planting of the skidroads and log yards, and enrichment planting(stocking) of those areas with a low number ofseedlings of commercial tree species. Indonesiaalso promotes forest regeneration throughplantations, resulting in complete conversion ofthe original vegetation, which is normallyachieved through clearcutting followed byreplanting. In both TPTI and plantation forestrythere is a need for high-quality planting stock ofmany species, including dipterocarps.

One of the steps taken to overcome the problemof limited planting stock supply was an increasedinput into research and development. Researchon dipterocarp planting stock production istaking place through various institutions at anumber of places in Indonesia, including Bogor(Agricultural University BIOTROP and the Centerfor Forestry Research), Yogyakarta (Gajah MadaUniversity), Samarinda (Mulawarman University)and Samboja (Wanariset Dipterocarp ResearchStation of the Forest Research Institute inSamarinda) through the TROPENBOS-Kalimantanco-operative project. In this paper, we willdescribe some of this work and present someresults of its application. The work described isthat of the TROPENBOS-Kalimantan project, aimedat solving the problems of planting stock supplyThis project is a co-operative effort of theAgency for Forestry Research and Developmentof the Indonesian Ministry of Forestry theIndonesian state forestry enterprises PTINHUTANI I and PT INHUTANI II, theTROPENBOS foundation and the Institute forForestry and Nature Conservation 'DeDorschkamp'.

It should be mentioned that older examples ofdipterocarp planting do exist, eg the Shoreajavanica plantations in Krui, Sumatra, theTengkawang plantations in West Kalimantan, andthe Shorea platyclados plantations in Purbatonga,

268

north Sumatra, etc. These examples have shownclearly that it is possible to plant dipterocarpssuccessfully

PROBLEMS IN PLANTING STOCKPRODUCTION

One particular feature of the dipterocarps istheir irregular flowering behaviour. Most treesflower in so-called mast flowering years. Duringsuch years most individuals of most dipterocarpspecies in a given region will flower. Someindividuals or species may flower in years inbetween mast flowering years. Seeds of manyclipterocarps cannot be stored for extendedperiods of time. They are classified asrecalcitrant (Tompsett 1987; see also Tompsett,pp61-71). Propagation is further hampered bythe fact that many of the seeds are commonlyattacked by weevils and suffer other forms ofpremature death. Even in the instances whereseed can be collected, the seedlings producedin the nursery may turn yellow and eventuallydie. In addition, wildlings which appear healthyin the nursery often die after planting in openterrain. In the past, this combination of factorshas made it rather difficult to rely upon seedpropagation for any planting scheme involvingdipterocarps. The research described below wasdirected at solving these problems.

PRODUCTION TECHNIQUES

Wildlings

The simplest method of planting stock productioninvolves the collection of wildlings from naturalforest during or shortly after periods of heavyrainfall. After a short period in the nursery theplants are ready for planting. Advantages of thissystem are that it is relatively simple and cheap,and that the plants are genetically varied (whennot originating from single apomictic mothertrees, which may in certain instances be thecase). In addition, they may have been infectedby ectomycorrhizal fungi from the roots of theirmother trees. A description of this method ofdipterocarp planting stock production waspublished in a simple manual in Indonesian(Smits 1986), which was reprinted by theAssociation of Forest Concessions in Indonesiaand distributed to all Indonesian timberconcessions. This system is now the mostcommonly used in Indonesia. In 1991 more than30 million dipterocarps were produced by alimited number of concessions (Smits & Leppe1991). Preliminary estimates indicate that thisnumber has now doubled. Smits et al. (1993)indicate that, by agreement between a numberof countries in the Association of South-EastAsian Nations, it should be possible to guaranteea constant supply of wildlings by exchangingdipterocarp material between different regions

with different mast flowering seasons. Technicallysuch a scheme has been proved possible, butthere are a number of practical problems,including the quarantine regulations of thedifferent countries.

Attention is likely to concentrate on the mostcommercially important dipterocarp species foruse in large-scale planting. At the TROPENBOS-Kalimantan project, a priority list of 20dipterocarp species is used as a reference formost of the research activities (Table 1). Most ofthese species occur in the project area andwere selected on the basis of their high giowthrates, as determined by growth and yieldresearch, the use of their timber, and practical

Table 1. Priority list of dipterocarp species for future

practical research (TROPENBOS-Kalimantan propagation and

'stand establishment project)

experience. However, if attention concentrates onsuch a limited number of species, a shortage ofwildlings will eventually occur. Such shortagesalso arise because the number of seedlings inthe natural forests decreases sharply in the yearsafter a mast fruiting. Moreover, the older andlarger the seedlings, the higher their mortalityand the more difficult they are to handle (ie theyneed longer in the nursery larger containers,more nursery space, etc). Other means ofproviding planting stock are, therefore, requiredin such circumstances. For this reason, mucheffort was put into developing a system forvegetative propagation of dipterocarps whichwould make the planting stock supplyindependent of dipterocarp mast floweringseasons. However, the most practical andeconomical way of producing planting stock oftenproves to be a combination of the two systems,producing more plants from wildlings when theyare available and switching to cuttings whenwildling supplies become scarce. Wildlings canalso be used for vegetative propagation, thematerial still being very juvenile and relativelyeasily rooted.

Vegetative propagation: rooted cuttings andhedge orchards

Vegetative propagation of dipterocarps has beentried by various researchers, with varyingdegrees of success (Srivastava & Manggil 1981;

Ha 116 & Kamil 1981; Smits 1983; Yasman & Smits1988). One of the first papers to mention rootingof Shorea leprosula cuttings was byArdikoesoema and Noerkamal (1955). Vegetativepropagation of dipterocarps offers severaladvantages, such as the production of geneticallymore homogeneous planting stock, the possibilityof multiplying selected material, andindependence from mast flowering seasons inrelation to planting stock supply. There are,however, a number of important factors affectingthe vegetative propagation of dipterocarps by therooting of stem cuttings.

Although rooting is feasible in principle, thematter of juvenile shoot production is critical fOrlarge-scale application of vegetative propagationtechniques in the field. Cuttings of dipterocarpstaken from plagiotropic branches do not, or onlyvery seldom, produce vertically growing trees.Almost all dipterocarps grow according to thearchitectural models of Roux and Massart (cfOldeman & Sieben-Binnekamp, pp25-33), so thatonly one vertical axis (the main stem) isnormally available, and the branches are allplagiotropic. This growth pattern severely limitsthe number of vertical shoots available for takingcuttings. Therefore, it is necessary to set uphedge orchards in which the stockplants can beinduced to produce many vertical shoots.Previously developed methods (Leppe & Smits1988) have now been refined by more recentresearch. The newer method induces verticalgrowth of axillary sprouts low on the main stem,and is based upon suppressing apicaldominance. Vertical shoots are bent sidewaysunder a rope net. This induces orthotropic shootsto sprout, which emerge distributed evenlythrough the net. After some time, the bent shootsbecome woody and the rope net can beremoved. Cuttings can be harvested from thesevertical shoots. Providing that the shoots areregularly cropped for cuttings, new shoots arisefrom their axils. It is important to harvest thecuttings constantly to prevent one of the verticalshoots from redeveloping strong apicaldominance and suppressing the other orthotropicshoots on the same stockplant from furtherdevelopment.

Those stockplants of Shorea leprosula whichproduced the most axillary shoots were normallyfound to yield cuttings with quicker rooting andhigher total rooting percentages. Furthermore, theresulting plants showed better growth(W T M Smits & A Erwinsyah, unpublished data).If the same is true for all dipterocarp species,this method may present an easy way ofselecting superior-quality planting stock for massproduction by vegetative propagation. Some ofthe rooted cuttings of dipterocarps planted in thefield show growth over the first five yearsamounting to over 3 cm diameter increment per

269

year and 2 m height growth per year. Someindividuals produced from wildlings growing inthe Kiani Lestari concession had incrernents of 4cm in diameter per year for the first three years.Although these performance data refer to certainindividuals only and in no way present aguarantee for future production of thesedipterocarps, they show clearly the greatpotential for genetic improvement in dipterocarps(see later).

Rooting of dipterocarp cuttings is possible insolid media and in water. Currently the watermethod, developed at Wanariset, is preferred. Inthis method, use is made of a kind of 'bubblebath', which may be best compared to aJapanese jacuzzi. The cuttings are held betweentwo palm fibre brushes suspended over watercontaining IBA or some other auxin. The solutionis aerated by a pump via porous aquariumstones. The basal ends of the stem cuttings areinserted in the solution. The leafy parts of thecuttings are partly pruned to enable the stockingof more cuttings per pair of brushes and todecrease transpiration, and a cover is placedover them to maintain high air humidity Thewater used for rooting is circulated andconstantly filtered through a small commercialdrinking-water filter and warmed to over 50°Cby solar power.

There are a number of advantages to thisrooting method. The first is that the rootingconditions are very homogeneous, especially forresearch purposes. The large body of waterbuffers temperature changes in the rooting zone.The aerated water also constantly touches thebrushes through which much water evaporates,automatically maintaining a high air humidityaround the leafy parts of the cuttings. There isalmost no maintenance needed apart from achange of water every two weeks, andassessment of the cuttings is easily achieved bylifting the brushes containing the cuttings out ofthe water. Rooted cuttings can be removedwithout causing any disturbance. Anotheradvantage is that more roots develop fromlenticels, providing a good connection betweenstem and root system. In solid media, cuttingstend to produce adventitious roots from callus atthe stem base. Many of these root systems canbe easily severed at planting time. However,cuttings rooted in water have longer and morefragile roots, which may make them moresusceptible to damage, and the plants may needmore careful maintenance (watering, misting,hardening off) as a result.

A number of factors have proved to be importantfor successful rooting of dipterocarp cuttings. Oneof the most important is the juvenility of thesource material. Omon and Smits (1989) foundthat Shorea lamellata cuttings taken from plantsyounger than one year had 100% rooting within

270

two weeks without any application of auxins. Ingeneral, older dipterocarp material needs higherdoses of auxin application for successful rooting.Avoidance of extreme temperatures (<15°C or>35°C) is also critical for successful rooting.Relative humidity of the air should be keptconstantly high around the leafy parts of thecuttings, and aeration of the rooting medium isessential. When the cuttings are pruned so thatthe basal part of the stem is cut through a node,rooting occurs more quickly Large rootingdifferences are also found between differentdipterocarp species. Dipterocarpus spp. are themost difficult to root because of the resin whichflows from the cut surface of the cutting base.Nevertheless, Somkiat (personal communication)reported 95% rooting of Dipterocarpus" alatuscuttings from very young source plants inThailand, even without application of auxins. DeFraiture (unpublished data) found that the massof the cuttings, as well as the trimming of theleaves, had a significant effect upon rooting ofS. leprosula cuttings. The time of day at whichthe cuttings are taken also has some influenceupon speed and percentage of rooting (Siagan1991), as has the degree of irradiance duringthe rooting process (cf Leakey Newton & Dick,pp72-83). Given the right combination of thesefactors, it has now proven possible to root manydifferent dipterocarp species, albeit not all withthe same degree of success.

Cuttings of 46 dipterocarp species have beensuccessfully rooted at the TROPENBOS-Kalimantanproject (Table 2), and a number of others havebeen propagated by other researchers. This workindicates that many dipterocarp species areamenable to vegetative propagation, when theright combination of rooting factors is applied.The usefulness of vegetative propagation is amatter of economics. If the species has lowgrowth potential after planting, has very low

Table 2. List of successfully rooted dipterocarp species bymeans of stem cuttings and/or air layering at the AgriculturalUniversity Wageningen, The Netherlands, and in EastKalimantan, Indonesia

Anisoptera costataA. marginataCotylelobium

malayanaDipterocarpus

confertusD. cornutusD. gracilisD. grandiflorusD. hasseltnD. humeratusD. knbricatusD. lowiiD. mundusD. tempehesDryobalanops

aromatica

D lanceolataD. keithiiHopea mengerawanH. odorataH. pierriH. rudiformisParashorea

tomentellaShores albidaS. assamicaS. balangeranS. bracteolataS. dasyphyllaS. faguetianaS. johorensisS. laevisS. lamellata

S. leprosulaS. rnacrophyllaS. multiflora

.5. ovalisS. pachyphyllaS. parvifoliaS. paucifloraS. platycladosS. polyandraS. selanicaS. seminisS. srnithianaS. stenopteraVatica chartaceaV rassakV umbonata

rooting percentages, or yields few verticalshoots suitable for the production of stemcuttings, the costs of production will be veryhigh and vegetative propagation will not be aneconomical option. Fortunately members of thegenus Shorea are among the easiest dipterocarpsto propagate, as well as being some of the mostcommercially important species. Experience hasshown that for these species vegetativepropagation is a viable alternative to productionby wildlings.

Before transplanting, rooted cuttings should beinoculated with ectomycorrhizal fungi (Smits1983). Various methods for inoculation have beentested in East.Kalimantan. The simplest and mostpractical method is the maintenance of a liveinoculum collection. This collection consists ofseedlings or cuttings inoculated with spores,pieces of sporocarp, or pure cultures ofectomycorrhizal fungi. Once the establishment ofsuitable ectomycorrhizal infection is confirmed,these mycorrhizal root systems can be used toinoculate the rooted cuttings.

The use of these techniques should besupported by a clear assessment of theexpectations based on the results achieved. Forthis purpose, a computer cost comparison modelhas been developed at the project whichaccurately calculates the cost of dipterocarpplanting stock for all possible options andsituations. This model is now being extended toinclude models for planting and maintenance.Initial analyses show that the cost of growingdipterocarps from cuttings can be reasonably lowThe programme can assist in determining themost critical factors for affecting the end price ofa cutting and help a nursery manager tooptimise his operations.

APPLICATION OF VEGETATIVEPROPAGATION IN THE FIELD

As mentioned before, planting of dipterocarpshas become popular in Indonesia. A start hasbeen made in a small number of places toapply the vegetative propagation techniques fordipterocarps for production purposes. The firstcompany to apply this technique on anoperational scale was PT INHUTANI I in theirtimber estate in Longnah (Figure 1). Longnah islocated in the middle of the extensive area thatwas burnt during the long drought of 1982-83,resulting in the destruction of more than 3million ha of forest in the Indonesian part ofBorneo, and about 1 million ha in Sabah. In theneighbourhood of Longnah, very few stands ofdipterocarps remain and the shrub vegetationresulting from the fire is now being converted totimber plantations for construction wood. At first,fast-growing light-demanding tree species areused, followed by line planting with dipterocarps.

Wildings

Virginforest Hedge

orchardSeedstand

Seedlings

Without ectomycorrhizal inoculation

Mainstem

Bent part

Cuttings

With ECMinoculation

With ectomycorrhizal inoculation

Figure 1. Production of dipterocarp planting stock in theLongnah timber estate of 17. INHUTANI I

Because sources of wildlings are scarce aroundLongnah, some of the PT INHUTANI staff weretrained at the TROPENBOS-Kalimantan projectand subsequently applied the project'spropagation techniques in the field. They set upa large hedge orchard (more than 1 ha) andmany propagating tanks for rooting the cuttings.To start up the hedge orchards, they partly usedwildlings collected from beneath healthy trees ofthe faster-growing light-red Shorea (meranti)species, and their unit is now capable ofproducing more than 300 000 cuttings a year.The trees developing from the cuttings aregrowing well. PT INHUTANI I is making use of apermanent nursery with roofed rooting facilities,shaded areas with screen wire, electricity supplyetc.

However, it is now also possible to produce thedipterocarp cuttings under much more modestconditions appropriate to the average forestryconcession. A simple water bath can beconstructed out of wooden boards that surrounda hole in the ground. The lining of this hole andthe boards are covered with a sheet of plastic,and the propagation brushes can then be put onthe edges of the boards. Normally a temporarynursery will be located where a road crosses ariver, where vegetation is still present, providingthe right light environment. If the nursery islocated along a river, use can be made of waterpower to rotate a water wheel which can in turnpump air to the rooting tanks. The soil is mostlyrather wet along the rivers, and the air humidityas a consequence is also high. Operation is,therefore, possible without covering the rootingtank with a plastic sheet. Examples of suchnurseries have now been set up anddemonstrated to be practical.

PROSPECTS FOR GENETIC IMPROVEMENT

Although rare, hybridisation in dipterocarps doesoccur (P S Ashton, personal communication).Some of the induced or naturally producedhybrids show heterosis. One example in EastKalimantan is a probable hybrid between

271

Input

DevelOpment,ofstndardtechniques

Improvementof standardtechniques

Developmentdif:'6chjah.6.6.d:...teChniqUW!

Productionof

wildlings

Cuttings ofwildlingsSeedlings

and juvenilehedge orchards

Cuttings ofrejuvenatedadult trees

Output

Abundantproduction ofplanting stock

pereCted, high

Siedorchaiils

Figure 2. Research approach for vegetative propagation at theWanariset I Dipterocarp Research Station

Dipterocarpus confertus and D. tempehes. Thetree shows intermediate morphology between thetwo mentioned species and was considered ahybrid by Ashton (personal communication). Thehybrid is growing faster than plants of either ofthe two original species, has good self-pruning,and has a very straight stem. At the same timeit does not seem to suffer from shoot-borers asis common in both the original species (Smits etal. 1990), although shoot-borers were foundwithin a few metres from the hybrid. Vegetativepropagation may prove useful in the near futureto propagate such promising individuals, althoughit may not be so relevant in the case ofDipterocarpus species, as their wood containsmuch resin and the trees are generallyslower-growing than the light-red Shorea species.

A start with provenance trials has been made inEast Kalimantan and additional trials of someseven species will be established soon. Inaddition, some 50 different clones of two Shoreaspp. have been selected with good potentialgrowth for testing on a larger scale. Research isalso continuing on the rooting of mature shootsby air layering, and on other aspects of treeimprovement (Figure 2).

CONCLUSIONS

Recent developments in Indonesia have shownthat there are good prospects for use ofvegetative propagation techniques in tropicalforestry These techniques are not solely suitablefor typical plantation species like Eucalyptus spp.and Gmelina arborea, but can also be used forindigenous species such as the dipterocarps, in

272

a cost-effective way The research resultsreported here suggest that in the futurevegetative propagation of dipterocarps may beapplied to the production of highly productiveclones, providing a completely new economicoutlook on the use of dipterocarps in plantationforestry, as well as the enrichment of naturalstands.

REFERENCES

Ardikoesoema, R.I. & Noerkamal M.T. 1955. Pertjobaantanaman Shorea leprosula di Djawa. (Pengumuman BalaiPenjelidikan Kehutanan, no. 48.)

Hallé, F. & Kamil, H. 1981. Vegetative propagation ofdipterocarps by stem cuttings and air layering. MalayanForester 44, 314-318.

Leppe, D. & Smits, W.T.M. 1988. Methoda pembuatan danpemeliharaan Kebun Pa.ngkas Dipterocarpaceae. (ForestReseamh Institute of Samarinda, no. 4. Special edition.)Samarinda: Forest Research Institute.

MN. 1992. Forestry information paper Indonesian ForestryCommunity

Omon, M.R. & Smits, W.T.M. 1989. Pengaruh IBAterhadap pertumbuhan stek batang Shorea lamellate. BuletinPenelitian Kehutanan, 5 (2), 103-112.

Siagan, Y.T. 1991. Pengaruh waktu pengambilan bahan stekterhadap daya perakaran stek batang Shorea parvifolia Dyer.Forest Research Bulletin, 536, 35-42.

Smits, W.T.M. 1983. Vegetative propagation of Shorea cf.obtusa and Agathis dammara.by means of leaf cuttings andstem cuttings. Malayan Forester, 46, 175-185.

Smits, W.T.M. 1986. Pedoman siStern cabutan bibitDipterocarpaceae, edited by S. Priasukmana & J. Tangketasik.(Edisi Khusus, no. 2.)

Smits, W.T.M. & Leppe, D. 1991. Prospek penanaman jenispohon Dipterocarpaceae melalui peranan kerja samapenelitian dan pengembangan. Rimba Indonesia, 50-52.

Smits, W.T.M., Leppe, D., Yasman, I. & Noor, M. 1990.Some observations on diseases of Dipterocarpaceae. In:Proceedings of the 4th Round Table Conference onDipterocarps. Samarinda: Biotrop.

Smits, W.T.M., Leppe, D., Yasman, I. & Noor, M. 1993.Ecological approaches to commercial dipterocarp forestry(Keynote address to the International Conference on ForestBiology and Conservation in Borneo, Kota Kinabalu, 1990.) Inpress.

Srivastava, P.B.L. & Manggil, P. 1981. Vegetativepropagation of some diptenocarps by cuttings. MalayanForester, 44, 144-160.

Tompsett, P.B. 1987. Desiccation and storage studies onDiPterocarpus seeds. Annals of Applied Biology, 110, 371-379.

Yasman, I. & Smits, W.T.M. 1988. Metoda pernbuatan stekDipterocarpaceae. (Balai Penelitian Kehutanan Samarinda.)(Edisi Khusus, no. 3.)

Poster paper titles

Some wild Malaysian species - domesticationthrough agroforestry(A M Ahmad, Universiti Pertanian Malaysia, Serdang,Selangor, Malaysia)

An attempt to establish indigenous tropical foresttrees of economic potential on mining wastelandat the Kenya coast(S Baer, R D Haller & A Robertson, Baobob Farm Ltd,PO Box 81995, Mombasa, Kenya)

CONSEFORH - conserving and developing forestgenetic resources in Honduras(A J Barrance, CONSEFORH, Apartado Postal 45,Siguatepeque, Honduras)

Neem decline in West Africa(E R Boa, Natural Resources Institute, Chatham, KentNIE4 4TB, UK)

The domestication of Coula edulis, amultipurpose species of the humid tropicalforests of Africa: the germination problem(L Bonnéhin, Centre Neerlandais, BP V51, Abidjan 01,Côte d'Ivoire)

Native tree species for improvement ofsecondary forest vegetation(S Brienza, EMBRAPA-CPATU, Cx Postal 48, 66240,Belem, PA Brazil)

Agroforestry vs forest destruction in easternBolivia(C Carden, ABAFA SRL, Santa Cruz, Bolivia)

Survival and growth of seedlings of four tropicaldry forest tree species, Hymenaea courbaril,Swietenia macrophylla, Manilkara chicle andCedrela odorata in Costa Rica(K Gerhardt, Institute of Ecological Botany UppsalaUniversity Box 559, 751 22 Uppsala, Sweden)

Studies in the nitrogen nutrition of Acaciaa uriculiformis seedlings(S R Goi & J I Sprent, Department of BiologicalSciences, University of Dundee, Dundee DD1 4HN, UK)

Multipurpose trees: the requirements of thelocal community(S Harris, Henry Doubleday Research Association,Ryton-on-Dunsmore, Coventry CV8 3LG, UK)

Sesbanias: model woody legumes forbiotechnological development(P Harris, Department of Biological Sciences, CoventryUniversity Priory Street, Coventry CV1 5FB, UK)

Early growth and survival of four open planteddipterocarps in peninsular Malaysia(A L Hoe, Forest Research Institute of Malaysia, POBox 201, Kepong, 52109 Kuala Lumpur, Malaysia)

CSIRO Division of Forestry(A House, Australian Tree Seed Centre, CSIRO Divisionof Forestry PO Box 4008, Queen Victoria Terrace,Canberra ACT 2600, Australia)

Leucaena salvadorensis: conservation andutilisation in Central America(C E Hughes, Oxford Forestry Institute, Department ofPlant Sciences, University of Oxford, South Parks Road,Oxford OX1 3RB, UK)

Floral biology and vegetative propagation of theAfrican plum (Dacryodes edulis)(J Kengue, IRA/CRA, Nkolbisson, BP 2067,Yaoundé, Cameroon)

Application of clonal technology for treeimprovement in India - ITC Bhadrachalamexperience(P Lal, ITC Bhadrachalam Paperboards Ltd, 106 SarclarPatel Road, Secunderabad 500 003, Andhra Pradesh,India)

Low-technology propagation of tropical trees:appropriate technology for rural development(R R B Leakey J F Mesén, P N Shiembo, D Ofori, TNketiah, A Hamzah, Z Tchoundjeu, C Njoya, P Oduol,A C Newton, J McP Dick & K A Longman, Institute ofTerrestrial Ecology Bush Estate, Penicuik, MidlothianEH26 OQB, UK)

Calliandra calothyrsus and the domestication ofclose relatives(D Mac Queen, Oxford Forestry Institute, Department ofPlant Sciences, University of Oxford, South Parks Road,Oxford OX1 3RB, UK)

Selection for resistance to the mahogany shoot-borer in Central America and the Caribbean(A C Newton, J F Mesén, P Baker & R R B LeakeyInstitute of Terrestrial Ecology Bush Estate, Penicuik,Midlothian EH26 OQB, UK)

Nodulation and characteristics of rhizobialisolates trapped with various legume genera in asemi-arid tropical soil(D W Odee, J M Sutherland & J I Sprent, Departmentof Biological Sciences, University of Dundee, DundeeDD1 4HN, UK)

Study of Inga genetic resources in Central andSouth America(T D Pennington, Royal Botanic Gardens Kew,Richmond, Surrey TW9 3AB, UK)

The effects of moisture content on the lowtemperature responses of Araucaria hunsteiniiseed and embryos(H W Pritchard, Royal Botanic Gardens Kew, WakehurstPlace, Ardingly Haywards Heath, Sussex RH17 6TN, UK)

Strategies for the selection of salt tolerance intropical woody legumes(I Puddephat, Department of Biological Sciences,Coventry University Priory Street, Coventry CV1 5FB,UK)

Ecophysiology of seeds and storability problemsin seven valuable tropical rain forest treespecies from south-eastern Mexico(M C Rodriguez-Hernandez & C Vazquez-Yanes,Centro de Ecologia, National University of Mexico,Apartado Postal 70-275, Ciudad Universitaria, 04510Mexico DF, Mexico)

The forestry management and plantation project(J Sandom, c/o FCC (Colombo, Sri Lanka), KingCharles Street, London SW1A 2AH, UK)

Infection process and root nodule developmentin Anadenanthera falcata Speg.(M A Schiavinato & J I Sprent, Department of BiologicalSciences, University of Dundee, Dundee DD1 4HN, UK)

Domestication of indigenous fruit trees inCameroon(P Shiembo, A C Newton & R R B Leakey ForestResearch Centre, Agronomic Research Institute, PO Box29, Kumba, SW Province, Cameroon)

273

Nutrient cycling and competition: the potentialfor management in agroforestry(S Smyth, Dept of Plant Sciences, University ofCambridge, Downing Street, Cambridge CB2 3EA, UK)

Shrubby forage legumes for the acid soils ofLatin America(R J Thomas, C Lascana, P Perdomo & B Maass, CIAT,Apartado Areo 6713, Cali, Colombia)

The selection of Khaya senegalensis for use asshade trees and line plantations(G W Tolkamp, Centre National de SemencesForestieres, 01 BP 2682, Ouagadougou 01, BurkinaFaso, West Africa)

Effect of the Rhizobium inoculation in Prosopispallida seedlings(J A Vasquez, L Rivero & C Door, Department ofForestry North Carolina State University Box 8002,Raleigh, NC 27695, USA)

Comparative evaluation of agroforestry trees onvertisol soils under low rainfall(I Verinumbe, F L Sinclair, W P Adderley & D AJenkins, Department of Biological Sciences, Universityof Maiduguri, PO Box 1069, Maiduguri, Nigeria)

274

Optimisation of agroforestry systems withtraditional tree crops in SW province,Cameroon(J Wainwright, Fountain Renewable Resources Ltd,Mollington House, Mollington, Banbury OX17 1AX, UK)

Domestication of rainforest trees in Cameroon( J Wainwright & N Hutton, Fountain RenewableResources Ltd, Mollington House, Mollington, Banbury0X17 lAX, UK)

Application of vegetative propagationtechniques for Dipterocarpaceae in EastKalimantan, Indonesia(I Yasmin, W T M Smits, D Leppe, A De Fraiture & AErwinsyah, Forestry Research Project (TROPENBOS), j1Industri 1/66, RT 37, Balikpapan 76103, Indonesia)

Endomycorrhizal research in the Philippines(J T Zarate, R E de la Cruz & J C Dodd, NationalInstitutes of Biotechnology and Applied Microbiology(BIOTECH), University of the Philippines at Los BariosCollege, Lagima, The Philippines)

List of Conference participants

M Abu-JanOxford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

J AcworthFountain Renewable Resource LtdMollington HouseMollingtonBanbury 0X17 lAXUK

W AdelereWorld Wide Fund for Nature UKWeyside ParkGodalmingSurrey GU7 1XRUK

V AgyemanForestry Research Institute of GhanaUST PO Box 63KumasiGhana

Prof A M AhmadForestry CommissionAlice Holt LodgeWreccleshamFarnhamSurrey GU10 4LHUK

M M AhmedOxford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

M AhsanOxford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

B A AkinolaOxford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

E AlbrechtsenDLHLohdesveg 4DK-7442 EngesvangDenmark

A K AmaduOxford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

F S AmoahOxford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

J AnaneOxford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

Dr L AndersonInstitute of Terrestrial EcologyBangor Research UnitUniversity College of North WalesDeiniol RoadBangorGwynedd LL57 2UPUK

G D ArmstrongOverseas Development AdministrationForestry Advisorc/o The British CouncilSCRN 708/709 Gl. F 1/3Caixa Postal 610470740 Brasilia DFBrazil

L C AvellanedaOxford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

A J BarranceCONSEFORHApartado Postal 45SiguatepequeHonduras

275

Miss C BatchelorDepartment of Biological SciencesCoventry UniversityPriory StreetCoventry CV1 51-13UK

B Bigger12 WI Warrender Park TerraceMarchmontEdinburgh EH9 1EGUK

E R BoaNatural Resources InstituteChathamKent ME4 4TBUK

A S K Boachie-DapaahOxford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

Miss L BonnehinCentre NeerlandaisBP V51 Abidjan 01Côte d'Ivoire

Dr T BoothAustralian Tree Seed CentreCSIRO Division of ForestryPO Box 4008Queen Victoria TerraceCanberra ACT 2600Australia

Prof J L BrewbakerDepartment of HorticultureUniversity of Hawaii3190 Mai le WayHonoluluHawaii 96822USA

S BrienzaEMBRAPA-CPATUCx Postal 4866240 Belem PABrazil

E M Lammerts van BuerenThe Tropenbos FoundationPO Box 2326700 AE WageningenThe Netherlands

D FRP BurslemDepartment of Plant SciencesUniversity of CambridgeDowning StreetCambridge CB2 .3EAUK

276

K CampbellNatural Resources InstituteCentral AvenueChatham MaritimeChathamKent ME4 4TBUK

Dr M G R CarmenInstitute of Terrestrial EcologyBush EstatePenicuikMidlothian EH26 OQBUK

C CardenDirector ExecutiveABAFA SRLSanta CruzBolivia

J C CentenoWorld Wide Fund for Nature UKWeyside ParkGodalmingSurrey GU7 1XRUK

M S ChaudryOxford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

M ChebetOxford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

E Chevallierc/o IntercooperationMaulbeerstrasse 10CH-3001 BernePO Box 6724Switzerland

J A Reyes ChirinosOxford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

Dr C ClementsUniversity of Hawaii at Manoa3190 Maile WayHonolulu HI 96822USA

J D DeansInstitute of Terresliial EcologyBush EstatePenicuikMidlothian EH26 OQBUK

J Delaunay31 Rue de Luxembourg72000 Le MansFrance

Dr J McP DickInstitute of Terrestrial EcologyBush EstatePenicuikMidlothian EH26 OQBUK

Dr J C DoddBiological LaboratoryUniversity of KentCanterburyKent CT2 7NJUK

A K DuttOxford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

E DzobokuOxford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

0 A EffaOxford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

E A ElsheikhOxford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

R EffendiSchool of Agricultural and Forest Sciences Mrs A de Fraiture

University of Wales Gravenbroekseweg 69

Bangor 2811 CC

Gwynedd LL57 2UW Reeuwijk

UK The Netherlands

Prof F EngelmannOffice de la Recherche Scientifique et Technique

d'Outre-MerCentre de Montpellier2501 Avenue du Val de MontferrandBP 504534031 MontpellierFrance

Dr F E FasehunDepartment of Forestry and Resource ManagementUniversity of lbadanOyo RoadIbadanNigeria

Prof P Felker

Dr S Duguma Center for Semi-Arid Forest Resources

Institut National de Recherche Caesar Kieberg Wildlife Research Institute

Agronomique/International Centre for Research in Texas A&I UniversityAgroforestry Campus Box 156PB 2067 KingsvilleYaoundé TX 78363Cameroon USA

S Z FerdousOxford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

Dr E C M FernandesNorth Carolina State University/EMBRAPA-CPAACP4554A-201 Rua Paraiba 624Manaus 69001 AIVIBrazil

Dr G S FosterUS Department of Agriculture Forest ServiceSouthern Forestry Experimental StationPlant and Soil Science DepartmentPO Box 1328Normal •Alabama 35762USA

277

Dr H FriebergEuropean Tropical Forest Research Networkc/o ATSAFHans-Bockler Strasse 5D-5300 Bonn 3Germany

Miss K GerhardtInstitute of Ecological BotanyUppsala UniversityBox 559751 22 UppsalaSweden

G GibsonInstitute of Ecology and Resource ManagementUniversity of EclinburghDarwin BuildingKing's BuildingsMayfield RoadEdinburgh EH9 3j1JUK

A GithithoOxford Forestry InstituteDepartment of Plard,SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

Mrs S R GoiDepartment of Biological SciencesUniversity of DundeeDundee DD1 4HNUK

Dr A R GriffinSIPCNTBFShell CentreLondon SE1 7NAUK

C HallAracruz International LimitedNew Malden House1 Blagdon RoadNew MaldenSurrey KT3 4TBUK

Dr R D HallerBaobab Farm LtdPO Box 81995MombasaKenya

H HamzaUniversity of EdinburghKing's BuildingsMayfield RoadEdinburgh EH9 3jUUK

278

Mrs S HarrisOverseas DepartmentHenry Doubleday Research AssociationRyton-on-DunsmoreCoventry CV8 3LGUK

Dr P HarrisDepartment of Biological SciencesCoventry UniversityPriory StreetCoventry CV1 5FBUK

Dr M HarrisonLTS International Limited10 WoodhallMillbraeEdinburgh EH14 5BJUK

F J HarveyInstitute of Terrestrial EcologyBush EstatePenicuikMidlothian EH26 OQBUK

Prof 0 W HealInstitute of Terrestrial EcologyBush EstatePenicuikMidlothian EH26 OQBUK

Dr J HebertFaculté des Sciences AgronomiquesUER Foresterie des Pays Chauds2 Passage des DeportesB-5030 GemblouxBelgium

S HennehUniversity of EdinburghKing's BuildingsMayfield RoadEdinburgh EH9 3JUUK

Dr C Heygate49 Toddington DriveNortonStockton-on-Tees TS20 1RQUK

A L HoeForest Research Institute of MalaysiaPO Box 201Kepong52109 Kuala LumpurMalaysia

P H V HolWijma Kampen BVHaatlandhaven 3PO Box 2418260 AE KampenThe Netherlands

Dr G D Holmes A Kalinganire7 Cammon Road University of WalesBarnton School of Agricultural and Forest SciencesEdinburgh EH4 8EF BangorUK Gwynedd LL57 2UW

UKMiss I HoranDepartment of Biological SciencesCoventry UniversityPriory StreetCoventry CV1 5FBUK

Dr A HouseAustralian Ti-ee Seed CentreCSIRO Division of ForestryPO Box 4008Queen Victoria TerraceACT 2600Australia

C E HughesOxford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

C KaliappanUniversity of EdinburghKing's BuildingsMayfield RoadEdinburgh EH9 3JUUK

Dr P Kapoor-VijayCommonwealth Science CouncilMarlborough HousePall MallLondon SW1 51-1XUK

Dr J KengueIRA/CRA NkolbissonBP 2067YaoundeCameroon

C KiminiOxford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

N HuttonShuttleworth College A S K Korang

Old Warden Park Oxford Forestry Institute

Biggleswade Department of Plant Sciences

Bedfordshire SG18 9DX University of Oxford

UK South Parks RoadOxford OX1 3RB

K Ingleby UKInstitute of Terrestrial EcologyBush Estate E KorirPenicuik Oxford Forestry InstituteMidlothian EH26 OQB Department of Plant SciencesUK University of Oxford

N Jones South Parks Road

The World Bank (ASTAG) Oxford OX1 3RB

1818H St NW UK

Washington DC 20433USA

S KumarOxford Forestry Institute

M Msafiri Kabota Department of Plant SciencesDistrict Natural Resources Office University of OxfordBox 69 South Parks RoadSingida Oxford OX1 3RBrIlanzania UKEast Africa

Dr F KwesigaForest DepartmentPO Box 22099KitweRepublic of Zambia

Dr D 0 LadipoT M Kalifulla International Centre for Research in AgroforestryISO Swed Forest c/o International Institute of nopical Agriculture34 Victoria Crescent PMB 5320Madras 600-105 IbadanIndia Nigeria

279

C s R LakshmiOxford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

P LalVice-PresidentITC Bhadrachalam Paperboards Ltd106 Sardar Patel RoadSecunderabad 500 003Andhra PradeshIndia

Dr F LapeyrieInstitut National de la Recherches AgronomiquesCentre de Recherches Forestières de NancyLaboratoire de Microbiologie54280 ChampenouxSeichampsFrance

Ms LasminiUniversity of EdinburghKing's BuildingsMayfield RoadEdinburgh EH9 3JUUK

Ms A LawrenceBarn CottageOld Boars HillOxford OX1 5SJUK

G J Lawsonc/o Office National de Dévéloppement des ForêtsODA Forest Management and Regeneration ProjectBP 163MbalmayoCameroon

280

D J MacQueenOxford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

J M MafannyOxford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

Dr F M MalayaDivision of Forest ResearchPO Box 22099KitweZambia

Dr D C MalcolmInstitute of Ecology and Resource ManagementUniversity of EdinburghDarwin BuildingKing's BuildingsMayfield RoadEdinburgh EH9 3J1JUK

C M ManikantOxford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

S MariithiOxford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

Dr A F MascarenhasDr R R B Leakey Plant Tissue Culture DivisionInstitute of Terrestrial Ecology National Chemical LaboratoryBush Estate Pune 411008Penicuik IndiaMidlothian EH26 OQB

Dr P A MasonUKInstitute of Terrestrial Ecology

M Lima Bush EstatePenicuikc/o Institute of Terrestrial EcologyMidlothian EH26 OQBBush EstateUKPenicuik

Midlothian EH26 OQB D MauambetaUK Oxford Forestry Institute

Department of Plant SciencesDr B Lundgren University of OxfordAlagen 1 South Parks Road18245 Enebyberg Oxford OX1 3RBSweden UK

F Bengono MbidaOxford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

C Mc BeathInstitute of Terrestrial EcologyBush EstatePenicuikMidlothian EH26 OQBUK

S McGreevyInstitute of Terrestrial EcologyBush EstatePenicuikMidlothian EH26 OQBUK

J F MesénCentro Agronomico Ropical deInvestigacion y EnserianzaTurrialba 7170Costa Rica

P B MilimoAustralian National UniversityDepartment of ForestryGPO Box 4ACT 2601 CanberraAustralia

F R MillerOxford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

Dr S MoestrupDanida Forest Seed CentreKrogerupvej 3A3050 HumlebaekDenmark

R C MunroInstitute of Terrestrial EcologyBush EstatePenicuikMidlothian EH26 OQBUK

G M MutuambuguDistrict Agriculture OfficePO Box 12MebuKenya

S K NegiOxford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

Dr A C NewtonInstitute of Terrestrial EcologyBush EstatePenicuikMidlothian EH26 OQBUK

A K NigamUniversity of EdinburghKing's BuildingsMayfield RoadEdinburgh EH9 3jUUK

E ObungaSchool of BiologyUniversity of SussexBrighton BN1 2RFUK

D W OdeeDepartment of Biological SciencesUniversity of DundeeDundee DD1 4HNUK

Dr P A OduolInternational Centre for Research in AgroforestryPO Box 30677NairobiKenya

PO OjuederieUniversity of EdinburghKing's BuildingsMayfield RoadEdinburgh EH9 3JUUK

Dr J C Okafor3 Kingsway RoadPO Box 3856EnuguNigeria

M H OmariRegional Development DirectorPO Box 5SingidaTanzania

K OrgkeForestry Research Institute of GhanaUST PO Box 63KumasiGhana

281

Dr K Osei-BonsuSilsoe CollegeSilsoeBedfordshire MK45 4DTUK

D D OtuOxford-Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

Dr J R PalmerTropical Forestry and Computing Ltd93 Gidley WayHorspathOxford 0X9 1TQUK

I D PandeOxford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

Dr V PatilAgro Forestry FederationSandeep ApartmentThrnbak RoadNashik 422002India

Dr T D PenningtonHerbariumRoyal Botanic Gardens KewRichmondSurrey TW9 3ABUK

M Y Poku-MarboahForest Resources Management ProjectPO Box 1457 KumasiGhana

Prof G T PranceDirectorRoyal Botanic Gardens KewRichmondSurrey TW9 3ABUK

I PuddephatDepartment of Biological SciencesCoventry UniversityPriory StreetCoventry CV1 5FBUK

R PurwarUniversity of EdinburghKing's BuildingsMayfield RoadEdinburgh EH9 3JUUK

282

Z QureshiOxford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

R B S RawatOxford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

Dr D A RookForestry CommissionNorthern Research StationRoslinMidlothian EH25 9SYUK

T A S SampathkumarISO Swed Forest34 Victoria CrescentMadras 600-105India

S SamaiUniversity of EdinburghKing's BuildingsMayfield RoadEdinburgh EH9 3JUUK

I SamsoedinSchool of Agricultural and Forest SciencesUniversity of WalesBangorGwynedd LL57 2UWUK

J Sandomc/o FCO (Colombo, Sri Lanka)King Charles StreetLondon SW1A 2AHUK

Mrs M A SchiavinatoDepartment of Biological SciencesUniversity of DundeeDundee DD1 4HNUK

T I Mohammed ShahISO SWed Forest34 Victoria CrescentMadras 600-105India

S K SharmaOxford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

Mrs A Sieben-BinnekampHaarweg 2176709 RM WageningenThe Netherlands

Dr A J SimonsOKford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RB.

Dr F L SinclairSchool of Agricultural and Forest SciencesUniversity of WalesBangorGwynedd LL57 2UWUK

D SinghDepartment of BotanyPunjab UniversityCahbdigarh 160 014India

W T M SmitsForestry Research Project (TROPENBOS)PO Box 319Balikpapan 76103Indonesia

Miss S SmythDepartment of Plant SciencesUniversity of CambridgeDowning StreetCambridge CB2 3EAUK

M SolumanUniversity of EdinburghKing's BuildingsMayfield RoadEdinburgh EH9 3JUUK

Prof J I SprentDepartment of Biological SciencesUniversity of DundeeDundee DD1 4HNUK

Ms J L StewartOxford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

V SuangthoUniversity of EdinburghKing's BuildingsMayfield RoadEdinburgh EH9 3jUUK

F SullivanWorld Wide Fund for Nature UKWeyside ParkGodalmingSurrey GU7 DCRUK

SuryantoUniversity of EdinburghKing's BuildingsMayfield RoadEdinburgh EH9 3JUUK

Z TchoundjeuOxford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

Dr S ThomasSchool of BiologyUniversity of SussexBrighton BN1 2RFUK

Dr R J ThomasCIATApartado Aereo 6713CaliColombia

Dr J ThompsonForestry Research Institute of GhanaUST PO Box 63KumasiGhana

G W TolkampCentre National de Semences Forestieres01 BP 2682Ouagadougou 01Burkina FasoWest Africa

Dr P TompsettRoyal Botanic Gardens KewWakehurst PlaceArdinglyNr Haywards HeathWest Sussex RH17 6TNUK

Dr J W TurnbullACIARPO Box 1571Canberra ACT 2601Australia

S UdofiaUniversity of EdinburghKing's BuildingsMayfield RoadEdinburgh EH9 3JUUK

283

J A VasquezNorth Carolina State UniversityBox 8002Raleigh NC 27695USA

Dr C Vazquez-YanesCentro de EcologiaNational University of MexicoApartado Postal 70-275Ciudad Universitaria04510 Mexico DFMexico

Dr I VerinumbeDepartment of Biological SciencesUniversity of MaiduguriPO Box 1069MaiduguriNigeria

Dr R VicencioCommonwealth Science CouncilMarlborough HousePall MallLondon SW1 5HXUK

Miss J WainwrightFountain Renewable Resources LtdMollington HouseMollingtonBanburyOxon OX17 lAXUK

K WakaneneInstitute of Terrestrial EcologyBush EstatePenicuikMidlothian EH26 OQBUK

Prof P G WatermanPhytochemistry Research LaboratoriesDepartment of Pharmaceutical SciencesUniversity of Strathclyde204 George StreetGlasgow G1 1XWUK

0 J Web leyOxford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

Printed in the United Kingdom for HMSODd 297188 CIO 5/94 531/3 12521

284

Dr J WilsonInstitute of Terrestrial EcologyBush EstatePenicuikMidlothian EH26 OQBUK

Dr P WoodSenior Forestry AdvisorOverseas Development Administration94 Victoria StreetLondon SW1E 5F,UK

I YasmanIBN-DLOPostbus 236700 AA WageningenThe Netherlands

D Yupetc/o Swedish Free Mission—Juba SudanPO Box 43318NairobiKenya

H ZakariOxford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

A ZaidiUniversity of EdinburghKing's BuildingsMayfield RoadEdinburgh EH9 331JUK

Mrs J T ZarateBiological LaboratoryUniversity of KentCanterburyKent CT2 71\IJUK

R K ZutshiOxford Forestry InstituteDepartment of Plant SciencesUniversity of OxfordSouth Parks RoadOxford OX1 3RBUK

á

á

á

9

ISBN 01562-8

780117 15623