the political economy of land conflict in the - michigan state

The Political Economy of Land Conflict in theEastern Brazilian Amazon

Cynthia S. Simmons

Department of Geography, Central Michigan University

An important goal of regional development in the BrazilianAmazonwas to enhance social welfare and alleviate direpoverty in other parts of the country by providing land to the poor.Nevertheless, both poverty and landlessness havepersisted despite development policies that distributed billions of dollars on highway construction, loans, andoutright subsidies. Inequitable land distribution has been held as a prime factor in land conflict across the country.Although episodes of conflict over land are common in Brazilian history, this paper focuses on agrarian issues thatarose with the opening of the Amazon frontier in the 1970s. The paper presents a political economy approachthat considers the role of hierarchical forces interacting across spatial scales, in creating conditions ripe for landconflict at the local level. The premise is that theBrazilian government, intending to bring about economic and socialdevelopment, promoted contradictory strategies creating land scarcity. These strategies led to expansion of largeranching operations, creation of conservation units, and demarcation of indigenous reserves, which constrained thepool of land available for small farmer settlement. Empirical analysis employing regression and spatial statistics isused to test the proposed model, advancing previous efforts by applying spatial regression, incorporating improvedindicators of conflict and explanatory variables generated by a Geographic Information System (GIS). The findingsprovide support for some elements of the argument, demonstrating statistically significant relationships betweenland conflict and land concentration, cattle ranching, and road construction. Finally, a case study analysis ofa county in the heart of the land conflict zone is provided, illustrating the interaction of scalar forces, and thearticulation of land conflict at the local level. Key Words: Amazonian development, agrarian reform, social andenvironmental justice, conservation policy.

Land conflict was a significant driver of social,political, and environmental change throughoutthe 20th century (Homer-Dixon 1999; Carius

and Lietzmann 1999). Monumental political transforma-tions during this period, including the Chinese andRussian revolutions, were sparked in part by inequitabledistribution of land. The struggle for land has also beenidentified as a source of internal strife in countries such asBangladesh, the Philippines, South Africa, and Israeland the precipitate factor of war between Senegal andMauritania and El Salvador and Honduras. Even inthe present era of globalization, world cities, and rapidurbanization, land struggle persists. The 1980s and1990s witnessed revolutions springing from the ruralhinterlands of numerous Latin American nations, whererapid population growth, economic decline, and failingindustrial sectors left countless people disadvantaged(Brockett 1988, 1994; Booth 1991; LaFeber 1993). Inthis region of the world, pronounced land concentra-tion and entrenched power of land-based elites havelong been regarded as the cause of poverty, and desirefor agrarian reform continues to be a powerful sourceof political and social confrontation (Paige 1975, 1996;Tutino 1986; Seligson 1995; Diskin 1996; Mason 1998;Kay 2002).

This paper addresses land conflict in the BrazilianAmazon, better known for problems associated with de-forestation and environmental change. Land conflict is notnew to Brazil, and has accompanied the process of frontierexpansion that started more than 500 years ago with Por-tuguese settlement (Mueller, Alston, and Libecap 1994).Numerous accounts of early peasant resistance (e.g., theWarofCunudos in 1821, theRoncodeAbelha rebellion in1851, the Quebra-Quilos rebellion in 1874, the Contest-ado rebellion in 1912, and others) have pointed to themaldistribution of land.1 Other land-related conflictsoccurred during frontier expansion into Sao Paulo Statebeginning in the 19th century (Welch 1999; Bran-nstrom 2001), western Parana in the 1940s,Matto Grossoin the 1950s (see Velho 1972; Katzman 1977; Foweraker1981; Schmink and Wood 1992; Mueller, Alston, andLibecap 1994; Alston, Libecap, and Mueller 2000),and Southern Para at the turn of the 20th century asentrepreneurs started claiming large tracks of Brazil nut fo-rest. Avery recent account of land conflict garneringmuchinternational attention was the clash between the militarypolice and landless farmers in Eldorado do Carajas, Para,which led to 19 fatalities (New York Times, April 21, 1996).

Although episodes of conflict over land are common inBrazilian history andwidespread throughoutmany parts of

Annals of the Association of American Geographers, 94(1), 2004, pp. 183–206 r 2004 by Association of American GeographersInitial submission, January 2002; revised submission, March 2003; final acceptance, September 2003.

Published by Blackwell Publishing, 350 Main Street, Malden, MA 02148, and 9600 Garsington Road, Oxford OX4 2DQ, U.K.

the country today, this paper focuses on agrarian issuesthat arose with the opening of the Amazon frontier in the1970s. An important goal of regional development inAmazonia during this periodwas to enhance social welfareand alleviate dire poverty in other parts of the country byproviding land to the poor (Simmons2002).Nevertheless,both poverty and landlessness have persisted despitedevelopment policies that distributed billions of dollars onhighway construction, loans, and outright subsidies (Ianni1979; Mahar 1979; Moran 1983; Hecht 1985; Hecht,Norgaard, and Possio 1988; Hall 1987, 1989, 2000;Browder 1988; Browder and Godfrey 1997; Becker,Miranda, and Machado 1990; Foresta 1991; SchminkandWood 1992). It is the apparent contradiction betweenthese prodigious development efforts and the recalcitrantnature of the regional agrarian problem that motivatesthis work. While conflict in the Amazon is multifacetedand involves numerous actors (e.g., Amerindians, miners,loggers, ranchers, small farmers), this paper considers theplight of the rural landless and situates their currentstruggle within a political economy framework.

Many have addressed land conflict in Amazonia,attributing it to (1) the failure of land tenure institutions(Mueller et al. 1994; Alston, Libecap, and Schneider1995; Alston, Libecap, and Mueller 1997, 2000); (2) theconflictive nature of production and exchange relations(Foweraker 1981; Wood 1983; Schmink andWood 1992;Walker and Homma 1996; Walker 1999; Walker, Moran,and Anselin 2000); and (3) the contradictory nature ofAmazonian development efforts (see for example Mahar1979; Ianni 1979; Smith 1982; Hecht, Norgaard, andPossio 1988; Browder 1988; Hall 1989; Moran 1983;Becker, Miranda, and Machado 1990; Almeida 1992;Cleary 1993; Hecht and Cockburn 1989). The basicpremise of this paper regardingAmazonian conflict followsfrom the notion of development contradictions but isbroadened to include the literatures on spatial scales andenvironmental and resource scarcity. In essence, thispaper attempts to integrate thework that has been done todate and to provide a comprehensive understanding ofland struggle in light of a regional development processthat was influenced by global, national, regional, and localcircumstances.

This task is accomplished as follows. First, a theoreticaldiscussion considers the environmental security andresource scarcity positions and examines the diverse workaddressing Amazonian frontier development. An analy-tical framework incorporating this literature provides anoverview of the political economy of land conflict,illustrating the scalar interaction of processes and forcesthat created intense land competition in the region. Adetailed description of land conflict in the Brazilian

Amazon is provided next, and the historical patterns ofland distribution are assessed. Within this discussion, thesocial and spatial dimensions of the struggle are elabo-rated, which involves a description of historical trends,current actors, and areas of intense conflict.

This is followed by an empirical analysis using regres-sion and spatial statistics to test the proposed analyticalframework. The analysis advances previous efforts (seeAlston, Libecap, and Mueller 1997) by applying spatialregression and by incorporating improved indicators ofconflict and several explanatory variables generated by aGeographic Information System (GIS). Although theregression analysis provides insight into some importantinfluential macroscale processes, it does not address theinteraction of scalar forces, and the articulation of landconflict outcomes at the local level. To address this, aspecific place in the heart of the present conflict zoneis considered, the county of Eldorado do Carajas. Thiscounty, located in the so-called South of Para, was thescene of the 1996massacre.While the discussion in gener-al treats the entire Brazilian Amazon region, the analysesfocus on Para State (see Figure 1), which was the site ofmuch early development interest and ongoing advanceof the agricultural frontier. The paper concludes with adiscussion of research findings and policy implications.

Theoretical Constructs of Land Conflict

This paper argues that land conflict in the EasternBrazilian Amazon results from power struggles over theregion’s abundant resources, which have been structuredand restructured within a multifaceted political economy.Central to this contention are two bodies of literature.One will be referred to as the resource access litera-ture,whichexamines linksbetweenviolent conflict and re-source competition. The second is a literature addressingland conflict in Amazonia in terms of frontier develop-ment processes. The analytical framework deployedconsiders the dynamics of political relationships and in-teractions of agents along a spectrum of spatial scales fromglobal to local.

Resource Access

The connection between land scarcity and conflict haslongbeen the focus of studies examining rural unrest in thedeveloping world (Moore 1969; Johnson 1968; Davies1971; Gurr 1971; Huntington 1971; Paige 1975, 1996;Prosterman 1976; Midlarsky 1982; Tutino 1986; Brockett1988, 1994; Homer-Dixon, Boutwell, and Rathjens 1993;Seligson 1995; Diskin 1996). A recent literature on so-called environmental security suggests that such scarcity

Simmons184

may directly generate internal as well as cross-borderfights, as groups vie for access and control over limitedresources (Ullmann 1983; Mathews 1989; Myers 1989;Brock 1991). Likewise, environmental degradation, cre-ated by overexploitation of resources, may lead to wor-sening socioeconomic conditions that precipitate conflict,and, as a result, represents an indirect cause of struggle.

Some contemporary examples of resource-scarcity warsinclude the fight for control over water between Egypt,Sudan, and Ethiopia, and cross-border disputes over oilbetween Iraq and Kuwait (Homer-Dixon and Blitt 1998;Serageldin 1995). The concern that there will be anoutbreak of wars instigated by struggle for scarce resourceshas garnered much attention by the international

ARGENTINA

PARAGUAY

BOLIVIA

CHILE

PERU

COLUMBIA

VENEZUELA

GUYANA

SURINAMEFRENCHGUYANA

Amapa

Amazonas Pará

Mato Grasso

Tocantins

Bahia

Maranhao

Piaui

Ceara

do NorteRio Grande

Paraiba

Pernambuco

Alagoas

Sergipe

SantoEsperito

Rio de Janeiro

Santa Catarina

du SulRio Grande

URAGUAY

Acre

du SolMato Grasso

Parana

Sao Paulo

Goias

FederalDistrito

Rondonia

Minas Gerais

Roraima

National Capital

Study Region (Pará)

Brazilian Amazon

Brazil

0 300 600 kilometers

AtlanticOcean

PacificOcean

Figure 1. Study Region: Para, State, Brazil.

Political Economy of Land Conflict in the Eastern Brazilian Amazon 185

community, and, consequently the environment became aprimary topic for national security under the ClintonAdministration (Westing 1989; Myers 1993; Levy 1995).Central to the environmental scarcity contention is theneo-Malthusian notion that population growth is the rootcause of environmental problems, and therefore a primary‘‘trigger’’ to violent reactions.

In contrast to the scarcity school of thought stands acritique based on empirical observations (Levy 1995; LeBillon 2001; Peluso and Watts 2001). Indeed, numerouscases provide compelling evidence that conflict oftenoccurs under circumstances of resource abundance, such asthe violent struggles that emerged as groups vied forcontrol over oil in Ecuador andNigeria (Watts 2001), goldin the Democratic Republic of Congo, and diamonds inSierra Leone and Liberia (Fairhead 2001).

The resource abundance argument posits that re-source-based wealth incites greed-driven conflict (Fair-head 2001). Primary commodities, especially high-valueresources such as oil and diamonds, are attractive to rulingelites, and, consequently, their capturemay spark strugglesbetween and within states for territorial control (Le Billon2001). In fact, much evidence suggests that countrieswhose economies are dependent on the export of primarygoods are more prone to political instability and conflict(Ross 1999). Finally, resource abundance is linked togovernments plagued with poor economic growth andinstability, creating greater vulnerability to conflict(Auty 2001; Ross 1999). What remains unclear withinboth the resource scarcity and resource abundancepositions is how political and economic forces interactingacross spatial scales, in general, structure resource access.A comprehensive account of conflict over resources, inthis case land, requires that it be situated within thehistoric, social, political, and economic circumstances ofspecific places.

Frontier Processes

In this regard, a substantial literature has emergedaddressing land conflict in the Brazilian Amazon. Thisliterature, while not expressly influenced by resourceaccess arguments, does consider competition for resourcesoccurring within the frontier, a region defined by abundantresources and weak institutions. Although interest in thefrontier as a sociological concept began with the work ofFredrick Jackson Turner at the turn of the 20th century,later work incorporated the frontier thesis into generaltheories of economic and political development (Taylor1972; Turner 1920).

In the Brazilian Amazon, land conflict has been attri-buted to immature tenure institutions (Mueller, Alston,

and Libecap 1994; Alston, Libecap, and Schneider 1995;Alston, Libecap, and Mueller 1997, 2000) and theconflictive nature of production and exchange relationsfound in places far from the long armof the law (Foweraker1981; Wood 1983; Schmink andWood 1992; Walker andHomma 1996; Walker, Moran, and Anselin 2000). Theinstitutional account focuses on the failure of land titlingto keep pacewith frontier expansion (Mueller,Alston, andLibecap 1994; Alston, Libecap, and Schneider 1995;Alston, Libecap, Mueller 1997), with conflict emerging asindividuals compete for access to valuable land, albeitwithdubious tenure arrangements. Other accounts of landconflict in Amazonia point to social structures and, inparticular, antagonistic relations between large landown-ers and peasant farmers (Foweraker 1981; Wood 1983;Schmink and Wood 1992; Walker and Homma 1996).Land conflict emerges here as class struggle betweenindividuals differentiated bymaterial wealth and ability tomobilize political power (Schmink and Wood 1992).

Land conflict has also been described in terms ofexploitive relations between capitalist and noncapitalisteconomies during the process of frontier expansion(Foweraker 1981; Wood 1983). In contrast to thetraditional notion of dualism frequently used to describethe structural disconnect between these economies, inthe Amazon region the two coexist as integral parts of thesame regional production system (Wood 1983). Violentstruggle has existed during every phase of frontierexpansion in Amazonia (Foweraker 1981), both in theearly stages when physical coercion was used to forcenomadic nut gatherers to extract resources (Velho 1972)and later as large-holders expelled squatters in the processof land appropriation and consolidation.

Besides these institutional and structural accounts,it has also been argued that land accumulation is abehavioral response tomarket forces (Walker andHomma1996; Walker, Moran, and Anselin 2000). In particular,cattle ranching benefits from scale economies and, as aresult, small and large ranchers have incentives to increaseherd size and expand pastures. In the Amazonian case,land concentration emerges on the basis of both the evolu-tion of small ranching operations and the exercise of socialpower by privileged groups. At some point, however, landaccumulation creates social discord that may result inlandconflict ( Johnson1968;Huntington1971;Gurr1971;Homer-Dixon 1999), as Walker (1999) demonstrates inViseu, one of the counties of greatest conflict in Para.

Analytical Framework

The analytical framework in this paper incorporatesthe theoretical contributions of the resource access and the

Simmons186

frontier literature and situates Amazonian land conflictwithin a political economy perspective, demonstrating thelink between an interaction of various interests acrossspatial scales and violent land conflict. This paper alsocontends that resource scarcity and abundance interactedin the Eastern Brazilian Amazon to create conditionsconducive of conflict. Early settlement in the region wastriggered by the discovery of abundant valuable resources,including rubber and brazil nuts. In addition, maldistribu-tion of land in the northeast and southern parts of Brazilcreated resource scarcity, which served as a push factorand resulted in a large displaced population attracted tothe Amazon because of seemingly abundant land. Build-ing on the foundations of the current frontier literature,this paper also suggests that land conflict emerged asinstitutional failures interacted with social forces andconditions to create land tenure insecurity and resourcedeprivation, an artificial form of scarcity. Although landabundance characterizes the frontier, it is the land madeaccessible by government road building that is the targetarea under contestation. Thus, in the Amazon, scarcityis not due to an absolute lack of land per se, but toland ownership patterns reflecting the relative power ofdifferent groups.

To reach a comprehensive understanding of landconflict arising from resource competition, it must berecognized as a site-specific phenomena deeply rooted inlocal histories and social relations, yet connected to largerprocesses of material transformation, political power, andhistorical conjuncture (Peluso andWatts 2001, 5). To thisend, the framework adopted uses a political economy ap-proach to situate land conflict within the complex social,economic, and political circumstance of the BrazilianAmazon, an essential first step to understanding suchstruggle (Homer-Dixon 1999).

In an attempt to dissect the factors influencing relativeland scarcity, the conceptual design examines the inter-action of numerous forces at the global, national, regional,and local levels (Massey 1984, 1994; Massey and Allen1984; Kodras 1997). At the local level, land scarcity is theprecipitate factor to conflict. Nevertheless, this scarcitymust be assessed from within the larger regional andnational contexts of land reform anddevelopment policiesthat reenforce power differentials among social groups.Moreover, global forces have clearly affected the nationalagenda for land distribution, development, and theenvironment in Brazil. The local articulation of theseforces, however, is more than a mere residue of exogenousfactors and is conditioned by endogenous, local responses(Cooke 1989). In turn, these local responses feed backinto the process by conditioning regional, national, andinternational actions and reactions.

This paper recognizes that the political and social rela-tionships between scales are not fixed, but are constantlyrenegotiated (Cox 1997; Swyngedouw 1997). In fact, inrecent years the apparently centralized and hierarchicalstructure of relationships in Brazil has given way to a newrealm of power relations. Since democratization in 1985, aplethora of nongovernmental organizations have emerg-ed on the political scene. And, slowly, state and localgovernments have been reasserting their political control.Perhaps the most important change has been the ‘‘glocal-ization’’ of the Amazon as grassroots social movementshave joined forces with international interests (Swynge-douw 1997). One poignant example that politicized the‘‘Amazonian problem’’ was the death in 1989 of ChicoMendes, a local syndicate leader who became an interna-tional symbol representing the plight of Amazonia’s ruralpoor (Hecht and Cockburn 1989). As a result of inter-national outcry over his murder, the Brazilian governmentplaced an estimated three million hectares of land inextractive reserves for the rubber-tapper community (ISA2000). Another example of the renegotiation of powerrelations having a dramatic impact on land distribution inthe region was the organization of Amerindian tribes into anational association, with links to International NGOs.The pressure asserted by this local/national/internationalcoalition in the 1980s directly influenced Brazil’s policyshift on indigenous rights as reflected in the 1988Constitu-tionand indirectly influenced the international donor com-munity to include indigenous concerns in fundingdecisions(Turner 1996; Moore and Lemos 1999; Simmons 2002).

Figure 2 illustrates the influences of interacting forcesacross geographic scales that have created conditions ofland scarcity and conflict in the study region. The basicargument is that conflicting national development poli-cies, influenced by global concerns and interregional effectsof development, have resulted in increased land concen-tration and resource scarcity within the region (Waller-stein 1979, 1984; Foweraker 1981; Branford and Glock1985; Hall 1989; Schmink and Wood 1992). At the locallevel such patterns of concentration can lead to conflict,although outcomes are influenced by local circumstancethat may mitigate (or exacerbate) conditions. Many havedescribed in elaborate detail the evolution of develop-ment in Amazonia since the mid-1950s (See for exampleMahar 1979; Smith 1982; Moran 1983; Hecht 1985;Browder 1988; Hall 1989; Foresta 1991; Santana 1997);consequently, this paper only touches on the scope ofdevelopment. The discussion that follows highlights thepredominate feature at each level of analysis.

Global Influences. Figure 2 indicates that at the globalscale conflicting international interest in economic


development, on the one hand, and concern for indige-nous and environmental protection, on the other, influ-enced the pattern of land distribution in the BrazilianAmazon (Simmons 2002). In the 1960s and early 1970s,donor agencies and the international business community(e.g., theWorldBank, Inter-AmericanDevelopmentBank,and Food and Agriculture Organization [FAO], amongothers) were interested in the Third-World livestocksector, and, as a result, invested heavily in cattle ranching.In the late 1970s and early 1980s, development interestshifted to the exploitation of mineral resources, parti-cularly exploration of the Carajas Mountains in easternPara. Investments for these endeavors came from a widerange of international actors, such as U.S. Steel, Japan,the former USSR, the EEC, World Bank, and many U.S.financial institutions (Hall 1989; Hurrell 1992).

In the mid- to late 1980s, the environmental crisisbrought on by alarming rates of deforestation inAmazoniabecame a focal point in the international environmental

debate, and influenced the course of development andenvironmental policy in Brazil (Wetterberg, Pranceand Lovejoy 1981; Plumwood and Routley 1982; Ledec1985; Pallemaerts 1986; Hurrell 1992; Thacher 1992). Ina similar fashion, international interests began to considerthe rights of indigenous peoples to land (Turner 1996;Moore and Lemos 1999). As a result, many Amazoniandevelopment projects have an environmental and indi-genous rights focus, and much international aid has beentied to programs that emphasize appropriate remedies(e.g., the G7 Pilot Program to Conserve the BrazilianRainforest financed by the Rainforest Trust Fund withcontributions from the G7 nations, Germany, the WorldBank, and the United Nations Environmental Program).

National Scale. Aside from international pressure,Figure 2 suggests that conflicting national interestsinfluenced the course of development. Three importantconcerns for the Brazilian government during the past

GlobalLevel

Economic Development(i.e., livestock, mining)

EnvironmentalConcerns

Indigenous Rights

NationalLevel

RegionalLevel

Local Level

Agro-IndustrialPrograms

Land Reform Conservation/Indigenous Policy

Large LandholdingsEconomic Reserves

Spontaneous In-migration/Small Farmer Settlement

Conservation/Indigenous Reserves

ResourceScarcity/Competition Intraregional Level

Agricultural ModernizationDrought in the NE

Social and PoliticalMobilization

Conflict?

Figure 2. Political economy of land con-flict in the eastern Brazilian Amazon.Source: Figure modified from Figure 2 inSimmons (2002, 244).

Simmons188

four decades have been to secure the national borders,increase economic growth, and appease the growinglandless population. Geopolitical concerns necessitatedthe settlement of the Amazonian frontier to secure thenation’s borders (Foresta 1992). The rhetoric surroundingsocial goals involved agrarian reformmeasures, indigenousrights to land, and environmental protection. Conse-quently, colonization schemes in the Amazon were elabo-rated, social and environmental legislation was enacted,and vast tracts of land were designated for indigenousreserves and conservation areas. At the same time,programs emphasizing economic development providedfiscal incentives, tax exemptions, and subsidized credit forland acquisition, agricultural expansion, and mineralexploitation (Browder 1988; Hall 1989; Mahar 1979).Although later economic incentive programs emphasizedthe inclusion of small farmers, development in Amazoniaremained focused primarily on the exploitation of naturalresources and has served primarily to benefit large com-mercial interests (See Santana et al. 1997; Toni 1999).

Interregional Scale. Figure 2 indicates that events andcircumstances in other parts of the country influenced theprocess of development in the study region. In particular,agricultural modernization in the Northeast resulted inthe displacement ofmany farm laborers and the creationofa landless peasantry (Aguiar 1979; Fearnside 1989). Thistransformation, combined with the northeastern droughtin the early 1970s, led to intensified out-migration to bothurban centers in theSouth and theAmazon (SchminkandWood 1992). The plight of the Nordestinos, or North-easterners, forced the government to take action onagrarian reform. Part of their solution was populationresettlement and infrastructural development in theAmazon designed to provide land and job opportunities(Browder 1988; Browder and Godfrey 1997; Hall 1987,1989; Santana et al. 1997). To this day, Northeasternimmigrants continue their move into the Amazon region(Perz 2000; Fearnside 2001).

Regional Development and Land Allocation. In gen-eral, as the figure reflects, national development strategieshave conflicted, with intended priorities vacillating be-tween commercial interests on the one hand and populistconcerns on the other. The overwhelming recipient of aidduring this period was the commercial sector. From 1960to 1994 cattle herds in Para grew by an estimated 792%,and pastures expanded twofold to 75,000 km2 in 1995(IBGE 1996). In addition, as of 1991, more than40,000 km2 of land was set aside for government-ownedeconomic activities such as the Grand Carajas Program(PGC) and hydroelectric operations (IDESP 1992). An

additional concern was national security (Foresta 1992),and, consequently, nearly 70,000 km2 was designated forthe military (IDESP 1992).

Aclear consequence of the emphasis on environmentaland indigenous concerns in the late 1980s and 1990s hasbeen expansion of natural conservation units and in-digenous reserves. As of 1999, over 110,000 km2 of land inthe State of Para had been set aside for natural areasconservation (ISA 1999), and currently, there are 41indigenous reserves encompassing nearly 310,000 km,2 anarea greater than 23% of the state (ISA 2000). Althoughland has been declared for natural areas conservation, theactual fate of the environment is of concern to manywho are skeptical of Brazil’s capability and willingness toprotect these areas (Pallemaerts 1986; Guimaraes 1991).Likewise, the security and integrity of indigenous landsare questionable, and it is estimated that some formof encroachment (i.e., small farmers, miners, loggers) hasimpacted 80% of the indigenous reserve areas.2

By way of contrast, the small-farmer sector receivedcomparatively little benefit from development efforts.Although landwasmade available along the developmenthighways for small farmers, and the roads opened theregion tomillions of spontaneous colonists (Walker 2003),the majority of colonization projects remain isolated, lackessential infrastructure, rural credit, and technologicalassistance, and suffer from land titling disputes (O Liberal1997). Despite the expressed intent of developmentefforts to promote agrarian reform, land concentrationpatterns in other parts of the country appear to have beenreplicated in the Brazilian Amazon (Almeida 1992;McCracken et al. 1998). Contrary to earlier positionsthat land accumulation was unsustainable without gov-ernment subsidies (Cleary 1993; Hecht and Cockburn1993), land accumulation involving small, medium,and large operations has continued (Walker, Moran, andAnselin 2000).

In general, land concentration at the national level hasworsenedover the years, as indicated by the increase of thegini-coefficient from .836 in 1967 to .843 in 1998 (SeeTable 1). The gini-coefficient gauges equality in thedistribution of resources, in this case land, with valuesranging from 0, perfect equality, to 1, complete inequality.A comparison of land distribution data in Table 1 showthat conditions have improved slightly in most parts ofthe countrywith the exceptionof theNortheast,where thegini-coefficient increased from .809 to .811 over the sameperiod. The North, representing six Amazonian states,has witnessed some improvement since 1978, yet itsinequality measure of .871 in 1998 is greater thanelsewhere in the nation. In Para, the gini-coefficientindicates that land concentration increased from .871 in


1967 to .889 as of 1998. The recent agricultural censusshows that nearly 18% of the land in the state is inagriculture. Of this area, an estimated 51% is in holdingsgreater than 1,000 hectares, and nearly half of that land isin extremely large holdings greater than 10,000 hectares(IBGE 1996). To the contrary, small farmers with less than100-hectare plots, representing 82% of all establishments,occupy a limited 19% of the agricultural land.

Although the Brazilian Amazon is commonly consid-ered an agricultural frontier, the rapid pace of urbanizationhas possible implications for land conflict in the region. Infact, land concentration and consolidation has initiatedthe movement of small farmers from the countryside tonearby urban centers, and many recent immigrants seek-ing land have instead opted to stay in frontier cities.Indeed, urban population in the Amazon increased from35.5% in 1970 to 55% in 1991, with more than one-thirdof the population made up of migrants (Perz 2000).However, a substantial number of urban dwellers are stillconnected to the countryside through employment inrural extractive activities (Simmons and Perz, unpub-lished manuscript), and some see the city as a temporaryplace to live as they wait for opportunities enabling theirreturn to the land (Becker 1996).

Local Factors and Circumstances. The final compo-nent of the analysis depicted in Figure 2 suggests that thearticulation of interacting forces across geographic scalesresults in land scarcity at the local level. Such scarcity wascreated, in part, by the expansion of large ranchingoperations, which, due to its land demanding productionsystem, frequently resulted in land concentration (Walkerand Homma 1996; Faminow 1998; Walker, Moran, andAnselin 2000). Ranching may also aggravate conditionsfor the rural poor because of the limited job opportunitiesit generates (Mattos and Uhl 1994; Faminow 1998).Intensifying these effects is the creation of conservation

units and the demarcation of indigenous reserves, whichfurther reduce the supply of land available for small farmersettlement.As the case of land conflict illustrates, landlessfarmers do not always passively accept their loss of accessto land, and they often react to these circumstances.

The actual outbreak of conflict, however, is influencedby the presence and activism of social movements organ-ized around the issue of land reform. In Para, influentialorganizations involved in these efforts include theCommissao Pastoral da Terra (Pastoral Land Commission[CPT]), the Sindicato dos Trabalhadores Rurais (RuralWorkers Syndicate [STR]), the Federacao dos Trabalha-dores da Agricultura do Estado do Para e Amapa (Federationof Rural Agricultural Workers of Para and Amapa[FETAGRI]), and the Movimento de Trabalhadores RuraisSem Terra (Movement of the Landless Rural Workers[MST]). Each of these will now be considered in turn.

Organizational Responses at the Local Level. TheCPT, formally established in 1975 and linked withthe BrazilianNational Conference of Bishops, actively sup-ports rural workers, their organizations, and their move-ment for agrarian reform and economic development.Since its inception, the CPT has monitored land struggle,collected details of conflict, and reported victims on anannual basis to the national and international community.The STRs, gaining independence from the corporatiststate in the 1970s with CPTsupport, provide local, state,and national representation for small farmers, share-croppers, permanent and temporary employees, and daylabor (Toni 1999). This group, and its state representationFETAGRI, have organized to improve conditions for ruralworkers on important issues such as access to agriculturalcredit, infrastructure development and improvement (i.e.,roads and schools), wages and benefits, and agrarianreform (Toni 1999). In fact, these rural worker unionshave orchestrated numerous occupations of large proper-ties by poor individuals and farmers in the State of Para(Fearnside 2001).

A particularly powerful social force in Brazil at thepresent time is theMST,whichwas officially established in1984 when various leaders of independent landlessmovements joined forces. Although originally based inSouthern Brazil, the MST has become a nationwideorganization with representatives from 22 of Brazil’s 27states and an organizational structure extending toregional and local levels (Covert 1998). In 1989 theMST began to coordinate efforts in Para with assistancefrom the CPT and the STRs, and today there are threedefined operational regions concentrated in the south-eastern part of the state and one in the vicinity of Belem,the state capital (Fernandes 2000; Campos 2002).

Table 1. Regional Comparison of Land Distribution in Brazil:Gini-Coefficient Indices 1967 to 1998

Geographic Unit 1967 1972 1978 1992 1998

Brazil .836 .837 .854 .831 .843North .882 .889 .898 .878 .871Northeast .809 .799 .819 .792 .811Southeast .763 .754 .765 .749 .757South .722 .706 .701 .705 .712Centerwest .833 .842 .831 .797 .798

Source: Hoffman (1998).

Note: The data for 1967 came from the Brazilian Institute for AgrarianReform

(IBRA), whose responsibility for land reform later fell within the National

Institute of Colonization and Agrarian Reform (INCRA) charge. Data for the

remaining years came from INCRA.

Simmons190

Approximately 60% of participants in the MST aredisplaced farmers and farm laborers, unemployed manualworkers, and disillusioned civil servants (Kelson 1997).

The main objective of the MST is to mobilize thelandless and force government to follow through on landreform promises (See Kelson 1997; Stedile 1997; Fer-nandes 2000). To do so, the MSTorganizes encampmentsonpublic landwith around1,000 to1,500 landless familiesand assists the community in selection of target propertiesfor ‘‘land invasions,’’ usually large private holdings.Immediately following such an invasion, MST officialspetition the local and federal government for legalizationof squatters’ claims, and for food and supply provisions.Frequently, property owners, usually living elsewhere, arenot aware that the invasion has occurred until the camp isestablished and a legal process is underway.Often, govern-ment intervenes by purchasing landdeemedunproductiveand distributing it to the claimants. However, when inter-vention is not forthcoming, the occupation may evolveinto confrontation and violence, involving, among others,fazendeiros (ranchers), pistoleiros (gunmen), peaoes (un-skilled wage laborers), and posseiros (squatters).

An additional player in the land struggle is the UniaoDemocratica Ruralista (UDR), an association of largelandownerswhohaveunited in defense of private propertyrights. The UDR was established in 1985 in the state ofGoias, but today has representation across the country,including Para. Although the official objective for orga-nizing was to defend the rights of rural producers, somesuggest the actual motivation was to stop the MST’saggressive pursuit of land reform through the use ofpolitical power and repression (Payne 2000). At theextreme, this group has allegedly hired private militia toassassinate MST leaders and expel landless families frominvasion sites (Covert 1998).

At the local level the outcome of competition for landmay be greatly influenced by the presence and activism ofthese groups. In areas where there is inequality in strengthof organization, domination of one group over the othermay result in preemptive clashes. When organizations onboth sides are well organized, the strength in numbersmayraise the stakes of violent action and reaction, and, con-sequently, may provide incentives for nonviolent negotia-tion. The low land-conflict-related mortality rates in thelate 1990s may suggest overall effective political mobiliza-tion on both sides, despite extreme events such as themassacre in Eldorado do Carajas (Simmons et al. 2002).

In sum, the conceptual design reflects the manner bywhich international interests influenced the direction ofnational development, which attempted to address social,environmental, and economic problems. Nevertheless, atthe regional level, conflicting strategies resulted in inequi-

table allocation of development benefits, and, in somecases, land scarcity. At the local level, resulting land con-flict may be mitigated or exacerbated depending on thestrength and activism of numerous social organizationswith local, national, and international representation.

Land Conflict in Para State

Geography of Land Conflict

In the Brazilian case, land conflict is well defined byhuman rights groups, such as the CPT, to include verbalconfrontation, vandalism, expulsion, physical alterca-tions, and murders (Barata 1995), and such clashes havebeen pervasive. In 1996, a year for which there is anelaborate data set and conflict census, there were areported 398 land invasions involving 63,080 families inrural Brazil; of these invasions, 759 incidences of conflictemerged, resulting in 54 murders (Table 2). The greatestdegree of struggle appears to be occurring in the North-east, with 40% of Brazil’s cases of land invasion and 34%of land conflict incidents.TheSouthby far has the greatestnumber of people involved, with 54% of nationwideparticipation. However, violent conflict is most dramaticin the North, with 72% of the conflict-related fatalities.The occurrence of conflict in this region is concentratedin the State of Para, with 85% of the regional and 61% ofBrazil’s land conflict-related murders.

Two primary sources of land conflict data exist for Para.One is Carajas: A Guerra dos Mapas by Alfredo WagnerBerno de Almeida (1994), and the other, Inventario daViolencia: Crime e Impunidade no Campo Paraense byRonaldo Barata (1995). Almeida (1994), using dataderived from the CPT, provides the number of land-conflict-related murders at the county level annuallybetween 1964 and 1992.3 The CPT data result frominterviews conducted by CPT representatives of indivi-duals involved in specific conflict events (CPT 1992).Accordingly, therewere 565politicalmurders in the entirestate during this period. As Figure 3 reveals, land conflicthas intensified since the late 1960s with peak violenceoccurring in themid- to late 1980s. Figures 4a,b, and c givethe geography of land conflict in Para and show thegreatest incidence in the southeast corner of the state, inthe counties Santana do Araguaia, Sao Felix do Xingu,and Xinguara, a region commonly referred to as the‘‘South of Para.’’ Violence declines northward alongthe eastern boundary of the state, and is least intense in thenorth-center and northwest.

Barata (1995), using data derived from local news-papers, documented 581 land-conflict-related deathsbetween 1980 and 1989. Figure 5 shows that land conflict


and relatedmurders during this time period run parallel toAlmeida’s record, with rates peaking in themid-1980s andtapering off at the end of the decade. The spatial pattern,as with the Almeida data, reveals greatest violence in thesouthern part of the state and least in the north and west(See Figures 4a,b and c).

Both Almeida (1994) and Barata (1995) record land-conflict-related murders committed to make politicalpoints, and should not be confused with homicide rates(Veja 1996). Also, although the spatial pattern for bothof these data sources is similar, time span and number ofdeaths reported vary significantly. The Almeida recordappears to be seriously underreported, which may be dueto the fact that the CPT, the data source, has little or noformal presence in parts of the state. One example ofunderreporting is the exclusion of casualties from the so-called War of Araguaia in southern Para, involving aconfrontation between the military police and an urban-based Maoist group from Sao Paulo that used land reformas the basis for peasant mobilization. This incident

resulted in the death of all 70 activist guerillas and an un-known number of local residents (IstoE 2000).

The Human Side of Violence

Adisaggregation of detailed accounts provided by bothAlmeida (1994) and Barata (1995) allows for an assess-ment of the individuals involved in the Amazonianconflict. On first glance the struggle may appear to be aclear-cut battle between large landowners and landlesspeasantry; to the contrary, the actual myriad of actorsinvolved is quite complex. In discussions and publishedaccounts of the Brazilian conflict, the landless farmeris identified as a posseiro (squatter), colono (colonist),pequeno produtor (small producer), lavrador (agriculturist),and trabalhador rural (rural worker). In accounts ofconflict, terms are often used interchangeably to describethe same victim, despite the distinct definitional differ-ences, as outlined in Table 3. In addition to the farmer,

Table 3. Definitions for Landless Farmers

Name Definition

Posseiro someone without legal title to the land they occupyColono person whomigrated into the region from another part

of the countryPequenoprodutor

small producer using traditional technology and familylabor

Lavrador small farmer focused on agriculture as opposed toranching

Trabalhadorrural

agricultural worker thatmay ormay not work for wages

Source: Personal interview with MSTrepresentative.

Table 2. Land Conflicts in Brazil, 1996

Region

Land Invasions Families Involved Incidents of Conflict Participants Deaths

No. % No. % No. % No. % No. %

North: 53 13% 6,898 11% 114 15% 67,726 7% 39 72%Para 33 8% 5,215 8% 72 9% 42,121 5% 33 61%Acre 3 110 2 550 0Amapa 0 0 2 490 0Amazonas 1 24 7 9,455 5Rondonia 7 1,245 16 9,290 1Roraima 0 0 1 1,690 0Tocantins 9 304 14 4,130 0

Northeast 157 40% n/a 256 34% 179,960 19% 9 17%Southeast 68 17% n/a 142 19% 89,206 10% 0 0%South 52 13% n/a 85 11% 507,560 54% 0 0%Centerwest 68 17% n/a 153 21% 90,716 10% 6 11%

Brazil 398 – 100% 63,080 – 100% 759 – 100% 935,134 – 100% 54 – 100%

Source: CPT, 1996.

Note: The murder count for Para includes the 19 peasants murdered in Eldorado do Carajas on April 19,1996; excluding this incident, Para would still have the

highest degree of land-conflict-related murders for the nation.

Barata Conflict Rate

Barata Assassination Rate

Almeida Assassination Rate

1964-69 1970-74 1975-79 1980-84 1985-89 1990-920

100

200

300

400

500

600

Time Period

Num

ber

of In

cide

nts

Figure 3. Land Conflict Trends, Almeida (1994) for 1964 to 1992and Barata (1995) for 1980 to 1989.

Simmons192

many advocates for the landless have been caught up inthe conflict, most notably, lawyers, clergymen, and ruralsyndicate leaders.

The actors involved in the conflict onbehalf of the largelanded interests are likewise various, and encompass largelandowners and their families, ranch managers, adminis-trators, employees, and hired gunmen. Included withinthe employee classification are peaoes, who are unskilled

ranch hands who frequently serve as bodyguards for theranchers. The position of the military and police is also acomplicated matter because it is believed there arefactions organized around both sides of the issue. Finally,there are a variety of other categories for victims includinggarimpeiros (gold-miners), pescadores (fishermen), and,tragically, children. Although not immediately recogniz-able as agents of conflict, the vast majority of gold minerslisted as victims had in fact squatted on land when theirmining efforts proved unsuccessful (Godfrey 1992).

Table 4 presents the disaggregation of theAlmeida databy victim, revealing that most were landless farmersaccounting for 67%, and those sympathetic to their cause,an additional 4%. Nevertheless, in the 1980s, there was adistinct rise in the number of casualties from the largelanded interests category. Of the 19% representing thisgroup, nearly three-fourths were ranch hands acting asbodyguards and hired gunmen for the ranchers. Finally, an

São Felix do XinguSantana do Araguaia

Xinguara

LowModerateHighExtreme

Figure 4a. Geography of land-conflict-related murders (Almeida1994).14


Xinguara


Figure 4c. Geography of land conflict incidents (Barata 1995).


Xinguara


Figure 4b. Geography of land-conflict-related murders (Barata1995).

80 81 82 83 84 85 86 87 88 89

0

50

150

100

200

Barata Conflict Rate

Barata Assassination Rate

Almeida Assassination Rate

Time Period

Num

ber

of In

cide

nts

Figure 5. Land conflict trends (Almeida 1994; Barata 1995) from1980 to 1989.


additional 10% were individuals not easily identified asactors in the land conflict. Most of these victims were goldminers ondisputed land (6%), followedby children caughtin the crossfire (2%).

Overall, the data presented show that violent landstruggle in Para has persisted since the late 1960s, andintensified after a remission in thewake of peak violence inthe late 1980s (Oliveira 1988; Americas Watch 1991;CPT 1992, 1995; Figueira 1992; IstoE May 19, 1993;Almeida 1994; Carvalho 1994; Barata 1995; Stedile1997). The spatial pattern of land conflict highlights thekey regions of violence commonly referred to as the Southof Para and Beak of the Parrot (Kotscho 1981; AmericasWatch 1991); these areas lie in the eastern part of thestate where Para, Maranhao, and Tocantins share borders.Notorious events here include the widely publicizedmurder of five homesteaders (O Liberal March 15, 1996)and the El Dorado do Carajas massacre of 19 peasants,both in 1996 (New York Times April 21, 1996; EconomistApril 13, 1996; IstoE April 24, 1996; O Liberal April andMay 1996; Veja April 24, 1996).

The account of land conflict here omits discussion ofthe War of Araguaia. Although clearly linked to the issueof agrarian reform, it may be regarded in some respects asan exotic political insurrection. It should also be noted

that althoughmost of the conflict is in Para, Rondonia andAcre have both experienced incidents. The best-knowncase, of course, was the assassination of ChicoMendes, aninternational symbol of social and ecological justice(Hecht and Cockburn 1989). Another publicized in-cident was the confrontation between police and 500squatter families in Rondonia that left 15 dead andover 100 wounded in 1995 (New York Times September19, 1995).

Research Methodology

As a means to empirically examine elements fromthe proposed analytical framework, this paper combinesquantitative techniques, in particular regression, withcase study narrative. A regression approach providesinsight into county-level factors driving conflict, whichcan be deductively linked to broad scale forces. Thereis precedence for employing such statistical methods instudies of land conflict. In one of the first applications ofspatial regression, Doreian (1980, 1981) demonstrated asignificant relationship between land concentration, ex-port economy, and insurgency during the Huk rebellionin the Philippines. Another study, by Alston, Libecap,and Mueller (1997), used tobit regression techniquesto examine the role of inefficient property institutionsin Amazonian land conflict. In light of these previousstudies, regression analyses, and spatial statistics are per-formed with data at the county level for the state of Para.Specifically, the regressions consider the link betweenpolicy-driven land scarcity and land conflict outcomes.Nevertheless, it is not possible to address hierarchicalinteractions across spatial scales with such a methodologyor explain the impact of local-level forces on conflictoutcomes. To complete the analysis, a case study ispresented, focusing on the county of Eldorado do Carajas,where confrontation between military police and landlesspeasantry in 1996 left 19 dead.

Regression Analyses

Following Doreian (1980, 1981) and Alston et al.(1997), a statistical model may be stated as:

Y ¼ Xbþ e

where Y is a vector of observations on land conflict (e.g.,assassinations, land conflict rates); X is a matrix ofindependent variables (e.g., land concentration, cattledensity) and important controls (e.g., population density);b is the regression coefficient vector; and e is a vector ofrandom disturbances (error terms). The research hypoth-esis is that consequences of conflicting policies, such as

Table 4. Land Conflict Victims, Almeida Data 1964 to 1992

Victims Count %

Landless Farmers 376 67%squatter 181 32%colonist 7 1%small producer 2 0%agriculturalist 49 9%rural worker 136 24%

Movement Activists 23 4%lawyer 3 1%syndicate leader 18 3%syndicate president 1 0%member of the clergy 1 0%

Large Landowners’ Interests 104 19%large land owner – rancher 7 1%ranch administrator/manager 10 2%employee of ranch 8 1%ranch hand – pawn 44 8%land swindler 2 0%hired gunmen 33 6%

Police 6 1%

Others 54 10%gold miner 36 6%business owner 4 1%fisherman 1 0%minor/youth 13 2%

Source: Almeida, 1994.

Simmons194

land concentration, will be statistically linked (in terms ofsignificant regression coefficients) to the various depen-dent variables reflecting land conflict.Theunit of observa-tion is the county as of 1985, which provides a sample sizeof 87 for the state of Para. Although there are currently inexcess of 140 counties in Para, data for analysis purposesfor each have been reaggregated to the 1985 boundaries tocorrespond to the period of analysis.4

The regression models presented here incorporatelessons learned from Doreian (1980, 1981) and Alston,Libecap, and Mueller (1997), and they include expandedvariables allowing for improved explanatory potential.The approach begins with ordinary least squares (OLS) inorder to test for spatial autocorrelation and correct for it asnecessary. Since studies byDoreian (1980, 1981) revealedthat violent conflict is subject to this pathology, it isessential to perform spatial diagnostics and to correctestimators as necessary, using the method of maximumlikelihood. In addition to spatial diagnostics, tests forconformity to normality assumptions revealed that thedependent variables were nonnormally distributed. Con-sequently, power transformations were performed on thedependent variables prior to regression analysis.5 Finally,tobit regressions were conducted, similar to that per-formed by Alston, Libecap, and Mueller (1997), due tothe fact that the values of the dependent variables arealways positive.6

Model Variables

A significant contribution of the present analysis is thatit uses three land conflict indicators derived from both theAlmeida (1994) and Barata (1995) data. Alston, Libecap,and Mueller (1997) limit their analysis to the CPT data,which seems to underestimate land-conflict-relatedassassinations. The dependent variables used here in-clude: (1) land conflict (conflicts between 1980 and 1989divided by rural population); (2) violence intensity-B(murders between 1980 and 1989 divided by rural popula-tion) publishedbyBarata (1995); and (3) violence intensity-A (CPT-documented murders between 1964 and 1992divided by rural population) as reported in Almeida(1994). Descriptive statistics for the model variables(Table 5) show that the violence intensity indicators havesimilar values despite the differing time spans, withaverages ranging from 15 to 17 deaths per 100,000 people,minimums of 0, and maximum death rates of 251 and261, respectively. The conflict intensity indicator hadan average of 25 conflicts per 100,000 people, a minimumof 0, and a maximum of 226.

The frontier variable presented in this analysis isdifferent from that ofAlston, Libecap, andMueller (1997),

who use percent land area held by squatters from the 1985IBGE Agricultural Census. Instead, this paper uses a roadindicator to represent the frontier environment; this indi-cator is derived from visual inspection of a regional high-way map, which should be highly correlated with percentsquatters, rising land values, and INCRA jurisdiction.

The road indicator is a categorical variable reflectingwhether one of the major development highways passesthrough the county. The roads considered in this analysisare (1) the Belem-Brasilia Highway, BR-10, designed toconnect the capital of the State of Para with the FederalCapital, Brasilia; (2) the Cuiaba-Santarem Highway, BR-163, in the western portion of the state; (3) the Trans-Amazon Highway, BR-230, traversing the Amazon basineast to west; (4) the PA 150, linking the southeast part ofthe statewith the northeast; and (5)BR-316,which servesas the major artery connecting Para to the northeast(See Figure 6).7 Overall, about 23% of the counties haveroad access.

Additional independent variables are included toprovide a more nuanced explanation of land conflictderived from the analytical framework presented. Inparticular, a cattle density indicator (number of cattle perkm2) is included to detect direct effects of ranching, thesector receiving the bulk of development aid. As sugges-ted, the cattle economy may exacerbate conflict bydepressing the local labor market. The cattle density datacame from the 1990 Annual Statistics for Para published bythe Institute for Socio-Economic Development of Para.On average there were 11 head per km2, with a minimumof .973 and maximum of 100.

Two land allocationvariables, land concentration (percentcounty in holdings greater than 10,000 hectares) and land

Table 5. Descriptive Statistics

Variables MeanStandardDeviation Minimum Maximum

N5 87

Dependent Variables:Violence Intensity (A) 17 38 0.00 261Violence Intensity (B) 15 37 0.00 251Conflict Intensity 25 43 0.00 226

Explanatory Variables:Cattle Density 11 15 0.973 100Land Concentration 8 15 0.00 95Land Distribution .52 .16 .405 .953Indigenous Land 6 15 0.00 75Conservation Area 14 29 0.00 100

Control Variables:Population Density 24 45 3.53 357Frontier - Road 23 77 0.00 1.00


distribution (gini-coefficient) calculated from the IBGE1996Agricultural Census, are used to gauge the impact ofland concentration and distribution on land conflict. Twoindicators are used because the land concentrationindicator does not reveal anything about the allotmentof properties less than 10,000 hectares.Although a countyhas a large proportion of land in holdings greater than10,000 hectares, the remainder of the area may be evenlydistributed among its population. For this reason, gini-coefficients were calculated at county scale and used asland distribution indicators for the analyses.8 The de-scriptive statistics in Table 5 show that on average landconcentration is 8% and land distribution is .52. Theland concentration indicators range fromaminimumof 0%to a maximum of 95%, and the distribution measuresfrom .405 to .953.

The indigenous reserve (as a percent of the county) andconservation land (as a percent of the county) variables re-

present the percentage of a county dedicated to these pur-poses and, consequently, imply a reduction in the supply ofland available for small farmer settlement. These indica-tors were created from a GIS with coverages representingthe spatial extent of Indigenous Reserves reported byFUNAI 1999 and State and Federal Conservation Unitsreported by IBAMA 1999 (See Figure 6). A countyboundary coverage for the State of Para was layered withthe reserve coverages, which enabled the calculation ofland area in reserve for each county. On average, 6% of acounty’s land was classified as indigenous reserves with avariation across the study region from 0% to an estimated75%. Conservation areas averaged about 14 % of acounty’s land, and ranged from no land to almost 100% incertain counties.9

Finally, the population density (number of people perkm2) is included as a control variable, accounting for anyneo-Malthusian effects. This variable was calculated from

Santarem

Ruropolis

Altamira

Belem

Maraba

Indigenous Lands

State Conservation Unit

Federal Conservation Unit

Unpaved Federal Highway

Paved Federal Highway

Planned Federal Highways

0 100 200 Kilometers

´

Figure 6. Spatial patterns of developmenthighways, conservation units, and indi-genous reserves.

Simmons196

population and county land size data reported in the 1990Annual Statistics for Para by IDESP.The population densityaveraged 24 people per km2, with a minimum of 3.53 anda maximum of 357.

Regression and Spatial Statistics Results

The regression strategy was to apply ordinary leastsquares, test for normality of the dependent variables, andthenperformdiagnostics for spatial autocorrelation. In theevent that nonnormality was detected, Box-Cox powertransformations were performed on the dependentvariables, and then regressions and tests for spatial auto-correlation were conducted. Subsequently, spatial regres-sions were performed as appropriate (See Doreian 1980,1981 for relevant applications).10 Table 6 (a–c) shows theregression results for all of the models tested and theircorresponding diagnostics. The tests for normality (Jarque-Bera) indicate that all three original models were non-normally distributed.Consequently, power transformationswere performed, and new OLS regressions run with thetransformed data. The spatial diagnostics reveal that allthree models, the violence intensity-A (log-transformed),violence intensity-B, and conflict intensity, are subject tospatial autocorrelation, as indicated by the appropriatesignificance probability for the Moran’s I statistic. Inaddition, the more highly significant results for theLagrange Multiplier–Lag statistic calls for the use of aspatial lag model (Anselin and Rey 1991).11 Also note,once power transformations were performed on each ofthe models the significance level of the Lagrange Multi-plier–Error diminished.

The lag model used for violence intensity-A (log-transformed), violence intensity-B, and conflict intensity, isspecified as:

Y ¼ kWYþ Xbþ e;

where Y, X, b, and e, are defined as before. The l intro-duces the autoregressive effect of the dependent variableon itself, and W is a spatial weights matrix (see Anselin1988, 1995; Anselin and Rey 1991). Note that the spatialweights were normalized to one for counties with sharedboundaries.

Table 6 (a–c) gives results of OLS and spatial lagregressions usingmaximum likelihood estimation for boththe original and log transformed data.Overall, as shown inTable 6 (a–c), the R2 statistics for all the models aresubstantial, with values ranging from 41% for the logtransformed conflict intensity model, to 71% for spatialregression on the original conflict intensity model. Ofgreater importance for this paper is the statisticalsignificance of the independent variables, each of which

represent an important element in the overall analyticalargument presented. The cattle density, land concentra-tion, and road variables proved to be robust, revealingstatistically significant results for all three models. Thedirection of the estimators suggests that increased landconcentration and road accessibility may result in in-creased land conflict. The cattle density variable, likewise,intimates that ranching may directly impact and increaseland conflict. Another statistically significant finding wasthe conservation variable in the violence intensity B andconflict intensity model. However, increased land inconservation appeared to reduce land conflict, which iscounter to the argument presented in the analyticalframework. In contrast, land distribution, indigenousland, and population density results were not statisticallysignificant for any of the models.

Finally, tobit regressions using maximum likelihoodestimation, similar to those performed byAlston, Libecap,andMueller (1997), were conducted. The tobit regressionis appropriate given that land conflict does not occur inevery county, so the value of the dependent variable isoften zero, creating a left censored data set. As with theabove models, the tobit results in Table 6(d) show thatcattle density, land concentration, conservation, androad, are all statistically significant. These results providefurther strength to the analytical argument presented inthe paper.

Eldorado do Carajas: The Local Articulation ofLand Conflict

The regressionanalyses demonstratea linkbetween roadaccess, land concentration, and cattle ranching, and theoccurrence of violent land conflict at the county level.Nevertheless, such modeling techniques only provideinsight into general processes at work at the regional scale.To fully understand violent struggles, it is necessary to focuson place, considering the spatiohistorical circumstancesthat created conditions conducive to conflict, and themanner by which endogenous and exogenous forces cametogether to mitigate or exacerbate violent outcomes. In aneffort to explore the articulation of land conflict in place,this section considers the historic economic, political, andsocial circumstances that led to the massacre at Eldorado.

The history of Eldorado do Carajas is the history of the‘‘South of Para,’’ a region endowedwith valuable resourcesand plagued with violent struggle. This region, as withmuch of theAmazon, remained isolated from the political,cultural, and economic core of Brazil due to its inacces-sibility until the turnof the20thcentury. Initial political andeconomic linkage to the area came in a series of boomand bust cycles tied to extractive resources, in particular


rubber and brazil nuts, that attractedmassive in-migrationand instigated resource-driven competition. To accom-modate land demands, the state provided long-term leaseson large tracts of forests for rubber extraction and lateramended its land legislation in 1920, and again in 1954,providing long term leases (aforamento perpetuo) on largetracts of brazil nut forest (Schmink and Wood 1993).The collapse of these economies freed labor who eithermigrated out of the region, or turned to subsistence agri-culture creating a substantial small-farmer population(Foweraker 1981).

In addition, steps taken to developAmazonia, initiatedwith the Brazilian Constitution of 1946, which called for a

Comprehensive Plan for development, brought radicalchanges to the South of Para. In particular, the construc-tion of the Belem-Brasilia highway that began in 1956opened the region, instigating a newwave of in-migration.As highway construction proceeded north from Brasiliathrough southern Para, large ranching interests followed,staking claim to vast tracts of land, converting forest topasture, and evicting small farmers in their path (Fower-aker 1981; Schmink and Wood 1992). In 1964, the statebegan construction of PA150, which provided addition-al access.

Development efforts were hastened under the militarygovernment’s initiative Operation Amazonia in 1966. The

Table 6a. Regression Results—Coefficients (Significance Level)

Violence Intensity (A)

Nontransformed Log Transform Data Log Transform Data

Estimation Procedure OLS OLS MLE-Spatial Lag

Entitlement:Cattle Density 1.71 (.00) 0.05 (.00) 0.04 (.00)Land Concentration 58.02 (.01) 1.68 (.18) 1.4 (.20)Land Distribution � 13.39 (.52) 0.01 (.99) � 0.04 (.96)Indigenous Land � 0.11 (.58) � 0.01 (.61) � 0.004 (.63)Conservation Area � 0.17 (.09) � 0.01 (.17) � 0.01 (1.7)Population Density � 0.05 (.39) � 0.00 (.69) � 0.00 (.91)Road 17.98 (.01) 0.9 (.02) .61 (.08)Jarque-Bera (Normality) 96.6 (.00) 5.36 (.07) n/aMoran’s I � 0.12 (.90) 1.34 (.17) n/aLagrange Multiplier—Error 0.24 (.61) 0.76 (.38) n/aLagrange Multiplier—Lag 1.20 (.27) 5.65 (.01) n/aR2 .59 .34 .37

Table 6b. Regression Results—Coefficients (Significance Level)

Violence Intensity (B)

Nontransformed Nontransformed Log Transform Data Log Transform Data

Estimation Procedure OLS MLE-Spatial Lag OLS MLE-Spatial Lag

Entitlement:Cattle Density .74 (.00) .22 (.25) 0.03 (.001) 0.02 (.01)Land Concentration 64.9 (.02) 53.1 (.01) 0.7 (.54) 0.46 (.66)Land Distribution 1.7 (.94) 8.12 (.69) 0.08 (.93) 0.21 (.82)Indigenous Land � 0.14 (.55) � .23 (.25) 0.01 (.26) 0.00 (.34)Conservation Area � 0.21 (.09) � .17 (.08) � 0.014 (.007) � 0.01 (.007)Population Density � 0.04 (.56) .02 (.70) � 0.00 (.24) � 0.00 (.27)Road 25.7 (.00) 15.7 (.02) 1.23 (.0009) 0.85 (.01)Jarque-Bera (Normality) 783.0 (.00) n/a 4.0 (.13) n/aMoran’s I 3.32 (.0008) n/a 2.01 (.04) n/aLagrange Multiplier—Error 7.43 (.006) n/a 2.25 (.13) n/aLagrange Multiplier—Lag 18.68 (.00001) n/a 7.52 (.006) n/aR2 .34 0.44 .37 .41

Simmons198

South of Para received the lion’s share, by far, ofgovernment investment in the form of SUDAM projects,which provided fiscal incentives to large ranching ven-tures, and facilitated land accumulation. In 1970, thegovernment’s National Integration Program (PIN) ex-panded road building with the construction of theTransAmazon Highway, BR230, which brought the poornortheastern part of the country into direct connectionwith the south of Para. Between 1960 and 1970 thepopulation in the region more than doubled, and landconcentration intensified.

In response to the social turmoil created by the influx ofpoor Northeasterners, numerous social movements, mostnotably the CPTand the STR, began to mobilize aroundthe plight of this new landless peasantry. A myriad ofactivists were attracted to their cause, from the ParaenseSociety for the Defense of Human Rights (SPDH) thatfounded an office in the south of Para in the early 1970s, to

a militant Maoist student group from Sao Paulo thatattempted to instigate a revolution in the countryside(Campos 2002). Largely in reaction to these activities, themilitary government declared the county of Maraba,which at the time included Eldorado do Carajas, a nation-al security zone, eradicated the communist elementin the region, and established the Executive Group forthe Araguauia—Tocantins Lands (GETAT) to defuse theviolent land conflict.12 Despite the military government’sabsolute authority over land distribution, critics suggestthat conditions for the landless worsened (Hall 1989;Schmink and Wood 1993).

The late 1970s witnessed a second social upheavaltriggeredbynewresourcediscoveries,namelymineral richesin the Carajas Mountains and gold in the Serra Pelada.The primary development effort of the 1980s was theGrand Carajas Program (PGC), the focus of which was alarge-scale iron-ore mine operated by the government-ownedCompanhiaVale doRioDoce (CVRD). To supportmining operations, the Carajas railway was constructed, a900 km line connecting the region to the port in Sao Luis,Maranhao, effectively providing a mode of transportationfor iron ore and immigrants to the region (Hall 1987).With the gold strike in Serra Pelada, a new onslaughtof immigration impacted the region, and a new source ofcontention emerged (Godfrey 1990, 1992; Cleary 1990).Miners descended on the region, and local residents, smallfarmer and large rancher alike, took up mining for goldfound in river beds throughout the Tocantins watershed.In 1983, there were an estimated 80,000 to 100,000min-ers, extracting nearly 1 metric ton of gold per month. Bythe late 1980s, 89 tons of gold had been extracted, anamount greater than that discovered in the 19th century

Table 6c. Regression Results—Coefficients (Significance Level)

Conflict Intensity

Nontransformed Nontransformed Log Transform Data

Estimation Procedure OLS MLE-Spatial Lag OLS

Entitlement:Cattle Density 1.76 (.000) 1.05 (.00) 0.04 (.00004)Land Concentration 54.3 (.03) 36.37 (.04) 0.08 (.94)Land Distribution 1.09 (.96) 7.04 (.67) 1.7 (1.2)Indigenous Land 0.09 (.67) .05 (.73) 0.01 (.25)Conservation Area � 0.27 (.01) � 0.18 (.03) � 0.01 (.03)Population Density � 0.01 (.87) .01 (.84) 0.00 (.27)Road 26.8 (.001) 11.89 (.04) 1.35 (.0004)Jarque-Bera (Normality) 43.9 (.000) n/a 4.4 (.10)Moran’s I 4.21 (.00002) n/a 0.72 (.47)Lagrange Multiplier—Error 12.61 (.0003) n/a 0.08 (.77)Lagrange Multiplier—Lag 36.01 (.000000) n/a 1.16 (.20)R2 .58 0.71 .41

Table 6d. Tobit Regression Results—Coefficients(Significance Level)

ViolenceIntensity (A)

ViolenceIntensity (B)

ConflictIntensity

Variables

Entitlement:Cattle Density 2.12 (.0001) 1.18 (.003) 2.0 (.0001)Land Concentration 89.25 (.01) 65.18 (.21) 55.6 (.07)Land Distribution � 2.16 (.95) 16.53 (.76) 32.1 (.29)Indigenous Land � 0.37 (.35) 0.11 (.23) 0.22 (.42)Conservation Area � 0.17 (.09) � 1.02 (.008) � 0.49 (.003)Population Density 0.03 (.71) � 0.32 (.23) 0.03 (.72)Road 33.29 (.002) 46.5 (.001) 38.2 (.0001)


gold rush in California and the Klondike (Schmink andWood 1993, 220).

Essentially, it was the discovery of mineral wealth thatbrought about the foundation of Eldorado do Carajas,named after the legendary ‘‘City of Gold.’’ Historically,several Amerindian tribes occupied the region that todaymakes up the county of Eldorado do Carajas. The decadeof the 1950s, however, witnessed gradual occupation as afew large ranchers established fazendas (large ranches)in the area, most notably the Fazenda Macaxeira. Thediscovery of gold and other minerals in the Carajasmountains placed Eldorado at a strategically importantspot, on the main transportation artery in the eastern partof the state, PA150, and the route to Serra Pelada. The po-pulation explosion that eventually led to the partitioningof the county of Curionopolis in 1988 from Maraba, andthe creation of the county of Eldorado do Carajas fromCurionopolis in 1991, stemmed from the in-migration oflaborers who worked on the Carajas railway, gold miners,and landless farmers arrivingbybus, rail, or foot alongoneofthe region’smain transportation links. In the 1980s, privateand government initiatives to provide formal land settle-ment in the vicinity further acted as a pull for immigrants.

As indicated by Figure 3, land conflict peaked inthe 1980s with violent land conflict concentrated in thesouth of Para. In the wave of political liberation inthe 1980s, which culminated in the return to democracyin 1985, political and socialmobilization accelerated (Toni1999). Important in the struggle for land was the growingmomentum of the syndicate movement, which foundednine branch offices in the vicinity of Eldorado do Carajas(Campos 2002). By 1989, the MSTofficially moved intothe Amazon, specifically the south of Para, and today twoof its four regional offices are located in the Araguaia areacorresponding to the counties of Maraba and Sao Joao doAraguaia, and theCarajas area encompassing the countiesof Parauapebas and Eldorado do Carajas. In conjunctionwith the CPTand STR activists, theMST began its effortsat spontaneous agrarian reform in earnest. The firstoccupation involving 100 families occurred in January1990 on Fazenda Inga in Conceicao doAraguaia, followedby a second involving 150 families on Fazenda Canaranalater that same year.

This growing landless movement did not go unnoticedby government authorities, and in 1991,military, civil, andfederal lawenforcement attempted toquell themovementby arresting the leadership. The State of Para also becameincreasingly concerned with MST actions, and in re-sponse, the Secretary of State Public Security (SEGUP)created, in 1995, the Delegation of Land Conflict withinthe Division of Investigations and Special Operations(DIOE) to investigate land invasions and intervene on

behalf of landowners (Fernandes 2000; Campos 2002). Inaddition, oppositional groups, and in particular, largelandowners began to mobilize under the UDR banner.Although there is no official UDR headquarters in Para,the county of Xinguara, just south of Eldorado do Carajas,is said to be the group’s stronghold and recruitment center(Campos 2002, 57).

In 1995, theMSTpetitioned INCRAto investigate theFazenda Macaxeira in Eldorado do Carajas, in an effort tosettle some 1,400 families in search of land, and undertheir leadership. The INCRA investigation concludedthat the ranch was in fact productive and instead offeredthe families land in a settlement project in Tucuruı. Un-satisfied with the outcome, 2,000 MST members began amarch onApril 10, 1996 to the capital, Belem.When theyarrived at Eldorado do Carajas on the 16th of April, a sitereferred to as the ‘‘curva do S’’ on PA150, they decided tostop and blockade the road in protest. In exchange forclearing the road, the government agreed to provide theprotestors with food and transport to the city Maraba,where they were to meet with the superintendent ofINCRA. The next morning word came down that thegovernment would not follow through on its promise, sothe protestors once again blocked the road. Within a fewhours, two military battalions arrived on the scene, onefromParauapebasandanother fromMaraba, and theeventsthat transpired next are amatter of great controversy. Theoutcome, however, is well documented—an estimated 19killed and 69 injured, including women and children.

This incident highlights the atmosphere of violencethat has emerged in the region following years of intenseresource competition influenced by the interplay of forcesfrom global to local scale. Eldorado do Carajas demon-strates how government development efforts increaseddemand for land in the region through road building andpopulation resettlement and decreased supply by provid-ing fiscal incentives to large ranching interests, and laterby placing substantial tracts of land into reserve. Thesegovernment actions, in turn, have been strongly shaped byglobal forces, including early interest in developing a cattleeconomy, and subsequent federal and state responsesto international concerns about the environment andindigenous rights (Simmons 2002).

Indeed, Maraba, the original county from whichEldorado do Carajas was partitioned in 1991, had anestimated 22% of its land in holdings greater 10,000hectares in 1985, and its land concentration increasedfrom a gini-coefficient of .67 in 1985 to .70 in 1996 (IBGE1996). Furthermore, nearly 36% of the county area is setaside for conservation and indigenous holdings. Althoughtimber and mining were once important economic sectorshere, these activities have pushed westward with resource

Simmons200

exhaustion, leaving cattle ranching as the primary econo-mic activity. Unfortunately, ranching provides very few jobsgiven its structure of production (Mattos and Uhl 1997).

In response to the worsening circumstances of the ruralpopulation, numerous social movements targeted theregion for mobilization (i.e., CPT, STR, MST), sparkingreaction by both state and federal government and byopposition groups acting on behalf of large holders (e.g.,UDR). Given the place-specific confluence of thesevarious social and political processes, it is not surprisingthat the south of Para has experienced a disproportionateshare of violent land conflict, culminating in the massacreat Eldorado do Carajas.13

Conclusions and Policy Implications

Rural unrest stemming from problems associated withagrarian reform has long been addressed by socialscientists (Moore 1969; Johnson 1968; Davies 1971; Gurr1971; Huntington 1971; Paige 1975, 1996; Prosterman1976;Midlarsky 1982; Tutino 1986; Brockett 1988, 1994;Homer-Dixon, Boutwell, and Rathjens 1993; Seligson1995; Diskin 1996; Mason 1998). More recently, othershave suggested that resource scarcity, such as that relatedto land concentration, presents serious concern fornational and international security (Myers 1989;Ullmann1983; Carius and Lietzmann 1999; Homer-Dixon 1999).In contrast to this position, critics argue that, in fact,resource abundance incites greed-driven conflict (Levy1995; Le Billon 2001; Peluso and Watts 2001; Fairhead2001). In the Brazilian context, land conflict has beenviewed as part and parcel of frontier expansion (see Velho1972; Katzman 1977), stemming from immature tenureinstitutions (Mueller, Alston, and Libecap 1994; Alston,Libecap, and Schneider 1995; Alston, Libecap, andMueller 1997, 2000) and the conflictive nature of pro-duction and exchange relations (Foweraker 1981; Wood1983; Schmink and Wood 1992; Walker and Homma1996; Walker, Moran, and Anselin 2000).

This paper attempts to integrate the literature on landconflict into a political economy approach that considersthe role of hierarchical forces interacting across spatialscales in creating conditions of relative land scarcity andconflict at the local level. The primary contention is thatresource scarcity and abundance interacted in the EasternBrazilian Amazon to create conditions conducive forconflict. Early settlement in the Brazilian Amazon wastriggered by its abundant resources, including rubber,brazil nuts, timber, gold, and land. Later, the Braziliangovernment, intending to bring about economic andsocial development, promoted strategies that led to land

conflict. The programs pursued, influenced by interna-tional pressure and biased toward corporate interests,promoted a course of development that created relativeland scarcity, a prime ingredient to any agrarian problem.On the one hand, the government encouraged in-migration by promoting colonization schemes and theconstruction of a massive highway network, therebyincreasing demand for land.On theother, it also decreasedthe supply by encouraging a land-hungry cattle economythat did little to create jobs.

Regression results provide support for this argument,showing that land concentration is statistically linked toland conflict and conflict-related murders. Likewise, theroad indicator was statistically significant in suggestingthat land accessibility, which stimulated increased de-mand, may have been a key factor to land conflict.Furthermore, results indicate that ranching had directimpacts on land conflict, in addition to its indirect contri-bution through land concentration. This link may berelated to the structure of the cattle economy, which doesnot provide significant employment alternatives for thelandless population. Consequently, conditions in areaswith extreme land concentration and substantial ranchingoperationsmay be difficult for the rural poor, inciting themto action.While a great deal of aidwent to ranching activi-ties, progress toward land reformandoverall improvementof social conditions in the frontier were not as significant.

The findings from the regressions, however, do notsupport the proposition that increased land under con-servation and indigenous reserves increase land conflict.In fact, results show land conflict is less than expected incounties with land under conservation. Although on firstglance this findingmay appear counterintuitive, a possibleexplanation is that most of the conservation units asshown in Figure 6, with the exception of the Tapirape-Aquiri and Carajas Forest reserve, are far removed fromthe areas of accessible land in the southeastern part of thestate. Nevertheless, incursions into conservation areas areoccurring, and frequently go undetected or underreported(Simmons 2002).

Unlike conservation units, many indigenous reservesare located near areas of small farmer settlement.Although land set aside for reserves affects land avail-ability at the county level, the regression results showedthat it had no impact on conflict between large land-holders and landless farmers. However, violent incursiononto indigenous lands has been considerable (Simmons2002). A particular area of intense conflict is the Parakanaindigenous reserve in Southern Para, which has beeninvadedby loggers, squatters,miners, fisherman, and large-landed-interests. In fact, in April of 1999 INCRA officialsdiscovered the existence of a 13,000-hectare ranchwithin


the reserve that had no registration records. Many otherindigenous areas lie directly beside main developmentroads and, consequently, have been subject to invasions aswell. For instance, along the TransAmazon highway theCachoeira Seca do Iriri, Bacaja Trincheira, and the Ararareserves have been severely impacted, and encroachmentis also occurring along the Santarem-Cuiaba Highwaysouth of Ruropolis on the Bau indigenous reserve of theKayapo. As with conservation units, land conflict onindigenous lands is not considered in the data source usedin this study. Nevertheless, it is estimated that 15% of the980,000 hectares in reserve have been invaded, andviolent conflict has been evident (CIMI 1997).

Although the regression analysis provides insight intosome influential macroscale processes, it does not addressthe interaction of scalar forces and the articulation of landconflict outcomes at the local level. The case study ofEldorado do Carajas, however, explores the historiceconomic, political, and social circumstances that led tothe massacre in April 1996. The tragic events that dayreflect an atmosphere of violence, culminating from yearsof intense resource competition that was greatly influ-enced by the interplay of forces from global to local scale.This region was the target of government developmentefforts throughout the past five decades, which, in effect,increased demand for land in the region through roadbuilding and population resettlement, and decreasedsupply by providing fiscal incentives to large ranchers.Such efforts were partly influenced by a global interest incattle. At the same time, the conditions suffered by therural poor became a rallying cry for social and politicalmobilization that engaged activists from across the spatialspectrum. Government and opposition groups’ responseto the landless movement intensified over the years,creating a volatile environment ripe for conflict.Althoughthis paper argues that local organizing may mitigate landconflict, the events in April 1996 demonstrate that theymay have, in fact, exacerbated circumstances that led tothe violent response.

Current policy rhetoric in Brazil calls for environment-ally, economically, and socially sustainable development.However, in order to meet such objectives, planners andpolicymakers must consider how individual componentsof an overall development strategy interact. Cattleranching exemplifies this problem well. Although it maylead to a certain level of economic development, ranchingmay not be consistent with agrarian reform goals. Thispoint is of particular importance in regions of ecologicalsignificance like the Amazon basin, given the migratoryresponses of dispossessed peasants, who are likely to seeknew land in the absence of local jobs. Many dispossessedpeasants, however, migrate to one of the many emerging

frontier towns. Despite the movement of people from thecountryside to the urban areas, the regional economy isstill agrarian based, and the employment opportunitieslimited. Consequently, a balanced development programmust consider opportunities for the rural population, notsimply access to land, but also income-generating activ-ities and overall improvement of conditions in the citiesin an effort to impede an urban exodus back to thecountryside. Although a comprehensive strategy fordeveloping the Amazon is beyond the scope of this paper,it does seem clear that successful sustainable developmentin Amazonia will have to bring about social, economic,and environmental justice.

Acknowledgments

I am especially indebted to Robert Walker for his helpwith the elaboration of the main concepts of this paper. Iwould also like to thank Stephen Perz for comments on anearly version of the manuscript, and Patrick O’Sullivanand Barney Warf for their support of the initial phase ofresearch. I am especially grateful to my Brazilian collea-gues PedroMourao deOliveira,Marcos Pedlowski, EraldoMatricardi,MarcellusCaldas, andEugenioArima for theirvaluable insight on the issues addressed. Finally, Severalanonymous reviewers contributed significantly to thepaper, although I remain responsible for remaining errors.

Notes

1. For accounts of theWar ofCanudos seeOs Sertoes byEuclidesda Cunha (1985); for the Ronco de Abelha and the Ouebra-Quilos rebellions see Crise Agraria e Luta de Classes byMonteiro (1980); and for the Contestado Rebellion seeMillenarian Vision, Capitalist Reality: Brazil’s ContestadoRebellion, 1912–1916, by Diacon (1991).

2. Many small farmers view indigenous reserves as governmentland and therefore available for occupation as allowed bythe federal constitution. This is especially true in light of the1997 passage of decree 1775 that opened reserve demarca-tion for contestation by interested parties.

3. The county, as a geographical unit, is similar to a county in theUnited States, although Brazilian counties can be substan-tially larger in terms of land area.

4. This includes data for the county of Eldorado do Carajas,which was created in 1991 from land within the county ofCurionopolis, which in turn was created in 1988 from thecounty of Maraba.

5. Data transformation involved adding a 1 to each value andthen performing a Box-Cox power transformation.

6. A tobit model is a regression model for left censored data.Since a significantnumber of observations hada zerovalue fortheir dependent variable, such an approach was appropriate.

7. The construction of the road variable involved the visualinspection of the map Sistema Rodo-Aeroportuario, preparedby theDepartamento de Estradas de Rodagem do Para, 1987.

Simmons202

8. The coefficientswere calculated by combining the 15 standardland size classifications in the census to5 categories: (1) 0 to49hectares; (2) 50 to 99 hectares; (3) 100 to 999 hectares; (4)1000 to 9,999 hectares; and (5) greater than 10,000 hectares.

9. Several counties onMarajo Island are completely under someform of conservation unit.

10. Spatial autocorrelation occurs when variables are correlatedwith themselves in space (Odland 1988). In the regressioncontext, problems arise when the dependent variables or theerror term are spatially autocorrelated. The first case is‘‘substantive’’ and the second is ‘‘nuisance’’ spatial auto-correlation. When either case occurs, coefficient estimatesfrom OLS are either biased or inefficient (Anselin and Rey1991).

11. Note in Tables 6a–c, once the power transformations wereperformed, the significance of the Lagrange Multiplier Errorfor all the models diminished.

12. The war of Araguaia, as it has been termed, was the militarysiege on the leftist guerrillas from Sao Paulo who used landreform as the basis for peasant mobilization. This group wasviewed as a threat to national security, and consequently all70 activists, and an undocumented number of peasants, werekilled (IstoE 2000).

13. The violence intensity (A [Almeida 1994]) is 66; violenceintensity (B [Barata 1995]) is 33; and the conflict intensity is61. The averages for these same variables are 17, 15, and 25,respectively.

14. The county boundaries are for 1997, but the data isaggregated to 1985. Consequently, the blank (white) regionsare those counties created since that time.

References

Adriance, M. C. 1995. Promised land: Base Christian communitiesand the struggle for the Amazon. Albany: State University ofNew York Press.

Aguiar, N., ed. 1979. The structure of Brazilian development. NewBrunswick, NJ: Transaction Books.

Almeida, A. L. O. 1992. Colonization of the Amazon. Austin:University of Texas Press.

Almeida, B. de A.W. 1994. Carajas: A Guerra dos Mapas. Belem:Falangola.

Alston, L. J., G. D. Libecap, and B. Mueller. 1997. Violence anddevelopment of property rights to land in the BrazilianAmazon. In The frontiers of the new institutional economics,ed. J. N. Drobak and V. C. Nye. San Diego: Academic Press,145–63.

Alston, L. J., G. D. Libecap, and B. Mueller. 2000. Land reformpolicies, the sources of violent conflict, and implications fordeforestation in the Brazilian Amazon. Journal of Environ-mental Economics and Management 39 (2): 16288.

Alston, L. J., G. D. Libecap, and R. Schneider. 1995. Propertyrights and the precondition for markets: The case of theAmazonian frontier. Journal of Institutional and TheoreticalEconomics 151 (1): 89–107.

AmericasWatch. 1991.Rural violence in Brazil. NewYork:HumanRights Watch.

Andrade, M. P. 1997. Chacinas e Massacres no Campo. Sao Luis:Mestrado em Polıticas Publicas.

Anselin, L. 1988. Spatial econometrics: Methods and models.Dordrecht: Kluwer Academic Publishers.

FFF. 1995. SpaceStat Version1.8 User’s Guide. Morgantown:Regional Research Institute, West Virginia University.

Anselin, L., and S. Rey. 1991. Properties of tests for spatialdependence in linear regression models. Geographical Ana-lysis 23:112–31.

Auty, R. 2001. Resource abundance and economic development.Oxford: Oxford University Press.

Barata, R. 1995. Inventorio da Violencia: Crime e Impunidade noCampo Paraense. Belem: Cejup.

Becker, B. K. 1996. Brazil’s frontier experience and sustainabledevelopment: A geopolitical approach. InFrontiers in regionaldevelopment, ed. Y. Gradus and H. Lithick, 73–99. Lanham,MD: Rowman & Littlefield Publishers, Inc.

Becker, B. K., M. Miranda, and L. Machado. 1990. FronteiraAmazonica: Questoes sobre a gestao do territorio. Rio de Janeiro,Brasil: Editora da UnB-Editora da UFRJ.

Bobrow-Strain, A. 2001. Between a ranch and a hard place:Violence, scarcity, and meaning in Chiapas, Mexico. InViolent environments, ed. N. L. Peluso and M.Watts, 155–88.Ithaca, NY: Cornell University Press.

Booth, J. A. 1991. Socioeconomic and political roots of nationalrevolts in Central America. Latin American Research Review26 (1): 33–74.

Branford, S., and P.Glock. 1985.The last frontier: Fighting over landin the Amazon. London: Zed Press.

Brannstrom,C. 2001. Producing possession: Labour, law, and landon the Brazilian agricultural frontier, 1920–1945. PoliticalGeography 20:859–83.

Brock, L. 1991. Peace through parks: The environment onthe peace research agenda. Journal of Peace Research 28 (4):407–23.

Brockett, C. D. 1988. Land, power, and poverty: Agrariantransformation and political conflict in Central America. Boston:Allen and Unwin.

FFF. 1994. El Salvador in the eighties: Counterinsurgency andrevolution. Latin American Research Review 29 (3): 174–87.

Browder, J. 1988. Public policy and deforestation in the BrazilianAmazon. In Public policies and the misuse of forest resources, ed.R. Repetto and M. Gillis, 247–98. Washington, DC: WorldResources Institute and Cambridge University Press.

Browder, J., and B. Godfrey. 1997. Rainforest cities: Urbanization,development, and globalization of the Brazilian Amazon. NewYork: Columbia University Press.

Campos, A. S. 2002. O Confronto em Eldorado do Carajas. Belem:Promev Grafica e Editora.

Carius, A., andK.M. Lietzmann, eds. 1999.Environmental changeand security: A European perspective. Berlin: Springer.

Carvalho, L. M. 1994. Contido a Bala: A vida e a morte dePaulo Fonteles, advogado de posseiros no Sul no Para. Belem:Cejup.

Cehelsky,M. 1979. Land reform in Brazil: The management of socialchange. Boulder, CO: Westview Press.

Cleary, D. 1990. Anatomy of the Amazon gold rush. London:MacMillan.

FFF. 1993. After the frontier: Problems with the politicaleconomy in the modern Brazilian Amazon. Journal of LatinAmerican Studies 25:331–49.

Comissao Pastoral da Terra (CPT) 1992. Conflitos No Campo.Belem: CPT.

FFF. 1995. Conflitos No Campo. Belem: CPT.FFF. 1996. Conflitos No Campo. Belem: CPT.Conselho Indigenista Missionario (CIMI). 1997. Newsletter No.

245, 1 January.


Cooke, P. 1989. The contested terrain of locality studies. InHuman geography: An essential anthology, eds. J. Agnew, D.Livingstone, and A. Rogers, 476–91. Oxford: Blackwell,1996.

Covert, H. H. 1998. Fighting for agrarian reform in Brazil: Ananalysis of the strategies and dialogue of the Sem Terramovement. Latin Americanist 33 (2): 1–6.

Cox, K. R., ed. 1997. Spaces of globalization: Reasserting the power ofthe local. New York: Guilford Press.

Cunha, E. d. 1985. Os Sertoes. Sao Paulo, SP: Brasiliense.Davies, J. C., ed. 1971. When men revolt and why. New York: Free

Press.Diacon, T. 1991. Millenarian vision, capitalist reality: Brazil’s

Contestado rebellion, 1912–1916. Durham, NC: Duke Uni-versity Press.

Diskin, M. 1996. Distilled conclusions: The disappearance of theagrarian question in El Salvador. Latin American ResearchReview 31 (2): 111–26.

Doreian, P. 1980. Linear models with spatially distributed data:Spatial disturbances or spatial effects? Sociological Methodsand Research 9 (1): 29–60.

FFF. 1981. Estimating linear models with spatially distri-buted data. In Sociological methodology, ed. S. Leinhardt,359–88. San Francisco: Jossey-Bass.

Economist, 1996. April 13, 38.Fairhead, J. 2001. International dimensions of conflict over

natural and environmental resources. In Violent environ-ments, ed. N. L. Peluso and M. Watts, 213–36. Ithaca, NY:Cornell University Press.

Faminow, M. D. 1998. Cattle, deforestation, and development in theAmazon: An economic, agronomic, and environmental perspec-tive. Oxford: CAB International.

Fearnside, P. M. 1989. Deforestation and agricultural develop-ment in the Brazilian Amazon. Interciencia 14 (6): 291–97.

FFF. 2001. Land-tenure issues as factors in environmentaldestruction in Brazilian Amazonia: The case of SouthernPara. World Development 29 (8): 1361–72.

Fernandes, B.M. 2000.A Formacao do MST No Brasil. Petropolis:Vozes.

Figueira, R. R. 1992. Rio Maria: Canto da Terra. Petropolis: Vozes.Foresta, R. A. 1991.Amazon conservation in the age of development:

The limits of providence. Gainesville: University of FloridaPress.

FFF. 1992. Amazonia and the politics of geopolitics. Geo-graphical Review 82 (2): 128–42.

Foweraker, J. 1981. The struggle for land: A political economy of thepioneer frontier in Brazil, 1930 to present. Cambridge: Cam-bridge University Press.

Godfrey, B. J. 1990. Boom towns of the Amazon. GeographicalReview 80 (2): 103–17.

FFF. 1992. Migration to the gold-mining frontier in BrazilianAmazonia. The Geographical Review 82 (4): 458–69.

Guimaraes, Roberto P. 1991. The ecopolitics of development in thethird world: Politics and environment in Brazil. London: LynneRienner Publishers.

Gurr, T. 1971.Why men rebel. Princeton, NJ: PrincetonUniversityPress.

Hall, A. L. 1987. Agrarian crisis in Brazilian Amazonia: TheGrande Carajas programme. Journal of Development Studies23 (4): 522–52.

FFF. 1989. Developing Amazonia: Deforestation and socialconflict in Brazil’s Carajas Programme. Manchester, U.K.:Manchester University Press.

FFF, ed. 2000. Amazonia at the crossroads: The challenge ofsustainable development. London: Institute of LatinAmericanStudies.

Hartmann, B. 2001. Will the circle be broken? A critique of theProject on Environment, Population, and Security. InViolentenvironments, ed. N. L. Peluso and M. Watts, 39–64. Ithaca,NY: Cornell University Press.

Hauge, W., and T. Ellingsen. 1998. Beyond environmentalscarcity: Causal pathways to conflict. Journal of PeaceResearch 35 (3): 299–317.

Hecht, S. 1985. Environment, development, and politics: Capitalaccumulation and the livestock sector in Eastern Amazonia.World Development 13 (6): 663–84.

FFF. 1992. Logic of livestock and deforestation: The case ofAmazonia. In Development or destruction: The conversion oftropical forest to pasture in Latin America, ed. T. E. Downing, S.Hecht, H. Pearson, and C. Garcia-Downing, 7–26. Boulder,CO: Westview Press.

Hecht, S., and A. Cockburn. 1989. Defenders of the Amazon:Deforestation is not just about ecological destruction butsocial and political injustice.New Statesman & Society 2 (55):16–20.

Hecht, S., R. B. Norgaard, and G. Possio. 1988. The economicsof cattle ranching in Eastern Amazonia. Interciencia 13 (5):233–40.

Hinchberger, B. 1998. Land of no return? Not Brazil. The Nation,March 2, 20–24.

Hoffman, R. 1998. A estrutura fundiaria no Brasil de acordo com ocadastro do INCRA: 1967 a 1998. Working Paper: INCRA/UNICAMP.

Holston, J. 1991. The misrule of law: Land and usurpation inBrazil. Comparative Studies in Society and History 33 (4):695–725.

Homer-Dixon, T. F. 1999. Environment, scarcity, and violence.Princeton, NJ: Princeton University Press.

Homer-Dixon, T. F., and J. Blitt, eds. 1998. Ecoviolence: Linksamong environment, population and security. Lanham, MD:Rowman & Littlefield.

Homer-Dixon, T. F., J. H. Boutwell, and G. Rathjens. 1993.Environmental change and violent conflict: Growing scar-cities of renewable resources can contribute to social inst-ability and civil strife. Scientific American (February): 16–24.

Huntington, S. 1971. Political order in changing docieties. NewHaven, CT: Yale University Press.

Hurrell, A. 1992. Brazil and the international politics ofAmazonian deforestation. In The international politics of theenvironment: Actors, interests, and institutions, ed. A. Hurrelland B. Kingsbury. Oxford: Clarendon Press.

Hurrell, A., and B. Kingsbury, eds. 1992. The international politicsof the environment: Actors, interests, and institutions. Oxford:Clarendon Press.

Ianni, O. 1979. Colonizacao e Contra-Reforma na Amazonia.Petropolis: Vozes.

Instituto Brasileiro de Geografia e Estatistica (IBGE) 1985.Agricultural Census of 1985. Rio de Janeiro: Secrataria dePlanejamento, Orcamento e Coordinacao.

FFF. 1996. Demographic Census of 1996. Rio de Janeiro:Secrataria de Planejamento, Orcamento e Coordinacao.

Instituto de Desenvolvimento Economico-Social do Para(IDESP), 1992. Governo Federal Nao Prioriza ReformaAgraria. Para Agrario, July–December (8).

FFF. 1990.Anuario Estatıstico do Estado Do Para, Vol. 11 (T-1).Belem: IDESP.

Simmons204

FFF. 1990.Anuario Estatıstico do Estado Do Para, Vol. 11 (T-2).Belem: IDESP.

Instituto SocioAmbiental (ISA). 1999. Newsletter, 11 March.FFF. 2000. Amazonia Brasileira 2000. Sao Paulo: Instituto

SocioAmbiental.IstoE, 1993. May 19, 55–67.FFF. 1996. April 24, 20–24.Johnson, C. 1968. Revolutionary change. London: London Uni-

versity Press.Katzman, M. 1977. Cities and frontiers in Brazil. Cambridge:

Harvard University Press.Kay, C. 2002. Reflection on rural violence in LatinAmerica.Third

World Quarterly 22 (5): 741–75.Kelson, G. C. 1997. Brazil up-close. The Brasilians, May, 7E-8E.Kodras, J. E. 1997. The changing map of American poverty in an

era of economic restructuring and political realignment.Economic Geography 73 (1): 67–93.

Kotscho, R. 1981. O Massacre dos Posseiros: Conflicto de Terras noAraguaia- Tocantins. Sao Paulo: Brasiliense.

LaFeber, W. 1993. Inevitable revolutions: The U.S. in CentralAmerica. 2nd ed. New York: Norton.

Langevin, M. S., and P. Rosset. 2000. Land reform from below:The landless workers movement in Brazil. http://www.mstbrazil.org/rosset.html.

Le Billon, P. 2001. The political ecology of war: Natural re-sources and armed conflicts. Political Geography 20 (5):561–84.

Ledec, George. 1985. The political economy of tropical defor-estation. In Divesting nature’s capital: The political economy ofenvironmental abuse in the Third World, ed.H. Jeffrey Leonard.New York: Holmes & Meier.

Levy, M. A. 1995. Is the environment a national security issue?International Security 20 (2): 35–62.

Mahar, D. J. 1979. Frontier development policy in Brazil: A study ofAmazonia. New York: Praeger Publishers.

Mason, D. T. 1998. Take 2 Acres and call me in the morning: Island reform a prescription for peasant unrest? Journal ofPolitics 60 (1): 199–231.

Massey, D. 1984. Spatial divisions of labor: Social structures and thegeography of production. New York: Methuen.

FFF. 1994.Place, space, and gender.Minneapolis: University ofMinnesota.

Massey,D., and J.Allen, eds. 1984.Geography matters. Cambridge:Cambridge University Press.

Mathews, J. T. 1989. Redefining security. Foreign Affairs 68 (2):162–77.

Mattos, M., and C. Uhl. 1994. Economic and ecologicalperspectives on ranching in the eastern Amazon. WorldDevelopment 22 (2): 145–58.

McCracken, S., E. Brondizio, D. Nelson, E. Moran, and A.Siqueira. 1998. The use of remote sensing and GIS in thecollection of survey data on households and land use. Proceed-ings of the IX Brazilian Remote Sensing Congress.

Midlarsky, M. I. 1982. Scarcity and inequality: Prologue to theonset of mass revolution. Journal of Conflict Resolution 26 (1):3–38.

Monteiro, H. M. 1980. Crise agraria e luta de classes: O Nordestebrasileiro entre 1850 c 1889. Brasılia, Distrito Federal, Brasil:Horizonte Editora.

Moore, B. 1969. The social origins of dictatorship and democracy.Harmondsworth, U.K.: Penguin.

Moore, S. G., andM. C. Lemos. 1999. Indigenous policy in Brazil:The development of decree 1775 and the proposed Raposa/

Serra do Sol Reserve, Roraima, Brasil. Human RightsQuarterly 21 (2): 1–19.

Moran, E. F. 1983. The dilemma of Amazonian development.Boulder, CO: Westview Press.

Mueller, B., L. J. Alston, and G. D. Libecap 1994. Land, pro-perty rights, and privatization in Brazil. Quarterly Review ofEconomics and Finance 34 (special issue, summer): 261–80.

Myers, N. 1989. Environment and security. Foreign Policy (74):23–41.

FFF. 1993.Ultimate security: The environmental basis of politicalstability. New York: Norton.

NACLA.1994.Report on the Americas28 (2): 5.NewYork:NACLA.New York Times, 1995. September 19, A6.FFF. 1996. April 21, 8Y.Odland, J. 1988. Spatial autocorrelation. Washington, DC: Annals

of the Association of American Geographers.O Liberal, March, April, and May 1996.O Liberal 1997.Oliveira, A. U. 1988. A Geografia das Lutas no Campo. Sao Paulo:

Contexto.Paige, J. M. 1975. Agrarian revolution. New York: Free Press.FFF. 1996.Land reformandagrarian revolution inElSalvador.

Latin American Research Review 31 (2): 127–39.Pallemaerts, M. 1986. Development, conservation, and in-

digenous rights in Brazil. Human Rights Quarterly 8 (3):374–400.

Payne, L. A. 2000. Uncivil movements: The armed right wing anddemocracy in Latin America. Baltimore, MD: Johns HopkinsUniversity Press.

Peluso, N. L., and M. Watts. 2001. Violent environments. Ithaca,NY: Cornell University Press.

Perz, S. G. 2000. The rural exodus in the context of economiccrisis, globalization and reform in Brazil. International Migra-tion Review 34 (3): 842–81.

Plumwood, and V. Routley. 1982. World rainforest destruction:The social factors. Ecologist 12 (4): 5–10.

Prosterman, R. 1976. IRI, A simplified predictive index of ruralinstability. Comparative Politics 8 (3): 339–54.

Roberts, J. 1992. Squatters and urban growth in Amazonia. TheGeographical Review 82 (2): 441–69.

Ross, M. L. 1999. The paradox of plenty: Oil booms and petro-states. World Politics 51 (2): 297–322.

Santana, A. C., J. L. Davila, M. I. R. Alencar, P. N. Mattar, R. F.Souza, and R. M. Queiroz da Costa. 1997. ReestruturacaoProdutiva e Desenvolvimento na Amazonia: Condicionantes ePerspectivas. Belem: BASA.

Schmink,M., andC.Wood. 1992.Contested frontiers in Amazonia.New York: Columbia University Press.

Seligson, M. 1995. Thirty Years of transformation in the agrarianstructure of El Salvador. Latin American Research Review 30(3): 43–74.

Serageldin, I. 1995. Earth faces water crisis. Press Release, August6. WDC: World Bank.

Simmons, C. S. 2002.Development spaces: The local articulationof conflicting development, Amerindian rights, and envi-ronmental policy in Eastern Amazonia. Professional Geogra-pher 54 (2): 241–58.

Simmons, C. S., S. Perz,M.A. Pedlowski, and L.G.T. Silva, 2002.The changing dynamics of land conflict in the BrazilianAmazon: The rural-urban complex and its environmentalimplications. Urban Ecosystems 6 (1/2): 99–121.

Smith, N. 1982. Rainforest corridors: The Transamazon colonizationscheme. Berkeley: University of California Press.


Smith, N., E. A. Serrao, P. T. Alvim, and I. C. Falesi. 1995.Amazonia: Resiliency and dynamism of the land and its people.New York: United Nations University Press.

Stedile, J. P. 1997. A reforma agraria a Luta do MST. Petropolis:Vozes.

Swyngedouw, E. 1997. Neither global nor local: ‘‘Glocalization’’and the politics of scale. In Spaces of globalization: Reassert-ing the power of the local, ed. K. R. Cox. New York: GuilfordPress.

FFF. 1999. Modernity and hybridity: Nature, Regeneracio-nismo, and the production of the Spanish eaterscape,1890–1930. Annals of the Association of American Geo-graphers 89 (3): 443–65.

Taylor, G. R. 1972. The Turner thesis: Concerning the role of theFrontier in American history. Lexington, MA: D. C. Heath.

Thacher, P. S. 1992. The role of the United Nations. In Theinternational politics of the environment: Actors, interests, andinstitutions, ed. A. Hurrell and B. Kingsbury. Oxford:Clarendon Press.

Toni, F. 1999. State-society relations on the agricultural frontier: Thestruggle for credit in the Transamazonica region. Ph.D. disserta-tion, University of Florida.

Turner, F. J. 1920.The frontier in American history. New York: Holt.Turner, T. 1996. Brazil: Indigenous rights vs. neoliberalism.Dissent

(summer): 67–69.Tutino, J. 1986. From insurrection to revolution in Mexico: Social

bases of agrarian violence 1750–1940. Princeton, NJ: Prince-ton University Press.

Ullmann, J. E. 1983. The arms-race and the decline of UnitedStates technology. Journal of Economic Issues 17 (2): 565–74.

Veja. 1996. April 24, 34–44.Velho, O. 1972. Frentes de Expansaao e Estructura Agraria. Rio de

Janeiro: Zahar.

Walker, R. T. 1999. The structure of uncultivated wilderness:Land use beyond the extensive margin. Journal of RegionalScience 39 (2): 387–410.

Walker, R. T. 2003. Mapping process to pattern in the landscapechange of theAmazonian frontier. Annals of theAssociationof American Geographers 93 (2): 376–98.

Walker, R. T., and A. Homma. 1996. Land use and land coverdynamics in the Brazilian Amazon: An overview. EcologicalEconomics 18:67–80.

Walker., R. T., E.Moran, and L. Anselin. 2000. Deforestation andcattle ranching in the Brazilian Amazon: External capitaland household process. Forthcoming in World Development28 (4).

Wallerstein, I. 1979. The capitalist world-economy. Cambridge:Cambridge University Press.

FFF. 1984. The politics of the world-economy: The states, themovements, and the civilizations. Cambridge: Cambridge Uni-versity Press.

Watts, M. 2001. Petro-violence: Community, extraction, andpolitical ecology of a mythic commodity. In Violent environ-ments, ed. N. L. Peluso and M. Watts, 189–212. Ithaca, NY:Cornell University Press.

Welch, C. 1999. The seed was planted: The Sao Paulo riot of Brazil’slabor movement, 1924–1964. University Park: PennsylvaniaState University Press.

Westing, A. H., ed. 1989. Comprehensive security for the Baltic: Anenvironmental approach. London: Sage Publications.

Wetterberg, G. B., G. T. Prance, and T. E. Lovejoy. 1981.Conservation progress in Amazonia. Parks 6 (1): 4–22.

Wood, C. 1983. Peasant and capitalist production in the BrazilianAmazon: A conceptual framework for the study of frontierexpansion. In The dilemma of Amazonian development, ed.E. Moran, 259–77. Boulder, CO: Westview Press.

Correspondence: Department of Geography, Central Michigan University, 276 Dow Science Building, Mount Pleasant, MI 48859, e-mail:[email protected].

Simmons206

the political economy of land conflict in the - michigan state

Documents