Agrarian Reform in the Brazilian Amazon: Its Implications for Land Distribution and Deforestation

11
Agrarian Reform in the Brazilian Amazon: Its Implications for Land Distribution and Deforestation PABLO PACHECO * Center for International Forestry Research (CIFOR), Bogor, Indonesia Summary. This paper examines the interactions between state-led land reform, agrarian structures, and deforestation in the Brazilian Amazon. Land reform tends to promote land redistribution through regularization of smallholder land invasions of large-scale landhold- ings, and by redistribution of public lands to smallholders in existing colonization frontiers. The implications of state-led land reform on deforestation are heterogeneous. I argue that impacts of agrarian reform are strongly related to the pre-existing social and economic configuration of the frontiers where it takes place. While it leads to lower rates of deforestation in landscapes dominated by smallholders with diversified land use, its impact on forest conversion is higher in landscapes where extensive land use, mainly prompted by large-scale cattle ranching, tends to dominate. I provide an assessment for the whole Brazilian Amazon, and examine two research sites in the state of Para ´ , namely Uruara ´ and Redenc ßa ˜ o. The study is based on informant interviews, secondary information, agricultural census data, and remote sensing data. Ó 2009 Elsevier Ltd. All rights reserved. Key words — Brazil, Amazon, agrarian reform, deforestation, land tenure 1. INTRODUCTION In this study, I assess the dynamics of agrarian reform tak- ing place in the Brazilian Amazon, and evaluate their direct and indirect implications for land tenure and deforestation. Agrarian reform is a state-led redistributive land reform which aims at granting land rights to smallholders under two vari- ants. The first consist in distributing available public lands to smallholders, and the second in redistributing large land- holdings that have been occupied by smallholders through land invasions. I do not assess in this study other forms of land reform underway in the Brazilian Amazon aimed at the regis- tration of indigenous people and traditional communities land rights. The impact of agrarian reform on land distribution in the Brazilian Amazon still has little influence in changing the existing agrarian structure given its limited scope. There is still a highly skewed land distribution in the region. Smallholdings are more often in distant locations, whilst occupation by large- holdings occurs more actively in areas relatively closer to mar- kets, and where there are better roads. Agrarian reform, therefore, is mainly aimed to, on one side, distributing avail- able public lands and, on the other side, redistributing lands and formalizing property rights in invaded large private land- holdings. Agrarian reform benefits both smallholders and landless people. Much of the outcomes of agrarian reform on deforestation are contradictory in practice because they depend on the resul- tant agricultural systems. While in some cases smallholders de- velop complex land-use mosaics in which some forest cover is retained, in other cases they turn to more specialized activities such as cattle ranching, which leads to convert most of the landscape into pasture. In opposition to common wisdom, smallholders do not show a much greater propensity to defor- est than largeholders, and in some cases they do so at lower rates. In the latter cases, above all in the absence of effective mechanisms to halt land squatting and deforestation, state- led land reform may have a positive effect in ameliorating deforestation in areas that would otherwise be rapidly appro- priated and deforested by largeholders for cattle ranching. This is a very contentious argument. It is assumed that largeholders have a greater capacity to occupy larger areas than smallholder settlements, even in spite of agrarian reform, and that they deforest faster than smallholders to implement pasture. In opposition, land use in smallholder settlements tends to be more intensive, includes tree crops, and often in- volves rotational cultivation followed by secondary forests in fallows, which are environmentally better outcomes than large tracts of pasture (Browder, Pedlowski, & Summers, 2004; Picho ´ n, 1997; Walker, Perz, Caldas, & Silva, 2002). However, smallholders tend to generate many small clearings that frag- ment forest stands into irregular and small remnants, while many large landholders should in theory leave large forest blocs intact, but that does not necessarily happen in practice due to large legal forest reserves removal. This paper contrib- utes to the latter discussion, but no definitive conclusions can be drawn due to data limitations, and difficulties for making generalizations based on two sites. In order to assess the contribution to deforestation by differ- ent agents, I provide an analysis for the whole Brazilian Legal Amazon (BLA). 1 Furthermore, I examine two case studies in greater detail: Uruara ´ and Redenc ßa ˜ o, both in the state of Para ´. Uruara ´ is on the Transamazon Highway, and is a typical col- onization area in which smallholders are the main actors. Re- denc ßa ˜o is located in southern Para ´ and is an old cattle ranching frontier in which medium- and large-scale cattle ranchers tend to dominate the landscape. The diverging impli- cations of state-led land reform on land occupation and distri- bution in the two frontier areas are assessed using a political economy perspective. In turn, its land-use impact is assessed * I am indebted to David Kaimowitz and Gerald Nelson for very insigh- tful comments on an earlier version of this paper. I also thank the parti- cipants of a Berlin workshop on ‘‘Land reform, land tenure and land usefor feedback on the ideas that have motivated this paper. I gratefully acknowledge the receipt of research funding from the Center for Intern- ational Forestry Research (CIFOR), Bogor, Indonesia, and the support for field work of the Institute of Environmental Research for Amazo ˆ nia (IPAM), Belem, Brazil. Final revision accepted: August 7, 2008. World Development Vol. 37, No. 8, pp. 1337–1347, 2009 Ó 2009 Elsevier Ltd. All rights reserved 0305-750X/$ - see front matter www.elsevier.com/locate/worlddev doi:10.1016/j.worlddev.2008.08.019 1337

Transcript of Agrarian Reform in the Brazilian Amazon: Its Implications for Land Distribution and Deforestation

Page 1: Agrarian Reform in the Brazilian Amazon: Its Implications for Land Distribution and Deforestation

World Development Vol. 37, No. 8, pp. 1337–1347, 2009� 2009 Elsevier Ltd. All rights reserved

0305-750X/$ - see front matter

www.elsevier.com/locate/worlddevdoi:10.1016/j.worlddev.2008.08.019

Agrarian Reform in the Brazilian Amazon: Its Implications

for Land Distribution and Deforestation

PABLO PACHECO *

Center for International Forestry Research (CIFOR), Bogor, Indonesia

Summary. — This paper examines the interactions between state-led land reform, agrarian structures, and deforestation in the BrazilianAmazon. Land reform tends to promote land redistribution through regularization of smallholder land invasions of large-scale landhold-ings, and by redistribution of public lands to smallholders in existing colonization frontiers. The implications of state-led land reform ondeforestation are heterogeneous. I argue that impacts of agrarian reform are strongly related to the pre-existing social and economicconfiguration of the frontiers where it takes place. While it leads to lower rates of deforestation in landscapes dominated by smallholderswith diversified land use, its impact on forest conversion is higher in landscapes where extensive land use, mainly prompted by large-scalecattle ranching, tends to dominate. I provide an assessment for the whole Brazilian Amazon, and examine two research sites in the stateof Para, namely Uruara and Redenc�ao. The study is based on informant interviews, secondary information, agricultural census data, andremote sensing data.� 2009 Elsevier Ltd. All rights reserved.

Key words — Brazil, Amazon, agrarian reform, deforestation, land tenure

* I am indebted to David Kaimowitz and Gerald Nelson for very insigh-

tful comments on an earlier version of this paper. I also thank the parti-

cipants of a Berlin workshop on ‘‘Land reform, land tenure and land use”

for feedback on the ideas that have motivated this paper. I gratefully

acknowledge the receipt of research funding from the Center for Intern-

ational Forestry Research (CIFOR), Bogor, Indonesia, and the support

for field work of the Institute of Environmental Research for Amazonia

(IPAM), Belem, Brazil. Final revision accepted: August 7, 2008.

1. INTRODUCTION

In this study, I assess the dynamics of agrarian reform tak-ing place in the Brazilian Amazon, and evaluate their directand indirect implications for land tenure and deforestation.Agrarian reform is a state-led redistributive land reform whichaims at granting land rights to smallholders under two vari-ants. The first consist in distributing available public landsto smallholders, and the second in redistributing large land-holdings that have been occupied by smallholders throughland invasions. I do not assess in this study other forms of landreform underway in the Brazilian Amazon aimed at the regis-tration of indigenous people and traditional communities landrights.

The impact of agrarian reform on land distribution in theBrazilian Amazon still has little influence in changing theexisting agrarian structure given its limited scope. There is stilla highly skewed land distribution in the region. Smallholdingsare more often in distant locations, whilst occupation by large-holdings occurs more actively in areas relatively closer to mar-kets, and where there are better roads. Agrarian reform,therefore, is mainly aimed to, on one side, distributing avail-able public lands and, on the other side, redistributing landsand formalizing property rights in invaded large private land-holdings. Agrarian reform benefits both smallholders andlandless people.

Much of the outcomes of agrarian reform on deforestationare contradictory in practice because they depend on the resul-tant agricultural systems. While in some cases smallholders de-velop complex land-use mosaics in which some forest cover isretained, in other cases they turn to more specialized activitiessuch as cattle ranching, which leads to convert most of thelandscape into pasture. In opposition to common wisdom,smallholders do not show a much greater propensity to defor-est than largeholders, and in some cases they do so at lowerrates. In the latter cases, above all in the absence of effectivemechanisms to halt land squatting and deforestation, state-led land reform may have a positive effect in amelioratingdeforestation in areas that would otherwise be rapidly appro-priated and deforested by largeholders for cattle ranching.

1337

This is a very contentious argument. It is assumed thatlargeholders have a greater capacity to occupy larger areasthan smallholder settlements, even in spite of agrarian reform,and that they deforest faster than smallholders to implementpasture. In opposition, land use in smallholder settlementstends to be more intensive, includes tree crops, and often in-volves rotational cultivation followed by secondary forests infallows, which are environmentally better outcomes than largetracts of pasture (Browder, Pedlowski, & Summers, 2004;Pichon, 1997; Walker, Perz, Caldas, & Silva, 2002). However,smallholders tend to generate many small clearings that frag-ment forest stands into irregular and small remnants, whilemany large landholders should in theory leave large forestblocs intact, but that does not necessarily happen in practicedue to large legal forest reserves removal. This paper contrib-utes to the latter discussion, but no definitive conclusions canbe drawn due to data limitations, and difficulties for makinggeneralizations based on two sites.

In order to assess the contribution to deforestation by differ-ent agents, I provide an analysis for the whole Brazilian LegalAmazon (BLA). 1 Furthermore, I examine two case studies ingreater detail: Uruara and Redenc�ao, both in the state of Para.Uruara is on the Transamazon Highway, and is a typical col-onization area in which smallholders are the main actors. Re-denc�ao is located in southern Para and is an old cattleranching frontier in which medium- and large-scale cattleranchers tend to dominate the landscape. The diverging impli-cations of state-led land reform on land occupation and distri-bution in the two frontier areas are assessed using a politicaleconomy perspective. In turn, its land-use impact is assessed

Page 2: Agrarian Reform in the Brazilian Amazon: Its Implications for Land Distribution and Deforestation

1338 WORLD DEVELOPMENT

through land-use change analysis at the regional level by merg-ing agricultural census data and remote sensing information,and in the two study sites by drawing on original data. Thisanalysis provides a comprehensive assessment of the temporaland a spatial perspective of land-use change trends.

In the second part of the paper, I review the discussionabout the interactions between land tenure and deforestationand the implications of land reform in this interplay. In thethird part, I provide a brief summary of the dynamics of agrar-ian reform in Brazil, highlighting its main outcomes in landtenure in the Brazilian Amazon. In the fourth part, I assessthe implications of agrarian reform on deforestation for thewhole BLA region, distinguished by the size of the landhold-ings involved, based on up-to-date empirical data. In the nextpart, I discuss the interactions between land occupation,agrarian reform, and deforestation based on two case studies.Finally, in the last part I synthesize the main conclusions ofthis study.

2. FRONTIERS, LAND TENURE, AND LAND USE:ASSESSING THEIR INTERACTIONS

Sikor and Muller (2009), this issue, suggest that there are di-verse modalities by which state-land reforms are implemented(i.e., redistributive land reform, land registration, and post-totalitarian land redistribution), while some community-ledtypes of land reform are emerging. In this paper, I focus onsome variants of state-led land reform, under the label ofagrarian reform, which constitute part of a broader processof political and economic reform. There are various types ofstate-led land reform. In Latin America, the most commonmodality entails redistributive land reform in favor of small-holders and the landless on the basis of expropriated largelandholdings (usually with compensation), distribution ofstate property, and ”market-assisted” land reform. Otherstate-led land reform modalities promote privatization ofownership or user rights in the transforming countries, andit may even imply legal and institutional reforms in favor oftraditional local groups (Kirk, Loffler, & Zimmermann,1998). In most of the cases, state-led land reforms are aimedto promote land redistribution and secure the rights of mar-ginalized social groups but do not always succeed in achievingsuch objectives. Furthermore, in Brazil, there has been anextensive process of registration of land rights held by individ-ual and collective actors under customary arrangements as astep to enhancing the recognition of indigenous territoriesand the creation of extractive reserves for securing the rightsof extractive populations (CNS, 2005). Although the lattermodalities are also embraced within the state-led land reforms,they are not part of agrarian reform on which this paper is fo-cused, which is redistribution of large estates and state prop-erty to smallholders.

When implemented in the tropics, agrarian reform tends tofavor two types of processes. The first consists in the formal-ization of land rights held by smallholders in public lands,and the second in the redistribution of large landholdings tosmallholders. The first type of actions was implemented inthe past as part of colonization programs, but they were notnecessarily motivated by agrarian reform aims but for otherpolitical and economic goals associated with forested lands(Nelson, 1977). Land redistribution in the tropics, and partic-ularly in the Amazon, has acquired preeminence in the recentpast. Its aims are to reverse an unequal land tenure structure inwhich extensive land grabbing has led to the concentration ofland in a few hands (Alston, Libecap, & Mueller, 2000), and to

respond to increasing claims from local actors for formal landrights (Simmons, 2002).

There is a perverse interaction in the tropics, in contexts inwhich there is still available public land—particularly in theBrazilian Amazon—in which the existence of open accesslands motivates speculative appropriation. This in turn pro-motes deforestation in expanding agricultural frontiers. In thislogic, the expectation is that the public land will increase in va-lue in the future mainly due to road construction and marketdevelopment. Furthermore, the easiest way to justify landownership is by converting the forest to agriculture uses,mainly pasture that is less costly than by establishing cashcrops (Hecht, 1993; Margulis, 2004). Thus, extensive cattleranching is the most common livestock production system un-der cheaply available land and scarcity of capital and labor(Kaimowitz, 1995). In this context, the government has notbeen able to enforce environmental law that mandates that80% of each property be set aside as legal forest reserves(Alencar et al., 2004; Mueller & Alston, 2007).

Who wins access to land and forest resources has decisiveimplications for the way in which such resources are used, aswell as for the distribution of the economic benefits and thetransformation of nature by deforestation. Smallholders useresources differently than medium- and large-scale landhold-ers, and thus the landscape imprint of the small-scale agricul-ture is different to that produced by large-scale cattle ranching.

Smallholder economies are dominated by more intensiveproduction systems in terms of the use of labor per unit ofland, and they rely mostly on agricultural on-farm incomes,which result in complex land-use mosaics due to a combina-tion of temporal and perennial crops and fallows, with less for-est being converted to pasture (Browder et al., 2004).Nonetheless, in some cases pressures on the land are greaterdue to increasing population affluence that in turn pushesexpansion of deforestation. Diffuse and/or fishbone spatialpatterns of deforestation dominate in the smallholder land-scapes, depending on the type of settlement (Geist & Lambin,2001). In contrast, largeholders often specialize, because theyhave access to machinery and capital (Kaimowitz, 2002). Inthe Brazilian Amazon, large landholdings are often devotedto cattle ranching to produce beef mainly under extensiveproduction systems (Faminow, 1998), but soybean croppingis expanding (Nepstad, Stickler, & Almeida, 2006; Vera,Kaufmann, Nepstad, & Schlesinger, 2008). Large-scalegeometric clearings are most common in the largeholder land-scapes (Geist & Lambin, 2001).

Thus, any state attempt to reallocate forest land from large-scale landholdings to smallholders will tend to favor the devel-opment of more complex land-use mosaics. Nevertheless, thespeed with which these actors will convert forest to other usesdepends on a set of factors not directly related to land policiesor landholding size, which are geographical and economic innature such as quality of soils, access to markets and capital,among others (Kaimowitz & Angelsen, 1998). In this regard,the contribution of the different actors to deforestation needsto be examined across the frontiers.

3. LAND TENURE POLICY AND AGRARIAN REFORMIN BRAZIL

It is possible to discern four different historical periods in re-cent Brazilian history. The first is associated with the militaryregimes of the mid-1960s to the mid-1980s. The second, from1985 to 1994, is labeled the ‘‘New Republic” and features thereturn of democracy. The third coincides with the administra-

Page 3: Agrarian Reform in the Brazilian Amazon: Its Implications for Land Distribution and Deforestation

AGRARIAN REFORM IN THE BRAZILIAN AMAZON 1339

tion of F.H. Cardoso (1995–2002) in which important macro-economic and social transformations were achieved. Thefourth is the period of the administration of L.I. da Silva(2003—present) in which more attention was paid to the distri-butional issues related to land and economic wealth.

In Brazil, there are two ways in which public land can passinto private ownership. The first is through sealed tenders (lic-itac�oes) to large private owners, and the second by sellingsmall lots (lotes) to colonists in government-sponsored settle-ment areas (with five years of grace and 6% annual interest).In the 1970s, small lots were 100 ha, decreasing to 50 ha inthe 1980s. It is argued that this type of land distributionshas not occurred since 1987 (Fearnside, 2001). While small-holder colonization was privileged during the military regimes,state-led agrarian reform in the form of sponsored settlementprojects labeled PAs (Projetos de Assentamento) was initiatedin the mid-1980s to facilitate access to land by small farmers.

The most effective way to claim land in practice is by squat-ting. The Land Statute (No. 4504, 1964) and the ‘‘usucapion”law (No. 6969, 1981), both applicable to small tracts of land,ratified the right of squatters to claim land ownership. TheLand Statute states that a squatter (posseiro) who lives on un-claimed public land (terra devoluta) and has farmed it for ayear and a day, can claim the right of occupation for a furtherperiod of four years. This right applies for up to three timesthe land farmed and up to a maximum of 100 ha. Thereafter,a deed or title may be issued if the farmer can prove land man-agement capability (World Bank, 1991). Under the 1981 Law,a person who has farmed uncontested land up to 25 ha (or upto the rural module) for more than a year can also acquire titleto those lands. In lands under federal control, up to 3000 hamay be claimed by using the mechanism described above(Binswanger, 1991).

In the period 1970–84, the state used a discrimination pro-cess to identify areas for colonization programs by differenti-ating public from private lands and by legally registeringpublic land in the name of the Federal Government. Thismechanism was employed less after 1985, since when land dis-tribution has been accomplished through redistributing largeprivate landholdings that did not fulfill a social and economicfunction by the Brazilian Agency of Agrarian Reform(INCRA). The legal procedure for this redistribution has beenone of indemnification and expropriation as stated in the LandStatute of 1964 (Fearnside, 2001). Indemnification was paid inAgrarian Debt Bonds (TDAs), which mature at varying times

Table 1. Number of families settled by agrarian

Settled families

1964–94 1995–2002 20

Total Brazil 218,534 635,035 33Total BLA 161,562 319,514 26Acre 11,088 14,974 1Amapa 2,990 7,012Amazonas 16,210 15,498 1Para 43,851 89,082 12Rondonia 42,701 22,512Roraima 14,715 13,152Tocantins 3,583 20,805Mato Grosso 16,219 68,491 3Maranhao 10,205 67,988 4BLA/Brazil (%) 73.9 50.3

Notes: (a) Includes the nine states comprising the Brazilian Legal AmazonMaranhao, Tocantins, and Mato Grosso.Source: Adapted by author. Based on INCRA (2002,2007).

depending on the area expropriated. The expropriation oflandholdings proceeded slowly until the mid-1990s, when itwas stimulated by social pressure from the landless people or-ganized into the Landless Social Movement, MST (Almeida,Sanchez, & Hallewell, 2000; Wright & Wolford, 2003).

In 1985, the National Plan for Agrarian Reform (PNRA)was implemented. It included ambitious goals such as thatof settling up to 1.4 million families in five years. The PNRAwas, in theory, a priority in the government agenda. It soughtto reorient the process of agrarian reform through the imple-mentation of several sub-programs (i.e., colonization, landregularization, land taxation, and rural cadastre, among oth-ers). Yet, after five years, only 90,000 families were settled,and very little progress was achieved in the process of land reg-ularization and titling. From 1990 to 1992, an emergency pro-gram to settle 80,000 families succeeded in settling only 22,000families (INCRA, 2000).

In 1993, when Law 8.629 was passed, the Agrarian Reformacquired priority within the government. This law regulatedsome mechanisms related to land tenure included in the Consti-tution of 1988. From 1993 onwards, INCRA begun to privilegethe process of land titling in the new settlements created inexpropriated landholdings. As result, in the 1985–94 period,150,000 families were settled in about 8 million ha, andINCRA expropriated a total of 12.7 million ha (INCRA,2000).

The government of F.H. Cardoso established ambitiousgoals for agrarian reform, and made relative progress inexpropriating and allocating land to INCRA settlements. Itsobjective was to settle up 280,000 families in four years. Thisgoal was exceeded in about 8,000 families in 1998. In 1999,373,000 families were settled on about 9 million ha of expro-priated land (INCRA, 2000). In total, this administration set-tled 80,000 families per year during 1995–2002, much morethan the average of about 7,200 families per year during1964–94 (Table 1). Yet, some argue that the last numbers in-clude not only the newly settled families but also old landre-distributions, and regular land titling to smallholders wasalso considered as land redistribution (Domingos, 2003).

Lula’s administration continued to implement agrarian re-form, but there is much criticism of its achievements. ThePNRA II, elaborated under the Lula’s administration in late2003, was expected to settle 530,000 families by late 2006, ofwhich 400,000 would be through settlement projects. Officialinformation shows that 340,000 families were settled during

reform in the Brazilian Legal Amazon (a)

Average per year

03–06 1964–94 1995–2002 2003–06

8,525 7,284 79,379 84,6315,164 5,385 39,939 66,2914,136 370 1,872 3,5345,642 100 877 1,4118,666 540 1,937 4,6675,891 1,462 11,135 31,4736,951 1,423 2,814 1,7387,452 491 1,644 1,8638,966 119 2,601 2,2421,604 541 8,561 7,9015,856 340 8,499 11,464

78.3

(BLA), namely: Para, Rondonia, Roraima, Amapa, Amazonas, Acre,

Page 4: Agrarian Reform in the Brazilian Amazon: Its Implications for Land Distribution and Deforestation

1340 WORLD DEVELOPMENT

2003–06, but critics argue that the number of families settledwas much lower since enough land was not expropriated tosettle these people (Wittman, 2005).

Table 1 summarizes families settled by state. The two formerperiods mentioned above have been merged together due tothe lack of detailed data. According to these data, the propor-tion of families settled in the Amazon with respect to Brazil asa whole decreased from 74% to 50% in the periods 1964–94 to1995–2002, but they increased to about 78% in the period2003–06. The average number of families annually settled inthe BLA increased from 40,000 in the period 1995–2002 to66,000 in the period 2003–06. Unfortunately, there is no infor-mation about how much private land was expropriated and di-verted to small farmers’ settlements in the Amazon states andhow this compares with other Brazilian states, which makes itdifficult to determine the impact of agrarian reform on landdistribution. Furthermore, there is no detailed informationabout how much public land was allocated to settlements un-der different modalities. 2 By 2004, the areas under agrarian re-form reached about 21 million ha, but it could have increasedsignificantly under the new modalities employed by agrarianreform.

The latter makes it difficult to estimate the impact of agrar-ian reform on the land tenure structure. The only availabledata for land tenure is that contained in the agricultural census1995/96 produced by the Brazilian Institute of Geography andStatistics (IBGE). The census comprises 120 million ha occu-pied by the so-labeled agricultural establishments. Accordingto these data, most of the land in the BLA is held in large land-holdings, and a small portion is under the control of smallfarmers. One percent of landholders held more than2,000 ha each and controlled 52.7% of the private land. Incontrast, 83% of landholders held less than 100 ha each andcontrolled only 11.3% of all agricultural land. Census informa-tion shows that, in the BLA, about 730,000 properties smallerthan 100 ha occupied 13.3 million ha. Considering the landdistributed under agrarian reform settlements mentioned ear-lier, and these census data, the impact of agrarian reform inland distribution could have been significant.

4. AGRARIAN REFORM AND DEFORESTATION INTHE BRAZILIAN AMAZON

Brazil holds the largest continuous tropical forest in theworld, but it has also lost the greatest extent of tropical forestamong all tropical countries. 3 Currently, the Brazilian Insti-tute of Spatial Research (INPE) is the single source providingannual estimates of deforestation (excepting 1993). 4 INPEdefines deforestation as ‘‘the conversion of areas of primaryforest by anthropogenic activity for the development of agri-culture and cattle raising, detected from orbital platforms”(INPE, 2005). 5 According to this source, the total deforested

Table 2. Annual rates of deforestation by state (

1978–88 1989–94

Amazonas 180 70 (±34)Maranhao 266 84 (±44)Mato Grosso 515 499 (±139)Para 751 446 (±75)Rondonia 258 194 (±63)Others 283 141 (±40)

Total BLA 2,253 1,435 (±219)

Source: INPE (2008). Adapted by author. Excludes year 1995.

area in the BLA grew from 15.2 million ha in 1978 to 72.5 mil-lion ha in 2007 (INPE, 2008). Nepstad and colleagues (1999)argued that cryptic deforestation from logging crews severelydamages an additional 1.0–1.5 million ha of forest per year.

Rates of deforestation follow an oscillatory rather than lin-ear trend. INPE estimates place the highest rates of deforesta-tion (excluding the increases in 1995 and 2002) in the period1978–88 (2.2 million ha/yr). The annual area of forest clearingdecreased from the mid-1980s to the early 1990s. It remainedbelow 1.5 million ha/yr during a major part of the 1990s ex-cept for a spike in 1995, but increased in the late 1990s andthe first half of 2000, when it reached a similar level to thatin the 1980s. Rates of deforestation were markedly lower at1.3 million ha/yr in 2006–07 (Table 2).

The unequal spatial distribution of deforestation is relativelywell known due to an explosion of remote sensing analysesand the use of geographic information systems (GIS). Somespatial analysis on the location of deforestation has revealedthat it is relatively concentrated in a few geographical areas(Alves, 2003; Skole & Tucker, 1993). The major area of defor-estation has been labeled the ‘‘arch of deforestation” due to itseast–west orientation (Figure 1).

The smallholder presence tends to dominate in the northeastof Para, the Transamazon, and Rondonia, mainly in the areasof implanted colonization projects. In contrast, the medium-scale landholders have tended to concentrate in the easternAmazon around intermediate urban centers such as Altamira,Maraba, and Redenc�ao. In turn, largeholders are dominantparticularly in the center and northeastern Mato Grosso, andsome areas in southern Para (Pacheco, 2005) (Figure 2). Assess-ment of the implications of agrarian reform for land-use changeis thus strongly contingent on the regions considered.

There are several estimates about the contribution to defor-estation made by largeholder farmers and ranchers as opposedto smallholders. Fearnside (1993) suggests that 70% of defor-estation is attributable to large-scale ranching operations.Homma and colleagues (1995) mention that half of deforesta-tion in the Amazon is due to smallholder, shifting cultivators.Chomitz and Thomas (2001) claim that establishments largerthan 2,000 ha account for about half of all land convertedfrom forest or cerrado to agricultural use. Given the fact thatthese estimates constitute out-of-date figures on deforestationtaking place in the BLA, an updated assessment is providedbelow.

The land tenure implications for deforestation are related tothe extent to which different agents have deforested theirlands. This can be assessed in two ways. The first is by lookingat agricultural census data, and the other is by remote sensing.Both methods have advantages and disadvantages. Table 3shows land tenure information for 1995/96. These data revealsome land-use change patterns, namely: (1) smaller holdingsare proportionally more deforested than other holding types;(2) a major portion of cleared areas within smallholdings has

mean ± standard deviation in thousand ha)

1996–2000 2001–05 2006–07

72 (±18) 101 (±39) 68 (±14)96 (±32) 93 (±10) 64 (±1)

632 (±63) 899 (±202) 340 (±131)558 (±98) 676 (±131) 554 (±5)226 (±23) 329 (±46) 176 (±42)102 (±21) 118 (±30) 60 (±15)

1,685 (±207) 2,216 (±404) 1,263 (±199)

Page 5: Agrarian Reform in the Brazilian Amazon: Its Implications for Land Distribution and Deforestation

Figure 1. Land use in 2003 with INCRA settlements in the BLA. Based on Brazilian Institute of Geography and Statistics (IBGE) and Brazilian Agency of

Agrarian Reform (INCRA) information. Adapted by author.

Figure 2. Accumulated deforestation up to 2003 according to property size in the BLA. Based on Brazilian Institute of Geography and Statistics (IBGE) and

Brazilian Institute of Spatial Research (INPE) information. Adapted by author.

AGRARIAN REFORM IN THE BRAZILIAN AMAZON 1341

been devoted to permanent and agricultural crops, while lar-ger areas in medium- and large-scale landholdings have been

converted to pasture; and (3) considering total deforestation,most land clearing has taken place in medium- and large-scale

Page 6: Agrarian Reform in the Brazilian Amazon: Its Implications for Land Distribution and Deforestation

Table 3. Land use by size of establishment in the Brazilian Legal Amazon, 1995/96 (a)

Establishment size Establishments % of total area cleared (b) Contribution tototal area cleared

in the BLA (%) (c)No.(000) Cleared (%) Perm + temp

cropsPlantedpasture

Plantedforest

Fallows + Unutilized

Less than 10 416.7 77.7 77.9 8.1 0.1 13.9 1.410–100 313.5 51.3 18.7 48.9 0.6 31.8 13.4100–500 113.4 45.0 9.1 68.1 0.4 22.3 19.4500–1,000 14.9 45.4 9.9 72.2 0.2 17.7 9.61000–2,000 8.8 46.5 11.3 73.1 0.1 15.4 11.72000–5,000 5.8 43.1 10.4 74.4 0.2 14.9 15.65,000–10,000 1.8 39.8 9.6 77.0 0.4 13.0 10.410,000 and more 1.2 26.4 8.0 74.0 2.4 15.6 18.5

Notes: (a) Includes the nine states comprising the Brazilian Legal Amazon (Para, Rondonia, Roraima, Amapa, Amazonas, Acre, Maranhao, Tocantinsand Mato Grosso), although the last three include a large area of cerrado, (b) the cleared area comprises all the areas under either permanent or annualcrops, areas in fallow, planted pasture, planted forest and unutilized productive land regardless of original vegetation type, (c) this value has been obtainedby dividing the total cleared area within each class by the total cleared area for the whole BLA region.Source: Adapted by author based on Brazilian Institute of Geography and Statistics (IBGE), agricultural census 1995/96.

1342 WORLD DEVELOPMENT

landholdings. Only 15% of land-cover loss (most of it forest,some cerrado) was attributable to smallholders (less than100 ha), 29% to landholders with 100–1,000 ha, and 56% onlandholdings larger than 1,000 ha.

These data provide only a limited basis for assessing theimplications of agrarian reform. Further basis is provided inthis study by an analysis of agricultural census data fromIBGE combined with remote sensing land-use informationprocessed by INPE. A two-step analysis was undertaken.While the first aims at determining the contribution of the dif-ferent actors to deforestation, the second focuses on assessingthe influence of agrarian reform.

The actor type was derived from the establishment sizeexamining the average landholding size in census tracts usingIBGE agricultural census information for 1995/96. The out-come is a map showing the average property size in each cen-sus tracts, and hence it shows the dominance of eithersmallholders (less than 100 ha), medium-scale landholders(from 100 to 1,000 ha), or largeholders with more than1,000 ha. This map was overlaid with a land-use changemap taken from INPE dataset, including accumulated defor-estation by 1997, and deforestation for the periods 1997–

Table 4. Comparison of land use within and outside state-spon

INCRA settlements by 2004

Outside Inside Small

Forest 310,626 11,139 126,2Never forests 79,093 1,845 9,9Water 10,642 106 6,1Clouds 4,953 541 1,9Def. By 1997 47,152 5,579 20,3Def. 1997–2000 5,083 816 1,5Def. 2000–03 6,254 1,040 1,4Total 463,803 21,066 167,6

Contribution to total deforestation (%)

By 1997 89.4 10.61997–2000 86.2 13.82000–03 85.7 14.3By 2003 88.7 11.3

Notes: (a) corresponds to agricultural establishments lower than 100 ha, (b)establishments higher than 1,000 ha, and (d) there was no information on estAmazon (BLA).Source: Adapted by the author based on Brazilian Institute of Spatial ResearcInstitute of Geography and Statistics (IBGE).

2000 and 2000–03. 6 The derived information showing defores-tation by actor type is presented in Table 4. Figure 2 shows theaccumulated deforestation up to 2003 according to landhold-ing size. These data show a slightly different story than theagricultural census data. The contribution of smallholders todeforestation up to 2003 was 35%, medium-scale landholderscontributed 43%, and largeholders contributed 19% of totaldeforestation. The remaining 3% is for areas in which thereis no available land tenure information.

In order to assess the implications of agrarian reform ondeforestation, geographical information of agrarian reformsettlements in the BLA in 2004 was obtained from INCRA.These data were overlaid on land-use patterns in 2003 as de-scribed previously. INCRA settlements (PAs) embrace an areaof about 21 million ha, of which 35% (7.4 million ha) was al-ready deforested (Table 4).

Furthermore, only 11.3% of accumulated deforestation upuntil 2003 took place within agrarian reform settlements.Yet the proportion of forest removal has increased slowly overtime from 10.6% by 1997, to 13.8% from 1997 to 2000, and14.3% from 2000 to 2003. During the period 1997–2003, thisproportion was 14%, as 1.8 million ha of 13.2 million ha

sored settlements and by establishment size (in 1000 ha)

Average establishment size in 1995/96

(a) Medium (b) Large (c) No data (d)

66 69,445 30,285 95,76895 32,331 28,410 10,20194 1,914 482 2,15825 2,143 126 1,30039 21,849 9,004 1,53941 2,936 1,258 16432 3,374 2,207 28193 133,992 71,773 111,412

38.6 41.4 17.1 2.926.1 49.8 21.3 2.819.6 46.3 30.3 3.935.4 42.7 18.9 3.0

refers to properties ranging from 100 ha to 1,000 ha, (c) corresponds toablishments at the census tract level for these areas in the Brazilian Legal

h (INPE), Brazilian Agency of Agrarian Reform (INCRA), and Brazilian

Page 7: Agrarian Reform in the Brazilian Amazon: Its Implications for Land Distribution and Deforestation

AGRARIAN REFORM IN THE BRAZILIAN AMAZON 1343

within INCRA settlement areas were deforested. This estimateis quite similar to others obtained using the same data (Diariodo Para, March 18th, 2006). It is noteworthy that not all ofthis deforestation is attributable to smallholders since a por-tion of the land occupied by INCRA settlements might havebeen deforested before these lands were expropriated. Thus,more than 85% of deforestation is taking place outsideINCRA settlements, and the impact of agrarian reform ondeforestation is relatively small. Another interesting trend isthat, although the participation of smallholders in total defor-estation is decreasing slightly over time, the contribution of to-tal deforestation inside INCRA settlements is growing slowly.This is likely the result of the fact that most contemporaryland occupation by smallholder is taking place as a result ofthe establishment of INCRA settlements in the frontiers.

A more refined assessment of the effect of agrarian reformon deforestation needs to consider not only its influence inaccumulated gross deforestation, but also the pace and magni-tude of deforestation taking place within the PAs compared torates in the oldest colonization zones and areas dominated bylarge-scale landholders. In the next section, I perform such ananalysis considering two case studies in the state of Para.

5. TWO CASE STUDIES IN THE EASTERN AMAZON:URUARA AND REDENC� AO

The history of the municipality of Uruara is linked to theconstruction of the Transamazon Highway in the early1970s as part of the National Integration Program (PIN)(Moran, 1981). This large-scale, sponsored program aimed

Figure 3. Study sites in the State o

to colonize the Amazon by settling peasants from other re-gions of the nation, especially from the northeast. Uruarais located in the middle of the Altamira-Itaituba road sectionof the Transamazon Highway 180 km from Altamira (Figure3). The population of Uruara has increased explosively in re-cent years due to a combination of pull and push factors.Most of the rural population are colonists who arrived inthe region subsequent to the early 1970s in distinct immigra-tion waves. A second important wave raised the municipal-ity’s population to about 25,000 in 1991, with over 11,000migrants since 1980 (Perz & Walker, 2002). The populationwas about 37,400 in 1996, increasing to 45,000 in 2000(IBGE, 2004).

Originally, each colonist was granted a plot of 100 ha. Ruralworkers on large cattle ranches and land managers residing inrural areas comprise a minor portion of the rural population.Some immigrants squatted on remaining public lands or set-tled in the five new INCRA settlements installed in the munic-ipality during the 1990s (Pacheco, 2005). These settlementshave been created on public lands, except by one which has re-sulted from an invasion of a large-scale cattle ranch. The pub-lic lands, however, where under control of some loggers whobuild roads for timber extraction, which in turn stimulatedthe arrival of smallholders who occupy plots along the timberroads.

The colonists residing in Uruara mostly grow annual andperennial crops, although calf and milk production are playingan increasingly important role in their livelihoods (Tourrand,Veiga, Guia, Carvalho, & Pessoa, 1995). While most small-holders develop diversified livelihood portfolios, some special-ize in cattle ranching or other agricultural systems.

f Para, Uruara and Redenc�ao.

Page 8: Agrarian Reform in the Brazilian Amazon: Its Implications for Land Distribution and Deforestation

1344 WORLD DEVELOPMENT

According to data from INCRA, complemented withinformation gathered in the field, 7 old settlements occupy35% (399,000 ha) of the municipality, new settlements 10%(126,000 ha), medium and large-scale ranchers 11%(130,000 ha), indigenous areas 7% (83,000 ha), the remainderbeing lands with overlapping rights, and federal lands (Pacheco,2005). These figures show that land occupation in the munici-pality has taken place mainly through the colonization settle-ments, and only a small portion by new agrarian reformsettlements. Yet, much of these new PA settlements were al-ready occupied informally before INCRA begun their demarca-tion. There are no dramatic differences in the land-use systemsbetween old and new settlements, though older settlements havelarger areas in agriculture and livestock.

The Redenc�ao area covers the southeastern portion of abroader region of southern Para, from Maraba in the northto Mato Grosso in the south. In contrast to Uruara, much ofthis area is dominated by medium- and large-scale landholdings(fazendas) from the beginning of the process of land occupation,largely stimulated by the opening of the Belem-Brasılia highwayin the 1960s (Hecht & Cockburn, 1989; Pacheco, 2005).

The first wave of occupation of the Redenc�ao area began inthe late 19th century, driven by rubber resin extraction (Ianni,1979). After the extractive economy collapsed around the sec-ond decade of the 20th century, little settlement remained be-sides some scattered villages around the main rivers, withsome incipient agriculture and some small amounts of live-stock. It was in the late 1950s and the early 1960s that the open-ing of the Belem-Brasılia highway coincided with the firstexploration for gold in the Redenc�ao area by a group of pio-neers coming from the south of Brazil (Veiga et al., 2004).The latter events launched the occupation of the extremesouthern portion of Para for both mining and livestock devel-opment. Large corporations from the south of the nation ar-rived, supported by a program of fiscal incentives initiated inthe 1960s. Since then, land occupation in Redenc�ao has under-gone boom and bust cycles, including the mahogany cycle inthe 1970s and the gold mining cycle in the 1980s. Livestock pro-duction is the most persistent economic activity in the area(Pacheco, 2005).

Excluding the rural population of Conceic�ao do Araguaia,where there is a larger concentration of small farmers, the rur-

Table 5. Land-use change in two study sites

Uruara 1986 1991

Water 0 0Forest 1,086 1,063Bare soil 8 12Pasture 21 34Juquira 17 21Regrowth 22 15Clouds 5 1Annual crops 3 14Burned areas 0 2Total 1,163 1,163

Redenc�ao area 1986 1992

Water 29 11Forest 3,994 3,538Bare soil 283 326Pasture 416 661Regrowth 0 199Clouds 12 0Total 4,735 4,735

Source: Author’s estimates based on LANDSAT TM and ETM + imagery a

al population of the Redenc�ao area is about 31,700 peoplewith a low density of about 0.7 people/km2. The rural popula-tion mainly comprises workers employed on cattle ranches,but a major proportion consists of smallholders residing inthe settlements (PAs) sponsored by INCRA. The thirty PAsin southern Para are occupied by 6,000 families and approxi-mately 30,000 people. A small portion of the population ofsouthern Para seeks temporary jobs in pasture maintenanceor forest clearing. Many rural workers reside in the urbanareas.

The model of land allocation based on sealed tendersclosed the frontier for smallholders until the early 1990s,when invasion of large-scale cattle ranches began to takeplace. In the late 1980s, three colonies were created withthe support of the Land Agency of the State of Para,ITERPA–Arraiapora I and II, and Capitinga—but they wererelatively small. INCRA became more active in the zone inthe early 1990s, but it was not until the administration ofF.H. Cardoso that the process of formal land expropriationand the formalization of land invasions began in earnest.The property owners were paid in the form of debt securi-ties, even at levels above the market prices (Fearnside,2001).

In the invaded areas, INCRA granted 50 ha lots to eachhousehold. Several families were able to claim more thanone lot in the name of different members. In addition, someof the current land managers squatted or acquired from othersquatter areas that exceed the maximum allowed, particularlyin the locations with established pastures. The two factorsmentioned have triggered a process of relative land concentra-tion within the INCRA settlements which will be difficult toreverse. Unfortunately, data by which to measure this re-ported concentration are unavailable, although INCRA isworking on the issue (Pacheco, 2005).

According to data from INCRA and information gatheredby myself in the field, about 60% (2.8 million ha) of the Re-denc�ao area is occupied by cattle ranches, while 10%(483,000 ha) is INCRA settlements. A small proportion ofland (103,000 ha) is occupied by small farmers to the east ofthe city of Redenc�ao. Another 10% (480,000 ha) constitutespart of the indigenous jurisdictions of Kayapo to the northand Badjonkore to the south. These territories have suffered

in the Brazilian Amazon (in 1000 ha)

1997 1999 2001

0 0 11,029 1,029 1,008

16 3 274 83 8611 15 2217 15 260 5 515 12 110 0 2

1,163 1,163 1,163

1996 2000 2002

11 16 133,324 3,005 2,895217 241 224981 1,269 1,424201 202 790 0 100

4,735 4,735 4,735

nalysis. Land uses may not equal the total due to rounding.

Page 9: Agrarian Reform in the Brazilian Amazon: Its Implications for Land Distribution and Deforestation

Table 6. Comparative assessment of deforestation in two study sites within the Brazilian Legal Amazon (BLA), and for the entire BLA (a)

Annual deforestation (1991/92–2001/02) Accumulated by 2001/02

Deforestation (1000 ha) Annual rate (%) Land area (1000 ha) Contribution (%)

Uruara (b)

Smallholders 9.7 2.1 163.8 70.9Medium and large 2.2 2.2 50.1 21.7Other areas (c) 1.5 0.3 17.1 7.4Total 13.4 1.3 230.9 100.0

Redenc�ao area (d)

Smallholders 14.3 3.4 198.9 10.2Medium and large 67.1 3.3 1,300.9 66.8Other areas (c) 23.7 2.1 446.7 22.9Total 105.2 2.9 1,946.5 100.0

Brazilian Legal Amazon

Total 1,910.4 0.5 63,122.3 100.0

Note: (a) Information for Uruara and Redenc�ao are the author’s estimates based on analysis of satellite imagery, and for the Amazon as a whole based onBrazilian Institute of Spatial Research (INPE) dataset. This period has been chosen for comparison considering availability of remote sensing informationacross the sites, (b) annual deforestation for the period 1991–2001, and accumulated accounts for deforestation by 2001, (c) areas in which there was noavailable property grid, embracing indigenous territories, and (d) annual deforestation corresponds to the period 1992–2002, and accumulated accountsfor deforestation by 2002.

AGRARIAN REFORM IN THE BRAZILIAN AMAZON 1345

encroachment in the past by loggers, and currently by cattleranchers.

The Uruara area in the Transamazon and the Redenc�aoarea in southern Para have undergone quite distinct processesof land occupation, and the role of agrarian reform has beendifferent. While agrarian reform helped to consolidate anagrarian structure dominated by smallholders in the Transam-azon, it tended to promote a process of land redistributionfrom medium- and large-scale cattle ranchers to smallholdersin southern Para. As outlined below, the implications of thesedynamics for land-use change are disparate.

The landscape of Uruara is much more diversified than that ofRedenc�ao. While in the first there is a mixture of perennial andannual crops, and forest regrowth and pasture, the second ismainly dominated by pasture (Table 5). 8 Nevertheless, overtime pasture has increased in importance in Uruara. Table 6 of-fers deforestation data distinguishing the main agents (small-holders versus largeholders). They indicate higher annual ratesof deforestation in Redenc�ao than in Uruara in the period1991/92–2001/02, smallholders having relatively equal rates ofdeforestation within each of these two areas but deforestationbeing more strongly associated with the largeholders who occu-py proportionally more land in Redenc�ao. Differences in therates of deforestation correspond with location factors (i.e.,the feasibility of perennial crops production, beef market devel-opment, and land tenure structures), which are strongly relatedto the socio-economic and geographic configuration of the fron-tier. While there are better conditions for perennial crop produc-tion in the Transamazon linked to more small-scale diversifiedeconomies, more vigorous beef markets exist in southern Para.

It is important to stress that absolute numbers are not usefulfor a comparison across frontiers simply because larger areaswill tend to have more forest clearing, but they are useful toassess the proportional contribution to both annual deforesta-tion and accumulated deforestation of the different agentswithin either frontier. A comparison of propensity to deforestbetween the two locales and the different agents within each ofthem can only be undertaken by looking at proportional an-nual deforestation rates. The information provided shows thata greater propensity to deforest prevails in Redenc�ao, a fron-tier dominated by cattle ranching, as opposed to Uruara, inwhich more diversified land-use strategies persist.

6. CONCLUSIONS

It is a common wisdom that state-led land reform by eitherredistributing large-scale landholdings or transferring publiclands to smallholders tends to promote deforestation when itstimulates the occupation of forested land for frontier devel-opment. However, I have shown that the implications ofagrarian reform for land occupation and land-use changeare less significant than expected, with differentiated outcomesin practice depending on the specific nature-society conditionsprevailing at the frontiers.

The direct impact of agrarian reform on land occupation isrelatively small, though it is growing over time. The lattersince only a small fraction of the occupied land in the Amazonhas been allocated through agrarian reform, but new small-holder occupation in the frontier is prompted by state-led landdistribution. The state-led land reform has tended to favor thedevelopment of small farming in public lands, and it also hasstimulated the invasion of cattle ranches, hence favoring landredistribution. The impact of this process in land-use change isminor since only slightly over one tenth of total deforestationhas occurred within INCRA settlements. Furthermore, not allof the deforestation occurring within these settlements can beattributed to smallholders since an unknown portion of thecleared area took place before the settlements were imple-mented.

Increasing smallholder control over forest areas will inevita-bly increase their contribution to deforestation. This contribu-tion, however, is overall less than half of the total accumulateddeforestation, and it tended to decrease over time due to amore aggressive land occupation by medium- and large-scalelandholders who still constitute an important agent of defores-tation in the Brazilian Amazon. In this sense, agrarian reformimpacts have to be assessed as part of broader land-usechanges taking place in the region, in which smallholders arenot the primary agent of deforestation.

Regarding the resultant land uses, it is important to high-light that they are more complex in frontiers in which small-holders predominate—as in the Transamazon—compared tofrontiers dominated by medium- and large-scale cattle ranch-ers—as in southern Para. In the former, landscapes comprisea mixture of annual and perennial crops, forest regrowth,

Page 10: Agrarian Reform in the Brazilian Amazon: Its Implications for Land Distribution and Deforestation

1346 WORLD DEVELOPMENT

and pasture, while the dominance of pasture is overwhelmingin the second. Furthermore, the propensity to deforest is com-paratively higher in cattle ranching frontiers. This is mainly re-lated to site-related factors such as suitability for perennialcrops, proximity to beef markets, and land tenure structures.Furthermore, smallholders tend to deforest less to the extentthey tend to maintain diversified production systems combin-ing annuals, perennial, and cattle raising. However, their con-tribution to deforestation will tend to increase in areas inwhich they rely more on cattle production. The implicationsof this are diverse for agrarian reform.

Agrarian reform may reduce deforestation rates whensmallholders take over lands that in other circumstanceswould be rapidly deforested for medium- and large-scalesoybean and livestock production in the absence of mecha-nisms that prevent deforestation. In contrast, the effect ofagrarian reform on deforestation is negative to the extentthat they induce settlement in areas with low pressure fromother agents of deforestation. The outcome of agrarian re-form on deforestation depends very much on the pre-exist-ing socio-economic and geographic configuration of thefrontiers.

NOTES

1. The Amazon Basin of Brazil has been defined by government decree in1953, and it is referred as the Brazilian Legal Amazon (BLA). It covers anarea of 5,000,000 km2 embracing the six ‘‘North” states (Acre, Amapa,Amazonas, Para, Roraima, and Rondonia), plus part of three others(Tocantins, north of the 130 parallel; Mato Grosso, north of the 160parallel; and Maranhao, west of the 440 meridian) (cf. Alves, 2001). In1977, the entire State of Mato Grosso was included in the BLA by thecomplementary law n�31 (art. 45). The main reason for the creation of theLegal Amazon was to define an area for the administration of economicdevelopment, rather than to delimit a region according to a uniformecosystem which is why the BLA includes extensive areas of naturalsavanna (cerrado), and open forest in the transition zone between closedforest and cerrado (cf. Faminow, 1998; Fearnside, 1993).

2. It is noteworthy that over time, INCRA has been trying to includeenvironmental concerns on the settlement projects, and hence creating newmodalities of agrarian reform such as the Sustainable DevelopmentProject (PDS), the Forestry Settlement Project (PAF), or the Agroextrac-tive Settlements Project (PAE). All of these include the allocation not onlyof individual plots but also of collective areas to be devoted to forestresources’ use.

3. Until the early 1990s, the debate about land-use change in the Amazoninvolved a lot of uncertainties due to differing definitions and estimates ofdeforestation. The figures produced in that period are accumulated grossdeforestation. A review of those estimates can be found elsewhere(Machado & Pasquis, 2002).

4. The INPE date from 1988 to 2000 uses Thematic Mapper (LANDSATTM) processed to a scale of 1:250,000, which only allowed for theidentification of changes in forest cover areas larger than 6.25 ha.Currently INPE uses a method of digital processing of remote sensinginformation in place since 1997, which employs a mixed spectral model todetermine land cover. See http://www.obt.inpe.br/prodesdigital/metodo-logia.html for a detailed description of this technique. INPE has digitallyprocessed a data for 1997–2000, and annual data since 2001.

5. Under this approach, forest regrowth is treated as deforested to theextent they were counted in the year they are first detected, and hence areincluded in gross deforestation estimates. The exclusion of secondarysuccession from forest stocks tends to overestimates net deforestationtaking place in the BLA. Furthermore, INPE estimates include the9.1 million ha of the so-called ‘‘old deforestation” (prior to 1960) in 1978(its baseline period). A major part of this old deforestation took place inthe states of Maranhao and Para.

6. There is an evident temporal mismatch between the agricultural censusinformation, which corresponds only to one point in time (1995/96) withdeforestation trends registered for three points in time (1997, 2000, and2003). This entails that the property sizes, and hence actor types areconsidered as a fixed category over time, which is not necessarily the casein reality. In spite of this, and given the limitations of information,merging census data with remote sensing constitutes the best way todistinguish deforestation rates by actor type.

7. Field work was carried out during 2002 in Uruara, and in 2003 inRedenc�ao with the aim of understanding the evolution of the agriculturalfrontier and the dynamics of land- use change.

8. Land-use change information presented here is based on LANDSATTM and ETM + images from several dates. Two sets of images wereproduced. The first set consists of mosaics for the study area of Uruara forthe years 1986, 1991, 1997, 1999, and 2001. The second set consists ofmosaics for the Redenc�ao area for the years 1986, 1992, 1996, 2000, and2002. Every LANDSAT TM image was geo-referenced to its correspond-ing LANDSAT ETM + image (same path and row) of year 1999/2000. Asupervised classification was performed based on training sites identifiedduring field work. A maximum likelihood procedure was performed inIDRISI (bands 2, 3 4, and 5). Nine land-use classes were identified inUruara, and six classes in Redenc�ao. Land-use maps were overlaid togenerate land-change maps.

REFERENCES

Alencar, A., Nepstad, D., McGrath, D., Moutinho, P., Pacheco, P., Vera,M. C., et al. (2004). Desmatamento na Amazonia: Indo alem daemergencia cronica. Belem, Brasil: IPAM, WHRC.

Almeida, L. F. d., Sanchez, F. R., & Hallewell, L. (2000). The landlessworkers’ movement and social struggles against neoliberalism. LatinAmerican Perspectives, 27, 11–32.

Alston, L. J., Libecap, G. D., & Mueller, B. (2000). Land reform policies,the sources of violent conflict, and implications for deforestation in theBrazilian Amazon. Journal of Environmental Economics and Manage-ment, 39(2), 162–188.

Alves, D. (2003). An analysis of the geographical patterns of defores-tation in Brazilian Amazonia in the 1991–1996 period. In C. Wood,& R. Porro (Eds.), Patterns and processes of land use and forest

change in the Amazon. Gainesville, Florida: University of FloridaPress.

Binswanger, H. P. (1991). Brazilian policies that encourage deforestationin the Amazon. World Development, 19(7), 821–829.

Browder, J. O., Pedlowski, M., & Summers, P. M. (2004). Land usepatterns in the Brazilian Amazon: Comparative farm-level evidencefrom Rondonia. Human Ecology, 32(2), 197–224.

Chomitz, K. M., & Thomas, T. S. (2001). Geographic patterns of land useand intensity in the Brazilian Amazon. Washington, DC: World Bank,Development Research Group, Infrastructure and Environment.

CNS (2005). Populac�oes Extrativistas da Amazonia: Processo historico,conquistas socio-ambientais e estrategia de desenvolvimento economicoManaus. Brasil: Conselho Nacional dos Seringueiros.

Page 11: Agrarian Reform in the Brazilian Amazon: Its Implications for Land Distribution and Deforestation

AGRARIAN REFORM IN THE BRAZILIAN AMAZON 1347

Domingos, M. (2003). Backgrounder part I: Land reform in Brazil. Ceara:Federal University of Ceara.

Faminow, M. D. (1998). Cattle, Deforestation, and development in theAmazon: An economic, agronomic, and environmental perspective.Wallingford, Oxon, UK; New York, NY: CAB International.

Fearnside, P. M. (1993). Deforestation in the Brazilian Amazon: The effectof population and land tenure. Ambio, 22(8), 537–545.

Fearnside, P. M. (2001). Land-tenure issues as factors in environmentaldestruction in Brazilian Amazonia: The case of southern para. WorldDevelopment, 29(8), 1361–1372.

Geist, H. J., & Lambin, E. F. (2001). What drives tropical deforestation? Ameta-analysis of proximate and underlying causes of deforestation basedon subnational case study evidence. LUCC Report Series No. 4. LUCC.Belgium, Louvain-la-Neuve: IHDP, IGBP.

Hecht, S. (1993). The logic of livestock and deforestation in Amazonia.Bioscience, 43(10), 687–695.

Hecht, S., & Cockburn, A. (1989). The fate of the forest: Developers,destroyers and defenders of the Amazon. London: Verso.

Homma, A. K. O., Walker, R. T., Scatena, F. N., Conto, A. J. de,Carvalho, A. de, Ferreira, C. A. P. et al. (1995). Reducao dosdesmatamentos na Amazonia: Polıtica agrıcola ou ambiental? Paperread at XXXIII Congresso Brasileiro de Economia e Sociologıa Rural,31 July–3 August, at Curitiba, Parana.

Ianni, O. (1979). A luta pela terra: Historia social da terra e da luta pelaterra numa area da Amazonia. 2da ed. Petropolis, RJ: Vozes.

IBGE (2004). Censo Demografico. Sistema IBGE de Recuperac�aoAutomatica 2004 [cited February 24, 2004]. Available from <http://www.sidra.ibge.gov.br/>.

INCRA (2000). Relatorio de Atividades INCRA 30 anos. Brasilia, Brasil:Ministerio do Desenvolvimento Agrario, Instituto nacional de Colo-nizacao e Reforma Agraria, Conselho Nacional do DesenvolvimentoRural Sustentavel.

INCRA (2002). Balanc�o da Reforma Agraria 2002. Brasilia: Secretaria deReforma Agraria, Ministerio do Desenvolvimento Agrario.

INCRA (2007). Numeros da Reforma Agraria 1995–2006. Brasilia: INCRA.INPE (2005). Monitoring of the Brazilian Amazon forest by satellite 2004–

2005. Brasilia: Instituto Nacional de Pesquisas Espaciais.INPE (2008). Monitoring of the Brazilian Amazon forest by satellite 2006–

2007. Brasilia: Instituto Nacional de Pesquisas Espaciais.Kaimowitz, D. (1995). Livestock and deforestation in Central America in

the 1980s and 1990s: A policy perspective. Washington, DC: Interna-tional Food Policy Research Institute, Interamerican Institute forCooperation on Agriculture.

Kaimowitz, D. (2002). Frontier theory and practice. Bogor, Indonesia(unpublished draft).

Kaimowitz, D., & Angelsen, A. (1998). Economic models of tropicaldeforestation: A review. Bogor, Indonesia: Center for InternationalForestry Research.

Kirk, M., Loffler, U., & Zimmermann, W. (1998). Land tenure indevelopment cooperation. Eschborn, Germany: GTZ.

Machado, L. de O., & Pasquis, R. (2002). O Desflorestamento NaAmazonia: Conceitos e Metodos. Brasilia, Brazil: Centro de Desen-volvimento Sustentavel, Universidade de Brasılia.

Margulis, S. (2004). Causes of deforestation of the Brazilian Amazon.Report No. 22, 107. Washington, DC: World Bank.

Moran, E. F. (1981). Developing the Amazon. Bloomington: University ofIndiana Press.

Mueller, B., & Alston, L. J. (2007). Legal reserve requirements in Brazilianforests: Path dependent evolution of de facto legislation. In XXXVEncontro Nacional de Economia. Recife, Pernambuco: ANPEC.

Nelson, M. (1977). El aprovechamiento de las tierras tropicales en AmericaLatina. Mexico DF: Siglo XXI.

Nepstad, D. C., Moreira, A. G., & Alencar, A. A. (1999). Pilot program toconserve the Brazilian Rain Forest 1999. Flames in the rain forest:Origins, impacts, and alternatives to Amazonian fires. Brasilia, Brazil:Pilot Program to Conserve the Brazilian Rain Forest.

Nepstad, D. C., Stickler, C. M., & Almeida, O. T. (2006). Globalization ofthe Amazon soy and beef industries: Opportunities for conservation.Conservation Biology, 20(6), 1595–1603.

Pacheco, P. (2005). Populist and capitalist frontiers in the Amazon:Dynamics of agrarian and land-use change. Ph.D. dissertation, Grad-uate School of Geography, Clark University, Worcester, MA.

Perz, S. G., & Walker, R. (2002). Household life cycles and secondaryforest cover among small farm colonists in the Amazon. WorldDevelopment, 30(6), 1009–1027.

Pichon, F. J. (1997). Settler households and land-use patterns in theAmazon frontier: Farm-level evidence from Ecuador. World Develop-ment, 25(1), 67–91.

Sikor, T., & Muller, D. (2009). The limits of state-led land reform: Anintroduction. World Development, 37(8), 1307–1316.

Simmons, C. S. (2002). Local articulation of policy conflict: Land use,environment and Amerindian rights in Eastern Amazonia. TheProfessional Geographer, 54(2), 241–258.

Skole, D., & Tucker, C. (1993). Tropical deforestation and Habitatfragmentation in the Amazon: Satellite data from 1978 to 1988.Science, 260(5116), 1905–1910.

Tourrand, J. F., Veiga, J. B., Guia, A. P. O. M., Carvalho, S. A., &Pessoa, R. O. (1995). Strategies et Pratiques d’elevage en AmazonieBresilienne: Dynamisme e Diversite Dans l’agriculture Familiale:Fertilite du Milieu et Srategies Paysannes sus les Tropiques Humides.Montpellier: CIRAD.

Veiga, J. B. da, Tourrand, J. F., Piketty, M. G., Poccard-Chapuis, R.,Alves, A. M., & Thales, M. C. (2004). Expansao e Trajetorias daPecuaria na Amazonia. Brasilia: Editora UNB.

Vera, M. C., Kaufmann, R., Nepstad, D., & Schlesinger, P. (2008). Aninterdisciplinary model of soybean yield in the Amazon Basin: Theclimatic, edaphic, and economic determinants. Ecological Economics,65(1), 420–431.

Walker, R., Perz, R., Caldas, M., & Silva, L. T. (2002). Land-use andland-cover change in forest frontiers: The role of household life cycles.International Regional Science Review, 25(2), 169–199.

Wittman, H. (2005). Agrarian reform under Lula: Marching to a historiccrossroads. Ithaca, NY: Cornell University.

World Bank (1991). Brazil: Key policy issues in the livestock sector –Towards a framework for efficient and sustainable growth. Report No.8570-BR. Washington, DC: Agriculture Operations Division, TheWorld Bank.

Wright, A. L., & Wolford, W. (2003). To inherit the earth: The landlessmovement and the struggle for a New Brazil. Oakland, CA: Food FirstBooks.

Available online at www.sciencedirect.com