An Assessment of Trends in the Extent of Swiddenin Southeast Asia

Dietrich Schmidt-Vogt amp Stephen J Leisz amp Ole Mertz amp Andreas Heinimann amp

Thiha Thiha amp Peter Messerli amp Michael Epprecht amp Pham Van Cu amp Vu Kim Chi ampMartin Hardiono amp Truong M Dao

Published online 27 May 2009 Springer Science + Business Media LLC 2009

Abstract Swidden systems consisting of temporarily cul-tivated land and associated fallows often do not appear onland use maps or in statistical records This is partly due tothe fact that swidden is a diverse and dynamic land usesystem that is difficult to map and partly because of thepractice of grouping land covers associated with swiddensystems into land use or land cover categories that are notself-evidently linked to swiddening Additionally in manyparts of Southeast Asia swidden systems have changed or

are in the process of changing into other land use systemsThis paper assesses the extent of swidden on the basis ofregional and national sources for nine countries anddetermines the pattern of changes of swidden on the basisof 151 cases culled from 67 articles Findings include (1) amajority of the cases document swidden being replaced byother forms of agriculture or by other livelihood systems(2) in cases where swiddening is still practiced fallowlengths are usually but not always shorter and (3)

Hum Ecol (2009) 37269ndash280DOI 101007s10745-009-9239-0

D Schmidt-VogtSchool of Environment Resources and DevelopmentAsian Institute of TechnologyPathum Thani 12120 Thailande-mail schmidtaitacth

S J Leisz ()Department of Anthropology Colorado State UniversityFort Collins 80523 USAe-mail steveleiszcolostateedu

O MertzDepartment of Geography and GeologyUniversity of CopenhagenOslashster Voldgade 10 1350Copenhagen K Denmarke-mail omgeokudk

A Heinimann P Messerli M EpprechtSwiss National Centre of Competence in Research North SouthInstitute of Geography University of BerneHallerstrasse 10Berne 3012 Switzerland

A Heinimanne-mail andreasheinimanncdeunibech

P Messerlie-mail petermesserlicdeunibech

M Epprechte-mail michaelepprechtcdeunibech

T ThihaWalai Rukhavej Botanical Research InstituteMahasarakam UniversityMaha Sarakham Thailande-mail thihafnumsuacth

P V Cu V K ChiVietnam National UniversityHanoi Vietnam

P V Cue-mail pvchanoivnnvn

V K Chie-mail vukimchi2001yahoocom

M HardionoJl Bukit Nusa Indah No 70Ciputat 15414 Indonesiae-mail hdmartinindonetid

T M DaoCenter for Natural Resources and Environmental Studies19 Le Thanh TongHanoi Vietname-mail mr_truonghotmailcom

shortened fallow length does not necessarily indicate atrend away from swidden since it is observed that shortfallow swidden is sometimes maintained along with othermore intensive farming practices and not completelyabandoned The paper concludes that there is a surprisinglack of conclusive data on the extent of swidden inSoutheast Asia In order to remedy this methods arereviewed that may lead to more precise future assessments

Keywords Swidden cultivation

Land use and land cover change Fallow Southeast Asia

Introduction

Swidden cultivation while found throughout the uplands ofSoutheast Asia often does not show up on land use maps orin statistical records (Padoch et al 2007) While clearedfields in a swidden system may be identified on a map asagricultural land the land cover associated with the fallowphase eg herbs grasses bush bamboo and differentphases of tree cover is often not recorded on land usecover cadastral and land-use planning maps of the regionas belonging to the swidden cultivation complex A factorcomplicating the identification of these land cover types asbeing associated with swidden is that they are transitionalland covers changing from year-to-year as the vegetationmatures andor is replaced by other vegetation types in thecourse of succession Complexity and dynamics within theland cover create difficulties in accurately delineating andmapping even when digital image processing or visualinterpretation techniques are used in conjunction withsatellite imagery or air photos Land usecover classificationexercises are further complicated by the fact that swidden isoften not the sole land use in a region but occurs embeddedin a matrix of other land uses

Land covers associated with the fallow phase of aswidden field are often grouped into lsquounclassifiedrsquo lsquobar-renrsquo lsquowastelandrsquo and lsquocropping mosaicrsquo categories or intovegetation categories such as lsquowood- and shrub-landsrsquo andlsquoforest regrowthrsquo along with land covers that are notassociated with swidden Labelling land as lsquoshiftingcultivationrsquo is often meant to indicate the potential fordeveloping other forms of agriculture or forestry on an areaor to indicate a degraded forest area

For a definition of swidden cultivation please see theintroduction to this issue (Mertz et al 2009) Ourunderstanding of swidden in Southeast Asia is that it issynonymous with shifting cultivation and includes both thecultivated fieldsmdashthe lsquoswiddensrsquomdashand the fallow landsThis paper has three goals The first is to review thedifferent land use and land cover classifications used onnational and regional scale maps and in remote sensing

analyses of the region in order to identify how swidden isdealt with (or not dealt with) in these classifications Thesecond is to provide an overview of the evidence that canbe culled from these various classifications and surveys foridentifying the extent of swiddening within the region Toaccomplish this goal time series information will beexamined with the express intent of identifying trends inhow the extent of swidden is changing in the region Thethird is to examine prospective methods for identifying andquantifying swidden areas

The paper is based on data derived from the literaturearchives government reports and various completed and on-going analyses of land usecover in the region A broaderintent of the paper is to provide an assessment of the regionalstate of the art that may assist in identifying research gaps andneeds in Southeast Asia and which may be applied to otherregions of the world where similar situations exist We believethat precise or at least realistic estimates of the extent andchanges of land area under swidden which is still animportant land use in Southeast Asia are a prerequisite forreducing poverty improving livelihoods and maintainingecosystem services in the region

Swidden in Southeast AsiamdashRegional and CountryAssessments of Swidden Systems Issues and Problems

Swidden is a diverse complex and dynamic land use thatresearchers have difficulty seeing and defining much lessmeasuring and quantifying Because of this complexity orldquoillegibilityrdquo as well as other characteristics of the practiceand its practitioners government agents often relegateswidden lands to categories such as ldquowastelandaban-donedunusedrdquo or ldquoresidualmiscellaneousrdquo in reports pre-sented on provincial national and regional levels Thesame situation is found in regional land cover analysisprojects such as the FAOrsquos decadal forest inventoryassessment the European Space Agencyrsquos TREES Projectand NASArsquos Pathfinder Project Each of these projectsregularly collects remote sensing and other data that is usedto interpret land cover for Southeast Asia However Fox(2000) notes that in many of these large-scale land-useclassifications for Southeast Asia from one-third to one-half of ldquonon-forestedrdquo land may be placed into ldquounclassi-fiedrdquo or ldquootherrdquo categories Much of this land may representswidden or swidden-in-transition areas but because thedata is not interpreted with the goal of identifying swiddenareas and indeed the identification of swidden areas maynot be possible given the above noted complexity and theremote sensing techniques that these projects are using(Leisz and Rasmussen in press) the location of swiddenareas has not been extracted from the remote sensing dataavailable

270 Hum Ecol (2009) 37269ndash280

shortened fallow length does not necessarily indicate atrend away from swidden since it is observed that shortfallow swidden is sometimes maintained along with othermore intensive farming practices and not completelyabandoned The paper concludes that there is a surprisinglack of conclusive data on the extent of swidden inSoutheast Asia In order to remedy this methods arereviewed that may lead to more precise future assessments

Keywords Swidden cultivation

Land use and land cover change Fallow Southeast Asia

Introduction

Swidden cultivation while found throughout the uplands ofSoutheast Asia often does not show up on land use maps orin statistical records (Padoch et al 2007) While clearedfields in a swidden system may be identified on a map asagricultural land the land cover associated with the fallowphase eg herbs grasses bush bamboo and differentphases of tree cover is often not recorded on land usecover cadastral and land-use planning maps of the regionas belonging to the swidden cultivation complex A factorcomplicating the identification of these land cover types asbeing associated with swidden is that they are transitionalland covers changing from year-to-year as the vegetationmatures andor is replaced by other vegetation types in thecourse of succession Complexity and dynamics within theland cover create difficulties in accurately delineating andmapping even when digital image processing or visualinterpretation techniques are used in conjunction withsatellite imagery or air photos Land usecover classificationexercises are further complicated by the fact that swidden isoften not the sole land use in a region but occurs embeddedin a matrix of other land uses

Land covers associated with the fallow phase of aswidden field are often grouped into lsquounclassifiedrsquo lsquobar-renrsquo lsquowastelandrsquo and lsquocropping mosaicrsquo categories or intovegetation categories such as lsquowood- and shrub-landsrsquo andlsquoforest regrowthrsquo along with land covers that are notassociated with swidden Labelling land as lsquoshiftingcultivationrsquo is often meant to indicate the potential fordeveloping other forms of agriculture or forestry on an areaor to indicate a degraded forest area

For a definition of swidden cultivation please see theintroduction to this issue (Mertz et al 2009) Ourunderstanding of swidden in Southeast Asia is that it issynonymous with shifting cultivation and includes both thecultivated fieldsmdashthe lsquoswiddensrsquomdashand the fallow landsThis paper has three goals The first is to review thedifferent land use and land cover classifications used onnational and regional scale maps and in remote sensing

analyses of the region in order to identify how swidden isdealt with (or not dealt with) in these classifications Thesecond is to provide an overview of the evidence that canbe culled from these various classifications and surveys foridentifying the extent of swiddening within the region Toaccomplish this goal time series information will beexamined with the express intent of identifying trends inhow the extent of swidden is changing in the region Thethird is to examine prospective methods for identifying andquantifying swidden areas

The paper is based on data derived from the literaturearchives government reports and various completed and on-going analyses of land usecover in the region A broaderintent of the paper is to provide an assessment of the regionalstate of the art that may assist in identifying research gaps andneeds in Southeast Asia and which may be applied to otherregions of the world where similar situations exist We believethat precise or at least realistic estimates of the extent andchanges of land area under swidden which is still animportant land use in Southeast Asia are a prerequisite forreducing poverty improving livelihoods and maintainingecosystem services in the region

Swidden in Southeast AsiamdashRegional and CountryAssessments of Swidden Systems Issues and Problems

Swidden is a diverse complex and dynamic land use thatresearchers have difficulty seeing and defining much lessmeasuring and quantifying Because of this complexity orldquoillegibilityrdquo as well as other characteristics of the practiceand its practitioners government agents often relegateswidden lands to categories such as ldquowastelandaban-donedunusedrdquo or ldquoresidualmiscellaneousrdquo in reports pre-sented on provincial national and regional levels Thesame situation is found in regional land cover analysisprojects such as the FAOrsquos decadal forest inventoryassessment the European Space Agencyrsquos TREES Projectand NASArsquos Pathfinder Project Each of these projectsregularly collects remote sensing and other data that is usedto interpret land cover for Southeast Asia However Fox(2000) notes that in many of these large-scale land-useclassifications for Southeast Asia from one-third to one-half of ldquonon-forestedrdquo land may be placed into ldquounclassi-fiedrdquo or ldquootherrdquo categories Much of this land may representswidden or swidden-in-transition areas but because thedata is not interpreted with the goal of identifying swiddenareas and indeed the identification of swidden areas maynot be possible given the above noted complexity and theremote sensing techniques that these projects are using(Leisz and Rasmussen in press) the location of swiddenareas has not been extracted from the remote sensing dataavailable

270 Hum Ecol (2009) 37269ndash280

Quantifying land use change and mapping swiddencultivation in Southeast Asia have been tried before Amap of the distribution of swidden in the Asia-Pacificregion published in the Conservation Atlas of TropicalForests Asia and Pacific (Collins et al 1991 p 32) andbased on the regional analysis of swidden in Southeast Asiaby Spencer (1966) shows that swidden is the dominantthough not the exclusive land use on nearly 50 of the landarea of Southeast Asia A later study on land use changes inSoutheast Asia between 1880 and 1980 focused ontemporary crops permanent crops forestwoodland andldquointerruptedrdquo forest and thus gives few indications on thechanges in the extent or practice of swidden which areconcealed in the overlap between these categories (Richardsand Flint 1994) In the same book Uhlig et al (1994)estimated that in the early 1980s approximately 55 to 6million hectares were under swidden in Sarawak Sabahand Thailand

The most recent attempts at quantifying land coverchange in SE-Asia were conducted by Heinimann (2006) inhis analysis of land cover change patterns in the LowerMekong Basin (LMB) and by Heinimann et al (2007) intheir assessment of the extent and dynamics of secondaryforests in the LMB These studies rely on analysis of theMRCGTZ forest cover data sets for 19931997 for theLMB (MRCGTZ 1998) With respect to the assessment ofsecondary forests the problem in using this dataset is thatsecondary forests are not delineated as a separate aggregat-ed forest class For the purpose of Heinimannrsquos (2006) andHeinimann et alrsquos (2007) studies the categories ldquomedium tolow cover densityrdquo (FMD) and ldquoforest mosaicsrdquo (FM) areconsidered to be proxies for secondary forest They may beconsidered to be proxies for swidden fallow as well Thisassumption is however burdened with a certain margin oferror First not all secondary forests even those in theuplands are fallow forests Second some dry and decidu-ous forests in the foothills and lowlands which are notnecessarily secondary in nature can be quite open and maythus fall under the category ldquomedium to low cover densityrdquoand conversely some secondary forests can be quite denseA similar approach is followed by Mittelman (2001) whoreviews various data sets for changes in secondary forestcover in Cambodia Laos Thailand and Vietnam between1993 and 1997 The approach of Heinimann (2006) andHeinimann et al (2007) differ from that of Mittelman inthat their analysis of land cover and related changetrajectories is not limited to secondary forests but considersthe entire spectrum of land cover classes and changetrajectories However no attempt was made to assess theextent and dynamics of swidden mainly because accuratelydifferentiating wood and shrubland from a land covermosaic of cropping and other land covers is a verylaborious task when intermediate resolution Landsat TM

imagery is used An additional factor rendering thedifferentiation difficult is the tremendous seasonal variationof vegetation density ie the volume of vegetation in greenfoliage in non-forested areas under the monsoonal con-ditions of Southeast Asia As a result land covertransformation between these land cover classes eventhough characteristic of swidden systems cannot be usedas a proxy for swidden activities based on the respectiveregional land cover datasets

In the absence of a satisfactory regional assessment anattempt is made in the following paragraphs and in Table 1to summarize the information regarding the extent ofswidden on a country by country basis

In Thailand swidden cultivation is practiced mainly inthe northern uplands close to the border with Laos andMyanmar Rotational swidden cultivation can still be foundin a corridor along the ThailandndashMyanmar border fromMae Hong Son in the north to Tak in the south (Rerkasemand Rerkasem 1994) Estimates of the swidden area inThailand in the mid-1980s were about 1 million ha (Uhliget al 1994) but later data are not available The ForestryStatistics of Thailand use a classification system that isbased on floristic composition and phenology differentiat-ing types such as ldquotropical evergreen forestrdquo or ldquomixeddeciduous forestrdquo or ldquoscrub forestsrdquo which cannot belinked to swiddening Swidden fallow forests dependingon their condition and location could fall under each one ofthese categories

The proportional extent of swidden is likely to be greaterin Laos than in any other country in Southeast Asia Twoassessments are available for the early 1990s Based on theassumptions that 300000 families or 18 million people(equivalent to 40) of the population are engaged inswiddening and that each family plants about 15 ha peryear Hansen (1998) estimated that 450000 ha would beused annually for swiddening and that the total area in theswiddening cycle could be 2ndash25 million ha equal to about10 of the area of Laos Chazee (1994) based on acomparison of aerial photographs from 19811982 and19881989 estimated that 4864000 ha were used inshifting cultivation in 1989 (equal to 205 of the area ofLaos) with a decline of the average fallow period from12 years in 19811982 to 8 years in 1989 Messerli et al(2009) applied a recently developed landscape mosaicapproach which analyzes patterns in the spatial coexistenceof land cover types to the territory of the Lao PDR andidentified the extent of lsquoswidden cultivation landscapesrsquo atapproximately 6500000 ha or 282 of the area of Laos

In Cambodia swidden is mainly found in Ratanakiriprovince in the northeast of the country The Atlas ofCambodia National Environment and Poverty Maps (SWC2006) shows swidden areas on a map of land use (p 11)and on a map of terrestrial vegetation and land use patterns

Hum Ecol (2009) 37269ndash280 271

(p 37) The land use maprsquos legend indicates that the areaunder swidden cultivation is 349000 ha However it is notclear whether this figure refers to area under swiddencultivation in 2002 only or whether it includes theassociated fallow lands as well

There has been no official attempt to categorize or mapthe extent of swidden in Vietnam The land use and landcover maps do not have a category that officially includesswidden land The land use and land cover maps drafted bythe Forestry Department and published by the CartographyPublishing House in Hanoi divide forest land into fivecategories Translated from Vietnamese they are lsquonaturalforestmdashspecial usersquo lsquonatural forestmdashprotected forestrsquolsquonatural forestrsquo lsquoartificial forestrsquo and lsquohill and mountainbarren landrsquo or lsquohill and mountain land not yet in usersquo (alsosometimes translated as lsquowastelandrsquo or lsquounused landrsquo)Researchers who work in the mountainous regions ofVietnam note that this last category appears to coincidewith areas that are used for swidden (Leisz 2007) Thebarren land category covers about one fifth of the territory

of Vietnam In reality 98 of the land cover in barren areasconsists of secondary formations such as grasslandgrassland and bushes scrubland with sparse trees bambooand dense growth of small and medium size trees (dbh=5to 15 cm height up to 4 m) which can be associated withthe fallow stages of the swiddening cycle (Nicolic et al2008)

In Myanmar swidden agriculture is mainly practiced byethnic minorities in the Karen Shan Kachin and ChinStates of eastern and northern Myanmar The availablefigures for Myanmar are extremely variablemdasheven thoseprovided by the same government agencymdashand thusexemplary of the problem faced by anyone trying to assessthe extent of swidden on the basis of official records Uhliget al (1994) estimated that 256 million ha were undercombinations of swidden and permanent agriculture inMyanmar in the late 1980s The Ministry of Forestryestimates that 15 to 2 million families practice swidden inan area of about 243 million ha (Ministry of Forestry1995) Figures published by the Ministry of Forestry in

Table 1 Regional and national estimates of land under swidden in Southeast Asia

Country Estimate Source of data Reference

Thailand 10 million hectares Based on FAOUNEP study datafrom early 1980s

Uhlig et al (1994)

Laos 2ndash25 million hectares Assessment based on census data Hansen (1998)

48 million hectares 19811982ndash19881989 aerialphotographs of the ForestInventory and Management Office

Chazee (1994)

65 million hectares 2002 SPOT satellite images Messerli et al (2009)

Cambodia 349000 ha JICA dataset 2002 SCW (2006)

Vietnam 1 fifth of total land area General Department of LandAdministration 2001

Nicolic et al (2008)

Yunnan No estimates

Myanmar 243 million hectares Estimate based on forest inventorydata from the early 1990s

Ministry of Forestry (1995)

532 million hectares Analysis of Satellite imagery 1995ndash2000

Forest DepartmentMyanmar 2000 (cited inHlaing 2004)

10 18 million hectares Estimate based on forest inventorydata from the early 2000s

Forest DepartmentMyanmar 2003 (cited inHtun 2007)

029 million hectares Unpublished data circulated withinthe Department

Planning and StatisticsDivision ForestDepartment Myanmar2008

Malaysia Sarawak 23 million hectares and Sabah 14 millionhectares in 1980

Malaysian Department of Statistics Uhlig et al (1994)

Indonesia Sumatra 3428600 ha Kalimantan 5457400 ha Sulawesi526900 ha Maluku 216400 ha Papua 1287300 haNTB 63 600 ha NTT 323 500 ha E- amp W-Timor98600 ha Total (including Timor E and W) 11402300ha

RePPProT data 1982 Weinstock (1990)

Philippines No estimates

272 Hum Ecol (2009) 37269ndash280

2003 indicate that swidden is practiced on an area of 1018million ha which is about 15 of the total area of thecountry (Htun 2007) The most recent statement from thesame government agency reveals that the total area of ldquoYardquo(the Myanmar term for ldquohill agriculturerdquo which can beequated with swidden) within Reserved Forests andProtected Forests is only 029 million ha This figurecontrasts starkly with other available estimates mainlybecause it does not include the areas of hill agricultureoutside protected forests Hlaing (2004) cited the resultspublished by the Forest Department in 2000 that are basedon the analysis of satellite imagery (1995ndash2000) whichestimates the total area under swidden (including fallowland) at 532 million ha

Yunnan is considered to be a part of Southeast Asiamainly on account of Xishuangbanna Prefecture in southernYunnan which resembles neighboring regions of LaosMyanmar and Thailand with respect to ethnic affiliation anddominant land use practices Ethnic minorities such asHani Lahu and others have practiced swiddening exten-sively in the past Swiddening considered an unwantedland use came under pressure due to successive policies ofcollectivization and economic development (Sturgeon2005a b) and has been largely replaced by other land usesmost notably by rubber plantations (Xu 2006) There ishowever no reliable estimate of the past or present extentof swidden

The area under swidden in Peninsular Malaysia isnegligible as most of the agricultural lands have beenconverted to permanent agriculture or plantations Relativelylarge areas are still found in Sabah and Sarawak but nopublished estimates of swidden areas are available since thoseof Uhlig et al (1994) (see Table 1) This may be partly dueto the fact thatmdashespecially in Sarawakmdashswidden lands arecurrently under pressure for conversion to other land usessuch as oil palm plantations Hence there is little interestfrom either the government or local people in publishingland information that may limit the possibilities in negotia-tions regarding which areas are classified as customary landor classified as state land (see Fox et al 2009 and Ngidang2002 for further discussion on land issues in Sarawak)

In Indonesia where swiddening is mainly practiced inKalimantan Sumatra Sulawesi and Papua there is a similarpressure to convert swiddens to plantations Weinstock (1990)gives the following assessment of the area under swiddencultivation in hectares based on REPPProt data which heconsiders to be the best Sumatra 3428600 Kalimantan5457400 Sulawesi 526900 Maluku 216400 Irian Jaya(currently Papua and West Papua) 1287300 Nusa TenggaraBarat 63600 Nusa Tenggara Timur 323500 E- and W-Timor 98600 The total area under swidden cultivation inIndonesia in the 1980s was estimated at 11402300 haContemporary estimates have not been found

In the Philippines swidden is practiced in many parts ofthe country especially in the uplands There are howeverno reliable estimates of the extent of swidden areas

Trends in Swidden System Changes Analysis of 67Articles Focusing on Swidden Systems in Southeast Asia

A review of 67 articles describing research on swiddenacross Southeast Asia (Table 2) reveals the followingpatterns of change in the systems shorter average fallowlengths and the replacement of swidden systems withplantation agriculture or tree crops It should be noted thatthe articles used varying methodologies and were notconsistent in how many cases were examined in eacharticle A total of 151 cases across Southeast Asia aredocumented in the 67 articles and in some articles differentcases are documented that show contradicting changes tothe systems discussed

Changing Fallow Lengths

A total of 37 articles document 55 cases of active swiddensystems across the region Of these 40 cases report that theaverage length of the fallow cycle has gotten shorter while15 report the fallow cycle remaining at a constant lengthOverall more than half of these articles document decreasesof fallow length within the system The cases are spreadacross the region with nine articles documenting cases ofdecreasing fallow lengths in Vietnam seven in Laos eightin Thailand five in Malaysia and four in China Sevenarticles documented cases of decreasing fallow lengths inIndonesia but five articles also documented cases of nochange in the average fallow length of the system in otherparts of Indonesia There were no articles found document-ing changes in the fallow lengths in Cambodia Myanmarand the Philippines In Vietnam and Cambodia two articlesdiscussed cases of the replacement of swidden systems withmanaged forest These may be extreme cases where thefallow lands are being abandoned and replaced by forests

Replacement of Swidden Systems

A total of 46 articles reviewed 91 cases of swidden systemsbeing replaced with other forms of agriculture or other typesof livelihood systems Of these cases 52 document swiddenbeing replaced by tree crops or tree-related enterprises

amp 17 document replacement by rubber treesamp 14 document swidden replaced by fruit tree cultivation

(orchards)amp eight document replacement by oil palmamp six document replacement by plantation timber

Hum Ecol (2009) 37269ndash280 273

Tab

le2

Overview

ofarticlesdo

cumentin

gchange

insw

iddensystem

sby

coun

tryandactiv

ityacross

Sou

theastAsia

Swiddento

Vietnam

Laos

Cam

bodia

Myanm

arThailand

Yun

nan

China

Indo

nesia

Philip

pines

Malaysia

Un-managed

forest

4344

5764

Swidden

sameaverage

fallo

wleng

th32

457

64

3439

40

5964

66

1617181920

Swidden

shorterfallo

wleng

th29

30

31

32

43

57

60

63

64

23

67

857

58

3435

38

39

40

57

64

67

5759

64

66

1617

18

19

22

33

55

454647

4849

Coffee(1)

43

Permanentvegetables

(6)

73436383941

Oilpalm

(6)

5764

2233

4650

51

52

Rub

ber(9)

1110

42

5761

62

64

2557

59

64

66

2122232728

Fruittrees(12)

4361

62

3436

41

1617

18

19

22

7

3355

Plantationtim

ber(inclTeak

ecalup

tus

(varsp)pine

(varsp)etc)(6)

4344

16

724

Permanentfieldcrop

s(inclCotton

cornm

aizefodd

ercrop

speanutsgrou

ndnu

tslegu

mesetc)(9)

3032

43

59

3538395764

12

Paddy

saw

ah(6)

3032

2227

33

13

Sug

arcane

(1)

1

Tea

(2)

4326

Forestfallow

cattle(1)

32

Protected

areas(6)

4435

36

5765

5514

53

Pepper(3)

5554555

Log

ging

(2)

5656

Opium

popp

y(opium

replacem

entprog

rams)

(2)

3441

Terracedagricultu

re(2)

3438

Flower

farm

sgreenh

ouse

agricultu

re(2)

3536

1mdashRod

eret

al(199

5)2mdash

Pravo

ngviengk

ham

(199

8)3mdash

Rod

er(199

7)4mdash

Ducou

rtieux

etal(200

6)5mdash

LaffortandDufurmier(200

6)6mdash

Lestrelin

etal(200

5)7mdash

Lestrelin

andGiordano

(200

7)8mdash

deRou

wandGray(200

3)9mdash

Hansen(199

8)10

mdashSchipani(200

7)11mdash

Thiha

etal(200

7)12mdash

Dressler(200

6)13

mdashDressler(200

5)14mdashDressleret

al(200

6)15

mdashPenot

(200

4)16

mdashColferetal(19

97)17

mdashColferandDud

ley(199

3)18mdashColfer(200

8)19mdash

DennisandColfer(200

6)20

mdashDennisandColfer(200

1)21mdashColfer(199

1)22mdash

Potter(200

4)23

mdashColchesteret

al(200

6)24mdash

Sturgeon(200

5ab)25

mdashSturgeon(200

6)26

mdashSturgeon(200

5ab)27mdashPeluso(199

6)28

mdashLaw

renceet

al(199

8)29

mdashFox

etal(200

0)30mdash

Leisz

etal

(200

5)31mdash

Jako

bsen

etal(200

7)32mdashLeisz

etal(200

7)33mdash

PotterandBadock(200

7)34

mdashHurni

(198

2)35mdash(Ziegler

etal20

04)36mdash

Crook

er(200

5)37

mdashChaplot

etal(200

5)38

mdashTurkelboo

m(199

9)39mdash

Ruank

eaw

(200

4)40mdash

Schmidt-Vog

t(199

9)41mdashJones(200

2)42mdash

Ziegler

(personalcommun

ication)43mdash

Vu(200

7)44mdash

Zingerliet

al(200

2)45mdashMertz

and

Christensen

(199

7)46

mdashHansenandMertz

(200

6)47mdash

Cramb(200

7)48

mdashMertz

etal(199

9)49

mdashLim

andDou

glas

(199

8)50

mdashNielsen

etal(200

6)51mdash

Majid

Coo

ke(200

2)52

mdashNgidang

(200

2)53mdash

Horow

itz(199

8)54

mdashCramb(199

3)55mdash

WadleyandMertz

(200

5)56mdash

Pedersenet

al(200

6)57mdash

Fox

andVog

ler(200

5)58

mdashTho

ngmanivon

get

al(200

5)59

mdashXu

etal(200

5)60mdash

Fox

etal(200

3)61mdashFox

etal(200

8)62

mdashFox

(200

2)63

mdashFox

etal(200

3)64mdash

Fox

etal(200

1)65mdash

Lon

get

al(199

9)66mdash

Xuet

al(199

9)67mdash

Fox

etal(199

5)

274 Hum Ecol (2009) 37269ndash280

amp three document replacement by pepper cultivationamp two document replacement by teaamp two document replacement by logging

The next greatest number of articles 29 document swiddensystems being replaced by some form of permanent annualagriculture Eleven of these discuss cases of the replacement ofswidden systems by permanent cultivation of field crops suchas maize cotton peanuts and fodder crops eight detailreplacement by either paddy or terraced agriculture six giveexamples of swidden replaced by permanent cultivation ofvegetables two provide examples of opium replacementprograms replacing swidden and two discuss cases of flowerraising in greenhouses replacing swidden These cases arespread across the region though the greenhouse and opiumreplacement programs are specific to Thailand

The last major category of land use change replacingswidden systems is the replacement of these systems withprotected areas Eight articles discuss cases of this type andthese are spread amongst the countries of the region

Overall the review of articles on swidden case studiesdoes not show a definitive trend away from swiddeningeither for the region or for any of the individual countries inthe region as evidence of swidden systems persisting isfound in more than half of the articles (37 of 67) and in36 of the cases in the articles and in 10 of the cases thefallow period is stable and not decreasing in lengthHowever a majority of the cases discussed (60) docu-ment swidden systems being replaced by other forms ofagriculture or by other livelihood systems and the majorityof the cases where swiddening is still practiced note that thefallow length within the system is shortening The lattermay also be interpreted as a trend away from swiddenhowever in some cases it is observed that short fallowswidden is maintained along with other more intensivefarming practices and thus not completely abandoned

Prospective Methods for Identifying and QuantifyingSwidden Systems on a Country and Regional Basis

Four prospective approaches for identifying and quantifyingthe extent of swidden systems at the country and regionalscale are discussed These approaches range from making useof currently available government and regional land covermaps to using medium and small scale satellite imagery suchas Landsat SPOT and MODIS data to using governmentcensus data combined with maps or satellite imagery

Identifying Swidden Areas on Current Maps

Several researchers have noted that current governmentland use and land cover maps while not explicitly

identifying swidden areas include classes of land use orof land cover that can act as proxies for areas that aredominated by swidden systems In Vietnam the governmentdivides upland land cover into agriculture forest and othercategories ranging from urban land to mining and nationaldefense lands The forest land cover is divided into naturalforest protected forest specialized forest artificial forestand hill and mountain land not yet in use (CartographyPublishing House 2002) The last category covers thelargest area of the uplands and is also the area where mostof the upland villages are located (Fig 1)

Through comparing this area with known swidden areasfor different regions of the country it appears that thiscategory can act as a first proxy for estimating the extent ofswidden in the uplands of Vietnam

Time Series Analysis of Satellite Imagery to IdentifySwidden Areas

After years of having been collected and archived mediumscale satellite imagery (eg Landsat and SPOT) is nowavailable for many parts of the tropics in year-on-year timeseries for upwards of 10 years in some cases Increasinglythe utility of these data for analyzing land cover and landcover change is being considered Budreski et al (2007)use a 14 year time-series of Landsat images for the Amazonregion to produce yearly maps showing cleared re-vegetated and primary forest areas They show how theyearly analysis aids in distinguishing re-vegetated fromprimary forest areas Leisz and Rasmussen (in press)demonstrate how to make land cover maps that accuratelydelineate stages of regrowth on a field after it has beencultivated by using year-on-year medium-scale satelliteimagery in conjunction with a rule based classifier thatuses rules derived from observed vegetation succession fora region The stages of regrowth correspond to stages offallow vegetation in a swidden system Other studies(Heinimann 2006) use a robust dichotomous classificationapproach to land cover change Multitemporal Landsatimagery is classified into regrowth and clearing based onbiomass indices and using object and rule based classifica-tion systems At a certain scale of observation the resultingspatio-temporal pattern of clearingregrowth can be seen asan indicator for shifting cultivation (Fig 2)

Detection and Interpretation of Landscape Patternand Mosaics

The next step in using medium-scale satellite imagery toaccurately map swidden areas is to group land covers thatbelong to the cultivated phase of a swidden system and landcovers that belong to the fallow phase of the system under asingle classification label on a land use map The main

Hum Ecol (2009) 37269ndash280 275

amp three document replacement by pepper cultivationamp two document replacement by teaamp two document replacement by logging

The next greatest number of articles 29 document swiddensystems being replaced by some form of permanent annualagriculture Eleven of these discuss cases of the replacement ofswidden systems by permanent cultivation of field crops suchas maize cotton peanuts and fodder crops eight detailreplacement by either paddy or terraced agriculture six giveexamples of swidden replaced by permanent cultivation ofvegetables two provide examples of opium replacementprograms replacing swidden and two discuss cases of flowerraising in greenhouses replacing swidden These cases arespread across the region though the greenhouse and opiumreplacement programs are specific to Thailand

The last major category of land use change replacingswidden systems is the replacement of these systems withprotected areas Eight articles discuss cases of this type andthese are spread amongst the countries of the region

Overall the review of articles on swidden case studiesdoes not show a definitive trend away from swiddeningeither for the region or for any of the individual countries inthe region as evidence of swidden systems persisting isfound in more than half of the articles (37 of 67) and in36 of the cases in the articles and in 10 of the cases thefallow period is stable and not decreasing in lengthHowever a majority of the cases discussed (60) docu-ment swidden systems being replaced by other forms ofagriculture or by other livelihood systems and the majorityof the cases where swiddening is still practiced note that thefallow length within the system is shortening The lattermay also be interpreted as a trend away from swiddenhowever in some cases it is observed that short fallowswidden is maintained along with other more intensivefarming practices and thus not completely abandoned

Prospective Methods for Identifying and QuantifyingSwidden Systems on a Country and Regional Basis

Four prospective approaches for identifying and quantifyingthe extent of swidden systems at the country and regionalscale are discussed These approaches range from making useof currently available government and regional land covermaps to using medium and small scale satellite imagery suchas Landsat SPOT and MODIS data to using governmentcensus data combined with maps or satellite imagery

Identifying Swidden Areas on Current Maps

Several researchers have noted that current governmentland use and land cover maps while not explicitly

identifying swidden areas include classes of land use orof land cover that can act as proxies for areas that aredominated by swidden systems In Vietnam the governmentdivides upland land cover into agriculture forest and othercategories ranging from urban land to mining and nationaldefense lands The forest land cover is divided into naturalforest protected forest specialized forest artificial forestand hill and mountain land not yet in use (CartographyPublishing House 2002) The last category covers thelargest area of the uplands and is also the area where mostof the upland villages are located (Fig 1)

Through comparing this area with known swidden areasfor different regions of the country it appears that thiscategory can act as a first proxy for estimating the extent ofswidden in the uplands of Vietnam

Time Series Analysis of Satellite Imagery to IdentifySwidden Areas

After years of having been collected and archived mediumscale satellite imagery (eg Landsat and SPOT) is nowavailable for many parts of the tropics in year-on-year timeseries for upwards of 10 years in some cases Increasinglythe utility of these data for analyzing land cover and landcover change is being considered Budreski et al (2007)use a 14 year time-series of Landsat images for the Amazonregion to produce yearly maps showing cleared re-vegetated and primary forest areas They show how theyearly analysis aids in distinguishing re-vegetated fromprimary forest areas Leisz and Rasmussen (in press)demonstrate how to make land cover maps that accuratelydelineate stages of regrowth on a field after it has beencultivated by using year-on-year medium-scale satelliteimagery in conjunction with a rule based classifier thatuses rules derived from observed vegetation succession fora region The stages of regrowth correspond to stages offallow vegetation in a swidden system Other studies(Heinimann 2006) use a robust dichotomous classificationapproach to land cover change Multitemporal Landsatimagery is classified into regrowth and clearing based onbiomass indices and using object and rule based classifica-tion systems At a certain scale of observation the resultingspatio-temporal pattern of clearingregrowth can be seen asan indicator for shifting cultivation (Fig 2)

Detection and Interpretation of Landscape Patternand Mosaics

The next step in using medium-scale satellite imagery toaccurately map swidden areas is to group land covers thatbelong to the cultivated phase of a swidden system and landcovers that belong to the fallow phase of the system under asingle classification label on a land use map The main

Hum Ecol (2009) 37269ndash280 275

challenge to doing this is to link spatiotemporal patterns toprocesses In the case of the above mentioned dichotomousclassification approach the geometric shapes of land coverchanges and their spatial and temporal arrangement in thelandscape (ie patterns)need to be linked to prevailingfarming systems or even categories of proximate causes

driving these changes The issue of land cover patternrecognition is not a new topic especially in the realm oflandscape ecology (eg ONeill et al 1988 Turner et al1989 Turner 1990 Wickham and Norton 1994) Whenlooking at images such as Fig 2 eyes and brain of theobserver have the capability to automatically segment the

Fig 2 Spatiotemporal patterns indicative of shifting cultivation systems

Fig 1 Example of upland land cover and village locations in Vietnams Nghe An Province

276 Hum Ecol (2009) 37269ndash280

image at an appropriate scale and detect patterns whichbased on pre-existing knowledge can be related to process-es Some studies have used this approach to delineatefarming system areas for parts of Southeast Asia (Leisz etal 2005) The main question however is whether it ispossible to replicate this procedure in an automatedquantitative way The answer to date is that no operationalmethodological framework has been developed in landscience (Heinimann 2006 Gustafson 1998) The complexityof quantifying such patterns is to a large extent related toissues of scale resolution (spatial and temporal) and extent(ONeill et al 1996 Heinimann 2006 Riitters et al 1997Southworth et al 2002 Turner and Meisel 1998 Wu 1999)

A novel approach for detecting and quantifying swiddenfarming systems for large areas is currently being developedby a team lead by the University of Bern Switzerland and isbeing applied to Lao PDR This lsquoland cover mosaicrsquo approachanalyzes the spatial juxtaposition of different land cover typesusing a spatially weighted moving windows methodology Inthe case of Lao PDR the existing national land cover dataset of2002 (based on SPOT imagery) is processed and mosaics ofdifferent land cover types (eg shrub grass with differentforest types and patches of cultivated areas) are identified atdifferent scales Messerli et al (2009) demonstrate that byapplying this new methodology to processed land cover datait is possible to detect and spatially delineate swidden areasin the uplands of Lao PDR

Integrating Census Information Into LandUse Classifications

Most current population censuses classify people by theiroccupation People both in the lowlands and uplands areclassified as lsquofarmersrsquo when their activities are tied toagricultural practices It is hypothesized here that swiddenfarming systems could be delineated by analyzing currentland cover maps in conjunction with population mapsproviding information on the percentage of farmers per unitof area Swidden areas could be estimated for uplandregions based on an analysis of the extent of land classifiedas lsquoagricultural landrsquo the extent of land classified asdifferent types of lsquoforest coverrsquo and the number of farmersTo lsquosharpenrsquo the identification of swiddeners based onhousehold characteristics agricultural censuses can betapped Integrating such analysis with either the land covermosaic analysis or the year-on-year time series satelliteanalysis could provide an even more robust analysis of theextent of swidden at a country or regional scale

The four approaches presented show that the quantifica-tion of swidden on a wider spatial scale necessitatesinnovative methods that allow combining environmentalas well as socio-economic data at different scales Thesuccess of such endeavors not only depends on novel

approaches but also on the availability of accurate dataparticularly when it comes to national or regional studiesThe development of prospective methods should thereforebe driven by the type and scale of data that is beinggenerated by other national or regional initiatives In thisregard the recent selection of Vietnam and Lao PDR toparticipate in the World Bankrsquos forest carbon fund isparticularly noteworthy Such countries are given thedifficult task of trying to determine the present state oftheir forests in order to measure future deforestation ratesLinking the quantification of swidden systems to suchinitiatives would not only offer methodological andfinancial synergies but also bears the potential of revealingecosystem services provided by swidden systems that haveso far gone unnoticed and hence unrewarded The con-sequences of trends in swidden cultivation on carbonstorage are discussed in Bruun et al (2009)

Conclusion and Research Needs

We have shown that there is a surprising lack of conclusivedata on the extent of swidden systems in Southeast AsiaSuch data are essential if development in rural areascurrently and formerly dominated by swidden is to achievethe international goals of reducing poverty strengtheninglocal livelihoods and maintaining ecosystem servicesWithout an adequate understanding of the extent of thearea under swidden and of the current trends of swiddendevelopment organizations will be working in the dark andcould propose solutions that are not viable

A wide range of remote sensing based techniques areavailable for assessment of local level change in land coverand potentially as described above in swidden land usebut the complexity and dynamic nature of swidden makesassessments on a wider scale more complicated Combiningsuch tools with demographic data ethnographic studiesand spatial information databases will make it possible toobtain a better picture of the current area under swidden aswell as the number of people depending fully or partially onthis system for their livelihoods A further need is toimprove the knowledge awareness and commitment tounderstanding land cover and land use complexity thatgovernment agents and scholars bring to the task ofcategorizing land uses found in the uplands of SoutheastAsia With a concerted effort we can document the range ofareas where swidden is found and the changes that aretaking place on these landscapes and their profoundenvironmental and social implications

Acknowledgements We would like to thank the Ford FoundationVietnam which provided funding for a workshop in Hanoi in March2008 where the data in this paper were discussed

Hum Ecol (2009) 37269ndash280 277

References

Budreski K A Wynne R H Browder J O and Campbell J B(2007) Comparison of Segment and Pixel-Based Non-parametricLand Cover Classification in the Brazilian Amazon UsingMultitemporal LANDSAT TMETM+ Imagery PhotogrammetricEngineering and Remote Sensing 73 7813ndash827

Bruun T B de Neergaard A Lawrence D Ziegler A (2009)Environmental Consequences of the Demise in Swidden Agri-culture in Southeast Asia Carbon Storage and Soil QualityHuman Ecology this issue

Cartography Publishing House (2002) Nghe An Province Land UseCover Map Cartography Publishing House Hanoi Vietnam

Chaplot V Coadou le Brozec E Silvera N and Valentin C(2005) Spatial and Temporal Assessment of Linear Erosion inCatchments Under Sloping Lands of Northern Laos Catena 63167ndash18

Chazee L (1994) Shifting Cultivation Practices in Laos PresentSystems and Their Future Vientiane Laos

Colchester M et al (2006) Tanh yan Diankan Hinyah Saint danPembebasan Tanah si Indoneais Implikasi terbasap Hasyarahatdokal san Hasyarakat Adat Forest People Programme Perkum-pulan Sawit Watch HUMA and the World Agroforestry CentreBogor Indonesia

Colfer C J P (1991) Toward Sustainable Agriculture in the HumidTropics Building on the TropSoils Experiences in IndonesiaNorth Carolina State University Raleigh North Carolina

Colfer C J P (2008) Longhouse of the Tarsier Changing Land-scapes Gender and Well-being in Borneo Borneo ResearchCouncil Phillips Maine

Colfer C J P and Dudley R G (1993) Shifting Cultivators ofIndonesia Managers or Marauders of the Forest FAO Rome

Colfer C J P Peluso N and Chin S C (1997) Beyond Slash andBurn Building on Indigenous Management of Borneos TropicalRain Forests New York Botanical Gardens New York

Collins N M Sayer J A Whitmore T C (eds) (1991) TheConservation Atlas of Tropical Forests Asia and the PacificMacMillan London

Cramb R A (1993) Shifting Cultivation and Sustainable Agriculturein East Malaysia A Longitudinal Case Study AgriculturalSystems 42 209ndash226 doi1010160308-521X(93)90055-7

Cramb R A (2007) Land and Longhouse Agrarian Transformationin the Uplands of Sarawak NIAS Copenhagen

Crooker R A (2005) Life After Opium Mountain Research andDevelopment 25 3289ndash292 doi1016590276-4741(2005)025[0289LAOITH]20CO2

de Rouw K K and Gray I (2003) Four Farming Systems AComparative Test for Erosion Weeds and Labour Input in LuangPrabang Region Juth Pakai 1 14ndash22

Dennis R and Colfer C J P (2001) Forest cover change analysis as aproxy sustainability assessment using remote sensing and GIS inWest Kalimantan Indonesia In Colfer C J P and Byron Y(eds) People Managing Forests RFFCIFOR Washington DCpp 262ndash287

Dennis R and Colfer C J P (2006) Impacts of Land Use and Fireon the Loss and Degradation of Lowland Forest in 1983ndash2000 inEast Kutai District East Kalimantan Indonesia SingaporeJournal of Tropical Geography 27 30ndash48 doi101111j1467-9493200600238x

Dressler W (2005) Disentangling Tagbanva Lifeways Swidden andConservation on Palawan Human Ecology Review 12 121ndash29

Dressler W (2006) Co-opting Conservation Migrant ResourceControl and Access to National Park Management in thePhilippine Uplands Development and Change 37 2401ndash426doi101111j0012-155X200600483x

Dressler W Kull A K and Meredith T C (2006) The Politics ofDecentralizing National Parks Management in the PhilippinesPolitical Geography 25 7789ndash816 doi101016jpolgeo200607005

Ducourtieux O Visonavong P and Rossard J (2006) IntroducingCash Crops in Shifting Cultivation RegionsmdashThe Experiencewith Cardamon in Laos Agroforestry Systems 66 65ndash76doi101007s10457-005-6645-1

Fox J (2000) How Blaming Slash and Burn Farmers Is DeforestingMainland Southeast Asia Asia Pacific Issues 47 1ndash8

Fox J (2002) Understanding a dynamic landscape land use landcover and resources tenure in Northeastern Cambodia In WalshS and Crews-Meyer K (eds) Linking People Place and PolicyA GIScience Approach Kluwer Academic Boston pp 113ndash130

Fox J and Vogler J (2005) Land Use and Land Cover Change inMontane Mainland Southeast Asia Environmental Management36 3394ndash403 doi101007s00267-003-0288-7

Fox J Krummel J Ekasingh M Yarnasarn S and Podger N(1995) Land-Use and Landscape Dynamics in Northern Thai-land Assessing Change in Three Upland Watersheds Since 1954Ambio 24 6328ndash334

Fox J Truong D M Rambo A T Tuyen N P Cuc L T andLeisz S (2000) Shifting Cultivation A New Paradigm forManaging Tropical Forests BioScience 50 6521ndash528doi1016410006-3568(2000)050[0521SCANOP]20CO2

Fox J Leisz S Truong D M Rambo A T Tuyen N P andCuc L T (2001) Shifting cultivation without deforestation acase study in the mountains of northwestern Vietnam InMillington A C Walsh S J and Osborne P E (eds)Applications of GIS and Remote Sensing in Biogeography andEcology Kluwer Academic Boston pp 289ndash308

Fox J Rambo A T Donovan D Cuc L T Giambelluca TZiegler A Plondke D Vien T D Leisz S and TruongD M (2003) Linking household and remotely sensed data forunderstanding forest fragmentation in northern Vietnam In FoxJ Rindfuss R Walsh S and Mishra V (eds) People and theEnvironment Approaches for Linking Household and Commu-nity Surveys to Remote Sensing and GIS Kluwer AcademicBoston pp 201ndash221

Fox J McMahon D Poffenberger M and Vogler J (2008) Landfor My Grandchildren Land Use and Land Tenure Change inRatanakiri 1989ndash2007 Community Forestry InternationalPhnom Penh Cambodia

Fox J Fujita Y Ngidang D Peluso N L Potter L SakuntaladewiN Sturgeon J Thomas D (2009) Policies Political-Economyand Swidden in Southeast Asia Human Ecology doi101007s10745-009-9240-7

Gustafson E J (1998) Quantifying Landscape Spatial Pattern WhatIs the State of the Art Ecosystems 1 2143ndash156 doi101007s100219900011

Hansen P K (1998) Shifting cultivation development in NorthernLaos In Chapman E C Bouaham B and Hansen P K (eds)Upland Farming System in the Laos PDR Problems andOpportunities for Livestock ACIAR Proceedings Series 87Canberra pp 34ndash42

Hansen T S and Mertz O (2006) Extinction or Adaptation ThreeDecades of Change in Shifting Cultivation in Sarawak MalaysiaLand Degradation and Development 17 135ndash148 doi101002ldr720

Heinimann A (2006) Patterns of Land Cover Change in the LowerMekong Basin The Relevance of Mesoscale ApproachesDoctoral dissertation Department of Geography Berne Univer-sity Berne Switzerland

Heinimann A Messerli P Schmidt-Vogt D and Wiesmann U(2007) The Dynamics of Secondary Forest Landscapes in theLower Mekong Basin A Regional-Scale Analysis Mountain

278 Hum Ecol (2009) 37269ndash280

Research and Development 27 3232ndash241 doi101659mrd0875

Hlaing M (2004) Analysis of Impacts of Taung-Ya Program onLivelihoods of Local People A Case Study in Bago YomaMyanmar Tokyo University of Agriculture and TechnologyTokyo

Horowitz L S (1998) Integrating Indigenous Resource Managementwith Wildlife Conservation A Case Study of Batang Ai NationalPark Sarawak Malaysia Human Ecology 26 371ndash403doi101023A1018752115074

Htun A A (2007) Local Communities and Environmental Conser-vation in Panglaung Watershed Area PhD Dissertation Depart-ment of Botany University of Yangon Yangon Myanmar

Hurni H (1982) Soil Erosion in Huai Thung Choa-NorthernThailand Concerns and Constraints Mountain Research andDevelopment 2 2141ndash156 doi1023073672960

Jakobsen J Rasmussen K Leisz S Folving R and Quang N V(2007) The Effects of Land Tenure Policy on Rural Livelihoodsand Food Sufficiency in the Upland Village of Que NorthCentral Vietnam Agricultural Systems 94 2309ndash319doi101016jagsy200609007

Jones S (2002) A Framework for Understanding EnvironmentalDegradation and Constraints to the Adoption of Soil Conserva-tion Measures Case Studies from Highland Tanzania andThailand World Development 30 91609ndash1620 doi101016S0305-750X(02)00056-6

Laffort J R and Dufurmier M (2006) From Slash and Burn to DiskPloughing The Land Policy and Tractors Behind Erosion andForest Pioneer Farming in Southern Xayaboury Province (Laos)Moussons 9ndash10 109ndash130

Lawrence D Pearl D R and Leighton M (1998) The Impact ofShifting Cultivation on a Rainforest Landscape in West Kali-mantan Spatial and Temporal Dynamics Landscape Ecology 13135ndash148 doi101023A1007985915187

Leisz S J (2007) Reinterpreting the Uplands of Vietnam TowardsUnderstanding and Correcting aMisreading of Its Land Cover LandUse and Resource Management System PhD DissertationGeography Department University of Copenhagen CopenhagenDenmark

Leisz S J Ha N T Yen N T B Lam N T and Vien T D(2005) Developing a Methodology for Identifying Mapping andPotentially Monitoring the Distribution of General FarmingSystem Types in Vietnamrsquos Northern Mountain Region Agricul-tural Systems 85 340ndash363 doi101016jagsy200506015

Leisz S J Rasmussen K Olesen J E Vien T D Elberling BJacobsen J and Christiansen L (2007) The Impacts of LocalFarming System Development Trajectories on Greenhouse GasEmissions in the Northern Mountains of Vietnam RegionalEnvironmental Change 7 4187ndash208 doi101007s10113-007-0037-1

Leisz S J Rasmussen M S (in press) Mapping Fallow Lands inVietnams North Central Mountains Combining Year-on-YearMultispectral Classifications with a Simple Land Cover Model toAccurately Map Previously Unclassified Areas InternationalJournal of Remote Sensing

Lestrelin G and Giordano M (2007) Upland Development PolicyLivelihood Change and Land Degradation Interactions from aLaotian Village Land Degradation and Development 18 55ndash76doi101002ldr756

Lestrelin G Giordano M and Keohavong B (2005) WhenConservation Leads to Land Degradation Lessons from BanLak Sip Laos International Watershed Management Institute(IWMI) Colombo Sri Lanka

Lim J N W and Douglas I (1998) The Impact of Cash Croppingon Shifting Cultivation in Sabah Malaysia Asia PacificViewpoint 39 315ndash326 doi1011111467-837300072

Long C L Fox J Lu X Gao L Cai Kui and Wang J (1999)State Policies Markets Land Use Practices and CommonProperty Fifty Years of Change in Yunnan China MountainResearch and Development 19 2133ndash139 doi1023073674254

Majid Cooke F (2002) Vulnerability Control and Oil Palm inSarawak Globalization and a New Era Development andChange 33 189ndash211 doi1011111467-766000247

MRC (Mekong River Commission) GTZ (Deutsche Gesellschaft fuerTechnische Zusammenarbeit) (1998) Forest Cover Data Set forthe Lower Mekong Basin MRCGTZ Forest Cover MonitoringProject Phnom Penh Cambodia

Mertz O and Christensen H (1997) Land Use and Crop Diversityin Two Iban Communities Sarawak Malaysia GeografiskTidsskrift Danish Journal of Geography 97 135ndash148

Mertz O Christensen A E Hoslashjskov P and Birch-Thomsen T(1999) Subsistence or Cash Strategies for Change in ShiftingCultivation Geografisk Tidsskrift-Danish Journal of Geography(special Issue) 133ndash142

Mertz O Padoch C Fox J Cramb R A Leisz S J Nguyen T LVien T D (2009) Swidden change in Southeast Asia under-standing causes and consequences Human Ecology this issue

Messerli P Heinimann A Epprecht M (2009) Finding homoge-neity in heterogeneitymdasha new approach to quantifying landscapemosaics developed for Lao PDR Human Ecology doi101007s10745-009-9238-1

Ministry of Forestry (1995) Myanmar Forest Policy 1995 YangonMyanmar

Mittelman A (2001) Secondary Forests in the Lower MekongSubregion An Overview of Their Extent Roles and ImportanceJournal of Tropical Forest Science 13 4671ndash690

Ngidang D (2002) Contradictions in Land Development SchemesThe Case of Joint Ventures in Sarawak Malaysia Asia PacificViewpoint 43 157ndash180 doi1011111467-837300171

Nicolic N Schultze-Kraft R Nicolic M Boecker R and Holz I(2008) Land Degradation on Barren Hills A Case Study inNortheast Vietnam Environmental Management 42 19ndash36doi101007s00267-008-9099-1

Nielsen U Mertz O and Noweg G T (2006) The Rationality ofShifting Cultivation Systems Labor Productivity RevisitedHuman Ecology 34 210ndash218 doi101007s10745-006-9014-4

ONeill R V Gardner R H and Turner M G (1988) Indices ofLandscape Pattern Landscape Ecology 1 3153ndash162doi101007BF00162741

ONeill R V Hunsaker C T Timmins S P Jackson B LRiitters K H and Wickham J D (1996) Scale Problems inReporting Landscape Pattern at the Regional Scale LandscapeEcology 11 3169ndash180 doi101007BF02447515

Padoch C Coffey K Mertz O Leisz S Fox J and Wadley R L(2007) The Demise of Swidden in Southeast Asia LocalRealities and Regional Ambiguities Geografisk Tidsskrift-Danish Journal of Geography 107 29ndash41

Pedersen M Mertz O and Hummelmose G (2006) The potentialfor co-existence between shifting cultivation and commerciallogging in Sarawak In Cooke F M (ed) State Communitiesand Forests in Contemporary Borneo Australian NationalUniversity Press Canberra pp 181ndash193

Peluso N L (1996) Fruit Trees and Family Trees in an IndonesianRainforest Ethics of Access Property Rights and EnvironmentalChange Comparative Studies in Society and History 38 3510ndash548

Penot E (2004) From shifting cultivation to sustainable complexrubber agroforestry systems (jungle rubber) in Indonesia ahistory of innovation processes In Babin D (ed) BeyondTropical Deforestation UNESCOCIRAD Paris pp 221ndash250

Potter L (2004) Can Indonesias Complex Agroforests SurviveGlobalisation and Decentralisation A Study in Sanggau DistrictWest Kalimantan 15th Biennial Conference of the Asian Studies

Hum Ecol (2009) 37269ndash280 279

Association of Australia Research School of Pacific and AsianStudies The Australian National University Canberra Australia

Potter L and Badock S (2007) Can Indonesias complex agro-forests survive globalisation and decentralisation A study inSanggau District West Kalimantan In Connel E W J (ed)Environment Development and Change in Rural Asia-PacificBetween Local and Global Routledge Pacific Rim GeographiesLondon pp 167ndash187

Pravongviengkham P (1998) Swidden-based farm economics innorthern Laos Diversity constraints and opportunities forlivestock In Chapman E C Bouahom B and Hansen P K(eds) Upland Farming Systems in the Lao PDR Problems andOpportunities for Livestock ACIAR Proceedings Series 87Canberra pp 89ndash102

Rerkasem K and Rerkasem B (1994) Shifting Cultivation inThailand Its Current Situation and Dynamics in the Context ofHighland Development IIED London

Richards J F and Flint E (1994) A century of land use change in Southand Southeast Asia In Lange O LMooney H A and Remmert H(eds) Effects of Land Use Change on Atmospheric CO2 Concen-trations Ecological Studies 101 Springer New York pp 15ndash57

Riitters K H Cain D H and Orvis K (1997) A Multi-scaleAnalysis of Landscape Statistics Landscape Ecology 12 199ndash212 doi101023A1007938619068

Roder W Keoboulapha B andMamivanh V (1995) Teak Fruit Treesand Other Perennials Used by Hill Farmers in Northern LaosAgroforestry Systems 29 47ndash60 doi101007BF00711281

Roder W (1997) Slash and Burn Systems in Transition Challenges forAgricultural Development in the Hills of Northern Laos MountainResearch and Development 17 11ndash10 doi1023073673908

Ruankeaw N (2004) Changing Land Use Patterns in NorthernThailand Effects of Agricultural Practices in Mae Chaem PhDDissertation Princeton New Jersey Princeton

Save Cambodias Wildlife (2006) The Atlas of Cambodia NationalPoverty and Environment Maps Save Cambodias WildlifePhnom Penh Cambodia

Schipani S (2007) Ecotourism as an Alternative to Upland RubberCultivation in Nam Ha National Protected Area Luang NamthaJuth Pakai 8 5ndash17

Schmidt-Vogt D (1999) Swidden Farming and Fallow Vegetation inNorthern Thailand Geoecological Research 8 Steiner VerlagStuttgart Germany

Southworth C Nagendra H and Tucker C (2002) Fragmentationof a Landscape Incorporating Landscape Metrics into SatelliteAnalysis of Land Cover Change Landscape Research 27 3253ndash269 doi10108001426390220149511

Spencer J E (1966) Shifting Cultivation in Southeastern AsiaUniversity of California Press Berkeley

Sturgeon J C (2005a) Border Landscapes The Politics of AkhaLand Use in China and Thailand University of WashingtonPress Seattle Washington

Sturgeon J C (2005b) Tea for Trees The Impact of State Policieson the Akha in Yunnan Indigenous Affairs 2 38ndash44

Sturgeon J C (2006) Report on NSF Grant Understanding DynamicResource Management Systems and Land Cover Transitions inMontane Mainland Southeast Asia East-West Center HonoluluHawaii

Thiha T Webb E L and Honda K (2007) Biophysical and PolicyDrivers of Land Use-Cover Change in a Central VietnameseDistrict Environmental Conservation 33 2164ndash172doi101017S037689290700389X

Thongmanivong S Fujita Y and Fox J (2005) Resource UseDynamics and Land Cover Change in Ang Nhai Village and

Phou Phanang National Reserve Forest Lao PDR Environmen-tal Management 36 3382ndash393 doi101007s00267-003-0291-z

Turkelboom F (1999) On-Farm Diagnosis of Steep Land Erosion inNorthern Thailand PhD-Dissertation Faculty of Agricultural andApplied Biological Sciences Leuven Katholieke UniversiteitLeuven Belgium

Turner M G (1990) Spatial and Temporal Analysis of LandscapePatterns Landscape Ecology 4 121ndash30 doi101007BF02573948

Turner M G and Meisel J (1998) Scale Detection in Real andArtificial Landscapes Using Semivariance Analysis LandscapeEcology 13 347ndash362 doi101023A1008065627847

Turner M G ONeill V Gardner R H and Milne B (1989)Effects of Changing Spatial Scale on the Analysis of LandscapePattern Landscape Ecology 3 34153ndash162 doi101007BF00131534

Uhlig J Hall C A S and Nyo T (1994) Changing patterns of shiftingcultivation in selected countries in Southeast Asia and their effect onthe global carbon cycle In Lange O L Mooney H A andRemmert H (eds) Effects of Land-Use Change on AtmosphericCO2 Concentrations Springer New York pp 145ndash200

Vu K C (2007) Land Use Change in the Suoi Muoi CatchmentVietnam Disentangling the Role of Natural and Cultural Factors PhD-Dissertation Leuven Katholieke Universiteit Leuven Belgium

Wadley R L and Mertz O (2005) Pepper in a Time of CrisisSmallholder Buffering Strategies in Sarawak Malaysia and WestKalimantan Indonesia Agricultural Systems 85 289ndash305doi101016jagsy200506012

Weinstock J A (1990) Situation and Outlook of the Forestry Sectorin Indonesia Volume 4 Social Environmental and InstitutionalAspects UtfIns065Ins Forestry Studies Technical Report No1 Directorate General of Forest Utilization Ministry of ForestryGovernment of Indonesia FAO Jakarta

Wickham J D and Norton D J (1994) Mapping and AnalyzingLandscape Patterns Landscape Ecology 9 17ndash23 doi101007BF00135075

Wu J (1999) Hierarchy and Scaling Extrapolating InformationAlong a Scaling Ladder Canadian Journal of Remote Sensing25 3367ndash380

Xu J (2006) The Political Social and Ecological Transformation ofa Landscape The Case of Rubber in Xishuanbanna ChinaMountain Research and Development 26 3254ndash262doi1016590276-4741(2006)26[254TPSAET]20CO2

Xu J Fox J Lu X Podger N Leisz S and Ai X (1999)Effects of Swidden Cultivation Population Growth and StatePolicies on Land Cover in Yunnan China Mountain Researchand Development 19 2123ndash132 doi1023073674253

Xu J Fox J Zhang P Fu Y Yang L Qian J Leisz S andVogler J (2005) Land UseLand Cover Change and FarmerVulnerability in Xishuangbanna Environmental Management 363404ndash413 doi101007s00267-003-0289-6

Ziegler A D Giambelluca T W Sutherland R A Nullet M AYarnasarn S Pinthong J Preechapanya P and Jaiarree S(2004) Toward Understanding the Cumulative Impacts of Roadsin Agricultural Watersheds of Montane Mainland Southeast AsiaAgriculture Ecosystems and Environment 104 145ndash158doi101016jagee200401012

Zingerli C Castella J -C Manh P H and Cu P V (2002)Contesting policies rural development versus biodiversityconservation in the Ba Be National Park Area In Castella JC and Quang D D (eds) Doi Moi in the Mountains LandUse Changes and Farming Systems Differentiations in Bac KanProvince Vietnam Agricultural Publishing House HanoiVietnam pp 249ndash275

280 Hum Ecol (2009) 37269ndash280