We are unable to display your institutional affiliation without JavaScript turned on.
Browse Book and Journal Content on Project MUSE

Download PDF

Settlement Formation and Land Cover and Land Use Change
A Case Study in the Brazilian Amazon
Marcellus M. Caldas
Department of Geography, Kansas State University
Cynthia Simmons
Department of Geography, Michigan State University
Robert Walker
Department of Geography, Michigan State University
Stephen Perz
Department of Sociology, University of Florida
Stephen Aldrich
Department of Geography, Geology and Anthropology, Indiana State University
Ritaumaria Pereira
Department of Geography, Michigan State University
Flavia Leite
Department of Sociology, University of Florida
Eugenio Arima
Environmental Studies Program, Hobart & William Smith College

The present paper describes the contentious process of settlement formation of a particular type of land reform settlement, which we call "spontaneous" direct action land reform. In addition, the paper place the settlement formation process within a land cover and land use framework by describing the underlying processes that lead to spontaneous settlement formation in terra devoluta in the Brazilian Amazon. The paper concludes with policy recommendations.


O presente artigo descreve o processo de criação de um tipo particular de assentamento de reforma agrária, no qual o denominamos "espontâneo". Em adição, o artigo coloca o processo de formação de assentamento dentro de uma conceptualização de mudança de cobertura vegetal e uso da terra, e descreve as causas que levam a formação de assentamento espontâneos em terra devoluta na Amazônia Brasileira. O estudo conclue com recomendações de política agrária.

Land Reform, Terra Devoluta, Amazon, Environment
Reforma Agrária, Terra Devoluta, Amazônia, Meio-Ambiente


Concern about loss of the Amazon forest has mounted over the past few decades, given its significance as a reservoir of biodiversity and carbon and given what appears to be an inexorable process of forest destruction. These concerns have spawned a great deal of research and even international programs attempting to elucidate the role of Amazonia in the global climate system (e.g., NASA's Large Biosphere Atmosphere program). Research on Amazonian deforestation—and development—has addressed the farming systems of smallholder households (e.g., Moran 1981; Walker and Homma 1996; Pichón 1997; McCracken et al. 1999; Perz and Walker 2002; Walker 2003; Pan et al. 2004; Aldrich et al. 2006; Caldas et al. 2007), the political economy of opening the region to colonization (e.g., Moran 1981; Schmink and Wood 1984, 1992; Bunker 1985; Hecht 1985; Browder 1988, 1994), and the motivations driving colonist migration decisions (e.g., Skole et al. 1994). [End Page 125]

Although much research has examined key drivers of deforestation, the large literature available has paid little attention to impacts associated with settlement formation, with one recent exception (Simmons et al. forthcoming). For instance, between 2003 and 2005, landless families organized 552 settlements in the Brazilian Amazon (INCRA 2007), and these settlements represent one-third of all new settlements formed throughout Brazil during 2003-2005 (INCRA 2007), pointing to the significance of what has been identified as direct action land reform, or DALR. Direct action land reform is a social and political process involving mobilization of the poor, the contentious occupation of public or private lands, and the formalization of land holdings in the wake of occupation (Simmons et al. forthcoming). However, the settlement formation process is not well understood, given its relative newness, as many of the settlements in the Amazon date from the mid-1990s and are still in the process of active development. Although their land cover impacts are necessarily linked to the farming decisions of small holders and stem from prior colonization and migration decisions, the specific factors leading to the spontaneous formation of settlements remain unknown.

This paper attempts to fill this gap in the literature by presenting an analysis of a particular type of DALR, which we call "spontaneous" or "organic" DALR, to be distinguished from those settlements created during state-led colonization or those orchestrated by landless movements, such as the Movement of Rural Landless Workers (MST). The goals of this article are (a) to understand the social processes leading to spontaneously formed settlements in terra devoluta (unclaimed, unused public land, not yet dedicated to specific use) in the State of Pará; (b) to identify factors affecting spontaneous settlement formation; and (c) to evaluate the importance of such settlements for deforestation.

This article is organized as follows. First, we approach contemporary DALR in Brazil by focusing on the concept of contentious politics and institutional opportunities in Brazil. In the next section, we describe the contention process of settlement formation and place this within a land cover and land use change framework by describing the underlying processes that lead to spontaneous settlement formation. Next, we describe the study site and the data collection, followed by specification and estimation of a set of statistical models addressing settlement formation—in other words, we identify factors affecting spontaneous settlement formation. Finally, we analyze the effects of settlement formation in the landscape through land cover and land use change analysis, and we conclude by considering implications of the findings for both research and policy.

Contentious Politics and Institutional Opportunities

Settlement formation can be approached by invoking the concept of contentious politics as a key underlying process that results in land occupations and settlement creation. Contentious politics refers to collective political struggle that is episodic rather than routine and differs from "politics as usual" by virtue of its innovative, often conflictive, tactics deployed strategically and sporadically to redress a social wrong (McAdam et al. 2001; Sewell 2001). Direct action land reform may be regarded as contentious because it seeks to redress the maldistribution of land and because its premier tactic, land occupation, falls outside conventional societal grievance channels.

Contention in the struggle for land results from a convergence of three important factors: political interests, social needs, and institutional opportunities. Political interests have been well described by research on Brazil's social movements (Cardoso 1983; Mainwaring 1989; Assies 1994; Fernandes 2000, 2001). These studies [End Page 126] have demonstrated that political interest, such as the democratization of Brazil, facilitated the emergence of social movement organizations (SMOs) with specific political agendas, including those focused on land reform. The social need for land reform in Brazil has been the subject of extensive commentary and much research (Muller et al. 1994; Alston et al. 2000, 2005; Fearnside 2001; Simmons et al. 2002; Wolford 2003a; Simmons 2004). In the case of institutional opportunities, we refer here to the structure of land law as well as to the status of certain lands in federal holdings, the terra devoluta. These three factors seem equal in importance to the current situation in Amazonia and, more generally, throughout Brazil. However, we call attention to the particular nature of the institutional opportunity that results from a definition of property stemming from the earliest days of the colony and from a relatively novel statement of land rights found in the 1988 constitution. Under this set of laws, individuals wanting land have clearly stated rights and may possess land under constitutional stipulations. Private land can be expropriated if it does not meet its social use function—in other words, it is not in use or is vacant (article 184 of the 1988 Brazilian Constitution). Also, land can be expropriated if occupied without contestation for an uninterrupted period of 5 years (article 191 of the 1988 Brazilian Constitution). Public land, which is vulnerable to actions of physical possession by land-desirous individuals, cannot be expropriated (article 191, §1 of the 1988 Brazilian Constitution). However, public land in the Brazilian Amazon is always considered vacant land because of its forest. In other words, service and goods provided by standing forests are not taken in consideration as productive use. Consequently, forest land presently is the most vulnerable to occupation. Thus, in the Amazon a sense of "free for all" often prevails, with people taking what they want and paying little or no attention to the legal niceties. In fact, this sense of "free for all" corroborates the spontaneous actions of individuals desiring land through DALR. As a result, landless people initiated their own land reform movement by increasing the number of invasions of private and public lands, leading to settlement formation.

The Conceptual Framework of Settlement Formation

This paper links theory on contentious politics, which provides an understanding of the process of DALR, to the land cover and land use change (LCLUC) literature, which seeks to account for deforestation. To forge such a link, we develop a conceptualization that links underlying factors responsible for fundamental social processes in Brazil to immediate human actions—the proximate causation (e.g., Geist and Lambin 2002). Our underlying/proximate causation (UPC) framework is then used as a means of relating the process of DALR settlement formation to the consequences of deforestation in DALR settlements (Figure 1). It is important to highlight here that our conceptual framework is based, in part, on anecdotal information collected over the past few years of field visits to the region.

The underlying causes of social marginalization

Many factors can be pointed out as underlying causes of social marginalization in Brazil, such as macroeconomic policies, population growth, technological change, political and institutional factors, household demography, and rural poverty, to cite a few. However, two factors can be considered crucial underlying causes of social marginalization. The first is the secular land concentration in the country (Silva 1981), and second, the macroeconomic and agricultural policies that favor technological and economic development (Foweraker 1981, Barros et al. 2001). [End Page 127]

The conceptual framework
Click for larger view
Figure 1. 

The conceptual framework

In fact, Brazil has a long history of land concentration that goes back to colonial times where, from the very beginning of Brazil's colonization, colonization stimulated a disorganized territorial expansion and land acquisition to satisfy an agro-export economy. Consequently, the Brazilian economy could be characterized as an agro-export economy until the beginning of the twentieth century. However, the great depression of the 1930s and the decline in commodity prices in the 1940s led the country to an economic stagnation. Seeking to break this inertia, the government promoted an initial industrial growth that left a heritage of problems, with which policymakers of the 1960s would have to deal with in order to develop the country. First, a severe rural-urban [End Page 128] migration occurred, resulting in an urban population growth. Second, the urban population growth increased demand for food, thus increasing food prices that would lead to inflationary pressures, thus creating social tensions in urban centers. Finally, the industrial growth was accompanied by an accentuation of inequalities at a regional scale that consequently increased social political pressures for solutions (Silva et al. 1984).

To solve these problems, the government changed the focus of the macroeconomic policies in the 1960s to emphasize the "laissez-faire." These policies benefited large-scale enterprises through rural credits and fiscal incentives with high levels of capital and imported technologies. In addition, to facilitate the expansion of agriculture, the federal government developed road-building programs in an effort to bring more land into cultivation (de Barros and Graham 1978). As a result, in less than a decade the country moved from stagnation to economic growth. But, the economic growth did not last long, and the fruits of the rapid expansion were not evenly distributed among regions (Langoni 1973; Tolipan and Tirelly 1975). Also, the oil shocks in the 1970s resulted in an increase of international debt that stimulated inflation, slowed economic growth, and increased unemployment. The outcome was a new change in the macroeconomic policies in the mid-1980s, with an expansion of new agro-industrial production systems that were detrimental to household production.

Thus, one of the major costs of this policy was the displacement of domestic food products by export products in the south of Brazil, in addition to an enormous rural exodus, that changed the labor market from a family property rights to a seasonal work force. These changes led to the creation of a marginalized population without jobs and living in subhuman conditions. Popular dissatisfaction increased in the whole country; and throughout Brazil, social movements of rural landless workers emerged, tired of waiting for concrete action for land reform.

Spontaneous DALR in the State of Pará: The Case of Transamazon

Throughout Brazil, the appearance of DALR land occupation is politically motivated by SMOs, a phenomenon that has already gained the attention of many scholars (Branford and Rocha 2002; Deere 2002; Wright and Wolford 2003; Wolford 1996, 2004, 2005; Ondetti 2006; Simmons et al. forthcoming). In general, research on SMO-led DALR highlights the contentious performances of SMOs seeking land reform, as enacted via the recruiting of rural workers in urban centers, the empowering of landless people to realize their own agency via organized land occupations, the repression and opportunity, the regional differences in land reform, and so forth.

Spontaneous DALR settlement formation in the Amazon stands in sharp contrast to SMO-led DALR. We argue that three factors help to account for spontaneous land occupation in the Transamazon region, none of which is salient for SMO-led DALR. The first factor is linked to the colonization efforts initiated under the military regime in the 1970s; the second relates to the demographic life cycle of old colonists with grown children; and the third concerns the formation of local urban centers that facilitate sharing of information about available land via social networks.

State-led colonization, as along the Transamazon, started in the late 1960s as a military nationalist policy that envisioned the integration of the Amazon to the rest of the country (Moran 1981). Through new roads built in pristine forest the military government attracted poor landless people to colonization settlements. However, during the economic crisis of the 1980s, state budget cuts led to state withdrawal of support for colonization. Colonists therefore forged ties with other social actors and extended roads themselves, seeking land by their own means to expand settlements. Such "unofficial" [End Page 129] road building opened land beyond the original colonization areas but still followed the state property cadastre by replicating property size and spatial organization in the hopes of gaining state recognition of informal land claims (Perz et al. 2007).

Road extensions and settlement expansion went hand in hand with demographic life cycles. The first wave of poor landless migrants who arrived in the region in the late 1960s were often young rural families with young children (Moran 1981; Smith 1982). By the time of state withdrawal, those pioneers (by virtue of the demographic life cycle) had grown children who started to search for land in order to begin their own families (Walker and Homma 1996; Wolford 2003a; Caldas et al. 2007). Such "demographic pressure" thus impelled road building and other strategies to obtain land ideally close to the residence of parents and other kin.

Finally by the 1990s, many of the original agro-villages had indeed grown into small towns (Browder and Godfrey 1997). Thus, as new settlers arrived along the Transamazon, they generally went to such local urban centers. At the same time, established families constructed houses in the same locations. This and the importance of social capital via family and church ties made urban localities very important for social networks and the diffusion of information, including that regarding terra devoluta. Through social networks, information spread rapidly; and even recent migrants heard about terra devoluta, which minimizes the costs and risks associated with the move to an unknown area for new land settlement. Thus, unofficial road building, demographic life cycles, and social networks in urban localities together impelled a form of DALR that does not involve SMOs but nonetheless results in new land settlements without prior state sanction.

The process of spontaneous occupation

In contrast with SMO-led DALR, the process of spontaneous occupation emerges as a consequence of organically evolving rather than planned collective action. Key to understanding spontaneous DALR is that while social networks are important for providing information on which individuals and families may act in occupying land, individuals and families initially act in isolation (or disarticulated from each other) without political interest or organization in the actual act of land occupation. Indeed, in contrast to planned land occupations organized by SMOs to occur in one fell swoop, spontaneous DALR involves a series of surreptitious individual land occupations over time that accumulate incrementally.

For a given settler, the process of spontaneous DALR proceeds similar to the following. Settlers (or posseiros) identify parcels of land suitable for occupation through the use of friends and family's knowledge of the area (Caldas 2008). The new settler then follows a colonization road to its end and opens trails (picadas) into the forest beyond (Caldas 2008; Perz et al. 2008). The settler then identifies a desirable land parcel and engages in informal demarcation (marcação) of the land that reproduces the dimensions of the old colonization parcels, thus extending the original colonization settlement cadastre. By following the preexisting road network, settlers try to reproduce the original settlement pattern of the old colonization zones with the ultimate objective of facilitating state recognition. Because of the lack of money necessary to clear forest and thereby develop and claim the land, settlers make use of valuable trees by exchanging them for road construction services, which provides start-up capital .

The next step involves invocation of the Land Statute clause of morada habitual (habitual residence) that obligates settlers to show that they are living on the land, as their absence for a long period of time can provide legal justification for [End Page 130] another settler to claim the land. However, the morada habitual clause is not a sufficient condition to assure a definitive land title; settlers also need to make use of the land. Consequently, deforestation ensues to demonstrate productive use, a legal precondition for gaining title by the principle of usocapião . This is why selling trees and gaining start-up capital is crucial to spontaneous DALR settlement formation. Initial deforestation provides a concrete, demonstrable basis to pressure the federal government to demarcate the occupied land as habitual residence; and productive use fulfills the constitutionally mandated social functions of land in Brazil.

Over time, settlers accumulate; and the land area informally settled via spontaneous DALR expands, eventually leading to the last step, which involves mobilization. The shared experience of spontaneous DALR settlement on state lands helps constitute solidarity among neighbors, who accumulate trust and confidence and thus social capital. Settlers thus form small communities that can be mobilized and may lead to the creation of local associations and other forms of organization. Once a critical mass of settlers is in place, a collective awareness about the institutional opportunities emerges. Mobilization based on solidarity and an understanding of Brazilian agrarian law provides the basis to pressure the National Institute of Colonization and Land Reform (INCRA) for official recognition of the spontaneous DALR settlement. Thus, in spontaneous DALR, mobilization occurs after the land has been occupied; and mobilization itself occurs to demand land legalization (regulamentação fundiária).

In such a situation, INCRA recognizes the DALR occupation by officially "creating" a settlement project. This constitutes a state-sanctioned mechanism for entraining a series of benefits to the settlers, including the possibility of applying for and obtaining land title, as well as the liberation of state resources for road building and other social services useful to agricultural activity and other aspects of settlement development. Consequently, with time and the consolidation of the settlement, farming systems expand and deforested land area increases. In other words, the formation of the official settlement acts as a proximate cause of land cover change through official state recognition and provision of additional resources for farming activities.

The Study Area and Data Collection

Given that settlement formation is a relatively new activity, as many of the settlements are still in phases of development, we collected data in two phases for purposes of understanding the process of spontaneous settlement formation. Phase one involved interviewing key informants in the summer of 2005 in order to identify DALR settlements in the Amazon. In Brazil, settlement creation is the task of the federal government agency INCRA, which is responsible for implementing colonization and land reform. We therefore visited INCRA headquarters in Marabá and Altamira, which are responsible for settlement formation in the eastern Amazon, to discuss with its officers the best sites for our research. Our main interests were in locations where contention could be occurring and DALR would be present. Through these key informant interviews and due to time constrain, we identified randomly 13 DALR settlements in two locations where DALR is particularly prevalent in the eastern Amazônian state of Pará. The first location was in southern Pará (Figure 2), an area that has experienced longstanding conflict over land between SMOs and large landlords (Simmons et al. 2007). The second location was the Transamazon region where state-led colonization occurred in the 1970s.

The second phase of data collection occurred in the summer of 2006 and involved a field survey. Data collection in DALR settlements is not a simple task. INCRA [End Page 131] officers had to contact settlement leaders in advance and explain our research goals. In addition, leaders of social movement organization were introduced to team leaders. The long period of experience of the team leaders in working in Amazonian problems facilitate the development of trust and confidence between researchers, SMO leaders, and community leaders, which facilitate a 17-page survey of families in each of the 13 DALR settlements.

The study sites – the Trans-Amazon region and the South of Pará State
Click for larger view
Figure 2. 

The study sites – the Trans-Amazon region and the South of Pará State

Survey participants were identified by the opportunistic method (Caldas et al. 2007). In other words, the difficulties of field data collection complicated formal efforts at geospatial sample design (Griffith 2005). Moreover, random sampling by distance (every nth kilometer) and household count (every nth farmstead) proved to be impractical. Counting houses was problematic due to proprietor absence, and houses were sometimes hidden from view along the road. For reasons such as these, the analysis sample was said to be opportunistic; in essence we conducted interviews with people we could find, building geographic representation by daily updates of the survey geography of settlement maps. This survey collected information on lot history (including lot size, year of establishment, land title, length of residence, and area in forest at arrival), landownership history, household demography (family size, age composition, and migration pattern), socioeconomic conditions at arrival, previous experience with DALR, settlement history (including conflicts, violence, road building, and land demarcation), and farming system components. Altogether, data were collected for 396 settlers, consisting of 177 settlers located in the Transamazon region and 219 settlers located in southern Pará.

We also interviewed community and union leaders responsible for supporting settlement formation in the study areas as well as prominent settlers, such as presidents of settlement associations and older settlers living in the settlements. This approach [End Page 132] took into consideration the possibility that some interviewees did not know the history of their settlement. We thus compared responses from multiple interviewees in a given settlement in order to understand its entire history. In the key informant interviews, we discussed settlement history, paying special attention to how settlement locations were selected. We asked explicit questions in this regard, particularly about the legal issues involved in land occupation and about any reactions by municipal, state, or federal governments as well as by other interested parties, such as loggers and landowners. This method yielded histories of unofficial settlement formation by nonstate social actors. From these key informant interviews, separating three types of settlements was possible in our sample: spontaneous, SMO-led settlements, and hybrids that combined elements of SMO and spontaneous DALR . Consequently, four settlements were identified as spontaneous (all located in the Transamazon), five were identified as SMO-led (all in southern Pará), and four as hybrid.

In addition to data collection to understand the process of spontaneous settlement formation in the State of Pará, we also gathered data for purposes of evaluating the effects of spontaneous DALR settlement on deforestation. Thus, we conducted a remote sensing analysis of deforestation occurring in the four spontaneous DALR settlements along the Transamazon Highway. We acquired Landsat imagery (TM and ETM+, paths 227, 226 and rows 62, 63) covering a 13-year period (1986, 1991, 1999) for the Transamazon region around the municipality of Uruará. Prior to classification, a model was created in ERDAS Image Modelmaker for radiometric and atmospheric correction. Geometric correction accuracy was evaluated by matching common points on corrected images to points collected by GPS units from several locations within the Uruará area. Image rectification using 8- to 10-point coordinates were applied. We considered individual geometric correction acceptable only if the root-mean square (RMS) was <0.5 pixels. The last step was image classification. Landsat Thematic Mapper (1986 and 1991) and Enhanced Thematic Mapper (1999) images were individually classified. To carry out this task, we used a hybrid unsupervised/supervised classification method in each individual image. Through this method, four thematic classes were identified: forest, nonforest, clouds and shadows, and water. Comparisons of classified images over time allowed calculation of deforestation between the three time points. In turn, we can compare deforestation among time points to information on the timing of spontaneous DALR activity and the formal recognition of settlements, which is available in public records (portarias). All four spontaneous DALR occupations along the Transamazon were recognized as INCRA settlements during the time period encompassed by the imagery, allowing an appraisal of deforestation before and after regularization. Table 2 shows the date of recognition and the imagery year for the spontaneous settlements.

Framing the Regression Model of Spontaneous Settlement Formation

As stated before, our conceptual framework outlined in the previous sections was based, in part, on anecdotal information collected over the past few years of field visits to the region, and our discussion to this point has provided an overview of the spontaneous DALR land occupation that allows us to formulate two preliminary hypotheses about spontaneous settlement formation. The first hypothesis argues that spontaneous DALR seeks out terra devoluta beyond the frontier . In general, terra devoluta in the Amazon is found at great distances from cities and has low land value because of the existence of primary forest and lack of infrastructure. Consequently, terra devoluta may minimize the risks of contentions for individuals who occupy land without organizational support or assistance from the state. Thus, we expect that the existence of terra devoluta will [End Page 133] stimulate the appearance of spontaneous settlement formation. However, the exact location of terra devoluta is not easy to acquire. Prior knowledge about terra devoluta can be a consequence of an informal network that transmits information outside the region and, thus, attracts people and stimulates migration. Also, knowledge about available land can be related to length of residence in the area. Consequently, individuals with a long length of residence are expected to have more knowledge about vacant areas and consequently may have a positive impact on spontaneous settlement formation. Notably, the conceptual model also suggests a number of factors that indirectly or directly are affecting spontaneous settlement formation. For instance, it argues that settlers act in isolation and have no experience in land contention and mobilization and, consequently, have more involvement in spontaneous settlement formation than an SMO settlement. Also, social mobilization is a defining characteristic of SMOs; and SMOs, such as the MST, seek to recruit more people from urban places. Thus, we expect that people with more urban-to-rural migration may have a lower involvement in spontaneous settlement.

Finally, the conceptual framework also argues that an endogenous process could be occurring in these settlements, as grown children of early pioneers are searching for land to begin their own farm. Consequently, our second hypothesis argues that the existence of kin in colonization zones could increase the probability of spontaneous settlement formation.

The Statistical Model

The theoretical model used in this paper suggests a number of relationships between spontaneous settlement (to be regarded statistically as the dependent variable) and a set of independent variables reflecting elements of the contentious politic theory, network information, urban-to-rural migration, and the existence of terra devoluta. Because our goal is to identify which factors are stimulating spontaneous settlement formations, we developed a Logit model to test the first and second hypotheses stated in this article.

In specifying the Logit model, a dichotomous dependent variable was set equal to 1 (one) if the settlement is a spontaneous settlement, or 0 (zero) otherwise (SMO-led or hybrid settlement). Consequently, we expected the predicted value of the dependent variable to fall mainly within the interval between 0 and 1 and then assuming a response probability that is linear in a set of parameters, ßj, such that:

INSERT DESCRIPTION - inline graphic

where (ß0 + Xß) is equal to f, H being a function of f, thus, taking on values strictly between 0 and 1: 0<H( f )<1, for all real numbers f. This ensures that the estimated response probabilities are strictly between 0 and 1. Various nonlinear functions have been suggested for the function H in order to make sure that the probabilities are between 0 and 1. In the Logit model, H is assumed to be the logistic function, such that H( f ) =exp ( f )/[1+exp( f )] which is between 0 and 1 for all real number f. This Logit model suggests that the predicted value of the dependent variable could be interpreted as the probability of formation of a spontaneous settlement rather than an SMO-led (or hybrid settlement) given a set of explanatory variables.

Results and Discussion

The Regression Model

Table 1 presents three Logit models associated with our hypotheses. The first model considers only variables associated with settler's experience with contention and [End Page 134] urban to rural migration. In other words, it evaluates if involvements with SMOs and migration history have effects in spontaneous settlement formation. The second model seeks to incorporate variables that capture the effects of social networks in identifying land for occupation. In addition, it evaluates the effects of distance from cities in the formation of spontaneous settlements. Finally, the last model tries to capture the effects of terra devoluta and the existence of kin in colonization areas.

Logit regression of settlement formation
Click for larger view
Table 1. 

Logit regression of settlement formation

We start by analyzing in model 1 if settler experience with contention for land reform has any effect on spontaneous settlement formation in the Amazon. First, we observe that a settler's personal experience in SMOs DALR activities (such as barricades, invasion of public offices, land invasion, and so on) is in agreement with the conceptual model. In other words, contentious experiences in DALR activities do not serve as a stimulus to spontaneous settlement formation. In fact, settler's experience in SMO activities decreases the probability that spontaneous settlement will emerge. Also, the involvement of relatives in DALR activities may serve as influential examples of the success of the struggle for land reform, and this may influence landless people to join SMOs. Thus, our results show that kin experience in DALR activities will decrease the probability that a spontaneous settlement will materialize. In addition, our findings [End Page 135] confirm that urban-to-rural migration, measured by the number of times an individual moved from rural to urban areas, have a negative effect on spontaneous settlement formation. As expected, urban residential experience lowers the probability that an individual will participate in spontaneous DALR settlement formation.

In the second model, we examined the effects of prior knowledge about terra devoluta and distance from the city. This provides a means of evaluating risk tolerance among settlers, where risk is greater among settlers occupying previously occupied lands or private land. The variable "prior knowledge" shows a positive effect in the appearance of spontaneous settlement, and the findings confirm our expectation that settlers have social networks of information about terra devoluta. However, contrary to our expectations, distance to city has a negative effect on spontaneous DALR formation. In other words, we expected to see the formation of spontaneous settlements beyond the frontier. Notably, nevertheless, spontaneous settlements in the Transamazon region are located on average between 23 and 63 km from the nearest cities. These distances are less than those found for SMO settlements, most of which are found in southern Pará. However, distance measured in kilometers does not necessarily reflect the real cost associated with transportation given road conditions. In general, traveling 60 km takes less than an hour in southern Pará, in contrast to the more than 5 hours needed to cover the same distance in the Transamazon region. Consequently, a pure distance measure, as used in the regressions, may not reflect actual travel times, which is the decisive factor.

The third model adds the effects of length of residence and the occupation of terra devoluta in spontaneous settlement formation. Our findings corroborate our expectations. First, individuals with a long length of residence are expected to have more knowledge about vacant areas (terra devoluta), which consequently increases the probability of the appearance of spontaneous settlements. Second, our results confirm that settlers who occupied terra devoluta were more likely to create spontaneous DALR settlements than other types of DALR settlements. Thus, the type of land occupied is one of the main institutional factors differentiating spontaneous settlements from those created by SMOs. Finally, we tested the hypothesis that having relatives living in colonization zones would stimulate the appearance of spontaneous settlements. The results validate our expectations and suggest that early land reform policies that stimulate migration to the region are still working via the life-cycle effect.

In sum, all three models present robust results and provide important information about the complexity of spontaneous settlement formation in the Transamazon region, and they confirm the expectations that spontaneous settlements have some characteristics that differentiate them from SMO settlements. In addition, the results suggest and are reinforced by the idea that these settlements have, in some way, been affected by early land reform that attracted colonists to the region. Consequently, given that the new spontaneous settlements are located in areas of primary forest, we investigate their effects in the landscape.

Spontaneous Settlements and their Effect on Deforestation

The final goal of our article seeks to evaluate the effects of spontaneous settlement formation in the landscape and argues that spontaneous DALR are capable of increasing deforestation. This necessarily occurs as the establishment of new settlements requires forest clearing for agriculture to demonstrate productive land use, a requirement for state recognition under Brazilian agrarian law. Thus to see the effect of spontaneous settlement formation in the landscape, we made use of multitemporal satellite imagery [End Page 136] to calculate the deforestation magnitudes for four spontaneous settlements found in the Transamazon area. Table 2 presents the results by comparing deforested areas for 1986 to 1991 to 1999. The results show that area in forest decreased while deforestation increased. However, differences in the deforestation rates among the four settlements are notable. The oldest spontaneous DALR settlement, Surubim, has more area deforested than the newer settlements. What is more, in Surubim, deforestation more than doubled after official recognition by INCRA. This settlement has its own history linked to the region. It was recognized in 1988, and it is one of the oldest settlements along the Transamazon with its formation sometimes merging with the history of the Transamazon highway itself. Long-term residents show larger areas of cultivation or pasture as a function of length of residence. Consequently, the expansion of land use led to increases in deforestation over time.

Deforestation Measures by Settlement
Click for larger view
Table 2. 

Deforestation Measures by Settlement

By contrast, in the settlement Trairão, deforestation only increased by a few hectares. These different results may be related to lot demarcation. In other words, before 1993, parcel locations were defined by the settler's own desires. After law 8,629 of 25 February of 1993, parcel demarcations had to conform to the Plano de Desenvolvimento de Assentamento (PDA). Consequently, through this law the federal government enforced the necessity of studies as a way to plan the spatial organization of the community and its respective land production sites. However, delays in the implementation of Trairão's PDA were restricted to the delimitation of its boundaries but not its parcels. While both settlements have official recognition by INCRA, settlers in Trairão do not know where INCRA will (re)locate them within the settlement. Consequently, settlers do not want to invest in clearing an area without a guarantee that they will receive a title to that area. As a result, settlers in Trairão rely more on annual crops for subsistence, thus clearing small areas every year. Also, settlers in this area are not investing in perennial crops or pasture. They know where they would like to live, but insecurity plays a role in land use decisions and, consequently, in deforestation.

The other two settlements, Rio do Peixe and Uirapuru, show similar trends in deforestation, with deforestation increasing after government recognition. Also, deforestation might be associated with land use systems and parcel delimitations, since the settlements Rio do Peixe, Uirapurú, and Surubim have (on average) more area in pasture and perennial crops than the Trairão settlement. The results nonetheless suggest that spontaneous DALR occupation on terra devoluta affects the landscape, especially when occupations are officially recognized by INCRA. Before government authorization and parcel delimitation, few tracts of deforested land inside the settlements existed. After settlement creation, infrastructure improved through credit for road building and agriculture development, increasing deforestation. Consequently, over time this process can bring serious consequences to the environment. As forest patch size decreases and as patches of habitat become more isolated from one another, species population sizes dependent upon contiguous blocks of forest may decline below the threshold needed to maintain their survival, reducing biodiversity. [End Page 137]


This paper calls for further attention to the underlying social and institutional circumstances contributing to tropical deforestation by focusing on spontaneous direct action land reform (DALR). In the Brazilian context, political interests and institutional opportunities constitute a circumstance that motivates various types of land occupations, including the spontaneous DALR on which we have focused here. Findings from the Logit models in the first part of our analysis confirm our expectations concerning the factors influencing participation in spontaneous land reform (length of residence, terra devoluta, kin ties, etc.), thus supporting our first and second hypotheses. Further, analysis of multitemporal satellite imagery for four spontaneous DALR settlements confirms our hypothesis that deforestation increases over time in spontaneous DALR settlements, especially after INCRA regularization and in those cases where lots are clearly allocated to settlers. However, different from other parts of the country, the occupation that follows seems to minimize the cost and risk of occupying land without organizational support. The previous knowledge of the area, even though not robust across the models, suggests that a network of information may exist. Also, apparently, length of residence in the settlement is an important factor in explaining spontaneous settlement appearance. Long periods of residence allow better knowledge of the area and favor the appearance of social capital among settlers; and the existence of kin in a colonization zone may serve as a stimulus to demarcate land for spontaneous settlement formation.

Contention in spontaneous settlements is nearly nonexistent, and when it occurs, it is ex-post. In other words, land reform occurs after individuals acting in isolation or disarticulated from each other have occupied their lands. Contention has a lower probability in influencing spontaneous settlement formation. However, this result does not mean that the region containing these settlements will always be "contention free." Every day, a dozen migrant families arrive in Uruará, a county in the sample of spontaneous settlements (key informant interview 2004). Many of them will probably have previous experience with DALR occupation, and this experience will affect the way that land occupation occurs in the region.

Migration to the Transamazon region is still small when compared to southern Pará where hundreds of landless families arrive every day (key informant interview 2006) and where a paved transportation system serves as a hub linking the Amazon to the northeast and west-central areas of Brazil. Thus, imagining that migration will increase as soon as the road infrastructure along the Transamazon improves is not difficult. In addition, the improvements in road infrastructure will decrease distances and, thus, will facilitate SMOs to recruit and mobilize people for DALR occupations and settlement formation. However, without control, the new settlements may increase land cover change. Apparently, from the deforestation analysis presented here, settlement formation on terra devoluta is modifying the landscape. Before government authorization for settlement formation, the areas where the settlements are located showed few large tracts of forest inside their boundaries. After settlement creation, forested areas decrease and deforestation increases. However, the magnitude of deforestation varies with settlement socioeconomic history. The discussion suggests that residents in spontaneous DALR settlements may be developing farming systems that involve expanded land use for crops and pasture over time, with the same environment impacts as previously seen in older colonization sites where a similar process of farm development occurred. This implies that spontaneous DALR yields expanded deforestation over time. [End Page 138]

Government approval generates incentives for demands, such as credit for pasture formation, cattle acquisition, and infrastructure improvements. Despite this, government approval for settlement formation is not a sufficient condition to stimulate deforestation. Posseiros need to organize and pressure INCRA to demarcate their lots for documentation of their land rights, which facilitates loans from credit agencies. Here, the policy aspect will play an important role in Amazon land reform. Many studies have shown that credit lines in colonization areas have favored cattle acquisition and pasture formation, and evidence is not available that will change in the future. Terras devolutas along the Transamazon highway have immense areas in primary forest; and if this same approach continues, it suggests that ranch formation will increase deforestation in this region. From a policy perspective, increasing credit lines would be easier for perennials crops, such as cocoa, rubber trees, and fruit trees (e.g. cupuaçu, Theobroma grandiflorum). This suggestion seems to be an "old-fashioned" idea. However, the failure of this policy in the past is linked to poor infrastructure and lack of technical assistance. The excuse that a market does not exist for these products sounds false. What makes these products noncompetitive with others are their higher transportation cost, low productivity, and lack of technical assistance that will change in the future.

Finally, the role of the state agency INCRA needs to change. This agency was created in the 1970s to implement agrarian reform in Brazil. However, since its creation, INCRA has faced many accusations of irregularities and corruption that have undermined the public confidence in this institution. These problems are deteriorating INCRA's capacity to stay one-step ahead of the problem. On the contrary, INCRA is now accused to be behind the problem. In conclusion, if the goal of the land reform program in Brazil is to bring social justice and dignity to the poor, then it is time to recognize past mistakes and adapt the land policy to the new reality in the Amazon that takes into consideration the environmental problems that current laws are causing. If we do not act now, the future of the poor in the region will not change; and the same cyclic processes of land occupation and degradation will occur until no forest will remain to support life in the region.


The research presented in this article was supported by the National Science Foundation (Geography and Regional Science) under the project (0522062) "Brazil's Direct Action Land Reform Movement: Environmental Impacts and Socio-Spatial Dynamics." Any opinions, findings, conclusions, or recommendations expressed in this article are those of the authors and do not necessarily reflect the views of the National Science Foundation. We would like to thank the anonymous reviewers and editor for their many useful comments made on an earlier version of this article.


1. An important distinction exists between occupation and invasion. From the perspective of the landless people, they occupy a property. From the perspective of the landowners, the landless people invade a property. In this article, we choose to use occupation because we are attempting to describe the actions of landless people.

2. See Arima et al. (2005) and Perz et al. (2007) for more details about road building along the Transamazon highway. [End Page 139]

3. Usocapião is a form of land acquisition of a property by using the land for many years without interruption.

4. Although the literature enables us to classify settlement history in two categories, the contemporary DALR in the Amazon allows us to identify a third category. A hybrid settlement is a combination of spontaneous settlement (at least at the beginning of the occupation), with the appearance of radical leadership that seeks alliance with government officials sympathetic to land reform, as well as with SMOs to facilitate settlement creation (for more details see Simmons et al. forthcoming).

5. The concept of frontier is open to intellectual interpretation. For instance, it has been viewed as the historical rupture marking the arrival of modernity (Watts 1992), the perennial reoccurrence of resource territorialization by competing groups (Little 2001), and the spatial limit between subsistence and market-oriented agriculture (Katzman 1977). In this paper, we adopt the concept developed in Simmons et al. (2007).


Aldrich, S. P., Walker, R.T., Arima, E., Caldas, M., Browder, J. and Perz, S. 2006. Land Cover and Land Use Change: Processes of Social Stratification in Agricultural Expansion. Economic Geography 82(3): 265-288.
Alston, L. J., Libecap, G. D. and Mueller B. 2000. Land Reform Policies, the Sources of Violent Conflict, and Implications for Deforestation in the Brazilian Amazon. Journal of Environmental Economics and Management 39(2): 162-188. [CrossRef]
———. 2005. How Interest Groups with Limited Resources can Influence Political Outcomes: Information Control and the Landless Peasant Movement in Brazil. Working paper EB2005-0005. Institute of Behavior Science, Boulder, USA.
Arima, E. Y. , Walker, R., Perz, S. G., Caldas, M. 2005. Loggers and Forest Fragmentation: Behavioral Models of Road Building in the Amazon Basin. Annals of the Association of American Geographers 95: 525-541. [CrossRef]
Assies, W. 1994. Urban Social Movements in Brazil: A Debate and Its Dynamics. Latin American Perspectives 21(2): 81-105. [CrossRef]
Branford, S. and Rocha, J. 2002. Cutting the Wire: The Story of the Landless Movement in Brazil. London: Latin America Bureau.
Browder, J. 1988. Public Policy and Deforestation in the Brazilian Amazon, pp. 247-298, in R. Repetto and M. Gillis, M., (eds.), Public Polices and the Misuse of Forest Resources. Cambridge, UK: Cambridge University Press.
———. 1994. Surviving in Rondonia: The Dynamics of Colonist Farming Strategies in Brazil's Northwest Frontier. Studies in Comparative International Development 29(3): 45-69. [End Page 140] [CrossRef]
Bunker, S. 1985. Understanding the Amazon: Extraction, Unequal Exchange, and the Failure of the Modern State. University of Illinois Press, Urbana.
Caldas, M. 2008. Settlement Formation and Land Cover and Land Use Change: A Case Study in the Brazilian Amazon. Unpublished Ph.D. dissertation, Department of Geography, Michigan State University, East Lansing.
Caldas, M., Walker, R., Arima, E., Perz, S., Wood, C., Aldrich, S. and Simmons, C. 2007. Theorizing Land Cover and Land Use Change: The Peasant Economy of Amazonian Deforestation. Annals of the Association of American Geographers 97(1): 86-110. [CrossRef]
Cardoso, R. C. 1983. Movimentos sociais urbanos: balanço critico, pp. 215-239, in Bernardo Sorj and Maria H. T. de Almeida (eds.), Sociedade e Política no Brasil pós 1964, São Paulo: Brasiliense.
de Barros, J. R. M. and Graham, D. H. 1978. A agricultura brasileira e o problema da produção de alimentos. Pesquisa e Planejamento Econômico 8(3): 695-726.
Deere, C. D. 2002. Diferenças regionais na reforma agrária brasileira: gênero, direitos à terra e movimentos sociais rurais. Estudos Sociedade e Agricultura 18: 112-146.
Fearnside, P. M. 2001. Land-Tenure Issues as Factors in Environmental Destruction in Brazilian Amazonia: The Case of Southern Para. World Development 29(8): 1361-1372. [CrossRef]
Fernandes, B. M. 2000. A Formação do MST No Brasil. Petrópolis: Vozes.
———. 2001. The Occupation as a Form of Access to Land. Paper presented at the meeting of the Latin American Studies Association, Washington, DC.
Foweraker, J. 1981. The Struggle for Land: A Political Economy of the Pioneer Frontier in Brazil, 1930 to Present. Cambridge, U.K.: Cambridge University Press. [CrossRef]
Geist, H. J. and Lambin, E. F. 2002. Proximate Causes and Underlying Driving Forces of Tropical Deforestation. BioScience 52(2): 143-150. [CrossRef]
Griffith, D. A. 2005. Effective Geographic Sample Size in the Presence of Spatial Autocorrelation. Annals of the Association of American Geographers 95: 740-60. [CrossRef]
Hecht, S. B. 1985. Environment, Development, and Politics: Capital Accumulation and the Livestock Sector in Eastern Amazonia. World Development 13: 663-684. [CrossRef]
INCRA (Instituto Nacional de Colonizaçãe Reforma Agrária). 2007. Diretoria de Obtenção de Terras e Implantação de Projetos de Assentamento, Relatório 227. Brasília: Brazil.
Langoni, C. 1973. Distribuição da Renda e Desenvolvimento Econômico do Brasil. Rio de Janeiro, Brazil. Editora Expressao e Cultura. [End Page 141]
Mainwaring, S. 1989. Grassroots Popular Movements and the Struggle for Democracy: Nova Iguacu, pp. 168-204, in A. Stepan (ed.), Democratizing Brazil. New York: Oxford University Press.
McAdam, D., Tarrow, S. and Tilly, C. 2001. Dynamics of Contention. Cambridge, UK: Cambridge University Press.
McCracken, S. D., E. Brondizio, D. Nelson, E. Moran, A. Siqueira, and C. Rodriguez-Pedraaz, 1999. Remote Sensing and GIS at Farm Property Level: Demography and Deforestation in the Brazilian Amazon. Photogrammetric Engineering and Remote Sensing 65:1311-1320.
Moran, E. F. 1981. Developing the Amazon. Bloomington: Indiana University Press.
Muller, B., Alston, L., Libecap, G.D., Schneider, R. 1994. Land, Property Rights and Privatization in Brazil. The Quarterly Review of Economics and Finance 34: 261-280. [CrossRef]
Ondetti, G. 2006. Repression, Opportunity, and Protest Explaining the Takeoff of Brazil's Landless Movement. Latin American Politics & Society 48(2)(Summer): 61-94. [Project MUSE]
Pan, W., Walsh, S., Bilsborrow, R. E., Frizzelle, B., Erlien, C. and Baquero, F. 2004. Farm-Level Models of Spatial Patterns of Land Use and Land Cover Dynamics in the Ecuadorian Amazon. Agriculture, Ecosystems and Environment 101(2-3): 117-134. [CrossRef]
Perz, S. and Walker, R. T. 2002. Household Life Cycles and Secondary Forest Cover among Smallholders in the Amazon. World Development 30(6): 1009-1027. [CrossRef]
Perz, S., Caldas, M. M., Arima, E. Y. and Walker, R. T. 2007. Socio-Spatial Processes of Unofficial Road-Building in the Amazon: Socioeconomic and Biophysical Explanations. Development and Change 38: 529-551. [CrossRef]
Perz, S., Caldas, M., Walker, R., Arima, E. and Souza Jr., C. 2008. Road Networks and Forest Fragmentation in the Amazon: Explanations for Local Differences with Implications for Conservation and Development. Journal of Latin American Geography 7(2): 85-104. [Project MUSE]
Pichón, F. J. 1997. Colonist Land Allocation Decisions, Land Uses and Deforestation in the Ecuadorian Amazon Frontier. Economic Development Cultural and Change 35(3): 707-744.
Schmink, M. and Wood, C. 1992. Contested Frontiers in Amazonia. New York: Columbia University Press.
Sewell, W. H. 2001. Space in Contentious Politics, pp. 51-88, in R. R. Amizade, J. A. Goldstone, D. McAdam, E. J. Perry, W. H. Sewell, S. Tarrow and C. Tilly, (eds.), Silence and Voice in the Study of Contentious Politics. Cambridge, UK: Cambridge University Press. [End Page 142]
Silva, J.G. da. And B. Kohl. 1984. Capitalist "Modernization" and Employment in Brazilian Agriculture, 1960-1975. Latin American Perspectives Vol. 11(1): 117-136. [CrossRef]
Silva. J. G. da. 1981. Modernizacao dolorosa: estrutura agraria, fronteira agricola e trabalhadores rurais no Brasil. Rio de Janeiro: Zahar.
Simmons, C., Walker, R., Perz, S., Aldrich, S., Caldas, M., Pereira, R., Leite,F., Fernandes, L.C., and Arima, E. (in press). Doing it for Themselves: Direct Action Land Reform in the Brazilian Amazon. World Development (2009), doi:10.1016/j.worlddev.2009.06.003
Simmons, C. S., Walker, R., Arima, E., Aldrich, S., and Caldas, M. 2007. The Amazon Land War in the South of Pará. Annals of the Association of American Geographers Vol. 97(3): 567-592. [CrossRef]
Simmons, C. S. 2004. The Political Economy of Land Conflict in the Eastern Brazilian Amazon. Annals of the Association of American Geographers 94(1): 183-206. [CrossRef]
Simmons, C. S., Perz, S., Pedlowski, M. and Silva, L. G. T. 2002. The Changing Dynamics of Land Conflict in the Brazilian Amazon: The Rural-Urban Complex and Its Environmental Implications. Urban Ecosystems 6: 99-122. [CrossRef]
Skole, D., Chomentwoski, W. H., Salas, W. A. and Nobre, A. D. 1994. Physical and Human Dimension of Deforestation in the Amazon. BioScience 44(5): 314-322. [CrossRef]
Smith, Nigel. 1982. Rainforest Corridors: The Transamazon Colonization Scheme. Berkeley: University of California Press.
Tolipan, R. and Tirelly, A. C. 1975. A Controvérsia sobre Distribuição da Renda e Desenvolvimento. Rio de Janeiro: Zahar.
Walker, R. T. 2003. Mapping Process to Pattern in the Landscape Change of the Amazonian Frontier. Annals of the Association of American Geographers 93(2): 376-398. [CrossRef]
Walker, R. T. and Homma, A. 1996. Land Use and Land Cover Dynamics in the Brazilian Amazon: An Overview. Ecological Economics 18(1): 67-80. [CrossRef]
Wolford, W. 1996. Edible Ideology? Survival Strategies in Brazilian Land-Reform Settlements. Geographical Review 86(3): 457-461. [CrossRef]
———. 2003. Families, Fields, and Fighting for Land: The Spatial Dynamics of Contention in Rural Brazil. Mobilization: An International Journal 8(2): 201-215.
———. 2004. This Land is Ours Now: Spatial Imaginaries and the Struggle for Land in Brazil. Annals of the Association of American Geographers 94(2): 409-424. [End Page 143] [CrossRef]
———. 2005. Agrarian Moral Economies and Neo-Liberalism in Brazil: Competing World-Views and the State in the Struggle for Land. Environment and Planning A 37: 241-261. [CrossRef]
Wright, A. and Wolford, W. 2003. To Inherit the Earth: The Landless Movement and the Struggle for a New Brazil. Oakland, CA: Food First! Publications. [End Page 144]