Domestication of Irvingia gabonensis: 2. The selection of multiple traits forpotential cultivars from Cameroon and Nigeria.

A.R. Atangana1, V. Ukafor2, P. Anegbeh3, E. Asaah1, Z. Tchoundjeu1, J.-M. Fondoun4, M.Ndoumbe4 and R.R.B. Leakey5,6,*1IRAD/ICRAF Project, International Centre for Research in Agroforestry, B.P. 2067, Yaoundé, Cameroon;2Rivers State University of Science and Technology, P.M.B. 5080, Port Harcourt, Rivers State, Nigeria;3International Centre for Research in Agroforestry (ICRAF/IITA Onne), PMB 008, Nchia-Eleme, PortHarcourt, Rivers State, Nigeria; 4Institut de la Recherche Agricole pour le Développement (IRAD), B.P. 2067,Yaoundé, Cameroon; 5Centre for Ecology and Hydrology (CEH), Bush Estate, Penicuik, Midlothian, EH260QB, Scotland, UK; 6current address: Agroforestry and Novel Crops Unit, School of Tropical Biology, JamesCook University, Cairns, Queensland 4870, Australia; *Author for correspondence (e-mail: roger.@jcu.edu.au)

Received 14 March 2001; accepted in revised form 25 June 2001

Key words: Bush mango, Dika nut, Elite trees, Fruit and kernel traits, Phenotype

Abstract

Ten fruit and kernel traits were assessed in 24 fruits of each of 152 Irvingia gabonensis trees in three distinctpopulations in west and central Africa [2 populations of non-planted trees in Cameroon: Nko’ovos II (21 trees)and Elig-Nkouma (31 trees) and 1 population of planted trees in Nigeria: Ugwuaji (100 trees)]. Strong relation-ships were found between fruit weight and other fruit traits (e.g. flesh weight [r 2 = 0.99: P < 0.001], fruit length[r 2 = 0.74–0.83: P < 0.001], fruit width [r 2 = 0.77–0.88: P < 0.001]). In contrast, relationships between kernelweight and other kernel/nut traits (e.g. shell weight and nut weight) were found to be weak [r 2 = 0.009–0.37, P= 0.058–0.001], with the exception of nut weight at Nko’ovos II (r 2 = 0.65, P < 0.001). Relationships betweenfruit and kernel traits (fruit mass v. kernel mass, fruit mass v. shell mass, flesh mass v. kernel mass, nut mass v.fruit mass and flesh depth v. kernel mass) were found to be very weak. This indicates that domestication throughthe selection and vegetative propagation of multiple-trait superior phenotypes is unlikely to be able to combinegood fruit characteristics and good kernel characteristics within cultivars. Consequently, domestication activitiesshould independently focus on ideotypes representing: ‘fresh fruit’ traits, and ‘kernel’ traits, that combine highvalues of the different fruit and kernel characteristics respectively. Evidence from this study indicates that selec-tion of the three trees closest to the fruit ideotype per village as the mother plants for vegetative propagation andcultivar development, should give village level gains of 1.3 – 2-fold in fruit mass, and up to 1.5-fold in taste.Similarly for the kernel ideotype, selection of the three trees with the best fit would give potential gains in kernelmass of 1.4 – 1.6-fold.

Introduction

Domestication of indigenous fruit trees within agro-forestry practices offers important ‘Win-Win’ oppor-tunities for the transformation of land use in Africa(Sanchez and Leakey (1997); Leakey (2001a,2001b)). Consequently, the International Centre forResearch in Agroforestry (ICRAF) and its partnershave launched an Agroforestry Tree Domestication

Programme (Leakey and Simons (1998)). This pro-gramme is based on the vegetative propagation oftrees identified as ‘elite-trees’ by collaborating farm-ers developing nurseries at the village level(Tchoundjeu et al. 1998). To be successful, tree do-mestication should provide farmers with both foodsecurity and opportunities for cash generation.

Irvingia gabonensis (Aubry Lecomte ex O’Rorke)(Bush mango/Dika nut) is an indigenous tree of west

221Agroforestry Systems 55: 221–229, 2002.© 2002 Kluwer Academic Publishers. Printed in the Netherlands.

and central Africa, found from Nigeria to Congo,which produces commercially important non-timberforest products. Okafor (1975) identified two variet-ies, I. gabonensis var. gabonensis and I. gabonensisvar. excelsa. These were recently named Irvingia ga-bonensis and Irvingia wombulu respectively, by Har-ris (1996). The main widely consumed products fromI. gabonensis are kernels (Ndjouenkeu et al. 1996;Ndoye et al. 1997) and fresh fruit. In a recent surveyof farmer’s species priorities for agroforestry, I. ga-bonensis was identified as the top priority species fordomestication in the humid lowlands of West Africa(Franzel et al. 1996). Leakey (1999) reviewed the nu-tritional and the commercial values of I. gabonensisfruits and kernels.

Prior to this study, phenotypic variation in fruittraits in I. gabonensis has only been descriptive (La-dipo et al. 1996). However, recently, within a projectto examine the biophysical and socio-economic con-straints to domestication in indigenous fruits Atan-gana (2000) and Ukafor (2001) have quantified themorphological variability of fruit and kernel traits inCameroon and Nigeria respectively. In addition, ker-nels from these studies are being analysed for varia-tion in their food-thickening properties and fat con-tent. Together, these studies are indicating that in I.gabonensis there is considerable opportunity for theselection of elite trees for multiplication as cultivars(Atangana et al. 2001), using air layering, grafting orcuttings (Shiembo et al. 1996). Recent studies in I.gabonensis indicate that the best approach may be theselection of ideotypes that express desirable combi-nations of traits for either fresh fruit or for kernelproduction (Leakey et al. (2000)). The present studyexamines in more detail the opportunities for multipletrait selection, which can also take into account thedifferent preferences of men and women, even withina village (Mbosso 1999), as well as the preferencesof farmers from different villages. A future paper willexamine the extent to which the trees in this studyhave been subjected to genetic selection (Leakey etal. 2001).

Methods and materials

The study was conducted in two villages in Cameroonand one in Nigeria (Table 1). These sites were sepa-rated by 100–350 km and hence are clearly differentpopulations. The use of three geographically distinctsites should reduce the chance of finding correlatedtraits that may be due to random non-general associa-tions that can occur in a single isolated population.Two of the sites (Nko’ovos II in Cameroon and Ug-wuaji in Nigeria) were in fact within the two geneticdiversity hotspots identified by Lowe et al. (2000),using DNA markers. These sites were also geneticallydistinct. The I. gabonensis trees occurred naturally onthe farms in Cameroon (Atangana 2000), but wereplanted in home gardens in Nigeria Ukafor (2001).Twenty-four fruits were collected from each tree asthey ripened over the fruiting season (May–August)in 1999 as described by Atangana et al. (2001).

The 24 collected fruits were weighed fresh in thevillages using small portable kitchen scales graduatedto 2 g. They were then measured (length and width)using calipers graduated to 0.1 mm. A spike attachedto the same caliper was used to measure fruit flesh(mesocarp) depth in fruit thickness dimension. Afterthe measurements, 2–3 fruits per tree were assessedfor taste (scored 1 [bitter] – 5 [sweet]) and for fibros-ity (scored 1 [non-fibrous] – 5 [fibrous]). At the sametime, skin and flesh colours were assessed using theMethuen Code of Colour (Kornerup and Wanscher(1978)). Fruits were then taken to the laboratory fordepulping. The washed nuts (endocarp and cotyle-dons – see Figure 1) were then dried in the sun for2–3 days and weighed using an electronic balance(Mettler Toledo PB 3002). The kernels were storedfor chemical analysis. Since the ease with which nutscan be cracked to allow kernel extraction is seen asan important trait, shell weight was derived (fruitmass − nut mass).

Data were collated using Microsoft Excel 97. Sta-tistical analyses were done using Genstat 5 (3.2). Thisstudy particularly examined the relationships between

Table 1. Location of Irvingia gabonensis populations used in this study

Elig-Nkouma (Cameroon) Nko’ovos II (Cameroon) Ugwuaji (Nigeria)

Latitude (°N) 4° 07� 2° 56� 6°25�

Longitude (°E) 11°24� 11° 21� 7°32�

Altitude (m) 461 610 175

No. of trees collected 31 21 100

222

different fruit, nut and kernel characteristics using lin-ear regression analysis.

Results

Fruit traits

With the exception of taste, relationships betweenfruit characteristics were generally strong (r 2 > 0.70;P = 0.001) and linear (Table 2, Figure 2).

Nut and kernel traits

With the exception of nut mass v. shell mass and nutmass v. kernel mass in Nko’ovos II (r 2 = 0.65, P <0.001), relationships between kernel traits were foundto be weak in trees from Cameroon and Nigeria (Ta-ble 3, Figure 3).

Fruit and kernel traits

With the exception of nut mass v. fruit mass in Ug-wuaji, relationships between fruit, nut and kerneltraits were found to be weak (Table 4). The relation-ships between kernel mass and fruit mass, kernelmass and flesh thickness and kernel mass and fleshmass were particularly weak (Figure 4).

Discussion

Kernels are the most important product from I. ga-bonensis and are a good source of revenue for farm-ers (Ndoye et al. 1997; Anegbeh et al. 1997). How-ever, the fruits are a rich source of vitamins (Leakey(1999)) and are consumed fresh by children. Theyhave also been shown to have potential for wine mak-

ing. The selection of both kernel and fruit character-istics are therefore very important for the domestica-tion of I. gabonensis as an agroforestry tree.

Strong relationships were found between some ofthe traits that characterise a good fruit for eatingfresh, such as fruit mass and flesh depth, indicatingthat large fruits have more flesh, rather than biggernuts (Atangana 2000; Leakey et al. (2000)). Conse-quently, overall fruit size is evidently an importantand easily selected trait for cultivars that are targetedat markets such as fresh fruits, juices, or wine mak-ing. However, other traits such as taste and fibrosity,and even skin and flesh colour, are also importantcharacteristics of fruits being sold fresh in the mar-ket. Results of the present study indicate that whilethese quality attributes can be found in large fruits,they are not exclusively related to fruit size. In addi-tion, taste and fibrosity and taste and size are notstrongly related. Consequently any improvement pro-gramme must also take these quality traits into ac-count separately and select I. gabonensis trees thatcombine large fruit size, with sweetness and lack offibres.

Weak relationships were found between fruit massand kernel mass, indicating that big kernels are notnecessarily found in big fruits. Relationships betweenfruit mass and shell mass were also found to be weak,indicating that shell brittleness is not related to fruitsize. Consequently, as also found in a preliminarystudy (Leakey et al. (2000)), it is not possible to re-liably predict kernel characteristics from fruit charac-teristics.

The weak relationships between fruit, nut and shellmass and kernel mass indicate that kernels have to belaboriously extracted before it is possible to assesstheir value characteristics. Consequently, easily ex-tracted kernels, such at those with self-cracking nuts(Ladipo et al. 1996; Leakey et al. (2000)), would be avery desirable trait. However, in the villages surveyedin this study, self-cracking nuts were not found, al-though there was continuous variation in shell massand some shells were light and brittle (Atangana etal. 2001). Strong relationships found between shellmass and nut mass suggest that nuts with brittle/thinshells (low shell mass) will only be very rarely asso-ciated with large nuts. Thus the probability of identi-fying trees that produce large kernels in light andbrittle shells is very low. However, from currentknowledge it is not possible to rule out the possibilitythat the self-cracking trait is due to factors other thanlight shell mass, as during germination the nuts do

Figure 1. Diagrams of the flesh (mesocarp), nut (endocarp) andkernel (cotyledons) of an Irvingia gabonensis fruit in the width andthickness dimensions.

223

crack longitudinally, apparently along an ‘abscission’zone. It is thus probable that the cracking processduring germination is under the control of plantgrowth regulators, and hence potentially subject togenetic variation in very different factors independentof shell weight. Consequently, self-cracking due to

early abscission may offer greater opportunities forgenetic selection than low shell mass.

The results of the present study confirm the needfor domestication of I. gabonensis to be based on twoideotypes, one for fruit flesh and the other for kerneltraits, as suggested by Leakey et al. (2000). However,it is clear that for the fruit ideotype, consideration

Figure 2. The relationships between fruit flesh characteristics in I. gabonensis from Cameroon and Nigeria.

Table 2. Relationships between fruit traits in I. gabonensis from Cameroon and Nigeria

Traits Elig-Nkouma, Cameroon (n =

31)

Nko’ovos II, Cameroon (n = 21) Ugwuaji, Nigeria (n = 100)

Fruit mass v. flesh mass r 2 = 0.99 (P = 0.001) r 2 = 0.99 (P = 0.001) r 2 = 0.99 (P = 0.001)

Flesh depth v. fruit mass r 2 = 0.85 (P = 0.001) r 2 = 0.83 (P = 0.001) r 2 = 0.73 (P = 0.001)

Fruit length v. fruit mass r 2 = 0.74 (P = 0.001) r 2 = 0.83 (P = 0.001) r 2 = 0.78 (P = 0.001)

Fruit width v. fruit mass r 2 = 0.82 (P = 0.001) r 2 = 0.88 (P = 0.001) r 2 = 0.78 (P = 0.001

Fruit mass v. fruit length, fruit

width, flesh depth and taste

(multiple regression)

r 2 = 0.97 (P = 0.001) r 2 = 0.97 (P = 0.001) r 2 = 0.90 (P = 0.001)

Taste v. flesh mass r 2 = 0.07 (P = 0.08) r 2 = 0.07 (P = 0.12) r 2 = 0.29 (P = 0.001)

Taste v. flesh mass and fruit

mass (multiple regression)

r 2 = 0.23 (P = 0.018) r 2 = 0.26 (P = 0.08) r 2 = 0.29 (P = 0.001)

Taste v. fruit length, fruit width

and flesh depth (multiple

regression)

r 2 = 0.02 (P = 0.31) r 2 = 0.37 (P = 0.01) r 2 =0.26 (P = 0.001)

224

must be given to flesh characteristics, such as fruitsize, good taste and low fibrosity. Since these traitsmay not be combined in all trees, relatively large pop-ulations from a wider range of sites will have to bescreened to find large numbers of trees that conformto the two ideotypes. In the present study, only threetrees, all from the Nigerian village of Ugwuaji, do

conform well to the fruit ideotype (Ug 10, Ug 12 andUg 75), while in Cameroon one tree from Elig Nk-ouma (EN26) and 3 from Nko’ovos II (Nk 6, Nk28and Nk 31) were the trees that were closest to thekernel ideotype (Figure 5). Selection of the three treesthat best fit the fruit ideotype per village would givepotential gains at the village level of 1.3 – 2-fold in

Figure 3. The relationships between nut and kernel characteristics in I. gabonensis from Cameroon and Nigeria.

Table 3. Relationships between nut and kernel traits in Cameroon (Elig-Nkouma and Nko’ovos II) and Nigeria (Ugwuaji)

Elig-Nkouma (n = 31) Nko’ovos II (n = 21) Ugwuaji (n = 100)

Nut mass v. kernel mass r 2 = 0.32 (P < 0.001) r 2 = 0.65 (P < 0.001) r 2 = 0.37 (P < 0.001)

Shell mass v. kernel mass r 2 = 0.09 (P = 0.058) r 2 = 0.26 (P = 0.01) r 2 = 0.17 (P < 0.001)

Shell mass v. nut mass r 2 = 0.92 (P < 0.001) r 2 = 0.86 (P < 0.001) r 2 = 0.95 (P < 0.001)

Figure 4. The relationships between fruit and kernel characteristics in I. gabonensis from Cameroon and Nigeria.

225

fruit mass, and 1.0 to 1.5-fold in taste. The greatergains in Ugwuaji village result from the small num-ber of trees with very large fruits in this population(Table 5), which may be the consequence of farmerselection (Leakey et al. 2001). Similarly for the ker-nel ideotype, selection of the three trees with the bestfit would give potential gains in kernel mass of 1.4 –1.6-fold.

In West African cuisine, the kernels of I. gabonen-sis are used as a thickening agent in soups and stews(Ndjouenkeu et al. 1996), although most farmers andconsumers prefer Irvingia wombolu kernels. Thisfood thickening property is said to be due to muci-laginous polysaccharides in the kernels, which be-come more viscous during cooking (Ndjouenkeu et al.1996). The characteristic has been called ‘drawabil-ity’ and is an important trait for genetic selection. Allthe kernels from the present study are being analyzedfor their food thickening properties, and a later paperwill examine relationships between the morphologi-cal and food quality traits. It is hoped that with thisinformation it will be possible to identify trees that

conform to a kernel ideotype, that combines goodfood thickening properties with high kernel mass andshell brittleness.

In conclusion, it is clear that there is a very exten-sive phenotypic variation in fruit and kernel traits inI. gabonensis. It is not clear from this study to whatextent the environmental differences between the sitesplays in the expression of this phenotypic variation.Establishing provenance tests could test this. It isclear, however, that there is a great potential for se-lection of ‘elite’ trees with superiority in several traits.The methods developed in this study will allow re-searchers, in partnership with farmers, to make quan-titative assessments of the variation in wild and semi-domesticated populations, and so to select trees thatare appropriate candidates for vegetative propagationand the creation of cultivars for use in their ownneighbourhood (i.e. retaining a constant genotype xenvironment interaction). Consequently, rapidprogress could be made towards a village-based par-ticipatory domestication programme, like that cur-rently being implemented by IRAD/ICRAF in village

Table 4. Relationships between fruit and kernel traits in Cameroon (Elig-Nkouma and Nko’ovos II) and Nigeria (Ugwuaji)

Elig-Nkouma (n = 31) Nko’ovos II (n = 21) Ugwuaji (n = 100)

Kernel mass v. flesh mass r 2 = 0.23 (P = 0.004) r 2 = 0.20 (P = 0.024) r 2 = 0.39 (P < 0.001)

Nut mass v. fruit mass r 2 = 0.49 (P < 0.001) r 2 = 0.25 (P = 0.012) r 2 = 0.57 (P < 0.001)

Shell mass v. fruit mass r 2 = 0.40 (P < 0.001) r 2 = 0.17 (P = 0.034) r 2 = 0.47 (P < 0.001)

Flesh mass v. kernelmass r 2 = 0.19 (P = 0.008) r 2 = 0.14 (P = 0.053) r 2 = 0.37 (P < 0.001)

Flesh depth v. kernel mass r 2 = 0.10 (P = 0.046) r 2 = 0.02 (P = 0.24) r 2 = 0.30 (P < 0.001)

Figure 5. Fruit and kernel ideotypes for I. gabonensis, compared with the data from the best trees.

226

Tabl

e5.

Pote

ntia

lga

ins

inI.

gabo

nens

isfr

uit

mas

san

dfr

uit

tast

ere

sulti

ngfr

omth

ecr

eatio

nof

culti

vars

from

the

thre

etr

ees

clos

est

toth

efr

uit

ideo

type

per

villa

ge

Frui

tm

ass

Frui

tta

ste

Ker

nel

mas

s

Tre

esw

ith

best

fitfo

r

frui

tid

eoty

pe

per

villa

ge

Mea

nfr

uit

mas

spe

rtr

ee

(g)

Mea

nfr

uit

mas

svi

llage

(g)

Pote

ntia

lga

in

(%)

atth

evi

l-

lage

leve

l

Mea

nta

ste

scor

epe

rtr

ee

Mea

nta

ste

scor

epe

r

villa

ge

Pote

ntia

lga

in

(%)

atth

evi

l-

lage

leve

l

Tre

esw

ith

best

fitfo

r

kern

elid

eo-

type

per

villa

ge

Mea

nke

rnel

mas

spe

rtr

ee

(g)

Mea

nke

rnel

mas

svi

llage

(g)

Pote

ntia

lga

in

(%)

atth

evi

l-

lage

leve

l

Ug1

041

9.8

254

516

1U

g56

10.0

171

Ug1

225

8.0

156

516

1U

g75

9.7

166

Ug7

535

3.2

214

516

1U

g97

8.8

150

Mea

n34

416

5.0

208

53.

116

1M

ean

9.5

5.9

162

Nk6

195.

418

23.

511

7N

k32

6.9

167

Nk2

817

0.9

159

4.0

133

Nk3

16.

916

7

Nk2

914

5.7

136

2.5

83N

k25

6.5

157

Mea

n17

110

7.3

159

3.3

3.0

110

Mea

n6.

84.

116

3

EN

2413

2.1

133

2.5

71E

N22

6.3

151

EN

4413

2.6

134

4.0

114

EN

245.

713

7

EN

4012

8.9

130

3.8

107

EN

295.

212

6

Mea

n13

199

132

3.4

3.5

97M

ean

5.7

4.2

137

227

nurseries (Tchoundjeu et al. 1998). Such a pro-gramme would make efficient use of the phenotypesclosest to the suggested ideotypes, and also ensurethat, in concurrence with the Convention on Biologi-cal Diversity, villagers maintain their rights over in-digenous knowledge and the developed cultivars. Ini-tiating this programme in villages close to geneticdiversity hotspots (such as Nko’ovos II and Ugwuaji),is additionally very appropriate to the maintenance ofgenetic diversity in the population of cultivars, as I.gabonensis populations in these hotspots are also ge-netically distinct (Lowe et al. (2000)).

Acknowledgements

This publication is an output from a project partlyfunded by the United Kingdom Department for Inter-national Development (DFID) for the benefits of de-veloping countries. The authors are indebted to DFIDfor funding this project (Project No R7190 of theForestry Research Programme) and the views ex-pressed here are not necessarily those of DFID. Theauthors wish to thank Dr Kate Schreckenberg and MsCharlotte Boyd (Overseas Development Institute,London) for their help and encouragement. Drs JuliaWilson and Andy Lowe are thanked for their com-ments on a draft manuscript.

References

Anegbeh P.O., Simons A.J., Dawson I., Jaenicke H. and Austin P.1997. Household Prospects for fruit production from Irvingiagabonensis (Aubry-Lecomte ex O’ Rorke) Baill. In: Horticul-tural Production and Utilization for Food Security and Eco-nomic Recovery. Proceedings of the 15th Annual Conferenceof the Forestry Association of Nigeria, 22–26 September 1997,Ibadan, Nigeria., pp. 145–159.

Atangana A.R. 2000. Caractérisation Biophysique des fruits de Irv-ingia gabonensis A. Lecomte (Irvingiaceae) de forêt dense hu-mide de basse altitude du Cameroun. Mémoire de DESS. Uni-versité de Yaoundé I, Yaoundé, Cameroun, 44 pp.

Atangana A.R., Tchoundjeu Z., Fondoun J.-M., Asaah E., NdoumbeM. and Leakey R.R.B. 2001. Domestication of Irvingia gabon-ensis: 1. Phenotypic variation in fruit traits in 52 trees from twopopulations in the humid lowlands of Cameroon. AgroforestrySystems 0: 000.

Franzel S., Jaenicke H. and Janssen W. 1996. Choosing the RightTrees: Setting Priorities for Multipurpose Tree Improvement.Research Report, 8. ISNAR, The Hague, The Netherlands, 87pp.

Harris D.J. 1996. A revision of the Irvingiaceae in Africa. Bulletindu Jardin Botanique National de Belgique 65: 143–196.

ICRAF 1997. ICRAF Medium Term Plan 1998–2000. ICRAF,Nairobi, 73 pp.

Kornerup A. and Wanscher J.H. 1978. Methuen Handbook of Co-lour. 3rd edn. Eyre Methuen, London, 252 pp.

Ladipo D.O., Fondoun J.-M. and Ganga N. 1996. Domesticationof the bush mango (Irvingia spp.): some exploitable intraspe-cific variations in West and Central Africa. In: Leakey R.R.B.,Temu A.B., Melnyk M. and Vantomme P. (eds), Domesticationand Commercialization of Non-timber Forest Products forAgroforestry. Non-Wood Products No. 9. FAO, Rome, Italy, pp.193–205.

Leakey R.R.B. 1999. Potential for novel food products from agro-forestry trees: a review. Food Chemistry 66: 1–14.

Leakey R.R.B. 2001a. Win: Win land use strategies for Africa: 1.Building on experience with agroforests in Asia and LatinAmerica. International Forestry Review 3: 1–10.

Leakey R.R.B. 2001b. Win: Win land use strategies for Africa: 2.Capturing economic and environmental benefits with multi-strata agroforests. International Forestry Review 3: 11–18.

Leakey R.R.B. and Simons A.J. 1998. The domestication and com-mercialization of indigenous trees in agroforestry for the alle-viation of poverty. Agroforestry Systems 38: 165–176.

Leakey R.R.B., Fondoun J.-M., Atangana A. and Tchoundjeu Z.2000. Quantitative descriptors of variation in the fruits andseeds of Irvingia gabonensis. Agroforestry Systems 50: 47–58.

Leakey R.R.B., Tchoundjeu Z., Smith R.I., Munro R.C., FondounJ.-M., Kengue J. et al. 2001. Evidence for farmer-initiated do-mestication through truncated selection in Dacryodes edulisand Irvingia gabonensis from Cameroon and Nigeria? SilvaeGenetica.

Lowe A.J., Gillies A.C.M., Wilson J. and Dawson I.K. 2000. Con-servation genetics of bush mango from central/west Africa: im-plications from random amplified polymorphic DNA analysis.Molecular Ecology 9: 831–841.

Mbosso M.F.P.C. 1999. Opportunités et contraintes socio-économiques liées à la culture des arbres fruitiers locaux dansles basses terres humides du Cameroun: Dacryodes edulis etIrvingia gabonensis, Mémoire de fin d’études présenté pourl’obtention du diplôme d’Ingénieur Agronome. Université deDschang, 111 pp.

Ndjouenkeu R., Goycoolea F.M., Morris E.R. and Akingbala J.O.1996. Rheology of okra (Hibiscus esculentus L.) and dika nut(Irvingia gabonensis) polysaccharides. Carbohydrates Poly-mers 29: 263–269.

Ndoye O., Pérez M.R. and Eyebe A. 1997. The markets of non-timber forest products in the humid forest zone of Cameroon.Rural Development Forestry Network Paper 22c. ODI, London.

Okafor J.C. 1975. Varietal delimitation in Irvingia gabonensis (Irv-ingiaceae). Nigerian Journal of Forestry 4: 80–87.

Sanchez P.A. and Leakey R.R.B. 1997. Land use transformation inAfrica: three determinants for balancing food security with nat-ural resource utilization. European Journal of Agronomy 7: 15–23.

Shiembo P.N., Newton A.C. and Leakey R.R.B. 1996. Vegetativepropagation of Irvingia gabonensis, a West African fruit tree.Forest Ecology and Management 87: 185–192.

Tchoundjeu Z., Duguma B., Foudoun J.-M. and Kengue J. 1998.Strategy for the domestication of indigenous fruit trees of West

228

Africa: case of Irvingia gabonensis in southern Cameroon.Cameroon Journal of Biology and Biochemical Sciences 4: 21–28.

Ukafor V. 2001. Variability studies in fruit and tree characteristicsof Dacryodes edulis in southeastern Nigeria. MSc, Rivers StateUniversity of Science and Technology, Port Harcourt, Nigeria.

229