Shifting Cultivation: A New Old Pa radigm for Managing Tropical Fo

June 2000 / Vol. 50 No. 6 • BioScience 521

Roundtable

hifting cultivation, or swidden farming, is oftenh eld to be the principle driving force for defore s t a-ti on in tropical Asia (Myers 1993). Na ti onal govern-

m ents in So utheast As i a , n o t a bly in In don e s i a , the Philip-p i n e s , Th a i l a n d , and Vi etn a m , h ave been incl i n ed to bl a m es h i f ting cultiva tors , u su a lly mem bers of ethnic minori ti e s ,for rapid loss of forests (Dove 1984, Do Van Sam 1994, LeTrong Cuc 1996, Ra m bo 1996). In Vi etn a m , the official vi ewof s h i f ting cultiva ti on has been parti c u l a rly nega tive , ref l ect-ing a com bi n a ti on of the et h n ocen tric assu m pti ons of t h enu m eri c a lly dominant Kinh (lowland Vi etnamese) abo utthe cultu ral su peri ori ty of wet ri ce farming and the Ma rx i s tvi ew that swi d dening repre s ents a pri m i tive stage in the cul-tu ral evo luti on a ry sequ en ce (Ja m i e s on 1991, Ra m bo 1995).

Re s o u rce managers in these co u n tries inva ri a bly see shift-ing cultiva ti on as a singl e , simple sys tem of f a rming inwh i ch the forest or scrub is slashed and bu rn ed to makes wi d den s . These fields are cultiva ted for on ly one or two sea-s ons before soil ferti l i ty is ex h a u s ted or weed growth over-wh elms the crops (Padoch et al. 1 9 9 8 ) . The field is thena b a n don ed and the farm ers move on to clear a new fiel del s ewh ere in the fore s t . From this pers pective , s wi d denf a rm ers are “forest eaters” whose unending search for newforests to clear is a major cause of defore s t a ti on . However, tovi ew swi d dens as just tem pora ry fields su rro u n ded by aban-don ed land under wild growth is wron g. More than fo u rdec ades ago, Ha rold Conklin (1957) poi n ted out that “s h i f t-ing cultiva ti on may refer to any one of an undeterm i n ednu m ber of a gri c u l tu ral sys tem s” ( p. 1 ) . S pen cer (1966)de s c ri bed 18 disti n ct types of s h i f ting agri c u l tu re wi t h i nSo utheast Asia alon e . Broo k f i eld and Padoch (1994) arguedthat swi d den agri c u l tu re is not one sys tem but many hu n-d reds or thousands of s ys tem s .

This arti cle seeks to de s c ri be the agroeco s ys tems of ah a m l et in nort h ern Vi etnam in wh i ch trad i ti onal shifti n gc u l tiva ti on has not re su l ted in ex ten s ive defore s t a ti on , but it

has altered the ch a racter of the veget a tive cover. Th ro u gha n a lyses of this agroeco s ys tem , we seek to devel op a bet teru n derstanding of the swi d den agri c u l tu ral sys tems found inthe regi on and the ef fects of these sys tems on land-coverch a n ge in the area over the last 40 ye a rs . Our findings havei m p l i c a ti ons for understanding the role of s h i f ting cultiva-ti on at more mac ro level s , i n cluding its con tri buti on togl obal cl i m a te ch a n ge .

S h i f ting cultiva ti on , the su bj ect of this arti cl e , d i f fersf rom the com m on practi ce of using fire simply to cl e a rforests for perm a n ent cultiva ti on , p a s tu re , or furt h er devel-opm en t . The latter practi ce was em p l oyed on a massive scaledu ring the Eu ropean set t l em ent of the fron ti ers of the east-ern Un i ted States (Ra m bo 1990). This slash-and-bu rn culti-va ti on , as practi ced by the mill i ons of l owland Vi etn a m e s ere s et t l ed in upland areas since the 1960s, tu rn ed large are a si n to vi rtual lunar landscapes (Le Trong Cuc et al. 1 9 9 0 ) .O fco u rs e , s ome trad i ti onal forms of s h i f ting cultiva ti on canalso cause lon g - term envi ron m ental degrad a ti on .“ P i on eer ”s h i f ting cultiva ti on , for ex a m p l e , is a sys tem in wh i ch newa reas of forest are cl e a red for fiel d s ,a ll owed to remain underc u l tiva ti on too lon g, and then abandon ed in a degradedcon d i ti on (Ku n s t ad ter et al. 1 9 7 8 ) . In nort h ern Vi etn a m ,H m ong farm ers have converted large areas in the high l a n d sto grasslands as a re sult of p i on eer swi d den i n g.

Nevert h el e s s , in focusing on de s tru ctive forms of s l a s h -a n d - bu rn cultiva ti on , n a ti onal govern m ents and re s o u rce

Shifting Cultivation: A New Old Pa r a d i g mfor Managing Tropical Fo r e s t sBY JEFFERSON FOX , DAO MINH TRUONG, A. TERRY RAMBO, NGHIEM PHUONG TUYEN,LE TRONG CUC, AND STEPHEN LEISZ

S

Jef ferson Fox (e-mail: fox j @ ew c . h awa i i . edu) and A . Terry Rambo (e-

mail: ra m b ot @ ew c . h awa i i . edu) are senior fell ows in envi ro n m en t a l

s tu d i e s , Ea s t–West Cen ter, Ho n ol u l u , HI 96848. Dao Minh Tru o n g

(e-mail: tru o n g @ u plands.ac.vn) and Ngh i em Phuong Tuyen (e-

mail: tuyen @ u plands.ac.vn) are re se a rch ers at, and Le Trong Cuc (e-

m a i l :c u c @ u plands.ac.vn) is dire ctor of , the Cen ter for Re sou rce and

Envi ro n m ental Stu d i e s , Vi etnam Na tional Un ivers i ty, Ha n o i , 1 6 7

Bui Thi Xu a n , Ha n o i , Vi etn a m . Step h en Lei sz (e-mail:

s j lv m @ f pt.vn) is GIS/monito ring and eva l u a tion advi so r, C A R E

In tern a tional in Vi etn a m , 63 To Ngoc Va n , Ha n o i , Vi etn a m . © 2000

Am erican In s ti tu te of B i ol o gical Sci en ce s .

Fi g u re 1. The co m po s i te swi d d en sys tem in the drysea so n . Pa d dy fields are in the fo regrou n d , with re cen t lycl ea re d / bu rned swi d d en fields in the ba ck grou n dsu rrounded by se co n d a ry veget a ti o n .

Downloaded from https://academic.oup.com/bioscience/article-abstract/50/6/521/261059by gueston 30 January 2018

m a n a gers have ign ored the natu re of m ore com m on andm ore su s t a i n a ble forms of s wi d den cultiva ti on . Al corn(1990) calls swi d den farming “m a n a ged defore s t a ti on ,” as ys tem built around patchy, p u l s ed rem oval of trees but notof the fore s t . She su ggests that indigenous farm ers work tom a n a ge defore s t a ti on in sequ en tial agrofore s try sys tem sthat integra te secon d a ry su cce s s i onal veget a ti on — every-thing from grass and bushes to young open - c a n opy treecom mu n i ties to matu re cl o s ed - c a n opy tree com mu n i ti e s .Broo k f i eld and Padoch (1994) su ggest that the con cept ofa b a n don ed fall ows is being displaced by re s e a rch that showsthat the plants found in any stage of s econ d a ry su cce s s i on a lveget a ti on are in large measu re the re sult of conscious plan-n i n g. Even wh ere the forest that su cceeds farming is notcl o s ely managed , it is used . (The “f a ll ow ” is the peri od wh enland is left to rec u pera te and veget a ti on all owed to regen er-a te . It is not “a b a n don ed ” l a n d , wh i ch the swi d den farm erdoes not intend to use aga i n . Ra t h er, it is part of the land-use sys tem in wh i ch the farm er wi ll retu rn to this plot aga i n .Me a nwh i l e , while it lies fall ow, the farm er may use it to co l-l ect nu m erous food , ti m ber, and non ti m ber forest produ ct s .Popular con cepti on , h owever, is to vi ew this land as “a b a n-don ed fall ow.” )

Fa i lu re to understand the swi d den agri c u l tu ral sys tem ,and its assoc i a ted secon d a ry veget a ti on , has led scien tists toovere s ti m a te the amount of “defore s t a ti on” that hasocc u rred in So utheast As i a . Po t ter et al. (1994) su ggest thatas mu ch as 26% of a ll land in So utheast Asia falls into the“o t h er ” c a tegory, wh i ch inclu des scru b, bru s h , p a s tu re ,w a s te , and other land-use categories that of ten repre s ent notdefore s t a ti on but forest fall ow or secon d a ry regen era ti on .L i kewi s e , Ku m m er and Tu rn er (1994) su ggest that approx i-m a tely 33% of the land cover in the Philippines falls into the“o t h er ” c a tegory, and this category grew by more than20,000 ha bet ween 1948 and 1987. In Th a i l a n d , the fiel dplan of the Tropics Program of the GEWEX Asian Mon s oonEx peri m ent (GAME-Tropics) is based on a land-use bre a k-down in wh i ch fully 49% of the non fore s ted land in nort h-ern Thailand is “u n cl a s s i f i ed ”( G A M E - Tropics 1996).

Because swi d den agri c u l tu ral sys tems are so little under-s tood , m a ny govern m ents have implem en ted (mostlyu n su ccessful) large-scale re s et t l em ent programs that arei n ten ded to convert swi d den cultiva tors into farm ers of per-m a n ent agri c u l tu ral fiel d s . Moreover, f a i lu re to unders t a n dthe role played by secon d a ry su cce s s i onal veget a ti on ins wi d den sys tems has meant that re s o u rce managers havenot correct ly iden ti f i ed the impact s , both po s i tive and neg-a tive , of s wi d den agri c u l tu re on species divers i ty, w a ters h edhyd ro l ogy, and carbon sequ e s tra ti on (Skole et al. 1 9 9 8 ) , a llof wh i ch have important implicati ons for bi od ivers i ty con-s erva ti on and gl obal warm i n g.

Shifting cultivation in northernVietnamShifting cultivation has been practiced for centuries,if notmillennia, in the northwestern highlands of Vietnam. Yet,

contrary to the popular conception that shifting cultiva-tion always causes deforestation,the area under forest cov-er in many parts of this region has not changed signifi-cantly, despite rapid population growth, over the past 50years. Indeed, a recent study of the Da River watershed(Nguyen Duy Khiem and Van Der Poel 1993) found nocorrelation between the occurrence of shifting cultivationand the extent of deforestation. In those districts havingthe greatest extent of swiddening (12–36% of total area),the percentage of land under forest cover ranged from 6%to 48%, whereas in those districts having almost no shift-ing cultivation (less than 6% of total area), forest coverranged from 8% to 48%.

Tat, one of 10 hamlets in Tan Minh village (Da Bac dis-trict, Hoa Binh province) is representative of many swid-dening villages in Southeast Asia. The hamlet is positionedalongside the Muong River, a tributary of the Da River, atan altitude of approximately 360 m above sea level. TheMu ong va ll ey ra n ges from a few hu n d red meters toapproximately 2 km wide and is surrounded by peaksreaching from 800 m to over 1100 m. The valley walls areextremely steep, with slopes often exceeding 60% criss-crossed by many small streams flowing into the MuongRiver. The valley is shaped like an amphitheater, with but-tress-forming ridges extending up to the surroundingpeaks.

Our understanding of the hamlet’s ecosystem is basedon an integrated spatial database that incorporated topo-graphic maps, aerial photographs, satellite images, and adigital elevation model with information on elevation,slope, and aspect. Information on land-use practices col-lected through interviews with farmers and other keyinformants was georeferenced to the spatial database. Thisdatabase served as a framework for analyzing changes inland cover and forest patterns through time and as a toolfor analyzing the information and insights collected insemistructured informal interviews.

The spatial database was developed on the basis of aer-ial photographs (nominally 1:40,000) taken in 1952 by theFrench military, as well as a 1995 Landsat Thematic Map-per image. The aerial photographs were taken as part of acomprehensive mapping of Vietnam.1 The photographswere manually interpreted and classified into three land-cover categories—secondary regeneration, swidden, andpaddy. The secondary regeneration or successional vegeta-tion category was further subdivided into three classes:closed-canopy forest (closed-canopy forest mapped fromthe 1952 aerial photographs may or may not have beenprimary forest, but for our purposes we are classifyingthese forests as secondary vegetation); open-canopy forest;and grass, bamboo, and scrub. These photographs were

522 BioScience • June 2000 / Vol.50 No. 6

Roundtable

1We are gra teful to Dr. Je a n - Fra n cois Du pon ,s en i or scien tist of O R S TO M( O f f i ce de la rech erche scien ti f i que et tech n i que outre - m er ) , for assistancein obtaining access to these ph o togra ph s . We would also like to ack n owl-ed ge the staff of the In s ti tut Geogra ph i que Na ti onal for their prom pt andco u rteous re s ponse to our request for the ph o togra phs of Ban Ta t .


registered on a map base, and the land-cover categorieswere digitized and entered into a geographical informa-tion systems (GIS) database (Arc/Info software on a SunSparc workstation). The Landsat image was classified intothe same land-cover categories. Ground-truth points werecollected in the field using differential GPS (global posi-tioning system). The Landsat image was registered to thesame map base as the 1952 photographs. Spectral signa-tures of the different land-cover types were derived from asubset of the ground-truth points that were then overlainon the imagery. A supervised classification of the imagewas done using a maximu m - l i kel i h ood rule (ERDA SImagine software). An accuracy assessment of the 1995satellite-derived land-cover classification showed 95% ofthe 155 checkpoints correctly classified after accountingfor GPS surveying and image registration error.

The socioeconomic database was developed throughinterviews with many Tay residents of Tat hamlet andprovincial, district, village, and hamlet government offi-cials. Researchers documented changes in national andregional policies influencing land use (e.g., tenure, taxa-tion, credit, import and export regulations) as well aschanges in infrastructure (roads and markets). We inter-viewed residents of the village to learn more about thesocioeconomic factors contributing to their decision tocreate or maintain forest fragments in their area. Keyinformants were used to assess,among other factors, localpeoples’ perception of the forest. Researchers conductedsemistructured informal interviews with villagers to iden-tify the socioeconomic and institutional factors influenc-ing use and management decisions regarding forestlandand forest vegetation.

The people of Ban (hamlet) Tat are mostly members ofthe Tay ethnic minority.2 They speak a language belongingto the T’ai family. Local oral history maintains thatmigrants from Son La settled the hamlet a little over 100years ago. Mobility appears to have been relatively high,with several waves of immigration and emigration.A largegroup of people migrated from Ban Tat to Nghia Lo, justac ross the hamlet’s nort h ern bo u n d a ry in Phu Th oprovi n ce , and kinship ties are sti ll maintained tod aybetween these two hamlets.

According to one elderly informant, only seven house-holds, or ap proximately 50 people, lived in the village in1954. Today the hamlet has grown to 69 households witha population of 389 people. This represents a populationgrowth ra te of ro u gh ly 4.9% annu a lly — prob a bly theresult of a natural growth rate of 3–3.5% and in-migra-tion. During this period, population density grew fromapproximately 10 people to 75 people per km2, which isapproximately twice the average population density for

the Da River watershed.For as far back as any informant can remember, the Tay

of Tat hamlet have been “composite swiddeners” (Rambo1996), that is, households simultaneously manage perma-nent wet rice fields in the valley bottoms and shiftingswidden fields on the hillslopes, and they exploit wildresources of the forest. Swiddening as practiced by the Tayis an integral component of the total agricultural system,not an adaptation of an earlier, more primitive,pure swid-dening that is in the process of being replaced by moreadvanced irrigated farming. Neither is swiddening a recentresponse to rapid population growth that has exceeded thecarrying capacity of the wet rice fields and forced peopleto expand their farming onto the forested slopes. Instead,composite swiddeners such as the Tay have practiced bothwet rice farming and swidden agriculture together as anintegrated system of subsistence for generations and prob-ably centuries. Figure 1 is a typical landscape in Ban Tat atthe end of the dry season. Recently planted paddy fieldsare seen in the foreground, with newly cleared swiddenfields in the background surrounded by secondary vegeta-tion. Similar composite systems are found among the Shanof Burma and northern Thailand (Schmidt-Vogt 1998),the Hani of Xishuangbanna prefecture in southwesternChina (Pei Shengji 1985),and the Ifugao of the Cordillerain the Philippines (Dove 1983).

In the case of Ban Tat, elderly informants reported thattheir parents had told them that they had employed bothsystems when they first began to settle the valley at least100 years ago. At that time, the entire area was covered byclosed-canopy forest and there was no scarcity of land onwhich to make paddy fields in the valley bottoms. The areaof paddy fields was much smaller than it is now, and goodforestland was abundant and free for the taking. It wouldthus have been possible for households to have had onlycultivated paddy fields or only cleared swiddens, but noneare reported to have done so. Evidently, there are sur vivaladvantages in maintaining a more diversified agroecosys-tem.

Indeed, the household resource system of the Tay isnotable for its incorporation of a wide range of subsys-tems. A typical Tay household manages a complex agro-ecosystem. The landscape of Ban Tat is therefore a mosaicof cultivated and fallowed fields interspersed with forestareas protected by the community. Figure 2 shows thefragmented mosaic of swidden fields, secondary vegeta-tion, and older tree cover that compose the Ban Tat land-scape. There does not seem to be any regular pattern tothis mosaic. Swidden fields can be found anywhere on theslope from the bottom of the hill to the top. A village peo-ple’s committee that is responsible for allocating land andenforcing regulations regarding forest use administers thehamlet. This committee allocates land to villagers from thetwo large areas near Ban Tat where swiddening is permit-ted—Suoi Co San and Suoi Muong. In principle, villagersshould alternate their swidden fields between the two


Roundtable

2Al t h o u gh of f i c i a lly cl a s s i f i ed as Tay, the people of Ban Tat are cultu ra llyqu i te disti n ctive . Some Vi etnamese et h n o l ogists argue that they are actu a l-ly a bra n ch of the Wh i te Th a i . Th ey are more likely simply a unique loc a lpop u l a ti on , one of m a ny va riant groups that have evo lved under con d i-ti ons of rel a tive isolati on in the nort hwe s tern mountains.


areas every several years, leaving one of the areas to fallowand regen era te . In practi ce , h owever, because land isscarce, both parcels are used at the same time.

Within the allocated fields (not more than 2 ha perhousehold), the villagers manage their own patterns ofrotation between cultivated crops and fallow periods. Atclearing time, everyone in the hamlet travels together tothe designated forests to mark their fields under the obser-vation of the village people’s committee staff. The use of afield, however, establishes no long-term ownership or userights over that field,and after one cycle of cultivation it isnot necessary to return to the same field.

On plots relatively close to the settlement, when the soilis sufficiently fertile the most common current pattern ofswidden cultivation is 3 years of dryland rice and 2 yearsof cassava, followed by 3–4 years of fallow. The length ofthe fallow appears to be determined primarily by the needfor land, and hence smaller households can afford tomaintain a longer fallow. The fallow period has beendeclining rapidly in recent years, reflecting the scarcity ofland available for swiddening caused by increased popula-tion density and by government intervention to protectthe forests.

Home and tree gardens and livestock are also importantcomponents of the agricultural system. Home gardenslocated near people’s houses are used to grow vegetables,fruits, herbs, and ornamental plants for household use.More fruit trees, such as plum and apricot, have beenplanted to take advantage of recent market opportunities.Tree gardens are located upslope from the home gardensin areas used also for cassava swiddens. Trees are grown forlocal construction needs and for sale as timber and pulp-wood. Cattle and buffalo graze in harvested paddy fields,roadsides, and fallowed swiddens and secondary forestareas. Cattle also facilitate material and nutrient flowsbetween components of the agroecosystem (Rambo andLe Trong Cuc 1997). Thus, the generalized spatial pattern

of land use is settlement and wet rice fields located at thebottom of the valley, home gardens on the hillside abovethe houses, tree gardens and cassava swiddens farther upthe slope, dryland rice swiddens on the upper slopes, andsecondary forest on the crests.

Forests, which are officially managed by the forestrydepartment, can be found across the landscape as well asbeyond the home and tree gardens. The district forestrydepartment defines three types of protected forests: pri-mary forests (either primary or well-developed secondaryforests), regenerating secondary forests, and watershedforests. Under a national program, the government payshouseholds to protect regenerating secondary forest plotsassigned to their care. Forests and scrub not classified inone of these categories can be cleared for swiddens. Thefine for cutting protected forest is 800 dong (US$0.08 in1995) per square meter; however, the chances of beingcaught are relatively low and the potential gains high, sothe fine system is not an effective deterrent to forest clear-ing in the more inaccessible areas where government sur-veillance is weak.

Some protected forests are located on the top part of thehills, some are found on the lower part of the hills abovethe houses with swidden fields on the top, and some arefound on the middle of the hills with swidden fields onboth sides. Protected forests are also found in gullies run-ning down the slope. These gullies are under the control ofthe hamlet cooperative and swidden is not allowed. Insome areas,particularly on slopes immediately above theirhouses,farmers themselves preserve the forest to maintainwater supply for home consumption.

Deforestation or forest degradation?The Food and Agriculture Organization of the UnitedNations (FAO) defines deforestation as “the transfer offorest land to non-forest uses and includes all land wherethe forest cover has been stripped and the land convertedto such uses as permanent cultivation,shifting cultivation,human settlements, mining, and building of dams” (Rao1989, p. 6). Degradation, on the other hand, “refers to areduction in the extent and quality of the forest cover dueto such factors as indiscriminate logging, inappropriateroad-making methods, and forest fires” (Rao 1989, p. 6).Note that FAO defines deforestation as both a change inland cover (i.e., loss of forest cover) and a change in landuse (i.e., converted to other perm a n ent uses). Fore s tdegradation, on the other hand, is simply a change in thequality of forest cover.

Other researchers have noted that estimated rates oftropical deforestation vary for several reasons, includingambiguities surrounding the future of forests that havebeen cut down (Williams 1990, Myers 1991). If a substan-tial portion of cut forests is regenerating, the rate of defor-estation is overestimated in the calculation of the net rateof change in forested areas (Uhl et al. 1988, Turner et al.1993, Moran et al. 1994). In other words, if the cut forest

524 BioScience • June 2000 / Vol. 50 No. 6

Roundtable

Fi g u re 2. The landsc a pe of Ban Tat in the wet sea so ns h owing the fra gm en ted mosaic of s wi d d en fiel d s ,se co n d a ry veget a ti o n , and ol d er tree cover.


regenerates, the process should bec a ll ed forest degrad a ti on , n o tdeforestation.

Historical changes in land coverand fra gm en t a ti on in Ban Ta tbetween 1952 and 1995 are shownin Figure 3 and su m m a ri zed inTable 1. During this period,the areacovered by secondary regenerationor successional vegetation decreased from 92% to 84% ofthe landscape. The area covered by closed- and open-canopy forests decreased from 11% to 3% and 54% to15% of the landscape, respectively, while the area coveredby grass, bamboo, and scrub increased from 27% to 66%of the landscape.

Our finding that 84% of the landscape of Tat hamletremained under secondary vegetation—despite more thana century of shifting cultivation—shows that traditionalswiddening does not necessarily entail permanent conver-sion but only temporary use of forestland. Hence, we donot describe the secondary regeneration that occurs aftershifting cultivation as deforestation. We found only 5%deforestation (the increase in land used for paddies from1% to 6%) for the study area over the 43 years between1952 and 1995. Although the amount of land under sec-ondary regeneration remained relatively constant over the43-year period, closed- and open-canopy forests havedegraded to earlier stages of succession. What Ban Tat evi-dences, as do many swidden agricultural landscapes inSoutheast Asia, is not so much deforestation as a changefrom a fairly homogeneous forest cover (closed and opencanopy) to a highly heterogeneous cover of secondary veg-etation.

Our work suggests that the other major land-coverchange resulting from the long-term practice or traditionof shifting cultivation is a great increase in the degree offorest fra gm en t a ti on . Si gnificant ch a n ges have alsooccurred in the spatial distribution of that land cover.Between 1952 and 1995, the number of secondary regen-eration fragments grew from 18 to 292, and the mean sizedecreased from 37 ha to 2 ha (Table 1). Fox et al. (1995)found similar results in northern Thailand. The phenom-enon of forest fragmentation is well illustrated also in theliterature on South America (e.g., Nepstad et al. 1991,McClanahan and Wolfe 1993,Enoksson et al. 1995,Schel-has and Greenberg 1996).

Scientists have only recently begun to study the effectsof fragmentation on biodiversity in general, on speciescomposition of forest ecosystems, and on forest hydrolo-gy. Field surveys in Southeast Asia have shown that thediversity of species in forest fragments is often comparablewith that in more mature forests (see Xu et. al. 1993,Schmidt-Vogt 1998). Studies in Kalimantan, Indonesia(Padoch and Peters 1992, Lawrence et al. 1998), showedthat although diversity of both plants and birds is lower inswidden areas than in the primary forest, some sites have50–80% of the diversity of comparable natural forests.Large wild mammals (e.g., rhinoceros, tiger, elephant) aremore adversely affected by forest fragmentation, however.The decline in other, more adaptable mammals (e.g., wildpig, deer) at Ban Tat is probably more the result of over-hunting than it is a consequence of habitat fragmentation.

The hydrological effects of forest fragmentation dependon the size and degree of fragmentation. Discontinuity inland-surface properties across a forest edge gives rise tomicroclimatic effects extending toward the interior of aforest fragment (Collins and Pickett 1987, Murcia 1995).Furthermore, the spatial heterogeneity of land-surfacecharacteristics resulting from forest fragmentation cani n du ce ch a n ges in mesoscale circ u l a ti on , convecti on ,cloudcover, and rainfall (Yan and Anthes 1987, Giorgi1989, Henderson-Sellers et al. 1993).

Social, economic, and cultural factors play importantroles in both the creation and maintenance of forest frag-m ents (Sch elhas and Green berg 1996). Al t h o u gh thecausal associations between such factors and the creationand mainten a n ce of forest fra gm ents are not cl e a rlyunderstood, social forces appear to exist that promote themaintenance of a patchy landscape even when many ruralinhabitants have modified traditional agricultural sys-tems, increased yields, supplemented incomes with earn-ings from off-farm employment, or migrated to urbanareas in search of wage-paying jobs.

June 2000 / Vol.50 No. 6 • BioScience 525

Roundtable

Fi g u re 3. Land cover andfra gm en t a tion in Ban Ta tin 1952 and in 1995. Th e

Ban Tat land-covercl a s s i f i c a tion map is in

Un iversal Tra n s verseMerc a tor (UTM)

coo rd i n a te s .


Changes in land use and land cover in Ban Tat are atvariance with the conventional wisdom that shifting culti-vation,under conditions of increasing population density,inevitably results in extensive deforestation. Part of thevariance is attributable to how we define deforestation asopposed to forest degradation, but two other factors areequally, if not more, important. First, and perhaps mostsignificant, is the high sustainability of the compositeswiddening system employed by the Tay, which outsideobservers have not previously appreciated. Because paddyprodu cti on can be inten s i f i ed as pop u l a ti on den s i tyincreases, combining paddy fields with swiddens in a sin-gle subsistence system relieves some of the pressure toclear more fore s t . Ho u s eholds in Tat hamlet deriveapproximately half of their carbohydrate supply fromtheir paddy fields, even though these fields are muchsmaller in area than the swiddens. The second factor, alsorelated to the Tay land-management system, is the impor-tance of secondary regrowth and the Tay’s active efforts tomaintain a mosaic of land cover across the full spectrumof the various stages of forest regrowth. This complex,indigenous land-use system thus both maximizes the sta-bility of food production and the percentage of the land-scape dominated by secondary vegetation.

Our work leads us to question whether too muchemphasis has been placed on the effect swiddening hashad on land cover (changes from homogeneous forest tohighly heterogeneous cover of secondary vegetation) andtoo little emphasis on the stability of swidden agricultureas the main land-use system in this region. In other words,what would happen if deforestation—and not just forestdegrad a ti on — f i n a lly occ u rred? Perm a n ent agri c u l tu recould result in land cover dominated by trees (e.g., rubber,palm oil,cardamom, or tea) or in one composed of annu-als (e.g., maize, cassava, and upland rice). Current trendstoward tree gardens indicate that the hamlet may be ableto maintain a high percentage of tree cover. In either thetree-dominated or annual-crop scenario, however, biodi-versity, as measured by the number of species found on the

landscape, would decline (Lawrence et al. 1998). Hydro-logical impacts could be considerably more severe thanthose experienced under secondary vegetation of the tra-ditional swidden system (Zinke et al. 1978, Alford 1992,Forsyth 1994). Carbon sequestration might be approxi-mately the same under an intensive tree-crop system as itis under swiddening, but it would be negligible under anannuals system (Tomich et al. 1998). The short-term eco-nomic returns of converting from secondary vegetation toa market crop, however, would probably be higher thanunder the swidden system; indeed, the worldwide trendtoward a global economy might very well drive a movefrom swidden to permanent agriculture over the next fewdecades.

Tat hamlet has been fortunate in being able to survivethe changes of the last several decades with minor varia-tions in the total area covered by secondary vegetation (seeTable 1). Government officials and planners need to rec-ognize that perhaps the biggest effect of tropical forestdegradation has b een a change from a relatively homoge-neous forest to a highly heterogeneous and fragmentedcover of secondary vegetation. The land cover may bedegraded in terms of merchantable timber species. How-ever, this secondary “degraded” vegetation, a product ofthe swidden land-use system, might well be the mostspecies rich and water- and soil-holding land cover avail-able. In addition, composite swiddening is the land-usesystem most suitable for meeting the needs of the localcommunity.

A new old paradigm for managingtropical forestsChazdon (1998) suggests that tropical biodiversity conser-vation is undergoing a conceptual transition in which iso-l a ted forest fra gm en t s , l ogged fore s t s , and secon d a rygrowth forests are now being recognized for their value inthe conservation of biological diversity. These studies rein-force a new paradigm in the management of tropical bio-diversity that extends conservation to human-impacted


Roundtable

1952 1995 1952 1995

Number of Mean Number of Mean

Land Cover Ha % Ha % fragments size (ha) fragments size (ha)

Secondary regeneration 681 92 616 84 18 37 292 2

Closed canopy 81 11 19 3 4 20 19 1

Open canopy 400 54 110 15 5 78 40 3

Grass,bamboo,scrub 200 27 487 66 9 22 233 2

Swidden 52 7 73 10 35 1 75 1

Paddy 7 1 43 6 1 8 45 1

Total 740 100 732 100 54 13 412 2

aWhen we superimposed the Landsat image on the aerial photographs,Landsat pixels (picture elements or grids) at the edge of the image didnot align perfectly with the aerial photographs and were eliminated. This explains why the total area is slightly different for the two dates.

Ta ble 1. Land cover and fra gm en t a tion in Ban Tat in 1952 and 1995.a


lands (Pimentel et. al.1992, Janzen 1998).Failure to see the benefits as well as the costs of sec-

ondary vegetation and the swidden agricultural system hasled to government policies for settling swidden farmers—many of which have been failures.A more efficient,as wellas humane, policy would be to invest in research on meth-ods of maintaining the biodiversity associated with swid-den fallows while increasing their productivity and soil-sustaining properties. Failure to understand secondarysuccessional vegetation has also meant that resource man-agers have often failed to recognize the implications, bothpositive and negative, of swidden agriculture on biodiver-sity, watershed hydrology, and carbon sequestration (Skoleet al. 1998).Finally, models of global climatic change havebeen based on an extreme scenario of forest conversion todegraded pasture or impoverished grassland (Giambellu-ca 1996). Failure to account for the effects of landscapeheterogeneity may mean that significant effects of land-cover change are not being recognized.

Swidden cultivation is an old paradigm built around thetemporary removal of trees but not of the forest. As weenter the new millennium, we would do well to recognizethe power of this paradigm for managing tropical forestecosystems.

AcknowledgmentsThis paper is based on joint research conducted in Tathamlet since 1992 by the Center for Natural Resources andEnvironmental Studies (CRES) of the Vietnam NationalUniversity, Hanoi,and the East–West Center (EWC), Hon-olulu. Financial support for our field research has beenprovided by grants to EWC from the Japanese Ministry ofForeign Affairs, the Ford Foundation,the Global Environ-ment Forum, and the US National Science Foundation,and by a joint grant to CRES and the EWC from the JohnD. and Catherine T. MacArthur Foundation. Analysis ofremote sensing data on land-cover change in Tat hamletwas supported by a grant to CRES from the RockefellerBrothers Fund.

References citedAlcorn J. 1990. Indigenous agroforestry strategies meeting farmers’

needs. Pages 141–151 in Anderson A, ed. Alternatives to Deforesta-tion: Steps toward Sustainable Use o f the Amazon Rain Forest. NewYork: Columbia University Press.

Alford D. 1992. Streamflow and sediment transport from mountainwatersheds of the Chao Phraya Basin, northern Thailand: A recon-naissance study. Mountain Research and Development 12:257–268.

Brookfield H, Padoch C.1994. Appreciating agrodiversity: A look at thedynamism and diversity of indigenous farming practices. Environ-ment 36 (5):6–45.

Chazdon R. 1998. Tropical f orests (log ’em or leave ’em). Science 281:1295–1296.

Collins BS,Pickett STA.1987. Influence of canopy opening on the envi-ronment and herb layer in a northern hardwood forest. Vegetation70:3–10.

Conklin H. 1957. Hanunoo Agriculture. Rome: Food and AgricultureOrganization of the United Nations.

Do Van Sam.1994.Shifting Cultivation in Vietnam: Its Social,Econom-

ic,and Environmental Values Relative to Alternative Land Use. Lon-don: International Institute for Environment and Development.IIEDForestry and Land Use Series no. 3.

Dove MR.1983. Ethnographic atlas of the Ifugao: Implications of theo-ries of agricultural evolution in Southeast Asia. Current Anthropolo-gy 24: 516–520.

______. 1984. Government versus peasant beliefs c oncerning imperataand eupatorium:A structural analysis of knowledge, myth,and agri-c u l tu ral eco l ogy in In don e s i a . Hon o lu lu : E a s t–West Cen ter.East–West Center Working Paper no. 54.

Enoksson B, Angelstam P, Larsson K.1995. Deciduous forest and residentbirds: The problem of fragmentation within a coniferous forest land-scape.Landscape Ecology 10:267–276.

Forsyth T. 1994. The use of cesium-137 measurements of soil erosion andfarmers’ perceptions to indicate land degradation amongst shiftingcultivators in northern Thailand. Mountain Research and Develop-ment 14:229–244.

Fox J, Krummel J, Yarnasarn S, Ekasingh M, Podger N. 1995. Land useand landscape dynamics in northern Thailand: Assessing change inthree upland watersheds. Ameba 24:328–334.

Giambelluca T W. 1996. Tropical land cover change: Characterizing thepost-forest land surface. Pages 293–318 in Giambelluca TW, Hender-son-Sellers A, eds. Climate Change: Developing Southern Hemi-sphere Perspectives. London: John Wiley & Sons.

Giorgi F. 1989. Two-dimensional simulations of possible mesoscaleeffects of nuclear war fires.I. Model description. Journal of Geophys-ical Research 94: 1127–1144.

Henderson-Sellers A, Dickinson RE, Durbidge TB, Kennedy PJ, McGuffieK , P i tman A J. 1 9 9 3 . Tropical defore s t a ti on : Modeling local- toregional-scale climate change. Journal of Geophysical Research 98:7289–7315.

Japan Su bcom m i t tee for GAME-Trop i c s .1 9 9 6 . The Im p l em en t a ti on Planof G A M E - Trop i c s : D raft at Ju ly 1996. < cl i m a te . gs fc . n a s a . gov / ~ t a i k a n /G A M E - T / Im p l e 2 e / i m p l e 2 e . h tml> (18 Apr 2000).

Jamieson NL. 1991. Culture and development in Vietnam. Honolulu:East–West Center. East–West Center Working Papers IndochinaSeries no. 1.

Janzen D. 1998.Gardenification of wildland nature and the human foot-print.Science 279: 1312–1313.

Kummer DM, Turner BL II.1994. The human causes of deforestation inSoutheast Asia. BioScience 4: 323–328.

Kunstadter E, Chapman C, Sabhasri S, eds. 1978. Farmers in the Forest:Economic Development and Marginal Agriculture in Northern Thai-land. Honolulu:East–West Center Press.

Lawrence D, Peart D, Leighton M.1998. The impact of shifting cult iva-tion on a rainforest landscape in West Kalimantan:Spatial and tem-poral dynamics.Landscape Ecology 13:135–148.

Le Trong Cuc.1996. Swidden agriculture in Vietnam. Pages 104–119 inRerkasem B, ed. Montane Mainland Southeast Asia in Transition.Chiang Mai (Thailand): Chiang Mai University Consortium.

Le Trong Cuc, Gillogly K, Rambo AT, eds. 1990. Agroecosystems of themidlands of northern Vietnam. Honolulu: Environment and PolicyInstitute,East–West Center. Occasional Paper no. 12.

McClanahan TR, Wolfe RW. 1993. Accelerating forest succession in afragmented landscape: The role of birds and perches. ConservationBiology 7: 279–288.

Moran EF, Brondizio E, Mausel P, Wu Y. 1994. Integrating Amazonianvegetation,land-use,and satellite data. BioScience 44:329–338.

Murcia C.1995.Edge effects in fragmented forests: Implications for con-servation.TREE 10:58–62.

Myers N. 1991. Tropical f orests: Present status and future outlook. Cli-matic Change 19:3–32.

______. 1993. Tropical forests: The main deforestation fronts.Environ-mental Conservation 20:9–16.

Nepstad DC, Uhl C, Serrao EAS.1991. Recuperation of a degraded Ama-zonian landscape : Forest recovery and agri c u l tu ral re s tora ti on .Ambio 20:248–255.

Nguyen Duy Khiem, Van Der Poel P. 1993. Land Use in the Song Da


Roundtable


Watershed (Northwest of Vietnam). Hanoi ( Vietnam): Vietnamese-German Technical Cooperation Social Forestry Development Project(SFDP) Song Da.SFDP Baseline Study no. 2.

Padoch C, Peters C. 1992. Managed forest gardens in West Kalimantan,Indonesia. Pages 167–176 in Potter CS, Cohen JI, Janczewski D, eds.Perspectives on Biodiversity: Case Studies of Genetic Resource Con-servation and Development. Washington (DC): American Associa-tion for the Advancement of Science.

Padoch C, Harwell E, Sustanto A.1998. Swidden,sawah,and in-between:Agricultural transformation in Borneo. Human Ecology 26:3–20.

Pei Shengji. 1985. Some eff ects o f the Dai people’s cultural beliefs andpractices on the plant environment of Xishuangbanna, YunnanProvince, Southwest China. Pages 321–339 in Hutterer K, Rambo AT,Lovelace G, eds. Cultural Values and Human Ecology in SoutheastAsia. Ann Arbor (MI): University of Michigan Press. Michigan Paperson South and Southeast Asia,no. 27.

Pimentel D, Stachow U, Takacs DA, Brubaker HW, Dumas AR, MeaneyJJ, O’Neil JAS, Onsi DE, Corzilius DB . 1992. Conserving biolo gicaldiversity in agricultural/forestry systems. BioScience 42:354–362.

Potter L, Brookfield H, Byron Y. 1994. The Sundaland region of South-east Asia. Pages 460–518 in Kasperson JX, Kasperson RE, Turner BLII, eds. Regions at Risk: Comparisons on Threatened Environments.Tokyo: United Nations University Press.

Rambo AT. 1990. Slash-and-burn farmers: Villains or victims. Earth-watch 39: 10–12.

______. 1995. Defining highland development challenges in Vietnam:Some themes and issues em er ging from the con feren ce . Pa ge sxi–xxvii in Rambo AT, Reed RR, Le Trong Cuc, Di Gregorio MR, eds.The Challenges of Highland Development in Vietnam. Honolulu:East–West Center.

______.1996. The composite swiddening agroecosystem of the Tay eth-nic minority of the northwestern mountains of Vietnam. Pages 69–89in Rerkasem B, ed. Montane Mainland Southeast Asia in Transition.Chiang Mai (Thailand): Chiang Mai University Consortium.

Rambo AT, Le Trong Cuc. 1997. Some observations on the role of live-stock in composite swiddening systems in northern Vietnam. Paperpresented at the Workshop on Upland Farming Systems in the LaoPDR, Vientiane;18–23 May 1997; Vientiane,Lao PDR.

Rao Y. 1989. Forest resources of tropical Asia. Pages 1–20 in Environmentand Agriculture—Environmental Problems Affecting Agriculture inthe Asia-Pacific Region. World Food Day Symposium; 11 October1989; Bangkok (Thailand): Food and Agriculture Organization of the

United Nations.Schelhas J, Greenberg R, eds.1996. Forest Patches in Tropical Landscapes.

Washington (DC): Island Press.Schmidt-Vogt D. 1998. Defining degradation: The impacts of swidden on

forests in northern Thailand. Mountain Research and Development18:135–149.

Skole D, Salas W, Silapathong C. 1998. Interannual variation in the ter-restrial carbon cycle: Significance of Asian tropical forest conversionto imbalances in the global carbon budget. Pages 162–186 in Gal-loway J, Melillo J, eds. Asian Change in the Context of Global Change.New York: Cambridge University Press.

Spencer J. 1966.Shifting Cultivation in Southeast Asia. University of Cal-ifornia Publications in Geography, Vol.19. Berkeley (CA): Universityof California Press.

Tomich T, Noordwijk M, Budidarsono S,Gillison A, Kusumanto T, Mur-diyarso D, Stolle F, Fagi A, eds.1998. Alternatives to slash-and-bur nin Indonesia: Summary report and synthesis of Phase II. Bogor(Indonesia): International Centre for Research in Agroforestry. ASB-Indonesia Report no. 8.

Turner BL, Moss RH, Skole DL.1993. Relating land use and global land-cover change: A proposal for an IGBP-HDP core project. Interna-tional Geosphere-Biosphere Programme: A study of global changeand the human dimensions of global environmental change.Stock-holm: Royal Swedish Academy of Science.IGBP Report no. 24.

Uhl C, Buschbacher R, Serrao EAS.1988. Abandoned pastures in easternAmazonia. I. Patterns of plant succession. Journal of Ecology 76:663–681.

Williams M. 1990. Forests. Pages 179–201 in Turner BL II, Clark WC,Kates RW, Richards JF, Mathews JT, Meyer WB, eds. The Earth asTransformed by Human Action. Cambridge (UK): Cambridge Uni-versity Press.

Xu JC,Li YH, Pei SJ, Chen SY, Wang KL.1993. Swidden-fallow successionin the Mengsong area of Xishuangbanna, Yunnan P rovince, China.Pages 183–196 in Pei S, Sajise P, eds. Regional Study on Biodiversity:Concepts, Frameworks, and Methods. Kunming (China): YunnanUniversity Press.

Yan H, Anthes R. 1987. The effect of latitude on sea breeze. MonthlyWeather Review 115: 936–956.

Zinke P, Sabharsi S, Kunstadter P. 1978. Soil fertility aspects of the Lua’forest fallow system of shifting cultivation. Pages 134–159 in Kun-stadter P, Chapman E, Sabharsi S, eds. Farmers in the Forest. Hon-olulu: University Press of Hawaii


Roundtable


Documents

Shifting Cultivation: A New Old Pa radigm for Managing Tropical Fo