a livelihood perspective on natural resource management

41

A Livelihood Perspective onNatural Resource Management and

Environmental Change in Semiarid Tanzania*

Torben Birch-ThomsenInstitute of Geography, University of Copenhagen,

Copenhagen, [email protected]

Pia FrederiksenDepartment of Policy Analysis,

National Environmental Research Institute,Roskilde, Denmark

[email protected]

Hans-Otto SanoDanish Centre for Human Rights, Copenhagen

[email protected]

Abstract: The aim of this paper is to explore how social relations influence land useand natural resource management at the local level. Through empirical analysisthat tracks changes in land use and environment over 40 years, we present evidenceof a process of agrarianization based on commercialization of crops and expansionof cultivated land. With the concept of livelihood strategies as an analytical frame-work, subcommunity processes are analyzed for their impact on intensification anddegradation. Accumulating strategies are linked to expansion, commercial cropproduction, and selective intensification through high-value inputs, while at theother end of the scale, peasant-labor households endure exhausted or marginalpotential land resources combined with lack of flexibility in input consumption.The article shows how degradation and intensification occur simultaneously andhow incomes may increase even during processes of land degradation. We arguethat a livelihood approach can be useful in uncovering and explaining theseprocesses.

Key words: livelihood strategies, environmental change, land degradation, landuse intensification, Tanzania.

also how livelihood diversification may beseen as part of a strategy to retain access toor control over land. The purpose is partlymethodological, in the sense that the con-cept and application of a livelihoodapproach are examined.

This article is about environmentalchange and livelihood strategies. It seeks toexplore how social relations and strategiescan influence environmental change, but

* All authors are members of the SASAresearch program (Sustainable Agriculture inSemi-arid Africa), a research cooperationbetween the Centre for DevelopmentResearch, Copenhagen; Institute of Geographyand International Development Studies,University of Roskilde; Institute of Geography,University of Copenhagen; and Royal School ofVeterinary and Agricultural Studies,

Copenhagen; in collaboration with Institute ofResource Assessment, University of Dar esSalaam. The case village studied here was one ofsix investigated by an interdisciplinary SASAresearch team. A draft version of this paper wasoriginally presented at the IUAES/IGUCongress on “Livelihoods from ResourceFlows” in Linköping, 19–22 August 1996.

Marczak Business Services, Inc.

This article originally appeared in Economic Geography Volume 77, Number 1, January 2001. Copyright © 2001 by Clark University.

42 ECONOMIC GEOGRAPHY

The context is a semiarid area inTanzania, where land use and environmen-tal change over the last 40 years aredescribed. A case study methodology hasbeen applied, with the village of Ikuwalaforming the central background againstwhich environmental and social change arediscussed. Ikuwala is located in IringaDistrict, approximately 500 kilometerssouthwest of the capital of Dar es Salaam.It is a subvillage under the greater villageof Ilula, a commercially lively communitylocated on the Iringa–Dar es Salaam high-way. Ikuwala benefits from its proximity tothe road. Since the early 1980s, tomato cul-tivation has become a major source ofincome in Ikuwala as well as in other local-ities close to the road.

The interest in livelihood strategiesderives partly from the interdisciplinaryintegration between social and natural sci-ence perspectives. Complexity and diver-sity of change and a growing emphasis onlocal knowledge have brought social andnatural science together in new institu-tional frameworks of research. One impor-tant outcome of this integration is theeffort to develop interdisciplinary methods,while another is a growing sensitivity tolocal diversity. Such insights and endeavorshave inspired the approach we have takenin the present article and its analysis ofchange as seen from below. While manystudies of environmental change inTanzania have been conducted with theirpoint of departure in the nature of land usechange and possible degradation,1 theseprocesses are still so rudimentarily under-stood that case study methodologies arewarranted. One important finding from thestudy described here is that intensification,

conservation, and degradation occur simul-taneously. Such processes are, we argue,strongly dependent on the nature anddevelopment of diverse livelihood strate-gies at the subcommunity level.

Environmental Change andLivelihood Strategies

Stagnation in agriculture due to human-induced land degradation has been thesubject of debate for several decades, espe-cially concerning semiarid parts of sub-Saharan Africa. Earlier research effortswere devoted to the definition of processesof degradation and desertification, andexplanations were found mainly in poormanagement regimes leading to overgraz-ing, overcultivation, and deforestation.Research in the last two decades, however,has demonstrated that technicist and man-agement-oriented explanations are inade-quate, not least in explaining the diversityof change and in attributing sufficientimportance to social and political factors,both at the micro and the macro levels(Dahlberg 1994).

A growing integration of natural andsocial science perspectives has resulted inexplanations that are both less categoric indefining environmental change as degrada-tion and less conclusive in attributing envi-ronmental problems to people and com-munities (Thomas and Middleton 1994;Dahlberg 1994; Little 1994; Fairhead andLeach 1996). During the 1980s and 1990s,the social science perspective contributedto moving part of the explanation of degra-dation from peasants and pastoralists toeither the character of political economy(Blaikie 1985; Blaikie and Brookfield1987), the form of state intervention(Thomson 1994), or the institutions gov-erning natural resource management(Secher Marcussen 1993). Thus, not onlyhas the perspective on the nature of envi-ronmental change been altered, but causa-tion and policy intervention issues havebeen reassessed. From the concern withtechnical solutions and with restructuring

1 Land degradation and desertification havebeen discussed at length (see, e.g., Dahlberg(1994) and Little (1994) for summaries), andthe dynamics of vegetation succession andchange in drylands have been debated in argu-ments against the concepts of stability and equi-librium as characteristics of dryland ecosystems(Ellis and Swift 1988; Thomas and Middleton1994, 128).

A LIVELIHOOD PERSPECTIVE ON NATURAL RESOURCES AND ENVIRONMENT 43

herd sizes, production techniques, andinputs at the community level that pre-vailed during the 1970s, the debate aboutinterventions and/or remedies nowrevolves to a much larger degree aroundthe complexity of local and wider commu-nity interactions—whether the point ofdeparture is taken from above, in areassuch as market liberalization, or frombelow, in reinterpreting local change.2

While local change and diversity andindigenous knowledge are increasinglyemphasized, the “local” discourse oftenremains at the level of the community, asfor instance in the Boserupian reinterpre-tation of conservation and intensificationprocesses in West and East Africa (Netting1993; Tiffen, Mortimore, and Gichuki1994; Mortimore 1995). Subcommunityprocesses such as social differentiation,growing poverty, and a diversity of liveli-hood strategies, however, render such gen-eralized community-level interpretationsinsufficient for understanding howprocesses of natural resource conservationare sustained from below. In fact, conser-vation success stories are often written witha strong focus on the outcome of success,while less attention is paid to the stories ofloss, marginalization, and poverty in theprocess. It is our conviction that, due toheterogeneity within the rural communi-ties in terms of access to and control overresources, conservation and degradationmay take place simultaneously within asmallholder community owing to the diver-sity of household strategies.

The livelihood perspective as a way ofstudying change from below and as a wayof emphasizing actor diversity has recentlygained importance. One example is when

the United Nations DevelopmentProgramme (UNDP), toward the end ofthe 1990s, made “sustainable livelihoods” afocus area for poverty reduction (UNDP1999), another when the BritishDepartment for International Develop-ment (DFID) in 1998 redefined itsapproach to rural development as “sustain-able rural livelihoods” (Carney 1998; seealso Bryceson 1999). The interest in liveli-hoods derives from many quarters: fromstudies of coping in arid and semiarid envi-ronments (Reitsma, Dietz, and Haan 1992;Davies 1996), from studies emphasizingincome diversification in rural communi-ties (Preston 1992; Reardon 1997; Ellis1998; Bryceson 1999), from studies ofpoverty and environmental degradation(Reardon and Vosti 1995; Murton 1997;Scoones 1998; Bebbington 1999), andfinally from studies of changing genderroles (Joekes et al. 1994; Francis 1998).UNDP draws on the approaches devel-oped in institutions like the Institute ofDevelopment Studies and theInternational Institute for SustainableDevelopment (UNDP 1999).

A key notion that has been central inmany of the previous discussions of liveli-hoods is that of rural diversity, and withthat, an emphasis on the importance ofnonagricultural income sources in consti-tuting rural livelihoods (see, for instance,Ellis 1998). Use and management of nat-ural resources, moreover, is increasinglylinked to broadly conceived resourceaccess (Bebbington 1999). Studies of nat-ural resource management could thereforebenefit from moving the focus from agrar-ian livelihoods and farming systems to rurallivelihoods, incorporating the wholeresource endowment.

A point of departure in our effort toelaborate on livelihood strategies, however,is an understanding of livelihoods as anapproach to the study of rural change andenvironmental management. Three dimen-sions are inherent in this approach: first,the provision of a framework whereby itbecomes possible to examine how widersocioeconomic and sociopolitical change

2 In Tanzania, much of the literature dealingwith environmental change has been conducted“from above,” with the point of departure inmarketing reforms. See Bagachwa and Limbu(1995), Reed (1996), Aune (1997), Øygård(1997). Studies of environmental change with abottom-up perspective are found in Östberg(1995) and Loiske (1995).


relates to local change; second, an empha-sis on the importance of social differentia-tion and of agency in determining out-comes of local change; and third, theimportance lent to both physical and socialresources employed by individuals orhouseholds in shaping strategies and valuesfor living. Thus, the livelihood study at thecore level examines household or individ-ual access to, utilization of, and allocationof resources, while at the secondary level itfocuses on how local livelihood strategiesinteract with wider changes in markets andpolitical institutions and with environmen-tal and social conditions.

The aim of this article is thus to elabo-rate empirically on livelihood and environ-mental change in order to explore how con-servation and intensification3 practices aresocially related. A secondary purpose is tothrow light on the relative importance ofagricultural incomes in this particular localsetting of Ikuwala, where diversificationoccurs as a strategy of defending access toand control over land even by the poorestgroups.

A historical perspective is warranted (seealso Scoones 1997). Longitudinal studies oflivelihood remain difficult, however, sincethe lack of comparable data inhibits solidinterpretations of livelihood change (Ellis1998, 5; see also Christiansson et al. 1993,14). The present study considers a periodof 40 years, examining changes in popula-

tion, production strategies, land use, andenvironment in order to ground the liveli-hood strategies of contemporary Ikuwala inbroader processes of changes in markets,resource scarcities, and social relations.

Land Use and EnvironmentalChange in Mazombe Division

The following section concerns theprocesses of change in land use and envi-ronment that have taken place in MazombeDivision during the last four decades, withemphasis on the case study area in Ikuwala.The changes will be analyzed in a contextof population change and settlement andinclude aspects of intensification/expansionand land degradation. Degradation isunderstood as processes of environmentalchange that lower the potential of produc-tion and other natural resource use withinthe present systems of farming and naturalresource management, and which do notseem reversible within the framework ofthe present institutional and socioeco-nomic setting. Land degradation thereforerelates to the opportunities of the currentfarming population and is not necessarily aquestion of ultimate irreversibility or cross-generational time frames. The interpreta-tion of change builds on a variety ofsources, as combinations of land use andland cover have been related to the per-ception of change that the farmers havepresented to us in interviews. This allowsadditional information on the importanceattributed to the various processes ofchange.

The Local Setting

Iringa District is one of the five districtsof Iringa Region and covers a little morethan one-third of the area of the region. Interms of population, it is the largest district,even when excluding Iringa town, while interms of population density, Ludewa,Njombe, and Mufindi districts farther tothe south are more densely settled thanIringa Rural, according to the 1988 popula-

3 Intensification is a concept often used in twodifferent ways. At a farming systems or commu-nity level, it is understood as an increase in landuse, i.e., as the cultivation of a growing share ofthe land to which the community has access.This is land use intensification, which is associ-ated with Ruthenberg’s R-factor, indicating theproportion of land used in a given season of thetotal land available (Ruthenberg 1980).However, it may also be understood at thehousehold or field level, where intensificationdenotes a higher resource input of labor or cap-ital per unit area. In the following sections, werefer to both these forms of intensification butattempt to make clear which meaning is given tothe concept in the context.


tion census (Lulandala 1994, 67–68). Thedistrict has been divided into four agro-ecological zones: the northern fringe, witha warm tropical climate and low annualrainfall of 550 millimeters; the Ismani-Nduli Flats, including Ismani Division,characterized by unreliable rainfall varyingbetween 600 and 700 millimeters; thesemiarid parts of Mazombe Division,including the village of Ikuwala (see Fig.1);4 and finally the Kilolo MountainousArea, with rainfall above 1,200 millimeters.Ismani and part of Mazombe Division thusconstitute the semiarid parts of IringaDistrict.

The Iringa–Dodoma road was com-pleted in the 1930s, linking Iringa to cen-tral areas of Tanzania, whereas the all-weather Iringa–Dar es Salaam road wascompleted in 1954 (Feldman 1983, 72).Infrastructure development and a growingdemand for maize were therefore impor-tant in stimulating immigration to andcommercializing maize production inIsmani Division in Iringa District. During1973, when the tarmac road connected theSouthern Highlands to Dar es Salaam,some of the national focus on maize pro-duction shifted from Ismani in IringaDistrict to higher-altitude areas with morereliable rainfall like Njombe in IringaRegion and to other parts of the SouthernHighlands (see Rasmussen 1986; Raikes1986; TRDB 1976).

With the increasing traffic on the tarmacroad to Dar es Salaam, Mazombe Divisionwas located much more favorably than

Figure 1. Location of study area, the semiarid parts of Mazombe Division. Source: T. Birch-Thomsen.

4 In geomorphological terms, the drier partsof Mazombe Division are part of the Ismani-Nduli Flats and Iringa Plain, but with a margin-ally more reliable precipitation than, forinstance, the Ismani Division, located to thewest of Mazombe Division.


before. Cultivation of tomatoes apparentlyincreased during the 1970s in villages closeto the road, such as Imalutwa and Ilula(Kiwasila and Odgaard 1992, 57–59).However, public efforts to promote maizeproduction bypassed the semiarid areas.Whatever commercialization took place inMazombe was then mainly based on effortssustained “from below.”

Methods of Analysis

Methods used to analyze land use andcover, settlement patterns, and environ-mental change rely on fieldwork conductedduring the period 1994–98, in combinationwith interpretation of remotely sensedinformation in the form of aerial pho-tographs from 1955–56 and 1977–78 aswell as satellite images from 1966, 1986,1993, and 1996.5 The fieldwork consistedof group and individual interviews coveringsuch issues as farming systems, village his-tory, and environmental change.Moreover, extensive walks were carried outin the main study village, Ikuwala, with insitu discussions in fields concerning farm-ing methods and land management.Transects yielded ground truth for thesatellite image interpretation.

The interpretations of settlement pat-terns and changes in natural vegetation arebased on interpretations of land use pat-terns, including areas under cultivation andgrazing areas, as well as major off-field veg-etation types. This analysis covers theMazombe Plains (for a detailed analysis,see Birch-Thomsen (1996)). Land coverhas been analyzed in terms of major vege-tation types and covers, while soil erosionhas been interpreted through analysis of

major features (gullies, development ofsedimentation fans), as well as throughindicators like soil colors and vegetationcover and type.6 The image interpretationsare supported by field observation from1994 and 1998 and cover Ikuwala subvil-lage only. Village interviews—group andindividual—supplement the interpretationof the time-series.

Demographic Change

The growing commercial importanceand availability of arable land in MazombeDivision are reflected in the population fig-ures shown in Table 1. According to thecensus figures, the growth rate of the pop-ulation in Ismani almost stagnated in themid-1960s to mid-1970s, while MazombeDivision far surpassed the district average.Even during the decade 1978–88, markeddifferences in rates of growth are evidentbetween the two neighboring divisions,with Mazombe exhibiting a growth rate ofnearly 1 percent above that of Ismani. Aprocess of virtual emigration took place inIsmani after the 1960s due to the shiftingpriorities of state-supported maize produc-tion, whereas Mazombe was characterizedby immigration in 1967–78 and populationgrowth similar to the district average in1978–88. During 1988 population densityin Ismani Division was 19 persons persquare kilometer, whereas in MazombeDivision it was 36 persons per square kilo-meter.

The subvillage of Ikuwala, located 5 kilo-meters to the northwest of Ilula Mwaya onthe Iringa–Dar es Salaam road, was sur-veyed extensively for this study (see Figs. 1and 2). According to our survey data (Sano1999), in 1994 Ikuwala was a subvillage of206 households, with an average size of 5.5members. This gives an estimated popula-tion density of 60 people per square kilo-meter. Maize is the main staple crop, while

5 Aerial photos were obtained from theDepartment of Surveys and Mapping, Dar esSalaam. Satellite images were DeclassifiedCorona Satellite Photography (B/W) fromNovember 1966, SPOT data from August 1986,TM data (bands 2,3,4) from December 1993,and TM data (bands 2,3,4) from August 1996.Refer to Images in the Reference list for fullsources.

6 For detailed analysis, see Frederiksen(1998). The analysis builds on image enhance-ments and visual interpretation.


tomatoes, sunflower, and maize are themajor cash crops.

Changes in Settlement, Land Use, andEnvironment

Since the mid-1950s, the socioeconomicconditions of the semiarid parts ofMazombe Division have gone throughmajor changes as a consequence of—or inaddition to—the population increasedescribed above. Aerial photos from1955–56 reveal that settlement in theMazombe semiarid area was sparse andlocated at the base of the hills and on the

upper foot slopes of the inselbergs in thearea. A more dense settlement structurewas located close to the main road to Dares Salaam. In general, cultivation tookplace around the settlements and alongrivers. The aerial photographs show thatnatural forest dominated the Mazombesemiarid area in 1955.7 A patchy forest

Table 1

Iringa District Population Data, 1967–1988Mazombe Rate of Ismani Rate of Iringa District Rate of

Division (N) Growth (%) Division (N) Growth (%) Total (N) Growth (%)

1967 32,121 — 34,044 — 230,881 —1978 47,397 3.6 35,848 0.5 290,101 2.11988 58,329 2.1 40,912 1.3 363,605 2.3

Sources: United Republic of Tanzania 1969, 48; 1978, 100; 1988, 105.Note: It is unclear whether Ilula Ward has been included in the figures provided by the 1988 census publication. Ifit has not, the rate of growth for Mazombe Division 1978–88 probably represents an underestimation.

7 A previous study (EEC 1986) has shown thatthe few undisturbed pockets of woodland con-sist of miombo and scrub vegetation of miomboand acacia. This vegetation type is likely to havedominated the natural vegetation at that time.

Figure 2. Map of Ikuwala subvillage, located in Mazombe Division, Iringa Region, northwest ofIlula Mwaya and the tarmac road between Iringa and Morogoro. Source: 1:50,000 map of Tanzania.

United Republic of Tanzania (1982).


clearing and settlement in the northernpart of Mazombe indicate that the settle-ment frontier was on the move, with newhomesteads located primarily around thesmall inselbergs. The area of present-dayIkuwala was still sparsely populated in themiddle of the 1950s. Apart from signs ofgrazing in the area closest to the IlulaMwaya settlement, the only signs of culti-vation were small plots surrounding thesesettlements (Fig. 3, 1955–56).

Generally, cattle were reported by vil-lagers to dominate the livelihood pattern.Each family was estimated to have in therange of 50 to 100 head of cattle living onfodder grasses from the hills and wateringin the swamp (mbuga). The effect of graz-ing on the natural vegetation in Ikuwalasubvillage is clearly revealed on the images.The eastern part of the Ikuwala foot slopeswas especially disturbed, since the cattlefrom Ilula trekked in this direction forgrazing and water. The hill plateau andremote hillsides were dominated by naturalvegetation, with patches of miombo wood-land8 on the steeper parts. Several deeplyincised gullies crossed the foot slopes (seeFig. 2). These large structures, however,were formed during an earlier period, andby the 1950s they were relatively stable,with vegetation covering the sides and thestreambed and with no signs of large sedi-mentation fans. In addition to livestockrearing, finger millet was cultivated onbushfields, while maize and beans wereintercropped on ridges—mainly for subsis-tence. The inhabitants explained this set-tlement and land use pattern mainly by thenatural resource availability at that time:water availability along the foot slopes andgrazing on the hillsides.

The 1960s and 1970s saw heavy immi-gration into Mazombe Plains by peoplefrom adjacent areas, mainly because ofland availability for cultivation and grazing(Birch-Thomsen 1996). The immigration

to Ikuwala took place during the 1960s andwas concentrated on the upper foot slopesalong the base of the hills. Settlement andcultivation dispersed over the upper footslopes throughout the 1960s, and expan-sion toward the swamp to the north startedin the late 1960s (Fig. 3, 1966). A similarexpansion of cultivation took place in thearea close to Ilula.9

In Ikuwala, the expansion of cultivationwas favored by the introduction of ox-ploughing, a new technology which raisedthe production potential of the heaviersoils close to the swamp. This change intechnology also affected the cropping pat-tern toward an increased production ofmaize and the cultivation method towardflat cultivation. According to interviews(Birch-Thomsen 1996) and image analysis,a similar but slower process of expansion ofcultivation took place in the hills. This con-version of the production system towardcultivation was accelerated by cattle dis-eases, which eradicated a large number ofcattle in 1961–62, as well as at the end ofthe 1960s, and changed natural resourcemanagement strategies immensely. Thereduced grazing pressure initially led toregrowth in certain parts of the area.Natural vegetation on less-sloping land wasincreasingly cleared for cultivation, how-ever, and hillsides became deforested as aresult of new grazing patterns and fuel-wood collection for household consump-tion.

8 The miombo is dominated by the generaBrachystegia, Julbernardia, and/or Isoberlinia(Campbell 1996).

9 During the late 1960s, a socioeconomic sur-vey providing information on demographic andproductive patterns in Iringa District was car-ried out in the Southern Highlands (Branner-Jespersen 1971). It included two enumerationareas in Mazombe Division close to the mainroad linking Iringa to Dar es Salaam. Accordingto this study, some of the basic elements thatprevail in present-day livelihood strategies werealready apparent in the productive patterns dur-ing the late 1960s: a reliance on crop produc-tion, with maize being the dominant crop,though often intercropped, and production oftomatoes, the cash importance of which was notdocumented, however.

Figure 3. Land use changes in Ikuwala subvillage, 1956–1996. Source: 1956 aerial photography;1966 declassified satellite photography; 1978 aerial photography; 1986 SPOT-image; 1996 TM-

image. Refer to Image reference list.


During the 1960s, animal traction wasrestricted to a minority of the population,whereas the process of adoption took placerapidly during the 1970s and 1980s. At theend of the 1970s, upland tomato cultivationon rainfed fields was adopted as part of theproduction system. A number of house-holds also acquired access to irrigated gar-den land where tomatoes could be grownthroughout the year. This crop spreadrapidly in the village, becoming an alterna-tive to maize as a cash crop.

Aerial photos from 1978 show that theMazombe Plains were now inhabited andcultivated extensively, with few pockets ofnatural forest left, except for the ImageForest Reserve. A major change in settle-ment pattern and localization took place asa result of the villagization process,1974–76, where people were movedtogether in nucleus villages, often with anelement of coercion. As formerly docu-mented from other parts of Tanzania (see,e.g., Birch-Thomsen 1993; Kikula 1997),this gave rise to intensified land use aroundthe newly settled villages and to moreextensive practices in the more distantfields. The immediate effect of villagizationin Ikuwala was that some of the peripheralareas in the hills experienced a period ofvegetation regeneration. This was short-lived, however, since farmers started torecultivate old and more remote fields afew years after the resettlement (Birch-Thomsen and Frederiksen 1995). By 1978,the area seems to have reached a minimumof vegetation cover: the foot slopes werecleared for cultivation, and to a large extentthe hillsides had been turned into bushedgrassland (Fig. 3, 1978). Cultivationexpanded upward from the bottom of thehills on slopes of 5–10 percent, and sheeterosion was probably accelerating,although it is not easily detectable on theimages. Bright soil colors indicate disap-pearance of topsoil. Similarly, cultivationspread to the very edges of the large gulliesintersecting the area, as well as to the edgeof the swamp. The large gullies had beenreactivated, vegetation in the bottom haddisappeared, and sedimentation was now

visible at the gully outlets. New gullieswere under development, especially in theold cattle tracks and on the newly openedland on the valley sides. Downslope, newrills were working their way toward theolder, large gully structures.

Satellite images from 1986 and 1996(Fig. 3) indicate pressure on land in theIkuwala area, where the edge of the swampretreated significantly, giving way to culti-vated fields. Expansion of cultivation tookplace in the hills during the late 1980s and1990s, as more people settled in these partsof the village area (Kagera Plateau).Observations made during the projectperiod show that vegetation cover is still onthe decrease in the hill areas. Charcoal pro-duction takes place on a commercial, if notlegal, basis, leaving hillsides deforested butcovered by grass and herbs. Some areas ofthe hilltops are cultivated after tree felling,leaving the soil bare for parts of the year.Steep hillsides and valleys are beingcleared for cultivation and cultivation isnow taking place upslope from the old set-tlements at the base of the hills.

Informants stated that fallowing wasincreasingly rare on the foot slope fieldsand that individual fields had often beencontinuously cultivated since the late1970s. In 1998, cultivation of marginalareas on the hillsides (> 35% slope) and inthe valleys along the small streams drainingthe hills was clearly on the increase.Villagers had a clear perception of adecrease in soil fertility within the lastthree decades, substantiating this by theamount of land required to feed a familythen and now. According to the socioeco-nomic survey carried out in the whole vil-lage in 1995, the average household yieldof maize was approximately 450 kilo-grams/hectare (2 bags/acre).10 Generally,fallow practices have decreased signifi-cantly and fields are only left fallow whenthe yields decrease “below what is accept-able.” This, in combination with currentpractices of trading and renting land as the

10 One bag is the equivalent of 90 kilograms.


now dominant way of acquiring new land,and added to the picture of land pressure,supports the villagers’ perception ofdecreasing soil fertility.

In 1998, a year of extremely high rainfall,a majority of farmers interviewed statedthat erosion had become worse during thepreceding years. They experienced soilerosion as a serious problem, causing eithera decrease in productivity followingremoval of nutrients or seeds and seedlingsto be washed away and crops buried undersandy sediments downslope. One elderlyfemale farmer, reporting that houses werebeing flooded during recent years,described accelerated erosion of footpathsby livestock, tractors, deforestation, andpopulation increase after villagization.Generally, however, there was no consen-sus among the villagers as to the causes ofsoil erosion, but deforestation in the hillswas often mentioned. According to the vil-lagers, trees were more abundant beforevillagization and deforestation in the hillswas related mainly to charcoal production.

Fieldwork conducted during 1994 and1995 confirmed this pressure. The areasnorth of the road were almost completelyunder cultivation, with new fields beingopened close to the swamp. The increasingnumber of fields in the swamp (especiallyvisible on the 1996 image, Fig. 3) weremade possible by digging ditches to drainthe area. The farmers explained that expan-sion was facilitated by a decrease in rainfall.However, identification of a generaldecrease in mean yearly rainfall in thestudy area has not been possible (pers.comm., J. R. Jensen, SASA researcher fromThe Royal Veterinary and AgriculturalUniversity, Copenhagen, November 1998).An alternative explanation to the percep-tion of a decrease in rainfall by farmerscould be a relative increase in the agricul-tural dryness as a result of changes in thehydrological regime. Changes in the vege-tation cover in the watershed may havereduced the infiltration rate and the waterretention capacity, leading to an increasedrunoff with a possible siltation of theswamp as a result. Evidence of siltation

linked to gully outlets near the swamp isvisible on recent images.

Evidence of accelerated runoff and ero-sion are abundant in the area. On the footslopes, some larger patches of land in theeastern part are left fallow, probablybecause of low productivity. Major areas onthe upper foot slopes are affected by seri-ous soil erosion, and rills are abundantthroughout the area, in some places devel-oping into gullies. Rills are also found onthe middle part, though less frequent,while sedimentation takes place simultane-ously in a general process of transportation.Along the lower course of the gullies,stream bank erosion removes agriculturalland. On a minor scale, farmers reclaimland within the gully floor for cultivation.On the lower foot slopes, the problem ofsedimentation is increasingly felt as rillsand gullies end up flowing into a field andwashing away or burying the crop. Shrubsand trees have disappeared from theswamp, and cultivation encroaches on theremaining grassland. Although this may notbe termed degradation, because it actual-izes a new potential for cultivation, it hasrepercussions for the grazing potentialwithin the village.

Hardly any soil conservation effort isobservable. When directly affected oneyear, some farmers tried to redirect theflow of water by digging a trench off thefield. Other farmers have ploughed a ridgeupslope, above the field, but this is only asmall effort compared to the extent of theproblem.

The changes within the last five decadesin Ikuwala are summarized in Table 2.During the late 1950s and early 1960s,agricultural practices began a process ofchange. A combination of permanent culti-vation on ridges close to settlements andshifting cultivation on more remote fieldschanged toward more widespread use ofpermanent cultivation on flat fieldsploughed by ox-ploughs or in a few casestractors—a process paralleled by adecrease in the number of livestock. Thischange was a reality by the mid-1980s,when most of the land on the foot slopes


was brought under cultivation. No changeor very little change in agricultural prac-tices in this part of the village area hastaken place since then. However, theexpansion of agriculture, observed withinthe last decade, into new areas such as theswamp and the hills supplement this pic-ture. New techniques for draining the wet-lands have been introduced, while old for-est slash-and-burn techniques incombination with charcoal burning wererevived. In terms of land use intensificationat the village/community level, a clear pic-ture of agricultural expansion is evident,whereas intensification of farming practiceis more questionable. The introduction ofanimal traction made an expansion of thearea under cultivation possible withoutnecessarily intensifying labor use due tothe substitution of manual labor by animals(oxen). On the other hand, through theintroduction of cash crops with additionalexternal inputs, farming became more cap-ital-intensive.

The most recent development, with agri-culture expanding into the wetlands, hasintroduced more labor-demanding meth-ods and resulted in labor intensification. Atthe same time, some farmers have reintro-duced the labor-extensive cultivation tech-niques of slash and burn in the hills, thoughoften accompanied by labor invested incharcoal burning. Thus, the case of Ikuwalarepresents a process of environmentalchange that in some respects can bedeemed land degradation. The foot slopesare exhausted to such a degree that longerfallow periods or additional inputs areneeded in combination with soil conserva-tion upslope in order to regenerate soilstructure and fertility. Gully developmentis taking its toll on agricultural land. Whilethe villagers blame charcoal production forthe state of natural resources, it may not bethe major cause of increased soil erosion.11

The grass layer under the trees and scrub isto a large extent undisturbed after thefelling of the trees unless cultivation fol-lows. When deforestation is combined withcultivation, however, it leads to sheet andgully erosion on steeper slopes to theextent that the bedrock is exposed. Theseprocesses can be termed irreversible withina time span of relevance to the users.

Current Livelihood Strategiesat the Community Level

The foregoing analysis has shown howmaintenance and intensification ofresource use have historically been limitedin the study area. In the following section,we seek to explain why this is the casethrough an analysis of the current liveli-hood strategies in the Ikuwala community.The livelihood analysis concerns questionsof how individuals and households mobi-lize, combine, and decide over resources,but also of how individuals or groups relateto each other, given their diverse resourceuse and management strategies. The liveli-hood analysis thus leads to a social analysisof strategies and patterns of adaptations tochanging conditions and hence of variousgroups’ room to maneuver (see alsoReitsma, Dietz, and de Haan 1992, 38–40;Bebbington 1999, 2022–23). The approachthus relates to the capacity to act andchange from below.

In this paper, the analysis of livelihoodstrategies in Ikuwala mainly concerns ananalysis of resource use and combinationsrelated to income sources, land use, andland management. The discussion of adap-tive strategies and room for maneuver isonly preliminary. Less emphasis is put onaccess to resources and mobilizationthrough, for example, networks and socialidentities. Institutional factors are thusdownplayed, while livelihood strategies areanalyzed primarily as income and alloca-tion strategies of respective groups. Thequantitative data used are based on datacollection over two years. Detailed house-

and that it may disappear altogether if protec-tive measures are not taken (Ndangalasi 1998).

11 While charcoal production may not be themajor cause of soil erosion, botanical studieshave shown that the miombo species(Brachystegia speciformis), preferred for char-coal and fuelwood, has been nearly depleted


hold interviews contributed importantqualitative information on the diverselivelihood strategies of the area.Concerning the interrelationship betweengroups and their different strategies, itshould be kept in mind that since the studyarea is semiarid, it is strongly influenced byannual variations, and such variations, mayinfluence both the proportion of house-

holds in different categories and their eco-nomic activities.

Socioeconomic Differentiation

A two-step process of fieldwork formsthe entry point for the following analysis. Asocioeconomic survey was undertaken witha random sample of 35 households in thevillage of Ikuwala (of which 34 are included

Table 2

History and Indicators of Environmental Change in Ikuwala

Image Source andFieldworka

Aerial photos1955–56

Corona satelliteimage 1966

Aerial photos 1978

SPOT image 1986

Landsat TM 1993

Fieldwork 1994–96

Landsat TM 1996

Fieldwork 1998

Village Level LandUse History

1950s: livestock themain livelihood,cultivation minor;few inhabitants

1960s: change inproduction systemfrom cattle to crop;ridge cultivation;introduction ofanimal traction

1970s: increasedadoption of animaltraction for plough-ing; flat cultivation;expansion of culti-vation on the footslopes; increasedmaize production

End of 1970s: intro-duction of tomatoes

1980s: agriculturalexpansion in thehills

1990s: adoption ofsunflower; trade inland; cultivation ofmarginal land

Agricultural expan-sion in the swampand hills

Context History

1954 all-weatherroad to Dar esSalaam completed

1961–62 cattledisease; immigra-tion and scatteredsettlements

1968–69 cattledisease

1970s: populationgrowth of 3.6%

1974 Villagization

1974–79 efforts toestablish pipedwater in the village

1980s: populationgrowth of 2.1%

1986 liberalizationof economy

1995–96 drought

1997–98 floods

Indicators of Environmental Change

a Refer to Images in References.

Clearing ofvegetation onfoot slopes forcultivation;decrease infallow period

Deforestationin the hills forcultivation andcharcoal

Increasingerosionproblems onfoot slopes(sheet andgullies)

Expansion ofcultivation intothe swamp;lack of grassingarea

Cultivationof marginalland


in the data base).12 This survey includedquestions on sources of income, but also onagricultural production and marketing. Asa next step, detailed and open-ended inter-views on earnings from diverse incomesources and on issues of decision makingwere undertaken with individual house-hold members in 10 households. Thesehouseholds were selected according totheir socioeconomic position and their pat-tern of income sources. Similar fieldworkwas undertaken in two other villageslocated in Mazombe, but information fromthese analyses is used only to support con-clusions drawn in the following.

Wealth ranking was used in identifyingcriteria of wealth and poverty. The methodhas been described in Sano (1999). Forother wealth-ranking exercises in Tanzania,see Boesen and Ravnborg (1992) andMung’ong’o (1995). On the basis of thewealth ranking and the interpretationemerging from the interviews, six indica-tors of inequality were used in differentiat-ing the households. These indicators are(1) acreage cultivated (farming ability); (2)tomato sales—that is, earnings from themost important cash crop in the respectivevillages (access to money); (3) whethermembers of the household worked as wagelaborers (the need to depend on others forwork); (4) payment for wage labor (theability to hire additional farm workers); (5)the number of cattle owned (access to cat-tle and capital); and (6) whether the house-hold was forced to buy or borrow food dur-ing the previous season (food security).

The villagers participating in the wealthranking divided their neighbors into two,three, or four groups: a group of impover-ished households disadvantaged on nearly

every score; an intermediary group thatincluded some with a certain potential fordevelopment and others likely to fail intheir endeavors; and an upper-middle oradvanced group that included a few out-standing households. Based on the detailedhousehold interviews, in combination withthe wider household survey, we formulatedthree hypotheses.

First, it became apparent that threemajor strategies characterized householdsin the area: a strategy of accumulation, withenterprising and commercially orientedfarmers in the center; a peasant strategy,which expressed both restraint toward themarket economy and failure to be success-fully involved in it; and finally, a strategy ofsometimes nearly desperate coping withwhatever means the household had at itsdisposal. Second, strategies of accumula-tion pertained to a larger group than thosefarmers who were characterized as well-to-do or wealthy in the wealth ranking.Accumulation strategies related not only tohousehold ownership, but also to a wideraccess to resources most often obtainedwithin the extended family. Moreover,choice of strategy was not only a questionof narrow resource access, but also of atti-tudes toward risk, market relations, andidentities within particular groups. Third,households’ strategic choices related insome way to age. Accumulation strategiescould thus be pursued by relatively youngunmarried men or young couples who werestill strong, whereas peasant restraint orcoping strategies were sometimes also dic-tated by age and relative weakness of thehouseholders.

Our argument is thus that livelihooddivisions say more about resource manage-ment than socioeconomic divisions,because inherent in the concept of liveli-hood strategies is a notion of actualresource use, whereas the socioeconomicdivision may only relate to this theoreti-cally. Differentiating according to liveli-hood strategies attaches importance toactual resource use and allocation ratherthan to formal access.

12 The SASA study comprised five villages inIringa District and one village in RujewaDistrict. Some of the dimensions of livelihoodstrategies discussed in the context of Ikuwalaalso prevailed in, for instance, Ilambilole andMkulula villages. For an early comparison oflivelihoods and poverty in Mkulula and Ikuwalarespectively, see Sano (1999).


Livelihood Strategies at theHousehold Level

Definitions. The criteria suggested inthe wealth rankings are all associated withthe agricultural economy. It also becameclear from the wealth ranking that wealthyor well-to-do households centered theirstrategy of accumulation on agriculturalproduction. However, as indicated above,the socioeconomic division based onhousehold resources blurred to someextent important differences related bothto the type of resources from which wealthwas derived and to the type of resourceallocation undertaken by the householders.Particularly in the intermediary groupdefined according to wealth criteria, someof the householders who were in the samecategory in a wealth ranking exhibited dif-ferent patterns of resource use and alloca-tion. Some had a pattern much moreclosely related to the well-to-do farmers,whereas others were almost as defensive intheir strategies as the poorest household-ers. Some of these intermediary house-holds were not well-to-do, but they wereenterprising.

When considering the households thatrevealed an enterprising and accumulatingstrategy in the detailed household inter-views, it became clear that they were allcharacterized by a fairly high agriculturalincome level and by a pattern of incomesources which implied that they obtainedincomes from business or crafts. Likewise,the poorest group was associated with a lowlevel of agricultural incomes and with anincome pattern where piecework, possiblycombined with incomes from naturalresources, played an important role. Thelevel of agricultural income in combinationwith the pattern of income sources of thehousehold were therefore used as indica-tors distinguishing one livelihood groupfrom another.

The peasant-labor coping strategy is thedesignation of the poorest livelihood strat-egy. It was given this designation to associ-ate this livelihood strategy with pieceworkand with efforts merely to cope. It is asso-

ciated with low agricultural income (fiveout of ten households had agriculturalincome below Tsh 5,000 in 1994)13 andwith piecework and income from naturalresources (mainly from the collection andsale of thatching grass and fuelwood) as theimportant cash earners.

The peasant strategy was associated withnonaccumulating strategies and with apropensity not to take too much risk andnot to engage strongly in the commercialeconomy. This group had intermediaryagricultural incomes, some quite low (5 outof 11 households had incomes below Tsh35,000). However, this group was alsocharacterized by a vacillating income pat-tern between, on the one hand, pieceworkand natural resources, and, on the other,business, rents, or crafts.

The accumulating farmer strategy wasassociated with a pattern of resource allo-cation clearly aimed at accumulation. Thisgroup had fairly high agricultural incomesand a pattern of income sources based onbusiness, crafts, and rents. However, mem-bers of this group were often quite younghousehold heads—in some cases unmar-ried men—who depended on help fromtheir fathers or the family network.Members of the accumulating group oftendisplayed a diversified income pattern.They sought to invest incomes derivedfrom agriculture in other income-generat-ing schemes such as kiosks, masonry, maizetrade, or investments in housing.

The divisions were undertaken in thefollowing manner: agricultural incomes(Tsh) were divided into four groups basedon frequency of distribution: levels from0–15,000 scored 0 points, from16,000–75,000 scored 1 point, from76,000–200,000 scored 2 points, and201,000 and above scored 3 points.Incomes from natural resources: yes scored0 points, no scored 1 point. Incomes frompiecework: yes scored 0 points, no scored 1point. Incomes from crafts: yes scored 1

13 The rate of exchange in 1994–95 was 1 U.S.dollar = 500 Tsh.


point, no scored 0 points. Incomes frombusiness: yes scored 1 point, no scored 0points. Incomes from rents: yes scored 1point, no scored 0 points. The livelihoodclassification is thus based on six variables,with a maximum score of 8 points and aminimum of 0 points. Based on the distrib-ution of particular income sources withinthe score groups, households scoring 0–2points were classified as peasant-laborers,households scoring 3–4 points were classi-fied as peasants, and households scoring5–8 points were classified as accumulatingfarmers. While the livelihood classificationis thus based on analysis of frequency, theselection of indicators was based on obser-vations made during qualitative interviews.Note that several methods of livelihoodclassification were tried with the same indi-cators. A core group of 18 households werealways placed in the same category, while16 households, mostly in the middle of thecontinuum, located themselves with somevariation according to the method used.

Characteristics. It is now possible toconsider the various livelihood strategies,as presented in Table 3. The peasant-labor-ers constituted only 29 percent of the sam-ple. Four out of ten households were

female-headed, and the average age of thegroup was 42, with little variance aroundthis mean. As indicated in the table, theyrelied strongly on their own labor as ameans of raising money or food. In addi-tion, they depended on off-field naturalresources, in the form of grass cutting andfirewood collection, for sale or for beerbrewing. Thus, the state of naturalresources and the distance to the resourcebecame important circumstances for theirresource allocation. Although some peas-ant-laborers might depend on charcoalburning, this was more difficult to ascer-tain, since charcoal burning is illegal.Moreover, they all cultivated land; landthus remained an important resource intheir livelihood strategy, not only because itprovided basic food, but also because itprovided grain for beer brewing. Indeed,beer brewing remained an importantincome source for all households across thesample. Sunflower and tomatoes were soldat low levels by peasant-laborers, but what-ever maize was grown was retained forhousehold consumption.

The group coped with the impossibleeither because household heads were oftentoo weak to work, because they were singlewith a number of dependants, or because

Table 3

Livelihood Strategies, Average Age of Household Head, and Income SourcesAverage

Average Crop Income Percentage of Households (HH) Reporting Incomes from

Age of in Tsh Natural HH Head (U.S. $s)a Resourceb Piecework Crafts Business Rents

Peasant-labor 42 15,500 60 70 20 20 0coping strategy (31)(N = 10)

Peasant 41 65,500 18 45 45 27 18strategy (131)(N = 11)

Accumulating 34 430,000 0 0 46 54 62farmer strategy (860)(N = 13)

Source: Village survey (1995) and qualitative household interviews (1996) carried out by SASA research team.a Rate of exchange in 1994–95 was U.S. $1 = 500 Tsh.b Includes activities such as grass-cutting, firewood collection, basket and storage facility production.


they were without access to resources. Oneexample in this group is a single womanaged around 35 with no dependants (HHNo. 11). She cultivated 0.4 hectares ofmaize intercropped with sunflower andcowpeas. The woman indicated that sale ofnatural resource products was her majorincome source, but it became apparentthat piecework had been an equally impor-tant income source during the previousseason. She had earned Tsh 6,000 fromdoing piecework three times, Tsh 2,000from selling ten bundles of firewood, andTsh 6,000 from selling one drum of beer.She had produced two bags of maize, butthis could hardly sustain her through theseason, as grain was also used for brewingbeer. Although she originally reported thatshe did not receive any help from otherpeople, it became apparent during theinterview that she had received finger mil-let from others for beer brewing.

The peasant strategy characterized 32percent of the households and can bedescribed by a limited market involvementand the absence of a long- or medium-termstrategy of change. Two out of 11 house-holds were female-headed. The averageage of the heads of households within thisgroup was 41, but with a greater number ofyoung or comparatively old households inthe group. Five household heads were thusbelow the age of 35 and three were morethan 55 years old. The majority of house-holds within this group may have opted forthe peasant strategy because they had noother choice. However, a minority withinthe group adopted a peasant strategybecause they shied away from a more com-mercialized pattern of income generationthat they found too demanding for variousreasons.

The level of crop marketing was notimpressive in the peasant group, althoughthey could hardly be characterized as sub-sistence producers. Five of the householdshad agricultural incomes below Tsh35,000. However, in contrast to the peas-ant-laborers, most households in this groupattempted to gain some earnings fromtomatoes, while also earning agricultural

incomes from sunflower. Of the averageagricultural earnings of Tsh 65,000, sale oftomatoes made up 52,000. Even in thisgroup, maize was retained for householdconsumption. Nonagricultural incomederived mainly from crafts and petty com-merce, but piecework and naturalresources in addition to beer brewing werealso important. While nearly half of thegroup participated in piecework, almostnone of the households, either in thisgroup or among the peasant-laborers,employed wage labor themselves. Only twohouseholds in each group did so, in verymodest amounts.

One example of the peasant group is afamily of five (HH No. 20), with one wife inthe household. Income from crop sales wascited as the major cash income source, butsupplementary incomes derived from beersales, sale of firewood and charcoal, inaddition to piecework. Tomatoes earnedthe family Tsh 80,000, while 52,000derived from the other income sourcesduring 1994–95. The family employed nopieceworkers themselves.

Substantial crop incomes and diversifiedincome sources from business, crafts, orrents characterized the accumulatingfarmer households. This group constituted38 percent of the sample. The average ageof household heads was 34, reflecting thefact that eight heads of households werebelow 35, while only one head of house-hold was more than 55. The majority of thegroup was thus young enterprising farmers,possibly supported by their fathers, whilethe remainder represented the small groupof wealthy farmers in the village, most ofthem middle-aged. No heads of house-holds were female.

The major income earner in this groupwas tomato production, possibly combinedwith maize marketing. Tomatoesaccounted for almost Tsh 350,000 of theaverage earnings of the accumulating farm-ers (see Table 2), while maize accountedfor nearly Tsh 80,000 of total agriculturalearnings. Growing tomatoes representedan avenue into and a source of accumula-tion, in contrast to the poorer livelihood


groups’ reliance on sunflower production.Apart from agricultural incomes, accumu-lating households typically diversified theircommercial activities. As a result, theyearned substantial incomes from rentingtractors or oxen to fellow farmers or fromtrading in maize. In addition, their wivescontributed with substantial earnings frombeer brewing and possibly from some othertrade. The accumulating farmers did notobtain any substantial cash incomes fromnatural resources, and they did not takepart in piecework. On the contrary, theaccumulating farmers were the employersin the village. Eight out of 13 householdshired wage labor, at an average of Tsh26,500 for the whole group.

The most striking example in the groupwas one of the wealthiest farmers in the vil-lage (HH No. 21). He cultivated 24hectares in the village and had access to anadditional 171 hectares, some in neighbor-ing Vitono. He had expenses of Tsh200,000 for wage labor, but earned Tsh 1.2million from tomatoes. Moreover, Tsh50,000 were earned from selling vegeta-bles, 700,000 from maize trade, and180,000 from renting his tractor to othervillagers. Both of his wives earned morethan Tsh 30,000 from selling beer. One ofthem also earned some funds from chickentrade.

This scale of operation is outstanding,but other members of the group wereengaged in similar accumulating operationson a lower scale. They strove to expandtheir agricultural production or other activ-ities. One such example was a younghousehold in their early 20s (HH No. 7).The husband worked as a mason and maizetrader, but earned additional income fromrenting a plough to neighbors. They culti-vated a comparatively small area—only 1.6hectares—but stated that they earned mostof their income from agriculture. Theyeven hired a laborer to help them clear theland. They earned Tsh 110,000 from sellingtomatoes and 19,500 from sunflower. Theyused all their maize for their own con-sumption, but the husband’s newly startedmasonry business allowed him to engage in

maize trade. For religious reasons, the wifebrewed no beer, but she engaged in fire-wood collection and sale. The very ener-getic husband described how he had firstearned money from selling water, thenfrom breeding goats. He invested theseearnings in pesticides for tomatoes. Theirpresent plan was to expand maize produc-tion while continuing to grow tomatoes andsunflower, along with the masonry.

The accumulating farmer strategy washighly dependent on access to land, whileother natural resources were not as impor-tant. Access to and control of capital wasalso important in this group. Control overcapital and connections provided thisgroup with good opportunities to rent ormonopolize good land in both Ikuwala andthe neighboring villages.

Livelihood Strategies, LandManagement, and Access toResources

A number of features reported by thefarmers in Ikuwala indicated an unsustain-able agricultural economy characterized bylow crop output per hectare, declining soilfertility, and erosion. In this section, wediscuss the relationship between differentlivelihood strategies and land managementpractices. The point of departure is mainlythe maize economy, as maize remains themost important food crop and an importantsource of income in the area.

As stated earlier, land productivity inIkuwala was very low—especially on thecentral village lands—and all farmersreported low yields of maize. Average yieldlevels of the three livelihood groups were310 kilograms/hectare (1.4 bags/acre) forpeasant-laborers, 400 kilograms/hectare forpeasants, and 645 kilograms/hectare foraccumulating farmers; land productivitythus corresponded to the livelihood strate-gies. A few outstanding farmers reachedsubstantially higher levels (approximately1,000 kilograms/hectare was the maximumyield in the sample), but a large majority ofthe farmers reported fertility declines andthat soil erosion was an ongoing phenome-


non on their land. Whatever measures ofintensification and soil conservation farm-ers undertook, they seemed to be insuffi-cient to restore reasonable land productiv-ity14 or even to halt the exhaustion anderosion of the land. Although some farmersclearly did not have the ability to investmany resources in their land, others didhave some capacity for investment.Therefore it was of interest to investigatewhy these investments were so poorlyreflected in land productivity.

In order to explore land managementwithin the three groups of farmers, weemployed data from the quantitative sur-vey as indicators of intensification and landmanagement. Two issues in particular wereexamined: access to resources for intensifi-cation and fertility management and flexi-bility in resource allocation. Intensificationis indicated by the inputs used: propensityto use manure, to use chemical fertilizers,and to apply improved seeds. Data on laborinputs in relation to farmed area were notcollected, but we considered investment inworking parties as well as hired labor asindicators of relative intensity of land use.Indicators of fertility improvement in farmpractices are residue incorporation, com-posting, and monocropped beans.

In addition, we explored the access toresources necessary for flexible manage-ment. Positive harvest results in semiaridareas are to a significant degree dependenton the ability of the farmers to have flexibleaccess to resources in order to put them touse at the right time (Adams andMortimore 1997). Thus, some resourcesbecome important at certain—variable—times, like access to tractors, oxen andploughs when the rain starts, seeds when itis time to sow, and labor for weeding.

The accumulating farmers were signifi-cantly better off on all indicators than peas-ant-laborers, who moreover lost flexibilityby hiring out their own labor at criticaltimes. The peasant group accessed laborthrough traditional beer parties to a higherdegree than the other groups. However,the differences in input use discussedbelow hardly seem to account for the dif-ferences in land productivity to any majorextent.

Manuring the maize fields was onlypracticed by a few of the peasant-laborers,whereas approximately 50 percent of theother groups used manure on the crops.Generally, small quantities of manure wereapplied, mostly in the planting holes.Improved maize seeds were not used toany large extent, and almost exclusively byabout one-third of the accumulating farm-ers. This practice was not reflected in thereported yields. The major users of artifi-cial fertilizer were the accumulating farm-ers (60 percent of the group using fertil-izer, but almost exclusively on tomatoes).Other fertility-improving practices werecarried out by less than 25 percent of thefarmers, but there was no clear pattern asto which farmers used these measures. Allfarmers stated that they now practiced onlyflat cultivation.

Peasant-laborers were to some extentrestrained in their access to labor, while thetwo other groups used different arrange-ments in order to gain access. Thesearrangements included traditional workingparties rewarded with serving of beer,which was still frequently arranged espe-cially by the peasants, but also hired labor,which was used by two-thirds of the accu-mulating farmers, but also by other farmerson a smaller scale.

Only two farmers, belonging to the accu-mulating group, owned a tractor, whileploughs were owned by two-thirds of theaccumulating farmers and by around one-third of the other farmers. Ploughs, oxen,or tractors were hired at one time oranother by a quarter of the farmers, butthere was no large differentiation amongthe groups. Since most fields were

14 In the village areas surrounding Ikuwala,the land productivity was much the same as inthe central village areas that had been undercultivation for many years. The best resultsshowed yields of around 2.2 tons per hectare (10bags/acre).


ploughed, it is likely that the peasant-labor-ers and peasants participated in ploughingsystems—that is, arrangements with theowner of the plough, wherein paymenttakes different forms (see also Birch-Thomsen 1993; McIntire, Bourzat, andPingali 1992). As we saw above, there was astrong relationship between livelihoodstrategy and the employment of wagelabor, on the one hand, and participation inpiecework on the other. Thus, there is astrong indication that the accumulatingfarmers were much more flexible in theirability to apply farming resources at theappropriate time than the peasant-labor-ers. The latter may have to work for othersat the optimal planting time or wait for avacant plough, making them more vulnera-ble in terms of dependence on soil condi-tions, rainfall, and timing in general.

Differences in manure application andaccess to labor, oxen, and ploughs may gosome way in explaining productivity differ-ences, but not all. Access to fresh land pro-vides an additional, and very important,explanation. Table 4 shows access to landamong the three groups, according to typeof access. Systematic differences emerge,as the accumulating farmers access largerareas and through various ways of appro-priation. Cultivated land in the group ofaccumulating farmers was double that ofthe other groups, and approximately 20–25percent of the total area was not cultivatedfor various reasons. Buying land remaineda more common strategy among the accu-mulating households, even though thepeasant-laborers additionally bought smallpieces of land. Even borrowing or rentingland seemed easier for the accumulatingfarmers. This was also valid for the practiceof acquiring land in other villages. Whilefarmers in all groups had access to someland in other villages, significantly moremembers of the accumulating householdsventured into that kind of arrangement.These opportunities may contain the mostsignificant factors in explaning productivitylevels. Land acquired in neighboring vil-lages is predominantly located close to theswamp. It is evident that the land near the

swamp both in Ikuwala and neighboringVitono is more fertile than the exhaustedfoot slope land because of its location onthe bottomland, where the nutrients accu-mulate. The water retention capacity isbetter, and the cultivation history shorter(see Fig. 3).15 These areas are likewise mostsuitable for tomato cultivation, but risky inflood years.

Thus, the accumulating farmers, whooperated with favorable conditions interms of access to ploughs, oxen, and labor,had been able to acquire land near theswamp, and they were capable of takingthe risk of cultivating the low-lying land. Asalready mentioned, some farmers had eveninvested in drainage canals. The majority ofthe tomato fields are also found in this areanorth of the road. The reasons for seekingout land in neighboring villages are muchthe same. Thus, strategies for maximizingoutput and income are more related toexpansion on fertile land suitable for pro-duction of high-value marketable cropsthan to intensification on central villageland.

It is this expansionist strategy that hasresulted in the heavy encroachment on theswamp, as evidenced by the remote-sens-ing data. What seems important in theaccumulating strategies is thus the abilityto mobilize such resources as fertile landfor expansion of cultivation; strategicmeans of production, including labor; and

15 Preliminary results from ongoing studieswithin SASA (Birch-Thomsen) indicate a varia-tion in maize yields related to the location andhistory of fields. On the upper foot slopes, andwith the longest cultivation history (from the1950s and 1960s), yields have been estimated tobe approximately 0.45–0.65 tons/ha (2–3 bags/acre). On the mid-foot slopes, typically fieldsallocated during the villagization process(1974–76), the estimated yield was 0.9–1.1tons/ha (4–5 bags/acre). Finally, on the season-ally flooded and sometimes drained lowlands,dominated by newly bought or hired fields,yields were approximately 1.5–1.8 tons/ha (7–8bags/acre). The first two locations are most fre-quently used by peasant-laborers.


capital allowing households to invest intomato production (to buy fertilizer, pesti-cides, and land of good quality), trade, orother accumulating ventures. While villageland use in general has intensified, it hashappened more through expansion of thecultivated land than through intensificationat the farm level.

Discussion and Conclusions

Contextual Change and Its Impact onLivelihood and Land Use

Livelihood opportunities and use of nat-ural resources in Ikuwala were influencedby external factors throughout the period.A combination of constraints on pastoralproduction (cattle diseases) and increasedmarket opportunities (infrastructure devel-opment and a market demand for foodcrops) favorably influenced a process ofagrarianization in Mazombe (i.e., a processin which agricultural production increas-ingly came to dominate income strategies).This process was subsequently coupledwith a process of commercialization, sus-tained by the adoption of technologicalinnovation (expansion of fields through ani-mal traction) and increased market oppor-tunities for crops with a local productionpotential (maize, tomatoes, sunflower) andfor charcoal, which is mostly produced ille-gally. Iringa villages are sometimesreferred to as subsistence villages (e.g.,Lulandala 1994), but this concept is hardly

applicable in Ikuwala, given that farmershere operate with three major cash cropsand sell even minor crops like cowpeas andfinger millet. With commercialization, aprocess of economic growth and social dif-ferentiation took place. Generally, Ikuwalahas become a more wealthy village, but inall likelihood (no baseline data exist) with agrowing number of peasant-laborers—pau-pers—among its households.

These changes, in combination with thevillagization settlement pattern, influencednatural resource management in severalimportant ways. The area under cultivationhas expanded throughout the period, andfallow areas have been reduced. Because ofthe villagization policies, production wasconcentrated on the flatter areas aroundthe new nucleus settlement, but it latermoved back to the hill plateaus.Meanwhile, increasingly new potentials(heavy and fertile soils) were explored astechnological adoption took place. Thevegetation cover has been characterized byoscillation dependent on how agriculturaland grazing potentials were realized. Ageneral decrease in land cover, includingdeforestation of hillslopes, has been a factsince the end of the 1970s, followed byboth reactivation and formation of new gul-lies, sheet erosion, and increased floodincidence in the nucleus settlement. Thus,the development of relatively scarce landhas implied increased use of marginal andrisk-prone land.

Table 4

Land Access and Livelihood Strategies

Accumulating farmersCultivated Land Uncultivated Land Borrowed or Rented Purchased Land

(Hectares) (Hectares) Land (%) (%)

Peasant-laborers (N = 10)

2.27 0.41 10 30

Peasants (N = 11)

2.27 0.57 54 09

Accumulating farmers (N = 12)a

4.62 1.22 61 46

Source: Village survey (1995).a One farmer has been left out of this analysis, as he owned 171 hectares of land that were fallow.


Differential Access to and Use ofNatural Resources

The study has identified a variety ofstrategies in relation to access to and useand management of natural resources.These strategies are related to the differ-ences in natural resource potentials. Thesepotentials result, as we have shown, fromthe spatial location and production historyof the resources in question. Historically,one way in which economic growth hastaken place in Ikuwala is by moving theagricultural frontier. This process contin-ues at present, in the sense that new areasare still drawn into cultivation in the hillsand along the swamp. Moreover, it is stillpossible to rent relatively fertile land innearby villages. The accumulating farmersthus have the option of renting out fieldsthat are exhausted from long periods ofcultivation and to hire or buy land of betterquality. They have access to hired laborand technology (oxen and plough), andthey are in a position to take risks withtomato production, which provides highreturns. They use commercial inputs butdo not add manure to any great extent.Thus, accumulating farmers have a rangeof opportunities without needing to ven-ture into illegal charcoal production, butthey do not have a strong incentive toembark on conservation measures, as theycan still obtain land with good productionpotential. Their maize cultivation is quitelow, but still higher than the other groups.A number of factors could explain thesedifferences in yields, among which are theaccumulating farmers’ more fertile soil andbetter access to labor.

In contrast, the peasant-laborers are in acoping situation. To the extent that theycan borrow, hire, or buy small pieces ofland to compensate for their exhaustedplots, these new plots may be located faraway, and they are not able to hire labor.Land along the foothills, used by the fami-lies before villagization, is increasinglyclaimed with reference to customary law,but cultivation of this land, along with moremarginal plots along streams, gullies, or on

steeper slopes, accelerates runoff and ero-sion. In general, land productivity for thisgroup is extremely low, and practices offertility management are almost absent. Anincrease in yields per land unit would makea difference, especially in terms of foodsecurity, but peasant-laborers do not com-mand the means to embark independentlyon conservation or productivity-enhancingefforts.

The middle group—the peasants—use avariety of strategies for the utilization ofnatural resources. They undertake variousmeasures of fertility management accord-ing to their access to resources (manure,labor parties, and small amounts of hiredlabor and chemical fertilizer). More thanhalf of the group borrow or rent land. Itwas also in this group that farmers movedback into the hills as a strategy of “market-aversion,” favoring access to more fertilesoil and water rather than according prior-ity to the market. Even this group ismarked by low returns per hectare, andfarmers within the group suffer some ofthe same constraints as the peasant-labor-ers in terms of land and labor access, albeitto a different degree.

The productivity of the exhausted landclose to the settlement and of the newlycultivated clay soil near the swamp differedby at least a factor of 4. Restoring fertilityto reasonable levels on the land around thesettlement would demand major soil con-servation and nutrient managementefforts, but within the respective house-holds resources do not seem to be availableand/or prioritized.

Livelihood, Income Strategies, andLand Degradation

The livelihood approach provides amethodological framework for understand-ing, for example, how natural resourcemanagement integrates with managementof other resources in a process of structuraldiversification. In Ikuwala, it became clearthat livelihood diversification to an impor-tant degree takes place within the agricul-tural sector and with a strong rural core.


What emerges from the examples above isthe relatively modest monetary importanceof noncrop incomes in comparison withcrop incomes. For all livelihood categories,the agricultural economy has thus evolvedas the most important sector in Ikuwala, towhich other income opportunities relate.This is also underscored by the importancegiven to agricultural factors in the choice ofwealth indicators. In Ikuwala, incomediversification among the poorest groups isa means of remaining in control of land.

All livelihood groups were keen to holdon to their lands and to their positions asfarmers. The income obtained from bothagricultural and nonagricultural sourcesprovides good reasons for this priority, asindicated in the case material. The accu-mulating households thus embarked onnonagricultural income sources such ascrafts, businesses, or rents to expand theirlevel of agricultural production and possi-bly also to obtain new land. The peasantgroup concentrated nonagricultural earn-ings in crafts and piecework, but incomesfrom these sources were mostly at a levelthat could only supplement the agriculturalearnings. In the peasant-laborer group,monetary income from nonagriculturalsources such as piecework, natural prod-ucts sales, or beer brewing often surpassedthat from agriculture, but agricultural pro-duction was nevertheless considered themain income source for food security rea-sons and for supplying grain for beer brew-ing. Among the two latter groups, liveli-hood diversification seemed moredefensive than offensive—that is, itseemed a component in maintaining livingstandards rather than in changing them—and for the peasant-laborers livelihooddiversification was to a considerable degreeconditioned by external demand.

The production and marketing of maize,tomatoes, and, more recently, the growingdemand for sunflower as a result of theestablishment of a new oil mill in Iringaimplies that even peasant households inIkuwala have managed to increase theirincomes during a period when naturalresources have been degraded. Thus, it

seems that the main avenues for obtainingwealth are agricultural, either by develop-ing tomato production or through the con-trol of more or new land. Conservation ofland belongs neither to the strategies ofaccumulation nor of survival.

The losers in this process were thosehouseholds unable to benefit from newaccess to land or from tomato cultivation.Nearly one-third of the households inIkuwala were impoverished and in no posi-tion to manage resources properly or to fol-low medium- or long-term strategies inorder to change their lot. Even some mem-bers of the peasant group were in an inse-cure and vulnerable situation. However,poverty is hardly the main explanation forland degradation. The land that is mostheavily degraded is or has been cultivatedby the accumulating farmers as well as byother groups.

Intensification and degradation thusoccur simultaneously. Even income growthamong accumulating, and sometimesamong peasant, households coincides withdegradation. In this situation, the incen-tives to adopt conservation strategies seeminsufficient.

The Livelihood Approach

From the foregoing discussion, it shouldbe evident that the ways in which liveli-hood strategies relate to natural resourcepotentials at the subcommunity level areimportant in explaining agricultural andenvironmental change. The bottom-upperspective on resource endowment is oneimportant advantage of a livelihoodapproach—here allowing the identificationof various livelihood and managementstrategies within a framework of localissues. Studies of livelihood strategies takeas axiomatic the heterogeneity of localfarming and other income-generating prac-tices, thereby allowing the possibility ofuncovering diversified and possibly contra-dictory strategies with respect to resourcemanagement and maintenance.

So far, however, there is no commonapproach to the application of livelihood


strategies. The one applied here has beendeveloped while working on the data, andour purpose has not been to present a blue-print, but rather to present one possibleway in which the approach can be applied.The approach is thus susceptible to differ-ent interpretations, and this can make gen-eralizations and comparisons regardingregions difficult. Further methodologicaldevelopment is warranted, and based onour experience two issues are of particularimportance.

First, heterogeneity is addressed differ-ently in various livelihood studies. The wayin which it has been addressed here, byestablishing fairly rigid livelihood groups,carries the risk of homogenizing and essen-tializing resource management practicesthat should be treated in a less rigid fash-ion. On the other hand, we would arguethat the groups identified in Ikuwala comeclose to capturing important managementpractices in the area. Second, the questionremains of what is at the core of the liveli-hood approach. We would argue that thelivelihood approach seeks to elaborate onmore than just income strategies. It seeksto gain an understanding of resourceaccess, use, and allocation and on the wayin which individuals and householderstransform resources into livelihoods.However, while this may be seen as thecore of the approach, it also problematizesthe context and the opportunities to whichhouseholders and individuals respond. Theapproach therefore represents an opportu-nity for gaining a more systematic approachto bottom-up studies and the interfacesbetween actors and structure.

ReferencesAdams, W. M., and Mortimore, M. J. 1997.

Agricultural intensification and flexibility inthe Nigerian Sahel. Geographical Journal163:150–60.

Aune, J. B. 1997. Structural adjustment and soildegradation in Tanzania. Discussion Papers189, Statistics Norway. Ås: NorwegianUniversity of Agricultural Science.

Bagachwa, M. S. D., and Limbu, F., eds. 1995.Policy reform and the environment in

Tanzania. Dar es Salaam: Dar es SalaamUniversity Press.

Bebbington, A. 1999. Capitals and capabilities:A framework for peasant viability, rural liveli-hoods and poverty. World Development27:2021–45.

Birch-Thomsen, T. 1993. Effects of land-useintensification—Introduction of animal trac-tion into farming systems of different intensi-ties. Geografica Hafniensia A2(Publications—Institute of Geography,University of Copenhagen).

———. 1996. Environment, people and landuse. SASA Notes and Working Papers, May1996. Centre for Research on SustainableAgriculture in Semi-Arid Africa, Copenhagenand Roskilde.

Birch-Thomsen, T., and Frederiksen, P. 1995.Report on fieldwork in Iringa District,Tanzania. SASA Notes and Working Papers,January 1995. Centre for Research onSustainable Agriculture in Semi-Arid Africa,Copenhagen and Roskilde.

Blaikie, P. 1985. The political economy of soilerosion in developing countries. New York:Longman.

Blaikie, P., and Brookfield, H., eds. 1987. Landdegradation and society. New York:Methuen.

Boesen, J., and Ravnborg, H. M. 1992. Peasantproduction in Iringa District, Tanzania.CDR Project Paper 93.1. Copenhagen:Centre for Development Research.

Branner-Jespersen, C. 1971. Southern high-lands socio-economic study. Final report.Uyole Agricultural Research Station, Mbeya.

Bryceson, D. F. 1999. African rural labour,income diversification and livelihoodapproaches: A long-term development per-spective. Review of African PoliticalEconomy 80:171–89.

Campbell, B. 1996. The miombo in transition:Woodlands and welfare in Africa. Center forInternational Forestry Research (CIFOR),Bogor, Indonesia.

Carney, D., ed. 1998. Sustainable rural liveli-hoods: What contributions can we make?London: Department for InternationalDevelopment.

Christiansson, C.; Dahlberg, A.; Loiske, M.-V.;and Östberg, W., eds. 1993. Environment-users-scholars: Exploring interfaces.Stockholm: EDSU, School of Geography,Stockholm University.

Dahlberg, A. 1994. Contesting views and chang-ing paradigms. The land degradation debate


in Southern Africa. Discussion Paper No. 6.Uppsala: Nordiska Afrikainstitutet.

Davies, S. 1996. Adaptable livelihoods. Copingwith food insecurity in the Malian Sahel.Chichester: Wiley.

Ellis, F. 1998. Household strategies and rurallivelihood diversification. Journal ofDevelopment Studies 35 (1):1–38.

Ellis, J. E., and Swift, D. M. 1988. Stability ofAfrican pastoral ecosystems: Alternate para-digms and implication for development.Journal of Range Management 41:450–59.

European Economic Commission (EEC). 1986.Regional agricultural development plan,Iringa Region, final report. Joint report of theUnited Republic of Tanzania and theEuropean Development Fund. Essen,Germany: Agrar-und-Hydrotechnik GMBH.

Fairhead, J., and Leach, M. 1996. Misreadingthe African landscape: Society and ecology ina forest-savanna mosaic. Cambridge:Cambridge University Press.

Feldman, R. 1983. Ismani: Agricultural changeand the politics of ujamaa. Monographs inDevelopment Studies No. 10. Norwich:School of Development Studies, Universityof East Anglia.

Francis, E. 1998. Gender and rural livelihoodsin Kenya. Journal of Development Studies 35(2):72–95.

Frederiksen, P. 1998. Environmental change—a case study from Ikuwala. Paper presentedat workshop on “Land Use Intensification inSemi-arid Iringa—Its Causes, Forms andConsequences,” 6 October, in Iringa,Tanzania, organized by Centre for Researchon Sustainable Agriculture in Semi-aridAfrica (SASA).

Joekes, S.; Heyzer, N.; Oniang’o, R.; and Salles,V. 1994. Gender, environment and popula-tion. In Development and environment:Sustaining people and nature, ed. D. Ghai,137–65. Oxford: Blackwell/UNRISD.

Kikula, I. S. 1997. Policy implications on envi-ronment—The case of villagization inTanzania. Uppsala, Sweden: Nordic AfricaInstitute.

Kiwasila, H., and Odgaard, R. 1992. Socio-cul-tural aspects of forest management in theUdzungwa. Report prepared for Danida.Copenhagen: Centre for DevelopmentResearch.

Little, P. D. 1994. The social context of landdegradation (“desertification”) in dry regions.In Population and environment—Rethinkingthe debate, ed. L. Arizpe, M. P. Stone, and D.C. Major, 209–51. Oxford: Westview Press.

Loiske, M.-V. 1995. The village that vanished—the roots of erosion in a Tanzanian village.Meddelanden Series B 94. Department ofHuman Geography, Stockholm University.

Lulandala, G. D. 1994. Analysis of factorsaffecting agricultural productivity in selectedsubsistence farming villages in Tanzania.Ph.D. diss., University of Michigan, AnnArbor.

McIntire, J.; Bourzat, D.; and Pingali, P. 1992.Crop-livestock interaction in sub-SaharanAfrica. World Bank Regional and SectoralStudies. Washington, D.C.: World Bank.

Mortimore, M. 1995. Caring for the soil:Agricultural expansion, population growth,and natural resource degradation in theSahel. In The Sahel—Ethnobotany, agricul-tural and pastoral strategies, developmentaid strategies, ed. A. Reenberg and H. SecherMarcussen. Proceedings of the 7th DanishSahel Workshop, 4–6 January 1995. Sahel-Sudan Environmental Research Initiative(SEREIN) Occasional Paper No. 1.Copenhagen.

Mung’ong’o, C. G. 1995. Social processes andecology in the Kondoa Irangi Hills, centralTanzania. Ph.D. diss., Department ofHuman Geography, Stockholm University.

Murton, J. 1997. Sustainable livelihoods in mar-ginal African environments? The social andeconomic impacts of agricultural intensifica-tion in Makueni District, Kenya. Paper pre-sented to the ESRC Conference onSustainable Livelihoods in Marginal AfricanEnvironments, Sheffield University, 10–11April.

Ndangalasi, H. 1998. Off-field vegetationstudy—Ikuwala, Mkulula, Ilambilole andKalenga Villages, Iringa Rural District,Tanzania. SASA Notes and Working Papers,October 1998. Centre for Research onSustainable Agriculture in Semi-Arid Africa,Copenhagen and Roskilde.

Netting, R. McC. 1993. Smallholders, house-holders—Farm families and the ecology ofintensive, sustainable agriculture. Stanford:Stanford University Press.

Preston, D. 1992. Households and the analysisof their livelihood strategies. Working Paper92(11). School of Geography, University ofLeeds.

Raikes, P. 1986. Eating the carrot and wieldingthe stick: The agricultural sector in Tanzania.In Tanzania: Crisis and struggle for survival,ed. J. Boesen, K. Havnevik, J. Koponen, andR. Odgaard, 105–41. Uppsala: ScandinavianInstitute of African Studies.


Rasmussen, T. 1986. The Green Revolution inthe Southern Highlands. In Tanzania: Crisisand struggle for survival, ed. J. Boesen, K.Havnevik, J. Koponen, and R. Odgaard,191–205. Uppsala: Scandinavian Institute ofAfrican Studies.

Reardon, T. 1997. Using evidence of householdincome diversification to inform study of therural non-farm labour market in Africa.World Development 25:735–47.

Reardon, T., and Vosti, S. A. 1995. Linksbetween rural poverty and the environmentin developing countries: Asset categories andinvestment poverty. World Development23:1495–1507.

Reed, D., ed. 1996. Structural adjustment, theenvironment, and sustainable development.London: Earthscan.

Reitsma, H.; Dietz, T.; and de Haan, L., eds.1992. Coping with semi-aridity: How ruralpoor survive in dry-season environments.Amsterdam: University of Amsterdam.

Ruthenberg, H. 1980. Farming systems in thetropics. 3d ed. Oxford: Oxford UniversityPress.

Sano, H.-O. 1999. Enabling strategies and stateintervention: Major Livelihood problems insemi-arid villages in Iringa District,Tanzania. In Sustainable agriculture in semi-arid Tanzania, ed. J. Boesen, I. Kikula, andF. Maganga. Dar es Salaam: Dar es SalaamUniversity Press.

Scoones, I. 1997. The dynamics of soil fertilitychange: Historical perspectives on environ-mental transformation from Zimbabwe.Geographical Journal 163:161–69.

———. 1998. Sustainable rural livelihoods: Aframework for analysis. IDS Working Paper72. Brighton, Institute of DevelopmentStudies.

Secher Marcussen, H., ed. 1993. Institutionalissues in natural resources management.Occasional Paper No. 9. InternationalDevelopment Studies, University ofRoskilde.

Tanzania Rural Development Bank (TRDB).1976. Iringa Region, Tanzania. Integratedrural development proposals for the thirdfive-year plan, 1976–81. Prepared byOverseas Development Group, University ofEast Anglia. Dar es Salaam: UNDP andFAO.

Thomas, D. S. G., and Middleton, N. J. 1994.Desertification: Exploding the myth.Chichester: John Wiley.

Thomson, J. T. 1994. The role of the state versusthe community in governance and manage-ment of renewable natural resources. Anargument and Sahelian examples. OccasionalPaper no. 12. International DevelopmentStudies, University of Roskilde.

Tiffen, M.; Mortimore, M.; and Gichuki, F.1994. More people less erosion—Environmental recovery in Kenya.Chichester: John Wiley.

United Nations Development Programme(UNDP). 1999. Sustainable livelihoods.http://www.undp.org/sl/.

United Republic of Tanzania. 1969. 1967 popu-lation census, vol. 1, Statistics for enumera-tion areas. Dar es Salaam.

———. 1978. 1978 population census.Preliminary report. Dar es Salaam.

———. 1988. Bureau of Statistics: TanzaniaSensa 1988, Preliminary. Dar es Salaam.

Östberg, V. 1995. Land is coming up: TheBurunge of central Tanzania and their envi-ronments. Stockholm: Stockholm Studies inSocial Anthropology.

Øygård, R. 1997. Structural adjustment policiesand land degradation in Tanzania. Forum forDevelopment Studies, No. 1. Oslo.

ImagesUnited Republic of Tanzania. Department of

Surveys and Mapping. 1955–56. Aerial pho-tographs (Re 1912.112-14). Dar es Salaam.

———. 1977–78. Aerial photographs (Re1842_02-04). Dar es Salaam.

———. 1:50,000 map of Tanzania. Series Y742,sheet 215/2 and 216/1, edition 1-TSO.

U.S. Geological Survey. 1966. DeclassifiedCorono Satellite Photograph (B/W),November (DS 1037-2154DF023). EROSData Center, Sioux Falls.

Système Probatoire d’Observation de la Terre(SPOT). 1986. SPOT-HRV. Scene id.140/366 XS 860812; Band 1, 2, 3.

Landsat Thematic Mapper (TM). 1993. Sceneid. 168-0-065 931210; Band 2, 3, 4.

———. 1996. Scene id. 168-0-065 960828;Band 2, 3, 4.

a livelihood perspective on natural resource management

Documents