18
Bulletin of Latin American Research, Vol. 29, No. 4, pp. 459 – 476, 2010 Intraregional Migration, Direct Action Land Reform, and New Land Settlements in the Brazilian Amazon STEPHEN G. PERZ AND FLAVIA LEITE University of Florida, USA CYNTHIA SIMMONS, ROBERT WALKER AND STEPHEN ALDRICH Michigan State University, USA MARCELLUS CALDAS Kansas State University, Kansas, USA This article analyses migration histories of residents in rural settlements of the Brazilian Amazon that resulted from direct action land reform (DALR), which involves organised land occupations. Our analysis eval- uates two hypotheses. The ‘urban migration’ expectation asserts that urban experience is important for DALR participation, which links rural and urban areas via migration for land occupations. The ‘DALR effi- cacy’ hypothesis argues that migration and DALR are complementary livelihood strategies, such that participation in DALR obviates the need for further migration. Our findings support both hypotheses and bear implications for regional development and environmental change in the Amazon. Keywords: Brazil, Amazon, migration, land, social movement. Introduction In the 1970s and 1980s, migration to roadside settlement frontiers of the Brazilian Amazon proceeded at a rapid pace (Arag ´ on and Mougeot, 1986; Perz, 2002; Oliveira and Sim ˜ oes, 2004). Most such migration was spontaneous, and the demographic flood in many road corridors overwhelmed state agencies and resulted in land conflicts. A large literature on the Amazon has documented cases of fraudulent land titles as well as threats and violence as means of securing land claims (Branford and Glock, 1985; Schmink and Wood, 1992; Wagner de Almeida, 1995; Alston, Libecap and Mueller, 1999; Simmons, Walker, Arima, Aldrich and Caldas, 2007). © 2010 The Authors. Bulletin of Latin American Research © 2010 Society for Latin American Studies. Published by Blackwell Publishing, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA. 459

Intraregional Migration, Direct Action Land Reform, and New Land Settlements in the Brazilian Amazon

Embed Size (px)

Citation preview

Bulletin of Latin American Research, Vol. 29, No. 4, pp. 459–476, 2010

Intraregional Migration, DirectAction Land Reform, and New LandSettlements in the Brazilian AmazonSTEPHEN G. PERZ AND FLAVIA LEITEUniversity of Florida, USA

CYNTHIA SIMMONS, ROBERT WALKER ANDSTEPHEN ALDRICHMichigan State University, USA

MARCELLUS CALDASKansas State University, Kansas, USA

This article analyses migration histories of residents in rural settlementsof the Brazilian Amazon that resulted from direct action land reform(DALR), which involves organised land occupations. Our analysis eval-uates two hypotheses. The ‘urban migration’ expectation asserts thaturban experience is important for DALR participation, which links ruraland urban areas via migration for land occupations. The ‘DALR effi-cacy’ hypothesis argues that migration and DALR are complementarylivelihood strategies, such that participation in DALR obviates the needfor further migration. Our findings support both hypotheses and bearimplications for regional development and environmental change in theAmazon.

Keywords: Brazil, Amazon, migration, land, social movement.

Introduction

In the 1970s and 1980s, migration to roadside settlement frontiers of the BrazilianAmazon proceeded at a rapid pace (Aragon and Mougeot, 1986; Perz, 2002; Oliveiraand Simoes, 2004). Most such migration was spontaneous, and the demographic floodin many road corridors overwhelmed state agencies and resulted in land conflicts.A large literature on the Amazon has documented cases of fraudulent land titles as wellas threats and violence as means of securing land claims (Branford and Glock, 1985;Schmink and Wood, 1992; Wagner de Almeida, 1995; Alston, Libecap and Mueller,1999; Simmons, Walker, Arima, Aldrich and Caldas, 2007).

© 2010 The Authors. Bulletin of Latin American Research © 2010 Society for Latin American Studies.Published by Blackwell Publishing, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street,Malden, MA 02148, USA. 459

Stephen G. Perz et al.

By the 1990s, interest in Amazonia expanded beyond questions of frontier expansionand land conflicts as deforestation and related environmental themes became pre-eminent (Wood and Porro, 2002; Becker, Alves and Costa, 2007; Malhi, Betts andRoberts, 2008). Migration and land settlement in the Amazon were consequentlysidelined to focus the spotlight on public policies to reconcile the triple bottom lines ofeconomic development, social equity and environmental sustainability.

While migration into the Amazon has slowed, migration within the region has con-tinued, and it has proceeded hand-in-glove with new tactics to achieve land settlement.Indeed, migratory circulation within the Amazon helps explain the continued expansionof roads and rural settlements in the region, which in turn constitute key determinantsof deforestation (Simmons, Perz, Pedlowski and Teixeira Silva, 2002; Caldas, 2008).The upshot is that we need to know more about contemporary migration patternsin the Brazilian Amazon because they are related to land settlements important forunderstanding environmental change.

In this article we present an empirical analysis of migration histories and participationin DALR, a set of tactics adopted by landless groups to constitute new rural landsettlements. We distinguish DALR from state-led agrarian reform (SLAR) becauseDALR is initiated by non-state actors, often as a form of contentious politics to contestSLAR as inadequate to meet popular demand for land (Simmons, Walker, Arima,Aldrich and Caldas, 2007). We focus on the state of Para in the eastern BrazilianAmazon, a heavily studied area, in part because of two prominent settlement frontierzones that we take up as our study cases, namely the Transamazon highway andsouthern Para. We document migration histories of families in various types of DALRsettlements to address hypotheses about (a) migration histories and residence in DALRsettlements; and (b) the relative timing of migration and DALR activity. Concerningmigration history and residence in DALR settlements, we argue that urban experienceand rural–urban–rural circulation are facilitators of participation in DALR. The firstpart of our analysis therefore focuses on the importance of urban experience andcircular migration among residents in rural DALR settlements. Regarding the secondquestion, we view migration and DALR as competing strategies to secure livelihoods,and suggest that DALR can halt chronic migration. Consequently, among those inDALR settlements, most migration comes first, as participation in DALR obviates theneed for continued migration. The second part of the analysis therefore focuses onthe relative timing of migration and DALR involvement. We find support for ourhypotheses, and conclude by setting the findings in the larger context of contemporaryurbanisation and environmental problems in the Brazilian Amazon.

Background

Frontier Expansion, Population Growth and Land Settlementin the Amazon

The 1960s bore witness to the onset of regional development projects to integratethe Amazon into Brazil’s economy (Cardoso and Muller, 1977; Mahar, 1979). Pairedwith infrastructure projects came state-directed colonisation projects in the 1970s thatattracted rural families from many regions of Brazil (Costa, 1992; Ozorio de Almeida,1992). By the 1980s, large-scale mining and dam projects provided additional targets for

© 2010 The Authors. Bulletin of Latin American Research © 2010 Society for Latin American Studies460 Bulletin of Latin American Research Vol. 29, No. 4

Intraregional Migration in the Brazilian Amazon

populations moving into the Amazon (Castro and Hebette, 1989; Hall, 1989; Wagnerde Almeida, 1995).

One key consequence of these initiatives was rapid interregional migration tosettlement frontiers. Total population in Brazil’s North region (the states of Rondonia,Acre, Amazonas, Roraima, Para, Amapa, and Tocantins, which encompass most ofAmazonia in Brazil) grew from roughly four million in 1970 to over 10 million by1991 (Oliveira and Simoes, 2004). Interregional migration to the North region summedto nearly 750,000 during the 1970s and roughly 1 million persons during the 1980s(Carvalho Brasil, 1997).

Analyses of migration during the 1970s and 1980s in the Amazon emphasise rapidurban growth (Lavinas, 1987; Browder and Godfrey, 1997). From the 1970s onward,urban migration and urban population growth in the Amazon have generally beenmore rapid than in rural areas (Carvalho Brasil, 1997; Perz, 2002). The result has beenurbanisation of settlement frontiers, as roadside towns sprang up, forming a networkof urban centers (Browder and Godfrey, 1997).

Urbanisation of the Amazon frontier has been explained in different ways, thoughavailable explanations are not mutually incommensurate. One explanation is thaturbanisation was a planned process, complete with state design of roadside settlements(Becker, 1995). This reflects an older view of urban-based regional development inBrazil, which proceeds by incorporating settlement frontiers via establishment of urbandistribution points (Katzman, 1977) to facilitate extraction of regional surplus forindustrial areas (Cardoso and Muller, 1977).

However, other processes suggest additional explanations for frontier urbanisationin the Amazon. Rapid population growth along the roadsides also led to intensecompetition to establish land claims, often among unequal groups, resulting in theexpulsion of the rural population (Branford and Glock, 1985; Schmink and Wood,1992; Wagner de Almeida, 1995; Simmons, Perz, Pedlowski and Teixeira Silva, 2002).Rural expulsion in turn resulted in intraregional circulation of population, sometimes toother sites along rural roadsides (Schmink and Wood, 1992), but most often to towns(Mitschein, Miranda and Paraense, 1989; Simmons, Perz, Pedlowski and Teixeira Silva,2002). By the 1980s, there was a ‘rural exodus’ in many parts of the Brazilian Amazon(Perz, 2000).

By the 1990s, several changes had taken hold in the Brazilian Amazon. Economicrecession beginning in the early 1980s led to cuts in state spending on road build-ing, directed colonisation, fiscal incentives and other state-led initiatives for frontierexpansion. The economic downturn also undermined the legitimacy of the militarydictatorship, which resulted in a shift from authoritarian rule to democratisation and amore open polity in Brazil in the mid-1980s. By the late 1980s, price fluctuations for keycommercial crops, gold strikes, the expansion of cattle ranching and other economicshifts were occurring in settlement frontiers of the Amazon. This context led to twochanges on which we focus in this article.

First, migration patterns shifted, from interregional movements into the Amazon tointraregional circulation within the region. Such circulation involves moves betweenrural and urban areas, often within the same state or even the same municipality.Circulation is tied to flexible livelihood strategies involving not only rural struggles forland but also participation in the urban informal sector (Roberts, 1992; MacMillan,1995; Browder and Godfrey, 1997; Simmons, Perz, Pedlowski and Teixeira Silva, 2002).

Second, state withdrawal and democratisation ceded space for the emergence oflocal politics in Amazon settlement frontiers. Given the precarious circumstances they

© 2010 The Authors. Bulletin of Latin American Research © 2010 Society for Latin American StudiesBulletin of Latin American Research Vol. 29, No. 4 461

Stephen G. Perz et al.

faced, popular groups in settlement frontiers began to organise politically (Allegretti,1994; Adriance, 1995; Hall, 1997). Mobilisation proceeded with various strategic endsas foci, but in many cases, it was a response to political pressures to expel ruralpopulations from land and other resource claims. Thus, state-led frontier expansion inBrazil stimulated considerable in-migration and population growth, especially in urbanareas, in part due to rural land conflicts, which in turn motivated social mobilisation toretain land claims.

Collective Mobilisation and Direct Action for Land Reform in the Amazon

Despite challenges such as threats and violence concerning land claims, difficulties withsustaining land productivity, and geographic disadvantages in marketing produce, ruralas well as urban populations in the Amazon continue to seek rural land (Simmons,Walker, Arima, Aldrich and Caldas, 2007; Caldas, 2008). Social mobilisation for landamounts to a political strategy that involves collective tactics for establishing land claimsin the face of obstacles that individual families are unable to overcome.

Brazilian law outlines procedures required to establish land claims (Alston, Libecapand Mueller, 1999; Simmons, Walker, Arima, Aldrich and Caldas, 2007; Caldas, 2008).Unused public land in Brazil legally belongs to the state (terra devoluta), but citizensmay claim such land as their own if they can demonstrate productive use for over oneyear (usocapiao). Land laws also recognise claims to private land by squatters if theycan demonstrate occupation and productive use, without contestation by the owner,for five years and a day. During that time, squatters pay taxes and initiate the processof registering their claim in order to receive a land title.

In Brazil, the National Institute for Colonisation and Agrarian Reform (INCRA)was charged with oversight of directed colonisation in the Amazon in the 1970s. Afterstate withdrawal and the onset of land conflicts and social mobilisation for land in theAmazon, INCRA’s focus shifted. By the mid-1990s, INCRA focused on processing landtitle applications in order to identify legitimate land claims. This shift in focus was aresponse to the mounting number of contested claims that necessitated investigationinto whether landholders were meeting the ‘social function’ of land productivity asstipulated by law (Simmons, Walker, Arima, Aldrich and Caldas, 2007; Caldas, 2008).

Landless groups seeking land in the Amazon seized on Brazilian land law and INCRAtitling activities by engaging in various forms of direct action land reform (DALR). OneDALR tactic resulted from social mobilisation elsewhere in Brazil, where landholdingsare highly consolidated (Branford and Rocha, 2002). In such circumstances, socialmovement organisations (SMOs) such as the Movimento dos Trabalhadores Rurais SemTerra (MST) (Landless Workers’ Movement) orchestrate land occupations (Branfordand Rocha, 2002; Simmons, Walker, Arima, Aldrich and Caldas, 2007). SMO-ledDALR involves considerable planning, beginning with articulation of a justice-basedcritique of unequal access to land, highlighting state failures to meet popular demandfor expropriation of unproductive lands. DALR SMOs then identify ‘unproductive’properties, recruit landless people from the countryside as well as urban areas, andconduct organised land occupations, complete with materials for temporary housing,plans for allocation of land among occupation participants and contingency plansin case of violence (ANCA, 2002). Alternatively, DALR may occur ‘spontaneously’or ‘organically’ without the leadership and planning of an SMO (Simmons, Walker,Arima, Aldrich and Caldas, 2007; Caldas, 2008). Spontaneous DALR draws on kin tiesor shared experiences by local families who have insufficient land, especially for grown

© 2010 The Authors. Bulletin of Latin American Research © 2010 Society for Latin American Studies462 Bulletin of Latin American Research Vol. 29, No. 4

Intraregional Migration in the Brazilian Amazon

children seeking to begin their own families. Individual families gradually populatean area, informally demarcating property boundaries as new families arrive. Onceenough families are present, they collectively mobilise to demand INCRA recognitionto regularise their land claims.1

In the Amazon, DALR SMOs arrived from outside the region by the end of the 1980s(Simmons, Walker, Arima, Aldrich and Caldas, 2007), which was also the momentwhen pioneer families who settled along the roadsides in the 1970s began to seek landfor their children (Caldas, 2008). Demand for rural land thus found expression in landoccupations beginning in the 1990s, whether via SMO-led or spontaneous DALR. Thus,despite challenges to land settlement and a largely urban population, popular groupsin the Amazon took advantage of Brazilian land law and INCRA titling to engage inDALR to gain access to land.

Migration and Direct Action Land Reform

Having discussed migration and DALR in the Amazon, we now focus on the relationshipbetween the two. By the 1990s, migration in the Amazon had become largely intrare-gional, involving local circulation among rural and urban areas as a livelihood strategy.DALR also provided a means of securing a livelihood that potentially obviated the needfor chronic migration, since collective mobilisation reduced the risks of securing land asa basis for productive activities.

We therefore offer hypotheses concerning migration experience and participation inDALR, shown in Table 1. One hypothesis concerns types of migration experience thatmay facilitate participation in DALR settlements (H-Mig). Of particular importance isexperience in urban areas (Simmons, Perz, Pedlowski and Teixeira Silva, 2002). Ruralland conflicts, expulsion from land claims, and rural exodus yield an urban populationwith a rural background that does not necessarily prefer the informal economy overlandownership. Further, DALR SMOs recruit in urban areas, which makes urbanresidence instrumental for participation in SMO-led DALR. We therefore expect thatamong residents in rural DALR settlements, most have lived in at least one urban areapreviously (H-Mig-1). In addition, whether participating in SMO-led or spontaneousDALR, families should also be likely to have engaged in circular (rural–urban–rural)migration, where the last urban–rural move is made in order to participate in DALR.This reflects circulation by families seeking land but expelled to urban areas, only toreturn to the countryside again via DALR. We therefore anticipate that most residentsin DALR settlements have experienced at least one circular migration cycle (H-Mig-2).We also entertain the alternate hypotheses that most residents in DALR settlements donot have urban or circular migration experience (H-Alt-1 and H-Alt-2).

In addition, we view migration and DALR participation as competing strategiesto secure livelihoods. This motivates additional expectations regarding migration andDALR. Specifically, the ‘efficacy’ hypothesis (H-Eff) is that DALR renders furthermigration unnecessary because DALR provides an efficacious alternative to chronic

1 Other forms of DALR are also possible (Simmons et al., 2010), such as state-facilitatedDALR where INCRA or other state agents advise local peoples seeking to make landclaims; there is also the possibility of endogenous DALR that resembles a hybrid ofspontaneous and SMO-led DALR in that it is organized by local people without SMOsupport but still involves a coordinated land occupation.

© 2010 The Authors. Bulletin of Latin American Research © 2010 Society for Latin American StudiesBulletin of Latin American Research Vol. 29, No. 4 463

Stephen G. Perz et al.

Table 1. Hypotheses Regarding Migration and Direct-Action Land Reform (DALR)

Number Hypothesis

Urban migration hypotheses (H-Mig): urban experience as a precursor of DALR participation

H-Mig-1 Most respondents in rural DALR settlements have lived in at least one urban areaH-Mig-2 Most respondents in DALR settlements have engaged in circular (rural–urban–rural)

migrationEfficacy hypotheses (H-Eff): DALR as a livelihood strategy to avoid further migration

H-Eff-1 Most moves come before DALR participationH-Eff-2 After DALR experience, there are few subsequent moves

migration as a means of securing a livelihood. One implication is that most movesin a family’s migration history come before participation in DALR (H-Eff-1); anotheris that among families with DALR experience, there are very few subsequent moves(H-Eff-2). An alternative hypothesis (H-Alt) is that DALR is not an effective alternativeto migration. In that case, families will make just as many moves after DALR activityas before (H-Alt-3), and among families with DALR experience, there are numeroussubsequent moves (H-Alt-4).

Study Sites, Methods and Data

For an analysis of migration and DALR in the Brazilian Amazon, there are few moreappropriate locations than the Transamazon corridor and southern Para. Both areaswere opened by new highways in the 1960s and 1970s. Both thereupon incurredrapid population growth because of in-migration. Both later experienced economicchanges, often including difficulties such as price declines in key crops, as well as newopportunities such as development projects or gold strikes. Both have exhibited ruralout-migration and urban population growth. Consequently, both exhibit continued landsettlement, involving various forms of collective mobilisation for DALR. Consequently,both serve as useful case studies of migration and DALR.

The Transamazon became famous as a key east–west corridor through the Amazonthat would facilitate westward expansion and integrate Brazil’s coast with its northerninterior. During the 1970s, the state invested in road construction and supported directedcolonisation along the corridor (Tavares, Considera and Castro e Silva, 1979; Moran,1981). Transamazon settlements followed a pre-planned property grid organised aroundthe familiar ‘fishbone’ road network, and featured nucleated settlements that wouldconstitute communities of small farm families. The fanfare of the 1970s gave way todisillusionment in the 1980s with economic crisis and withdrawal of state support forcolonisation. But in the mid-1980s, price increases in key local crops attracted moremigrants (Nascimento and Drummond, 2003). Price declines in the 1990s led to furtherproblems, but by then local producers had greater capacity to shift livelihood strategies,and cattle and other activities became more prevalent. By the 1990s, local organisationshad emerged or arrived, and they helped constitute new interest groups (Nascimentoand Drummond, 2003; Toni and Kaimowitz, 2003). That in turn facilitated the creationand (later) official recognition of new settlements beyond the original INCRA propertygrid, constituting a much larger area in agricultural production (Caldas, 2008). Newsettlement formation along the Transamazon was accomplished largely via spontaneousDALR based on kin and church ties (Caldas, 2008).

© 2010 The Authors. Bulletin of Latin American Research © 2010 Society for Latin American Studies464 Bulletin of Latin American Research Vol. 29, No. 4

Intraregional Migration in the Brazilian Amazon

Southern Para refers to the municipalities in Para south of the Transamazon highwayand east of the Xingu river (Simmons, Walker, Arima, Aldrich and Caldas, 2007). Thearea was opened via construction of several state and federal roads, and because ofits relative proximity to other regions of Brazil, in-migration was rapid (Schmink andWood, 1992). Southern Para was not, however, a remote area in which to showcasedirected colonisation; rather, it already had several riverside settlements and some largeranches, and became a key target not only for farm families but also for corporations andinvestors (Branford and Glock, 1985). By the 1980s, southern Para comprised a mosaicof land claims by interest groups with unequal political clout (Schmink and Wood,1992). The result was a tumultuous period of contested land settlement involving ruralviolence, including hundreds of assassinations and thousands of families displaced fromtheir land claims (Wagner de Almeida, 1995; Simmons, Walker, Arima, Aldrich andCaldas, 2007). One consequence was a greater state presence via emergency land titling,which only motivated further land occupations and contestation (Schmink and Wood,1992). Another consequence was widespread mobilisation of smallholders, supportedby unions, churches and DALR SMOs, contesting large private landholdings (Branfordand Rocha, 2002; Simmons, Walker, Arima, Aldrich and Caldas, 2007). As a result,land occupations, mostly SMO-led, have proceeded in many parts of southern Para(Simmons, Walker, Arima, Aldrich and Caldas, 2007).

During the summer field season of 2006, Brazilian and North American researchersfrom several universities visited 13 DALR settlements (assentamentos) in Para. Weselected settlements formed after initial road openings and pioneer settlement in orderto focus on the more recent period of settlement expansion. We also sought settlementsin different locations with distinct histories. Table 2 shows the five settlements we visitedalong the Transamazon highway where spontaneous DALR predominates, along withthe eight settlements visited in southern Para, where SMO-led DALR predominates.We present basic information about each settlement, noting its location, year whenregularised by INCRA, and an INCRA estimate of the number of families settled(Caldas, 2008).

Table 2. DALR Settlements included in analysis, Para, Brazil, 2006

Settlement Municipality Year INCRA recognised Families settled

TransamazonRio do Peixe Uruara 1995 241Surubim Medicilandia 1988 1527Trairao Uruara 1997 160Tutuí Norte Uruara 1999 341Uirapuru Uruara 1997 262

Southern Para1 de Marco Sao Joao do Araguaia 1998 35017 de Abril Eldorado dos Carajas 1997 688Alegria Maraba 1999 95Cabanos Eldorado dos Carajas 2003 81Canudos Eldorado dos Carajas 2004 58Castanhal Araras Sao Joao do Araguaia 1987 128Palmares II Parauapebas 2001 286Santa Maria do Pontal Eldorado dos Carajas 1997 67

Source: Caldas, 2008.

© 2010 The Authors. Bulletin of Latin American Research © 2010 Society for Latin American StudiesBulletin of Latin American Research Vol. 29, No. 4 465

Stephen G. Perz et al.

In each settlement, we sampled households. Because these settlements had beenregularised by INCRA, they had lists of residents, so we sought to sample basedon equal probability. Random sampling, however, proved impossible because someproperty owners no longer lived in the settlement or could not be found, and somehouses were not visible from the road. We therefore sampled as systematically aspossible to incorporate variation in terms of distance to market by including residentswith land claims spread throughout each settlement. While the samples are not entirelyrandom, they are representative insofar as respondents held land in different parts ofthe settlements visited.2

For each household interviewed, we applied a structured questionnaire includingitems to obtain a complete migration history as well as information about DALRparticipation. The migration questions included items on place of birth, and for eachmove, the year, location (state and municipality) and type of location (rural or urban).This allows construction of more detailed migration histories than those possible withcensus data and most previous surveys of settlement frontiers in the Amazon. With thisinformation we derived migration indicators for respondents including the total numberof moves, moves within Para, moves within the municipality of current residence, movesto urban and rural areas, and number of rural–urban–rural circular movements. Inaddition, we derived indicators of DALR experience, including whether the respondenthad participated in SMO-led DALR activities, and whether anyone in the respondent’sfamily of origin had participated in DALR activities. DALR activities refer to actions foragrarian reform, including protest marches, camp-outs in front of INCRA buildings andoccupations of contested land. Participants may have been involved in several DALRevents related to a given DALR settlement, or perhaps none if they arrived after thesettlement was recognised by INCRA. DALR data include information about year ofaction, which allows historical organisation of events over time.

Our analysis proceeds in two parts. The first part focuses on the migration histo-ries and evaluates the ‘urban migration’ hypotheses that most respondents in DALRsettlements have previous urban experience (H-Mig-1) and have engaged in circularmigration (H-Mig-2). The second part of the analysis features the relative timing ofmigration and DALR to test the efficacy hypotheses that most migration occurs beforeDALR (H-Eff-1) and DALR is followed by little subsequent migration (H-Eff-2).

Findings

Part 1: Urban Migration Analysis

We begin with an analysis of our migration history data as a means of evaluating theurban migration hypotheses (H-Mig-1, 2). We first characterise the geographic originsof the respondents, followed by discussion of their migration history in terms of totalmoves, moves to reach Para, moves to reach the municipality of residence at the time of

2 Comparisons of our sample sizes to the INCRA estimates of populations in the settlementsrevealed differences in the ratio of sample sizes to populations. To ensure that our analysisreflects the size distribution of populations among the settlements sampled, we weightedthe sampled cases by settlement. After weighting, we arrived at 235 respondents (families)along the Transamazon and 161 in southern Para.

© 2010 The Authors. Bulletin of Latin American Research © 2010 Society for Latin American Studies466 Bulletin of Latin American Research Vol. 29, No. 4

Intraregional Migration in the Brazilian Amazon

interview, and moves within a given municipality. We then focus on previous rural andurban residences to evaluate our ‘urban migration’ expectations.

For geographic origin, we consider respondent state of birth. Given what is knownabout populations in the Transamazon corridor and southern Para, we should expectmost respondents to have been born in other regions of Brazil. Table 3 presents a seriesof migration indicators for respondents in the sample, beginning with information onplace of birth. Table 3 confirms our expectation: roughly 11 per cent of respondentswere born in Para, somewhat less in southern Para than along the Transamazon.Table 3 also indicates that roughly 3 per cent more came from other states in Brazil’sNorth (Amazon) region, where Para is located. This confirms that residents in DALRsettlements in the two study areas largely came from other regions of Brazil.

Given that most respondents were born outside the Amazon, and based on priorliterature, we should expect several moves to be normative in the sample. We thereforecounted the total number of moves, defined as a change in residence between ruraland urban areas of the same municipality and/or between Brazilian municipalities orstates or regions. Table 3 confirms that respondents had moved multiple times. Theoverall average was roughly three moves, somewhat more in southern Para than alongthe Transamazon. We also call attention to the standard deviations, which indicatevariation among respondents; a handful in both study areas had moved seven or moretimes.

The foregoing findings raise questions about the number of moves to reach Para. Weexpect that initial moves among the respondents were interregional, but later moveswere intraregional. We calculated the percentage in Para after each move (first, second,etc.) out of the number of respondents with that many moves (or more) to see how

Table 3. Comparative Analysis of Migration, Transamazon and Southern Para DALRSettlements, Brazil, 2006

Migration indicator Transamazon Southern Para Total X2/F-test

State of birthPercentage born in Para 13.2 7.6 10.9 X2 = 77.63Percentage born in other states, north region 1.8 3.2 2.3 p < 0.001Percentage born in other regions of Brazil 85.0 89.2 86.8n 227 158 385

Number of total movesMean 2.90 3.71 3.23 F = 24.45Standard deviation 1.54 1.69 1.65 p < 0.001

Moves within municipality of residenceMean 1.42 1.16 1.31 F = 11.96Standard deviation 0.84 0.54 0.74 p < 0.001

Moves to urban areasPercentage with zero urban moves 60.4 31.9 48.9 X2 = 47.69Percentage with 1+ urban moves 39.6 68.1 51.1 p < 0.001n 235 160 395

Circular (Rural–Urban–Rural) migrationsPercentage with zero circular migrations 58.1 31.9 47.1 X2 = 33.76Percentage with 1+ circular migrations 41.9 68.1 52.9 p < 0.001n 222 160 382

© 2010 The Authors. Bulletin of Latin American Research © 2010 Society for Latin American StudiesBulletin of Latin American Research Vol. 29, No. 4 467

Stephen G. Perz et al.

100.00

80.00

60.00

40.00

20.00

Perc

enta

ge in

Par

á St

ate

0.001st 2nd 3rd 4th 5th

Transamazon

Southern Pará

Total

6th 7th

Ordinal Number of Move

Figure 1. Percentage of Respondents Migrating to the State of Para, by Ordinal Number ofMove, Transamazon and Southern Para DALR Settlements, Brazil, 2006

many moves were required before half of the respondents in the sample were in Para.Figure 1 shows how these percentages vary by number of moves. In both study areas,only 30–40 per cent of respondents had arrived in Para after the first move, but thisjumped to roughly 70 per cent after the second move. Compared to the values for totalmoves, these figures imply that most respondents had made at least one move (if notseveral) within Para.

We also examined the number of moves required for respondents to reach theirmunicipality of residence at the time of the 2006 survey. As with moves to Para, wecalculated the percentage of respondents who were in their municipality of currentresidence by number of moves. Figure 2 shows the results. After the first move, only10–20 per cent of respondents had reached their municipality of residence. This rose,more slowly among respondents in southern Para than the Transamazon, such thatmajorities had reached their municipality of residence after four moves. After that, thereare relatively few respondents who continued to move, so estimates become somewhatunstable. Compared to the data for moves to reach Para, the data for moves to reach themunicipality of current residence suggest one or more moves within Para after reachingthe state before coming to the municipality of current residence. Together, the data onmoves to Para and moves to municipality of current residence indicate that the earliestmoves are interregional, and later moves are more local.

We also examined moves within the municipality of current residence. We countedthe number of times a respondent indicated a rural–urban or urban–rural move butno change in municipality. We found that 91 per cent of respondents had made oneor more intra-municipal moves. This confirms the importance of localised circulationamong rural and urban areas in the study sites. Table 3 presents means for intra-municipal moves. Respondents along the Transamazon reported 1.4 intra-municipalmoves, whereas those in southern Para reported 1.2. Whereas mobility among statesand municipalities appears greater among respondents in southern Para, intra-municipalmobility is greater along the Transamazon.

© 2010 The Authors. Bulletin of Latin American Research © 2010 Society for Latin American Studies468 Bulletin of Latin American Research Vol. 29, No. 4

Intraregional Migration in the Brazilian Amazon

100.00

80.00

60.00

40.00

20.00

Perc

enta

ge in

Mun

icip

ality

of

Cur

rent

Res

iden

ce

0.00

1st 2nd 3rd 4th 5th 6th

Ordinal Number of Move

Transamazon

Southern Pará

Total

Figure 2. Percentage of Respondents Migrating to Municipality of Current Residence, by OrdinalNumber of Move, Transamazon and Southern Para DALR Settlements, Brazil, 2006

The data on intra-municipal moves beg questions about rural and urban movement,which brings us to the question of the ‘urban migration’ expectation among residents ofDALR settlements. At first glance, our data indicate that rural residences predominatein the migration histories in our sample. Figure 3 shows percentages migrating to urbanareas by number of move. For first moves, 34 per cent were to urban areas, and thisdeclined among later moves to 28 per cent of respondents making a second move, 21per cent who made a third move, and 17 per cent who made a fourth move. Estimates

50.00

40.00

30.00

20.00

10.00

Perc

enta

ge M

ovin

g to

Urb

an A

reas

0.00

1st 2nd 3rd 4th 5th 6th

Ordinal Number of Move

Transamazon

Southern Pará

Total

Figure 3. Percentage of Respondents Migrating to Urban Areas, by Ordinal Number of Move,Transamazon and Southern Para DALR Settlements, Brazil, 2006

© 2010 The Authors. Bulletin of Latin American Research © 2010 Society for Latin American StudiesBulletin of Latin American Research Vol. 29, No. 4 469

Stephen G. Perz et al.

become less stable after the fourth move because of relatively few respondents with fiveor more moves. Figure 3 also shows that these values differed among the two studyareas, with urban movement less common among respondents along the Transamazonthan in southern Para.

It is nonetheless the case that most respondents had lived in at least one urban area.Table 3 shows that a slight majority (51 per cent) had lived in one or more urban areasprior to settling in a DALR settlement. This confirms the ‘urban migration’ expectationand supports H-Mig-1. However, there are significant differences among the two studyareas, for a large majority in southern Para (68 per cent) had resided in one or moreurban areas, compared to a minority along the Transamazon (40 per cent). Thus, weconfirm H-Mig-1, but only for southern Para, where SMO-led DALR predominates.This likely reflects the fact that urban recruitment is instrumental to SMO-led DALR.

Findings for number of moves, intra-municipal moves, and rural and urban migrationalso raise questions about circular migration. Whereas the vast literature on internalmigration in Latin America has emphasised rural–urban movement, DALR tacticsprovide a mechanism for urban–rural returns. We identified circular migration asany rural–urban–rural migratory sequence, whether within or between municipalities,and whether there is one or more urban residences involved in the middle part ofthe cycle. Table 3 presents data for circular migrations made by respondents. Mostrespondents (53 per cent) had made one or more such circuits, which supports H-Mig-2that most respondents had made at least one circular migration. However, as withurban migration, significant differences in circular migration appeared among the twostudy areas. In southern Para, a large majority (68 per cent) had made at least onerural–urban–rural circulation, but along the Transamazon, only a minority (42 percent) had made one or more circular migrations.3 Consequently, we only supportH-Mig-2 for southern Para. This finding reflects the greater migration experience ofrespondents in southern Para, in turn a reflection of more difficult conditions for accessto land, which motivates participation in SMO-led DALR.

Part 2: Timing Analysis

We now turn to the second part of our analysis and focus on the efficacy hypothesis,which argues that migration and DALR are competing livelihood strategies, such thatparticipation in DALR may halt further moves and thereby stem chronic migration. Wetherefore pursue an analysis of the relative timing of migration and DALR activity. Weconsider the timing of migration, then that of DALR activity, and then intersect the twoby analysing the number of moves before and after DALR.

Figure 4 shows the average year of moves among respondents in the two studyareas. First moves on average occurred around 1970; second moves in the early 1980s;third moves in the late 1980s; and fourth, fifth, and sixth moves during the early1990s. We do not present moves after the fifth because sample sizes become small and

3 The percentage for circular migrations among Transamazon respondents should not bebut is higher than the percentage of the Transamazon respondents with urban experience.This is because of the loss of a few cases as a result of missing data when calculating circularmigration, where the lost cases were among those who did not have urban experience,which raised the relative importance of those who did, resulting in an apparently higherpercentage with circular migrations.

© 2010 The Authors. Bulletin of Latin American Research © 2010 Society for Latin American Studies470 Bulletin of Latin American Research Vol. 29, No. 4

Intraregional Migration in the Brazilian Amazon

Transamazon

SouthPA

Total

2000

1995

1990

1985

1980

1975

1970

1965

1960

1955

1950

Yea

r of

Mov

e

1st 2nd 3rd 4th 5th 6th

Ordinal Number of Moves

Figure 4. Year of Migration, by Ordinal Number of Move, Transamazon and Southern ParaDALR Settlements, Brazil, 2006

estimates become unstable. Migration among respondents thus began during the periodof frontier expansion in the Amazon (the 1970s and 1980s), when our respondents alsomade interregional moves (from other regions of Brazil to Para); afterwards, migrationintervals shortened as moves became increasingly more localised.

Because the nature of DALR participation differs along the Transamazon and insouthern Para, we analyse the timing of DALR in the two study sites in distinct ways. Wefirst consider the Transamazon and compare migration histories to the year when thesettlement where respondents resided at the time of interview was officially recognisedby INCRA. This separates migration from arrival in the settlement of residence at thetime of interview, leaving open the possibility of migration after initial arrival in thesettlement. This is still valid as regards the process of spontaneous DALR along theTransamazon because, there, DALR activity comes after informal settlement, whenthere are enough families to go to INCRA and demand regularisation. Such collectivemobilisation defines DALR activity and is instrumental for INCRA recognition, makingINCRA regularisation a useful marker for dating DALR activity in the context ofspontaneous DALR. Table 2 indicates that four of the five Transamazon settlementswere regularised during the late 1990s; only Surubim officially existed before 1995.Figure 4 indicates that, on average, first, second and in many cases third moveshad already occurred by the time of DALR among Transamazon respondents, whichsupports H-Eff-1 for that study area.

In southern Para, we consider participation in DALR activities. This is important inthe context of SMO-led DALR, which may involve a series of DALR actions to occupypreviously claimed lands and lobby INCRA to recognise the new occupation. It is worthnoting that only a subset of respondents in southern Para actively participated in DALRprotests, land occupations, and so forth (n = 106); some arrived in DALR settlementsafter they were established (n = 55), in some cases after INCRA regularisation. Insouthern Para, respondents averaged 1.2 DALR experiences, with a range of 0 to 4experiences. We focus on the timing of the first reported DALR experience among the

© 2010 The Authors. Bulletin of Latin American Research © 2010 Society for Latin American StudiesBulletin of Latin American Research Vol. 29, No. 4 471

Stephen G. Perz et al.

subset of respondents in southern Para who reported participation in one or more DALRactivities. This provides a more conservative test of the timing of DALR participationwith which to compare to migration histories, because first DALR participation comesearliest in time. The average year of first DALR experience among respondents insouthern Para is 1995. This is also substantially later than the average year of first,second, and third moves among respondents in southern Para, which supports H-Eff-1for that study area.

Based on the timing analysis thus far, one might be tempted to conclude that H-Eff-1is correct, that most migration comes before DALR participation, and also that H-Eff-2is supported, that is, there is little migration following DALR participation. However,the data presented so far only compare average times of migration and DALR experiencefor the two study areas, and make no comparisons for individual respondents. It couldstill be that there is considerable migration following DALR, hidden in the averagesfor the study areas. Therefore, for the Transamazon, we directly compared the timingof respondent moves to settlement regularisation dates, and for southern Para, wecompared the timing of respondent moves to their first DALR participation.

For the Transamazon, we focus on respondents who made at least one move andhad valid data for dates of their moves (n = 222 out of 235 cases). Table 4 presents theresults up to the fifth move. For the first move, over 90 per cent of respondents movedprior to DALR; for the second move, this was 76 per cent; for the third move, this figurewas 58 per cent. The mean difference in the timing of migration and DALR was nineteenyears before for first moves; nine years prior for second moves; almost five years priorfor third moves; and still nearly a year prior for fourth moves. Meanwhile, DALRregularisation first as a percentage of the Transamazon respondents rose from roughly 8per cent for the first move to 24 per cent for the second to 40 per cent for the third. It isonly with the fourth move that DALR comes first more often (41 per cent) compared tomigration (38 per cent). Nonetheless, these findings provide further support of H-Eff-1that migration largely occurs before DALR, because few respondents made four ormore moves. Based on these comparisons, we calculated that Transamazon respondentsreported on average 1.99 moves at least one year before DALR, compared to 0.08moves in the same years as DALR and 0.71 moves at least one year after DALR. Thesefindings support both H-Eff-1 and H-Eff-2 for the Transamazon study area.

Table 5 compares migration and DALR timing for southern Para, using the year offirst DALR experience as the benchmark. We again focus on the subset of respondentswho reported participation in DALR activities. Among those, we focus on respondentswho made at least one move and had valid data on dates for both migration and

Table 4. Timing Analysis of Migration and DALR as Settlement Regularisation, TransamazonDALR Settlements, Para, Brazil, 2006

PercentagesOrdinal numberof move DALR first Same year Migration first Mean difference (years) n

First 7.64 0.26 92.09 19.43 200Second 23.82 0.61 75.57 9.50 186Third 39.77 2.30 57.93 4.78 138Fourth 41.04 20.49 38.48 0.78 73Fifth 64.35 0.00 35.65 0.43 21

© 2010 The Authors. Bulletin of Latin American Research © 2010 Society for Latin American Studies472 Bulletin of Latin American Research Vol. 29, No. 4

Intraregional Migration in the Brazilian Amazon

Table 5. Timing Analysis of Migration and First DALR Experience, Southern Para DALRSettlements, Para, Brazil, 2006

PercentagesOrdinal numberof move DALR first Same year Migration first Mean difference (years) n

First 0.99 1.25 97.76 28.12 89Second 6.04 11.24 82.72 13.58 88Third 12.50 17.46 70.04 10.27 68Fourth 16.22 23.07 60.71 5.86 48Fifth 25.39 29.59 45.01 3.34 28

DALR participation (n = 89 out of 106 cases). Table 5 shows that for the first move,migration came before DALR for 98 per cent of respondents; for the second, migrationcame first 83 per cent of the time; for the third move, the same figure was 70 per cent;and for the fourth move, it was still nearly 61 per cent. Only with the fifth move didthis drop below 50 per cent, though migration still came first 45 per cent of the time,compared to DALR first 25 per cent. Increasingly, migration and DALR occurred in thesame year, perhaps a reflection of urban recruitment and moves to the countryside aspart and parcel of land occupations. Nonetheless, the mean difference in years betweenmigration and DALR was 28 for first moves, almost 14 for second moves, ten for thirdmoves, and almost six for fourth moves. Further calculations indicated that in southernPara, respondents made 2.6 moves prior to first DALR experience, 0.45 in the sameyear as DALR and 0.31 moves after DALR. These findings support both H-Eff-1 andH-Eff-2 for the southern Para study area.

Discussion

Intraregional migration and DALR both serve as useful tactics for securing livelihoodsby landless populations in the Amazon. Both tactics are well suited to an Amazonianlandscape now defined by a network of regional highways connecting numerous townsand cities, which have their own local road networks that integrate the countryside.Para is not the only state in the Brazilian Amazon with highways that now definecorridors with urban nuclei from which secondary roads emanate where land consoli-dation, economic dynamism and an urban informal sector can be found. Consequently,migration and DALR activities also prevail outside the state of Para, notably in theBrazilian Amazonian state of Rondonia, a comparison we consider elsewhere (Simmonset al., 2010).

The findings suggest that DALR can be an efficacious alternative to chronic migration.This conclusion needs to be tempered, however, because there remain importantchallenges to rural settlements, even if DALR could entirely resolve the question ofland concentration and rural violence. On the one hand, there is the issue of the socialand economic viability of DALR settlements. INCRA regularisation makes possiblesome forms of state assistance, but there remains the issue of sustaining the solidaritythat is crucial to land occupations themselves. Related to this broad question arespecific issues of securing definitive titles, adequate credit, market access and enduringcommunity organisation. Beyond the socio-economic questions are the environmentalissues. Relatively little is known about land management in recently formed DALR

© 2010 The Authors. Bulletin of Latin American Research © 2010 Society for Latin American StudiesBulletin of Latin American Research Vol. 29, No. 4 473

Stephen G. Perz et al.

settlements. Forested land continues to be defined as unproductive and is thereforetargeted for DALR occupations, and such occupations continue to exhibit forestclearing (Simmons, Walker, Arima, Aldrich and Caldas, 2007). Prior work on Amazonfrontier settlements indicated a mixed record of sustainability in land management(Serrao and Homma, 1993). If clearing leads to unsustainable land use, then DALRsettlements may indeed generate out-migration. This implies that, on an aggregate level,migration will not necessarily decline over time even if more DALR settlements appear.

On a larger scale, DALR bears implications for public policy and regional devel-opment as a latter-day rural-led development strategy. Even as the Amazon’s urbanpopulation grows larger than its rural counterpart, the increasingly service-orientedregional economy nonetheless relies in part on rural productive activities to generatelocal income that does not rely on state transfer payments and the like. In that light,DALR occupations provide a means of impelling agrarian reform and thereby intensi-fying rural land use by forcing the state to expropriate unproductive rural properties,which in turn may beget economic growth. This raises additional issues about policiesconcerning credit and extension support to farm families in DALR settlements as wellas others in the Amazon (Toni and Kaimowitz, 2003).

That said, DALR must also face up to other regional development initiativesthat also recognise opportunities implicit in intensification and expansion of ruralproduction. Perhaps ironically, the past several years have been witness to majornew infrastructure projects in the Amazon, part of national integration plans withinBrazil as well as continental integration strategies for South America as a whole(IIRSA, 2008). Infrastructure upgrades are intended to facilitate access of agribusinessto many portions of the basin in order to expand mechanised production for export(CEPEI, 2002). It is far from clear whether DALR settlements will remain in areasbeing opened for mechanised production, notably in roadsides with low-productivitycattle ranches. With expanding agribusiness activity, questions would be re-openedconcerning rural expulsion, chronic migration and urban circulation of population inthe Amazon.

Conclusions

In sum, the first part of the analysis examined migration histories among DALRsettlement residents and confirmed the urban migration expectation that most respon-dents had previously resided in urban areas (H-Mig-1) and had engaged in circular(rural–urban–rural) migration (H-Mig-2) for southern Para, though not the Transama-zon. The second part of the analysis confirmed the efficacy hypotheses concerningmigration and DALR by showing that migration largely occurs before first DALRparticipation (H-Eff-1), with relatively little migration occurring after first DALRparticipation (H-Eff-2).

Acknowledgements

This research was supported by NSF grants #0521794 and #0522062. The authorsthank the Universidade Federal da Bahia and the Universidade Federal do Para in Brazilfor in-country logistical support and field team members for conducting interviewsduring fieldwork. Errors of interpretation are the responsibility of the authors.

© 2010 The Authors. Bulletin of Latin American Research © 2010 Society for Latin American Studies474 Bulletin of Latin American Research Vol. 29, No. 4

Intraregional Migration in the Brazilian Amazon

References

Adriance, M. C. (1995) Promised Land: Base Christian Communities and the Struggle forthe Amazon. State University of New York Press: Albany.

Allegretti, M. H. (1994) ‘Reservas Extrativistas: Parametros Para Uma Política de Desen-volvimento Sustentavel na Amazonia’ in R. Arnt (ed.) O Destino da Floresta: ReservasExtrativistas e Desenvolvimento Sustentavel na Amazonia. Relume – Dumara: Curitiba,17–47.

Alston, L. J., Libecap, G. D. and Mueller, C. C. (1999) Titles, Conflict and Land Use: TheDevelopment of Property Rights and Land Reform on the Brazilian Frontier. Universityof Michigan Press: Ann Arbor.

Associacao Nacional de Cooperacao Agricola (ANCA). (2002) O Que Levar em ContaPara Organizacao do Assentamento: A Discussao no Acampamento. ANCA/MMA:Sao Paulo.

Aragon, L. E. and Mougeot, L. J. A. (eds.) (1986) Migracoes Internas na Amazonia:Contribuicoes Teoricas e Metodologicas. UFPa/NAEA: Belem.

Becker, B. K. (1995) ‘Undoing Myths: The Amazon – An Urbanized Forest’ in M. Clusener-Godt and I. Sachs (eds.) Brazilian Perspectives on Sustainable Development of theAmazon Region. UNESCO/Parthenon: Paris, 53–89.

Becker, B. K., Alves, D. and Costa, W. (eds.) (2007) Dimensoes Humanas da Biosfera-Atmosfera na Amazonia. Editora da Universidade de Sao Paulo: Sao Paulo.

Branford, S. and Glock, O. (1985) The Last Frontier: Fighting Over Land in the Amazon.Zed: London.

Branford, S. and Rocha, J. (2002) Cutting the Wire: The Story of the Landless Movement inBrazil. Latin American Bureau: London.

Browder, J. O. and Godfrey, B. (1997) Rainforest Cities: Urbanization, Development andGlobalization of the Brazilian Amazon. Columbia University Press: New York.

Caldas, M. M. (2008) Settlement Formation and Land Use and Land Cover Change: A CaseStudy in the Brazilian Amazon. Unpublished PhD dissertation, Michigan State Univer-sity, East Lansing, USA.

Cardoso, F. H. and Muller, G. (1977) Amazonia: Expansao do Capitalismo. Brasiliense: SaoPaulo.

Carvalho Brasil, M. (1997) ‘Os Fluxos Migratorios na Regiao Norte Nas Decadas de 70 e80: Uma Analise Exploratoria’. Caderno de Estudos Sociais 13(1): 61–84.

Castro, E. M. R. and Hebette, J. (eds.) (1989) Na Trilha dos Grandes Projetos: Modernizacaoe Conflito na Amazonia. Cadernos do NAEA: Belem.

Centro Peruano de Estudios Internacionales (CEPEI) (2002) in A. Wagner Tizon and S. SantaGadea Duarte (eds.) La Integracion Regional Entre Bolivia, Brasil y Peru. CEPEI: Lima.

Costa, J. M. M. (ed.) (1992) Amazonia: Desenvolvimento ou Retrocesso. Editores CEJUP:Belem.

Hall, A. L. (1989) Developing Amazonia: Deforestation and Social Conflict in Brazil’sCarajas Programme. Manchester University Press: Manchester.

Hall, A. L. (1997) Sustaining Amazonia: Grassroots Action for Productive Conservation.Manchester University Press: Manchester.

Iniciativa para la Integracion de la Infraestructura Regional Suramericana (IIRSA) (2008)[WWW document]. URL http://www.iirsa.org [accessed 1 June 2008].

Katzman, M. T. (1977) Cities and Frontiers in Brazil. Harvard University Press:Cambridge.

Lavinas, L. (ed.) (1987) A Urbanizacao da Fronteira. PUBLIPUR/UFRJ: Rio de Janeiro.MacMillan, G. (1995) At the End of the Rainbow? Gold, Land and People in the Brazilian

Amazon. Columbia University Press: New York.Mahar, D. J. (1979) Frontier Development Policy in Brazil: A Study of Amazonia. Praeger:

New York.

© 2010 The Authors. Bulletin of Latin American Research © 2010 Society for Latin American StudiesBulletin of Latin American Research Vol. 29, No. 4 475

Stephen G. Perz et al.

Malhi, Y., Betts, R. and Roberts, T. (eds.) (2008) ‘Climate Change and the Fate of theAmazon’. Theme Issue. Philosophical Transacions of the Royal Society B 363(1498):1725–1932.

Mitschein, T. A., Miranda, H. R. and Paraense, M. C. (1989) Urbanizacao Selvagem eProletarianizacao Passiva na Amazonia: o Caso de Belem. Edicoes CEJUP: Belem.

Moran, E. F. (1981) Developing the Amazon. Indiana University Press: Bloomington.Nascimento, E. P. and Drummond, J. A. (eds.) (2003) Amazonia: Dinamismo Economico e

Conservacao Ambiental. Garamond: Rio de Janeiro.Oliveira, A. D. and Simoes, A. G., (2004) Deslocamentos populacionais no Brasil: uma

analise dos censos demograficos de 1991 e 2000 [WWW document]. URL http://www.abep.org.br. [accessed 1 June 2008].

Ozorio de Almeida, A. L. (1992) The Colonization of the Amazon. University of Texas Press:Austin.

Perz, S. G. (2000) ‘The Rural Exodus in the Context of Economic Globalization, Crisis andReform in Brazil’. International Migration Review 34(3): 841–880.

Perz, S. G. (2002) ‘Population Growth and Net Migration in the Brazilian Legal Amazon,1970-1996’ in C. H. Wood and R. Porro (eds.) Deforestation and Land Use in theAmazon. University of Florida Press: Gainesville, 107–129.

Roberts, J. T. (1992) ‘Squatters and Urban Growth in Amazonia’. Geographical Review82(4): 441–457.

Schmink, M. and Wood, C. H. (1992) Contested Frontiers in Amazonia. Columbia UniversityPress: New York.

Serrao, E. A. S. and Homma, A. K. O. (1993) ‘Brazil’ in Sustainable Agriculture and theEnvironment in the Humid Tropics. National Academy Press: Washington, 265–351.

Simmons, C. S., Perz, S. G., Pedlowski, M. A. and Teixeira Silva, L. G. (2002) ‘The ChangingDynamics of Land Conflict in the Brazilian Amazon: The Rural-urban Complex and ItsEnvironmental Implications’. Urban Ecosystems 6: 99–121.

Simmons, C. S., Walker, R. T., Arima, E. Y., Aldrich, S. P. and Caldas, M. M. (2007) ‘TheAmazon Land War in the South of Para’. Annals of the Association of AmericanGeographers 97(3): 567–592.

Simmons, C. S., Walker, R. T., Perz, S. G., Aldrich, S., Caldas, M., Pereira, R., Leite, F. andFernandes, L. C. (2010) ‘Doing it for Themselves: Direct Action Land Reform in theBrazilian Amazon’. World Development 38(3): 429–444.

Tavares, V. P., Considera, C. M. and Castro e Silva, M. T. L. L. (1979) Colonizacao Dirigidano Brasil. 2nd edn. IPEA/IUPES: Rio de Janeiro.

Toni, F. and Kaimowitz, D. (eds.) (2003) Municípios e Gestao Florestal na Amazonia. A. S.Editores: Natal.

Wagner de Almeida, A. B. (1995) Carajas: a Guerra dos Mapas. 2nd edn. Editora Supercores:Belem.

Wood, C. H. and Porro, R. (eds.) (2002) Deforestation and Land Use in the Amazon.University of Florida Press: Gainesville.

© 2010 The Authors. Bulletin of Latin American Research © 2010 Society for Latin American Studies476 Bulletin of Latin American Research Vol. 29, No. 4