Navigation – Plan du site

AccueilRubriquesPolitique, Culture, Représentations2026Making environmental compensation...

2026
1103

Making environmental compensation visible. Biased data transparency and agrobusiness strategies in the Bahianese agricultural frontier of the Cerrado

Rendre visible la compensation environnementale. Transparence biaisée des données et stratégies de l’agrobusiness dans le front agricole de l’ouest de Bahia
Hacer visible la compensación ambiental. Transparencia sesgada de datos y estrategias del agronegocio en el Cerrado brasileño
Pierre Gautreau, Eve Anne Bühler, Ludivine Eloy et Sylvain Théry

Résumés

La compensation écologique, bien que devenue centrale dans les politiques environnementales, demeure peu visible dans les paysages et le débat public. Cet article analyse selon une perspective géographique les stratégies de (re)localisation des Réserves Légales (RLs) des exploitations rurales sur une frontière agricole de l’ouest de l’état de Bahia, au Brésil. Les Réserves Légales sont des portions des propriétés rurales, préservées de toute activité productive. Elles sont obligatoires et doivent couvrir 20 % de la surface des propriétés dans l’ouest de Bahia. À partir des données publiques du Cadastre Environnemental Rural (CAR), une plateforme numérique de suivi de la conformité environnementale, nous avons élaboré une méthodologie à base de SIG permettant de reconstituer les processus de compensation des RLs à grande échelle. Quatre tendances majeures émergent : 1) La compensation est massivement mobilisée par l’agrobusiness, avec des transferts à longue distance qui déplacent les obligations de conservation vers l’est ; 2) De nombreuses RLs compensées se superposent aux territoires collectifs des communautés paysannes, entraînant conflits et formes d’accaparement « vert » ; 3) Les zones compensées sont fréquemment déclassées pour autoriser l’expansion agricole ; 4) Environ 35 % des compensations enfreignent probablement le Code forestier en régularisant des propriétés qui n’étaient pas autorisées à recourir à ce mécanisme. Le CAR produit ainsi une « transparence biaisée » : il offre une visibilité inédite sur le comportement environnemental de l’agrobusiness, tout en masquant les liens de compensation. Cette transparence sélective permet aux grands exploitants d’utiliser la compensation de manière stratégique, légalisant la déforestation dans le Cerrado tout en affichant une conformité environnementale sur les marchés internationaux. L’étude montre comment la gouvernance numérique de l’environnement, bien qu’innovante, peut être détournée au profit de l’agrobusiness. Renforcer la transparence du CAR est essentiel pour garantir efficacité écologique et justice sociale, notamment à l’heure des nouvelles réglementations européennes contre la déforestation importée.

Haut de page

Texte intégral

This work was funded by the French National Research Agency (CNRS), first through the PICS project “AGROPOWER, Nouvelles pratiques, nouveaux enjeux de la régulation environnementale de l’agriculture au Brésil”, and successively within the framework of the ANR-23-CE55-0002 SOYLANDIA project. It was also supported by CNRS through its International Emerging Actions Project, and by the Brazilian National Council of Scientific and Technological Development – CNPq, through the 405635/2023-0 Soypiba project.

Introduction

1Environmental compensation has become a common practice globally and a cornerstone of contemporary environmental legislation (Droste et al., 2022). Designed to offset unavoidable ecological damage from development projects through the restoration or protection of natural habitats elsewhere, its objective is to balance or ideally neutralize the ecological loss (Levrel, 2020). However, critical scholarship, informed by Political Ecology, interrogates the social, discursive, and technical constructions of this instrument as a key process of neoliberal environmental governance (Castree, 2008). Empirical research across diverse national contexts has highlighted significant limitations in its effectiveness, transparency, and monitoring. Critics argue that compensation schemes often serve to legitimize, rather than prevent, environmental destruction by displacing its impacts spatially. This dynamic raise profound concerns regarding social and spatial justice, situating the practice within the broader debate on “green grabbing” (Fairhead et al., 2012; Devictor, 2018).

2Although the ecological and ethical debates surrounding environmental compensation remain unresolved, it is now materialized in several places, with dedicated tracts of land set aside to counterbalance land-use changes elsewhere (Devictor, 2015). Despite this materiality, its visibility in the landscape and its discussion in the public arena remain low: the specific locations, their extent, and the stakeholders involved in compensation are generally unknown to the public.

3This article addresses this gap by proposing a geographical methodology to enhance the landscape and political visibility of environmental compensation within the Brazilian context. In Brazil, beyond its role in mitigating impacts from infrastructure projects –as is common in Europe– environmental compensation is integrated into forest conservation policy for private properties. The specific scheme we examine facilitates the regularization of farms that violate the Forest Code (Código Florestal). This policy mandates that private properties maintain “Legal Reserves” (RL)– i.e. set aside areas of native vegetation initially conceived to curb deforestation. Following a series of reforms to the Forestry Code, these Legal Reserves, under specific conditions, can be placed outside the farm, enabling landowners to regularise their properties.

4This article aims to, first, provide a systematic regional analysis of environmental compensation via Legal Reserve (re)location and, second, assess how public digital data can illuminate this process, evaluating both its potential and limitations. We analysed a publicly available dataset from the Brazilian Rural Environmental Registry (Cadastro Ambiental Rural, or CAR), which documents farm and Legal Reserve boundaries. The data were processed through two complementary approaches: 1) mapping actual Legal Reserve compensation by applying GIS techniques to circumvent data anonymization constraints; and 2) characterizing the on-the-ground locations and stakeholders involved through field research. Our case study focuses on Western Bahia, the country’s foremost agricultural frontier during the 2000s and 2010s, located within the Cerrado (savanna) biome, where agribusiness interests significantly shape environmental governance (Brannstrom, 2009; Eloy, Da Silva et al., 2023).

5We contend that the CAR system embodies “biased transparency” – offering unprecedented synoptic visibility while obscuring crucial compensation linkages. This selective transparency enables large-scale farmers to use compensation strategically, facilitating legalized deforestation in the Cerrado while projecting an image of environmental compliance to international markets. Consequently, this paper engages with the political dimensions of both environmental compensation and Open Data practices in agricultural frontiers (Bühler, Gautreau et al., 2022). It concludes by offering policy perspectives to better assess the social and ecological consequences of compensation schemes and to enhance the transparency of environmental compensation.

6The remainder of this paper is structured as follows: section 1 reviews the literature on the Political Ecology of environmental compensation, and about environmental governance in private properties of Brazilian agricultural frontiers; section 2 details the case study and methodology; section 3 presents the results on the spatial dynamics of compensation in the region. Finally, section 4 discusses the interpretation of these compensation patterns and provides recommendations for Open Data public policies.

The Political Ecology of environmental compensation: from theory to reality in Brazil’s agricultural frontiers

Theory

7Although it has been a widespread policy globally, environmental compensation has, at best, limited effects on curbing biodiversity erosion (Droste et al., 2022). Despite a wide variety of standards and practices, it is governed by two core principles. First, ecological equivalence must be established between the impacted site and the compensation site. Second, the compensation must generate a net ecological gain. The entire process must be measurable to evaluate its effectiveness (Maron et al., 2025). Its application, however, varies significantly across jurisdictions, following distinct regulatory frameworks and spatial logics. Numerous studies have evaluated the policy and governance design of compensation mechanisms worldwide, consistently reporting implementation challenges, including a lack of ecological effectiveness (Reid et al., 2015; Rundcrantz, Skärbäck, 2003), inadequate monitoring (Gelot, Bigard, 2021; Padilla et al., 2024), and unsustainable economic models (Wang et al., 2022).

8The value of Political Ecology lies in its capacity to address the political and moral dimensions of environmental compensation, fundamentally challenging the principles of equivalence and commensurability (Apostolopoulou, Adams, 2017) – that is, the premise “that it is possible to assess, compare and equate what is destroyed and what is repaired” (Devictor, 2015). Scholars question the marginalization of ethical dimensions in compensation schemes (Moreno-Mateos, et al., 2015; Martinez-Alier, Munda, O’Neill, 1998), and their implications for environmental justice (Dauguet, 2015). In this vein, Political Ecology often frames compensation as part of the global trend of the “neoliberalization of nature” and its manifestation in public policies such as Payments for Ecosystem Services (McAfee, Shapiro, 2010). From a political economy perspective, compensation also represents the creation of ‘fictitious commodities’ in the sense of Marx and Polanyi (Boisvert, 2015), thereby commodifying land through the abstract valuation of natural processes.

9However, this literature only marginally addresses the spatial dimensions of environmental compensation, an area where geography can make a decisive contribution. This gap is largely due to confidentiality issues and restricted access to the spatial data required to map these processes. Furthermore, the technical complexity of compensation and its documentation in specialized databases gives a central role to experts and private consultancies, which in turn limits public data sharing and comprehension. With few exceptions, notably in France (Padilla et al., 2024), little attention has been paid to the precise spatial distribution of emitting and receiving sites on a regional scale. Yet, the manner in which stakeholders define and enforce this spatial permutation reveals major political and socio-economic issues, particularly as the land in question is often long-inhabited territory.

10To address this gap, we propose a geographical methodology to reveal the landscape and political visibility of environmental compensation through a case study in a Brazilian agricultural frontier region.

Context of the study

  • 1 For savannahs located inside the administrative region called “Legal Amazon”, the Legal Reserve is (...)

11In Brazil, the primary compensation mechanism is embedded in the Forest Code, the country’s main agro-environmental regulation governing land use and deforestation on private properties (Daugeard, Cunha, 2022). First enacted in 1965 and subsequently amended, most recently in 2012, the Forest Code obligates landowners to set aside or restore native vegetation in a Legal Reserve (RL) covering 20 to 80% of the property, depending on the biome. In the Brazilian Cerrado, this requirement is 20%.1 Historically, widespread non-compliance led to the deforestation of these reserves. However, since the late 1990s, government investments in satellite monitoring have significantly improved the capacity to detect illegal deforestation, leading farmers to protest and lobby for looser regulations (Bühler, Gautreau et al., 2022).

  • 2 In practice, restoration in Brazilian farms is reached mainly by enclosure of the Legal Reserve are (...)

12The Brazilian government’s efforts to regularize this situation have involved multiple strategies. Since 1998, the Forest Code has allowed farmers who have not maintained their Legal Reserve to compensate by acquiring and protecting an equivalent area of land outside their property (Rajão et al., 2021). Thus, compensation serves as an alternative to on-farm restoration.2 As farmers are reluctant to relinquish well-located agricultural land (in terms of topography, distance, soil), they prefer to establish their RLs in less accessible areas to comply with the law (Brito, 2020). Over time, the principles governing this compensation have become increasingly flexible. Initially, compensation was required within the same micro-watershed and biome, for the same area and with ecological equivalence. The 2012 reform extended this to the entire state territory (within the same biome) and even beyond state borders in areas designated as high-priority for conservation; furthermore, strict ecological equivalence is no longer legally mandatory (Avila Peterlini de Souza, 2021).

13Central to these efforts, the Rural Environmental Registry (Cadastro Ambiental Rural, or CAR) stands as a unique attempt in Latin America to enforce legal restrictions on every rural property. Created in 2012 as the primary tool for implementing the new Forest Code (Soares-Filho et al., 2014), this digital cartographic registry enables, in theory, public monitoring of Legal Reserves to detect non-compliance. The CAR dataset is partially accessible via a digital platform that allows landowners to self-delimit their property boundaries and RLs on satellite imagery. This allows the federated states to control these declarations. Despite its conceptual and technical originality, CAR has been subject to several debates. While there is a broad consensus that the CAR provides an unprecedented volume of structured data on the environmental compliance of Brazilian agriculture –encompassing over 7 million farms (Brazilian government, October 2023),– scholars have raised concerns. Some warn that the instrumental use of CAR could foster legalized deforestation (Azevedo et al., 2017; Rajão, Azevedo, Stabile, 2012; Rasmussen et al., 2016), while others argue that inadequate incentives and environmental control persist, preventing full compliance with the Forest Code (Azevedo et al., 2017; Freitas et al., 2018; Frederico et al., 2024). Moreover, a growing body of literature links the digitalization of environmental policies in Brazil with “green-grabbing,” citing cases where Legal Reserves overlap with protected areas or peasant territories (Aguiar, Bomfim, Correia, 2021; Sauer, Pereira Leite, 2012). However, no systematic regional analysis about RL compensation has been conducted.

Making compensation visible with incomplete data: materials and methods

Western Bahia: a vast region dominated by agribusiness

14Our analysis is grounded in a case study from Western Bahia, the country’s foremost agricultural frontier of the 2000 and 2010 decades (Brannstrom, 2009). Located within the Cerrado (savanna) biome, this area is critically relevant given the national context of escalating land cover change; annual vegetation loss across Brazil increased by 42% between 2012 and 2022. The Cerrado itself, which has lost approximately half of its native vegetation (Gibbs et al., 2015; Strassburg et al., 2017), accounted for 40% of the national cumulative loss during this period (248,187 km²) (INPE, 2023).

15The study region encompasses a sub-horizontal plateau, sloping eastward from the border with Goiás state to the São Francisco River (figure 2). Historically, this landscape has been occupied by diverse Indigenous and peasant communities practicing swidden cultivation, fruit gathering, hunting, and extensive cattle ranching. The northwestern sector, characterized by flat uplands, underwent the first conversion to cropland in the 1970s and has since lost most of its original Cerrado cover. Agricultural expansion progressed eastward, southward, and northward, with soybean, corn and cotton cultivation stabilizing in areas where irregular rainfall made production economically risky. Recently, the agricultural frontier has resumed its eastward advance through the adoption of center-pivot irrigation, confining peasant communities to valleys where most remaining natural vegetation persists (Eloy et al., 2023).

  • 3 One of the main mechanisms used to force farmers to register in the CAR database was to block their (...)

16This case study provides an ideal context to examine the landscape and political visibility of environmental compensation. First, the region is dominated by a highly organized agribusiness sector that emphasizes its compliance with environmental legislation, despite a long history of ongoing deforestation (Eloy et al., 2023). Indeed, the widespread adoption of the Rural Environmental Registry in western Bahia –unlike other parts of the state– can be attributed to two factors: the requirement of CAR registration for accessing bank loans and the sector’s strong organizational capacity (Pinto et al., 2018).3 Consequently, the region offers unique opportunities to quantify, across a large territory, a practice that appears central to its agricultural economy. Furthermore, Brazil stands out in the Global South for its long-standing Open Data culture concerning land use and the availability of public databases on farm-level environmental compliance. During the 2010s, Latin American environmental policies increasingly incorporated instruments for regulating agriculture on private lands using remote sensing (Gautreau et al., 2016). This context makes Western Bahia an exemplary case for evaluating the potential and limitations of public digital data in raising the profile of environmental compensation.

Data production and processing

17The study relies on the Rural Environmental Registry’s open database (SICAR), which includes geographic data layers of farm boundaries, Legal Reserve limits, and other environmental features4. SICAR is a user-sourced land register: data layers are drawn on-screen by cartography enterprises paid for by farmers. It is important to note that at the start of 2023, the SICAR database was not yet being systematically checked by the Bahia environmental authority. SICAR has been evolving over time because it is updated regularly (except between December 2021 and in April 2023). We downloaded the data layers for November 2016, February 2019, September 2021, and April 2023 to assess the spatial evolution of RL over time, in our study region (figure 2). SICAR layers were crossed with deforestation layers from Brazilian public institutions (INPE) or recognized civil society teams (MapBiomas).5 To assess the superimposition of RL on lands claimed by peasants, the Comissão Pastoral da Terra6 and the NGO project Tô No Mapa granted access to peasant tenure information. Tô No Mapa encourages peasants, indigenous peoples, and afro descendants to self-map the territories they claim in order to defend themselves against land grabbers. The terms of cooperation with Tô No Mapa required us to aggregate the results in order to avoid revealing visible conflict situations that could bring additional trouble to communities. Extensive fieldwork was performed between 2017 and 2023 in places where compensation occurred, using landscape observation and interviews with inhabitants affected by compensation. At last, a population of 17 peasant communities, all of them situated near of RL compensations, was interviewed to understand the effects produced by compensation in their neighbourhood. Interviews were also performed within the state of Bahia’s agricultural, land and environmental administration (states secretaries and INEMA7), the municipal public authorities of western Bahia (environmental and agricultural secretaries), and international conservation NGOs associated with agrobusiness organizations or with rural population right’s protection. In these cases, we wanted to understand the process of implementing the CAR (technical and political) in western Bahia, as well as the role that the different territorial actors had respectively played in it. We also asked them what assessment they had of these years of CAR registration in the region. All interviews were conducted in a semi-directive mode, and all of these persons asked to remain anonymous.

Farm typology according to the Legal Reserve compensation strategy

18Four different situations can be distinguished when regarding the relationship between farms and RL compensation (figure 1). The first type of situation is represented by farms that have no Legal Reserve within their perimeter, called “full compensators.” The second type concerns farms with a RL that covers less than 20% of their area, complemented by a compensated Legal Reserve outside the farm’s perimeter: these are referred to as “partial compensators.” Farms with one Legal Reserve covering 20% of their area correspond to the “normal situation,” and farms that receive a compensated RL from other farms are “compensation receivers.” A step by step GIS analysis of the CAR cartographic farms layer (see Supporting Information, S1, sections 1 to 3) allowed the cartography of this typology, shown in figure 2. We also summarized the concentration of each farm type in a grid, in order to detect concentration phenomena in the compensation strategies (figure 2). For the detailed GIS process, see (Supporting Information, S1, sections 1 to 3).

Figure 1: Typology of farms according to Legal Reserve compensation strategy

Figure 1: Typology of farms according to Legal Reserve compensation strategy

Legend: Black polygons represent farms; green polygons represent legal reserves. Arrows represent compensation direction.

Source: authors

A technique for identifying Legal Reserve compensation links

  • 8 This situation changed in 2025, 13 years after the first release of the CAR.

19SICAR database did not provided the code (IDF) that could link a compensated Legal Reserve with its owner’s farm8. To attribute an IDF code to farms that compensate, the following method was designed (Supporting Information, S1, section 4). We discovered that the compensated RL polygons had the same IDF as another feature that is always located inside the farm, the Area de Preservação Permanente (APP), and strip along rivers that should be preserved by farmers. As APPs are always located inside the farm, it is possible to give the APP IDF code to its corresponding farm polygon. When this operation is possible, a link can be established between the compensated RL and its origin farm when they share the same IDF code. One major limitation of this method is that compensation links cannot be identified when a farm has no APP (because there is no stream or water source). This is frequent in the very flat parts of the study region. In fact, of the 1,763 farms that compensate, we were only able to link 773 (43,87%) to their LR. Amongst the two thirds of links that remained undetected, we cannot rule out the possibility that some farms may compensate outside the state of Bahia, or made wrong declaration to SICAR.

Quantifying Legal Reserve loss or gain

20To study the spatial evolution of Legal Reserves over time, we used GIS data layers downloaded at four different times, from 2016 (four years after the official creation of the Rural Environmental Registry) to early 2023. This allowed us to identify the updates in the open database and to examine them in relation to agribusiness strategies. In particular, the 2021-2023 comparison covered the period when the relocation of Legal Reserves was authorized by Bahia’s environmental institution (INEMA), from January 2021 to November 2022. GIS analysis, by overlapping the layers for each period, allowed the identification of the differences in the area of Legal Reserves (stable areas, erased areas, new areas), for 2016-2019 (2 years and 3 months), 2019-2021 (2 years and 7 months), and 2021-2023 (1 year and 7 months). To determine the changing monthly rates for each period, the areas gained and lost were calculated and compared to the initial area covered by the Legal Reserves (for example, 2016 for the 2016-2019 period; 2019 for 2019-2021).

Results

The compensation mechanism involving legal reserves is a widespread practice in Bahia

21Figure 2 maps four practices linked to the RL compensation mechanism in the Cerrado region of Bahia: farms that use compensation for their entire LR; farms that apply compensation only partially; farms that do not apply the mechanism; and farms that receive compensated RLs from others. Figure 3 focuses on the spatial concentration of each practice, mapping the percentage of land occupied by farms that apply compensation for their RL (3a) and by farms that receive compensated RLs from others (3b). Among the 5,308 farms linked to agribusiness in the area, 1,196 (22,5%) compensate for their RL completely and 566 (10,6%) partially: one third of the farms use the compensation mechanism in some way. On the other hand, 1,665 farms (out of a total of 32,532, including family farms) receive compensated RL from others, which means that 5% of all the farms receive compensated RLs. This practice follows a clear geographical pattern (figure 3): on the western border of our study area, 28-100% of the land is occupied by landowners who use the compensation mechanism. The percentage drops to 15% at 60 km to the east and almost ceases after 100 km. Two other small areas where compensation is frequent are found in the northeast, corresponding to agribusiness enclaves in regions where family farming prevails. The practice of receiving compensated RL is a more diffuse phenomenon and tends to be concentrated in sectors near the deforestation frontier. In the northern, southern, and central parts of this frontier, over 20% of land is occupied by farms that receive compensated RLs, with peaks of up to 80% in some places. In formerly deforested areas (the western part of the region, deforested during the 1970s and 1980s), the practice is concentrated along stream valleys.

The direction of compensation and its social effects

22Figure 4 displays the links between farms that use compensation and their LR. This map confirms that the use of RL compensation is spreading from west to east. 61% of compensations are made at a distance of less than 10 km; 20,5% are made from 10 to 150 km away; and 18% are more than 150 km away. Figures 3 and 4 demonstrate that large farms use compensated RLs not only in their own neighbourhood, but also much farther away. In 610 identified cases, compensation is made using territories that are being disputed with peasant communities. Although these communities territories represent only 5,6% of the study area (454,237 ha out of 8,057,790), 9,13% of the total compensated area is made over them, proving that those territories are over affected by the phenomenon. Figure 5 maps the magnitude of such overlapping in one municipality, Correntina. These results show that the practice of RL compensation in western Bahia is spatially structured, with certain areas specializing in the reception of RL from distant farms.

Figure 2: Typology of farms according to their compensation strategy

Figure 2: Typology of farms according to their compensation strategy

One third of compensations declared to CAR do not meet the Forest Code requirements.

Source: authors, based on CAR dataset (2021)

23The Forest Code only allows compensation for deforestation prior to 22 July 2008, for farms that had either no vegetation left on this date or, for the Cerrado, less than 20% of their area covered with vegetation. Document S3 in the Supporting Information shows that 35% of the 1,763 agribusiness farms that compensate today had more than 20% their area covered with natural vegetation in 2008, and 22% of farms had more than 50%. Some farms were entirely covered by Cerrado vegetation. Most of these farms apparently infringe on the Forest Code: they should have not been able to compensate, because they deforested after 2008. These farms are located in a central strip running from north to south, and a particular concentration can be found in the southern pre-front zone (the Cocos and Jaborandi municipalities).

Figure 3: Spatial concentration of emission or reception of compensated areas

Figure 3: Spatial concentration of emission or reception of compensated areas

Source: authors, based on CAR dataset (2021)

Figure 4: Links between farms that emit and farms that receive compensated areas

Figure 4: Links between farms that emit and farms that receive compensated areas

Source: authors, based on CAR dataset (2021)

Figure 5: Overlapping of compensated Legal Reserves and territories claimed by traditional peasant communities in the Rio Corrente hydrographical basin

Figure 5: Overlapping of compensated Legal Reserves and territories claimed by traditional peasant communities in the Rio Corrente hydrographical basin

Source: authors, based on CAR dataset (2021) and Comisão Pastoral da Terra (Fundos e fechos de pasto limits)

The routine practice of erasing Legal Reserves

24We have observed that RLs are continuously being erased from SICAR in this region. Figure 6 represents the changes in the area covered by RLs between 2016 and 2023: their creation, erasure, and cross-processes (where RLs were created and then erased or vice-versa). Between 2016 and 2019, 7,2% of the former total RL area disappeared (68,781 ha out of a total of 948,979 ha in 2016); the figure rose to 12,1% between 2019 and 2021 (148,733 ha out of a total of 1,228,454 ha in 2019) and was 9,4% between 2021 and 2023 (137,361 ha out of a total of 1,457,608 ha in 2021). The process is gaining velocity, as shown by the increase in the monthly rate of RL erasure: from 0,26% for 2016-2019 to 0,39% for 2019-2021 and 0,5% for 2021-2023. This does not mean a reduction in the total area of RLs, since the rate of creation of RLs from 2016-2021 exceeds that of erasure (618,107 ha compared to 352,691 ha). However, it shows the volatility of the RL declarations made by farmers using the open data platform: during the same period, 118,153 ha of RLs were involved in a cross-process of creation-reclassification or vice-versa.

25It seems that since the farmers’ declarations on SICAR have not yet been systematically validated by the state of Bahia, they can easily change the boundaries and locations of their RLs in the informatic system. The crossing of reclassified RLs with deforestation data (Supporting Information, S2, section 6) does not provide much evidence that the reclassification of RLs is motivated by deforestation projects in the short term: only 4,2% of the deforestation during the period of 2016-2023 occurred in a declassified LR, and only 5% of the declassified RL area during the same period was deforested. However, it is worth noting that the greatest number of cases of declassification occur in areas along the region’s pioneering frontiers, which are still covered by the Cerrado. By 2021, a decree by the state of Bahia authorized farms to suppress their Legal Reserve if they could reallocate it elsewhere (No 22.078, 08/01/2021). The decree was suspended at the end of 2022 by the Brazilian Government Agency for Law Enforcement and Prosecution of Crimes. Nonetheless, a large number of RL relocations may have been legally authorized in the intervening period, and a new decree was published in December 2022 with more evasive language that could newly allow some RL relocations (No 22.646, 21/12/2022).

Figure 6: Evolution of Legal Reserve declared in the CAR between 2016 and 2023

Figure 6: Evolution of Legal Reserve declared in the CAR between 2016 and 2023

Source: authors, based on CAR datasets (2016 and 2021)

Discussion

How to explain the importance of RL compensation in western Bahia?

26The importance of RL compensation in western Bahia is the result of a two-period process: first the Forest Code’s enforcement through CAR at the beginning of the 2010 decade and then the current agricultural frontier’s expansion. According to our interviewees, even if RL compensation existed as a legal figure before the creation of CAR in 2012, it was not widely implemented in the region, due to very little enforcement of the Forest Code. This low level of enforcement concerned not only compensation, but all the features of this law, including illegal deforestation and restoration deficit (Azevedo et al., 2017). This explains why so many farms in the western part of the region had almost no RL when 2012 came: the high agricultural potential of these lands led to their systematic deforestation between the 70s and 2012. With the new Forest Code, these farms collectively sought to regularize their situation, looking massively for lands to use for compensation. They found them in the eastern part of the region, over steep slopes and at cheaper prices because of its lower potential for the highly mechanized agriculture they practiced (interviews). The new wave of agricultural expansion toward the east during the 2010 decade, over areas still covered by the Cerrado vegetation and frequently classified as RL, created new pressures in these lands.

  • 9 Information collected through fieldwork interviews.

27Furthermore, compensation is largely coordinated by Bahia’s agribusiness sector: during the early 2000s, its main association (AIBA) coordinated the digital drawing of property limits with TNC (an NGO);9 this was the basis for the charts later used by CAR in the region. After 2012, the task of finding land to use for compensated RLs, a task that required a high degree of coordination due to the huge amount of surface area needed, was informally organized by the agribusiness sector as well (interviews). Some properties (about 5%) have become specialized in the hosting of compensated RLs, and the existence of long-distance compensation shows a collective capacity to find and share territories for this purpose. It is then probable that the huge scale of the compensation process and the practice of declassification and creation of RLs may be linked to future agricultural and extractive projects involving irrigation and emerging agricultural techniques -new seeds, new soil conditioning methods, etc. (Eloy et al., 2023).

Compensation does not stabilize the agricultural frontier

28Results show that the concentration of RLs on the agricultural forefront does not stabilize this frontier. RLs are only stable when they are located on steep slopes (non-arable land), mainly along stream valleys. Under current conditions, RLs are precarious because they can easily be declassified. Consequently, still-forested lands located in slightly sloping areas may be vulnerable to future deforestation, even if they are currently designated as RLs.

29The new regulations brought by the Bahian reforms (Brannstrom, 2005), followed by the Forest Code reform, have led to the spatial reorganization of this territory. Before 2012, in a context of low enforcement of the Forest Code, this frontier was organized in two parallel stripes: a western strip, intensively deforested for agriculture, near an eastern strip covered by Cerrado, under uncertain legal tenure and mainly used for grazing by peasant communities. Ten years later, the enforcement of the Forest Code has led to the systematization of RL compensation along the frontier. The frontier is now organized into three strips: an almost completely deforested strip in the west, an intermediary strip with a high proportion of land covered by compensated RLs, and a third strip, where the remaining traditional grazing territories are under high pressure due to both threats from big farmers (who would like the eastern pastoralists to abandon their lands) and the imposition of compensated RLs from distant farms. The results suggest that the intermediary strip is not meant to endure and may be a temporary feature of a dynamic spatial organization. This complements previous observations in the same region (Eloy et al., 2016), showing that these areas can easily be changed into croplands when there is a shift in the economic context or in the provision of energy and infrastructure for irrigation. Recent reports from social movements and field data confirm the rapid emergence of new agricultural projects and land conflicts in these areas.

30Moreover, the spatial concentration of compensated RLs does not meet ecological planning objectives, despite the recommendations of the 2012 law (Silva, Ranieri, 2014). For example, there is little overlap between compensated RLs and Priority Areas for Biodiversity Conservation in the Bahianese Cerrado (Seehusen et al., 2017). On the other hand, as in other regions of Brazil, there is a huge overlap between RLs and protected or natural areas. These findings suggest that what attracts compensators is mainly the availability of cheap forested land. These results tend to corroborate the warnings expressed in many prospective studies that evaluated the potential conservation biases introduced by the new Forest Code of 2012 (Sparovek, 2012; Bernasconi et al., 2016; Freitas et al., 2017; Da Silva, Sauer, 2022; Mello et al., 2021).

Compensation in a context of lowering of environmental regulations

31Our results show that the massive use of RL compensation allows increasing deforestation in the most agriculturally suitable areas, transferring the conservation obligation to other areas. Indeed, the systematic use of compensation by the farms located in the western strip of the frontier led to huge tracts of land deprived of any natural vegetation. The agricultural qualities of the land in the west (flat uplands that are easy to mechanize) justified major investments in deforestation and modification of the soil with limestone inputs. It seems to be far more economic for farmers to use compensated RLs than to reforest 20% of the land locally and to lose the agricultural rent on this area. While not illegal, this situation could have huge environmental consequences, as most deforestation previous to 2008 took place in the aquifer catchment areas situated in the flat uplands (Aguiar, Bomfim, Correia, 2021), with major impacts on water infiltration and the hydrological cycle at a regional scale (Salmona et al., 2023; Gaspar, Campos, 2007). Furthermore, our findings show that 35% of the compensations contravene the new Forest Code, since some farms had sufficient natural vegetation in 2008 to meet the obligation to place their RL inside the property. Consequently, what should be a mechanism reserved for the regularization of deficits prior to 2008 then contributes, in this region, to farmers not complying with the obligation to maintain a minimum of vegetation on each property.

32This deregulation of RL location is intrinsically linked to how the CAR is used. There is a consensus that the Brazilian environmental regulation apparatus suffered a general flexibilization at the end of the 2010 decade (Sabourin et al., 2020; Ferraz da Fonseca et al., 2023; Coelho-Junior et al., 2022; Barbosa, Alves, Grelle, 2021), and CAR’s implementation took place in that context (Soares-Filho et al., 2014). Even if, in these circumstances, CAR represented an innovative instrument for the enforcement of environmental regulation, its interaction with other processes had, in our study region, paradoxical effects. Recent studies (Bühler, Gautreau, Oliveira, 2022) point out that agribusiness lobbies have used this instrument strategically, arguing that they take advantage of digital procedures that are not yet systematically checked by public authorities. In fact, although the information provided by farmers has generally not been validated by the state, farmers receive a certificate of environmental compliance that allows them to access public services and market facilities (Bühler, Gautreau, Oliveira, 2022; Da Silva et al., 2023). Despite these limitations, several authors point out that, if soundly implemented (Stabile et al., 2020; Luiz, Steinke, 2022) and coupled with other incentives or land tenure clarification (Azevedo, Stabile, Reis, 2015; L’Roe et al., 2016), CAR still has an unprecedented potential to enforce the environmental regulation of agriculture. Mainly, this is because it will (theoretically) greatly reduce the cost of monitoring farms (Azevedo et al., 2017).

33Moreover, the quest for compensation led to conflicts with the traditional users of the eastern strip by dispossessing them of their grazing territories and threatening their food security (Aguiar, Bomfim, Correia, 2021). Indeed, the large-scale deforestation and irrigation development of the flat uplands of western Bahia is disrupting preexisting forms of water and land use in downstream areas. Traditional community-owned grazing lands, locally called “gerais” or “fecho de pasto,” are also home to the natural springs, which feed gravity-led irrigation ditches that deliver water several kilometers away, in the valleys, for use in cultivation during the dry season (Eloy et al., 2023). “Gerais” are legally vulnerable to expropriation by agribusiness operations because peasant farmers rarely acquire land titles from the state despite having lived there for more than a century. Today, the last remaining plots of “gerais” are on the periphery of the main agribusiness region (figure 2). Although their occupation has generally occurred gradually and silently, violence has erupted in several places along the frontier. In the Correntina and Formosa do Rio Preto municipalities, groups of agribusiness farmers actively threaten to occupy cattle breeders’ lands and use them for agriculture, to set up RLs, or even to sell them to other farmers looking for compensation areas. Social movements in the region describe this process as “green-grabbing” (Aguiar, Bomfim, Correia, 2021), with effects that include making it impossible to access lands historically used by the community, physical violence toward persons, destruction of fences and houses, and general feelings of insecurity (Actionaid, 2017; Porto-Gonçalves, Britto Chagas, 2019; Comisão Pastoral da Terra, 2023).

34As a result, the inconsistent monitoring and enforcement of CAR declarations in western Bahia may have transformed compensated RLs into areas of speculation for agriculture expansion. We argue that compensated RLs can enhance agribusiness strategies for securing land tenure and speculation, as demonstrated in Bahia by the recent attempts to facilitate the compensation rules through the ad-hoc decree of 2021 already mentioned. In the short term, compensated RLs make it possible to award a digital certificate to those agro-industrial farms, even when they do not comply with the Forest Code (in 35% of the cases). In the long term, only strict control of CAR declarations will be able to stabilize the frontier and prevent RL surfaces from speculation and deforestation.

Conclusion: Challenges for the socio-environmental traceability of Brazilian agribusiness

  • 10 The conclusions of this article are valid for western Bahia and cannot be extrapolated to the whole (...)

35To enhance the landscape and political visibility of environmental compensation, this study conducted a systematic regional analysis of Legal Reserve (re)location in a primary agricultural frontier of the Brazilian Cerrado. Environmental compensation, as currently practiced by the agribusiness sector in western Bahia,10 appears paradoxically to contribute to environmental degradation, despite being designed to stabilize ecological impacts. This strategic deployment of compensation mechanisms, although widely documented globally (Maron et al., 2025), appears to reach an extreme display in this region. The situation is particularly paradoxical given that organizations representing regional agribusiness present themselves as pioneers of ecological transition in agriculture. Notably, compensation is largely coordinated by Bahia’s agribusiness sector. In a region heavily dependent on agricultural exports, compensation serves to guarantee to foreign buyers the environmental compliance of imported products. Thus, it is presented as evidence of strict compliance with the law and attributed to the collective virtuous behavior of the agribusiness sector. These results confirm that western Bahia exemplifies a new type of agricultural frontier, deeply embedded within conservation rationales, characteristic of South America’s neoliberal agricultural frontier (Brannstrom, 2005).

36This study also assessed the potential and limitations of public digital data in analyzing this phenomenon. Our methodology and findings highlight the challenges in rendering the social and ecological manifestations of compensation visible across large areas. The CAR exemplifies what may be termed “biased transparency,” offering an unprecedented synoptic view (Scott, 1998) of farm-level environmental conditions in Brazil while concealing crucial dimensions such as compensation linkages. Specifically, it lacks information connecting farms to their off-site compensation areas. While our study demonstrates the value of public digital platforms like the Rural Environmental Registry (CAR) for monitoring landowner compliance with the Forest Code, these platforms remain challenging for non-specialists to analyze and can obscure illegal or controversial practices. Although CAR has been recognized as a significant opportunity for Brazilian agriculture to meet global environmental transparency requirements (Rajão, Loconto, Gautreau, 2020), particularly regarding regulations to prevent importation of deforestation-linked goods (see EU 2023/1115 regulation), we argue that its biased transparency obscures the critical role of environmental compensation in facilitating legal deforestation in the Cerrado.

  • 11 This information was only released after the submission of this paper, 13 years after the 2012 refo (...)

37To strengthen compliance with EU requirements, CAR transparency should be enhanced to provide clearer information about RL compensation. Our analysis required complex GIS processing precisely because, until 2025, the dataset omitted the identification codes necessary to link compensated RLs with their originating farms.11 This limitation undermines CAR’s ability to inform international markets about potential socio-environmental issues in commodity production. Enhanced transparency could also help mitigate the negative effects of compensation schemes on the collective lands and food security of rural communities in Western Bahia. More generally, further research is essential to deepen understanding of the Political Ecology of deforestation within the context of increasingly digitized environmental policies.

Haut de page

Bibliographie

Actionaid, 2017, “Impactos da Expansão do Agronegócio no Matopiba: Comunidades e Meio Ambiente”, Rio de Janeiro, Actionaid.

Aguiar D., Bomfim J., Correia M., 2021, “Na fronteira da (i)legalidade: Desmatamento e grilagem no Matopiba”, Salvador da Bahia, AATR.

Apostolopoulou E., Adams W. M., 2017, “Biodiversity Offsetting and Conservation: Reframing Nature to Save It”, Oryx, Vol.51, No.1, 23‑31.

Avila Peterlini de Souza A. L., 2021, “A compensação ambiental de Reserva Legal”, in: Rajão R., Giudice R. del, Van der Hoff R., Bergo de Carvalho E., Uma breve história da legislação florestal brasileira: contém a Lei no 12.651, de 2012, com comentários criticos acerca da aplicação de seus artigos, 1ª edição, Florianópolis, SC, Expressáo, 141‑144.

Azevedo A. A., Rajão R., Costa M. A., Stabile M. C. C., Macedo M. N., dos Reis T. N. P., et al., 2017, “Limits of Brazil’s Forest Code as a Means to End Illegal Deforestation”, Proceedings of the National Academy of Sciences, Vol.114, No.29, 7653‑7658.

Azevedo A. A., Stabile M. C. C., Reis T. N. P., 2015, “Commodity Production in Brazil: Combining Zero Deforestation and Zero Illegality”, Kapuscinski A. R., Dooley K. (éd.), Elementa: Science of the Anthropocene, Vol.3, 000076.

Barbosa L. G., Alves M. A. S., Grelle C. E. V., 2021, “Actions against Sustainability: Dismantling of the Environmental Policies in Brazil”, Land Use Policy, Vol.104, 105384.

Bernasconi P., Blumentrath S., Barton D. N., Rusch G. M., Romeiro A. R., 2016, “Constraining Forest Certificate’s Market to Improve Cost-Effectiveness of Biodiversity Conservation in São Paulo State, Brazil”, Hewitt J. (éd.), PLOS ONE, Vol.11, No.10, e0164850.

Boisvert V., 2015, “Conclusion. Quel capitalisme pour la biodiversité ?”, in : Le pouvoir de la biodiversité. Néolibéralisation de la nature dans les pays émergents, Thomas F., Boisvert V. (éd.), Versailles, Quae, 247‑262.

Brannstrom C., 2005, “Environmental Policy Reform on North-Eastern Brazil’s Agricultural Frontier”, Geoforum, Vol.36, No.2, 257‑271.

Brannstrom C., 2009, “South America’s Neoliberal Agricultural Frontiers: Places of Environmental Sacrifice or Conservation Opportunity?”, Ambio, Vol.38, No.3, 141‑149.

Brito B., 2020, “The Pioneer Market for Forest Law Compliance in Paragominas, Eastern Brazilian Amazon”, Land Use Policy, Vol.94, No.104310.

Bühler È. A., Gautreau P., Oliveira V. L. de, 2022, “La revanche de l’agrobusiness brésilien : usages et paradoxes de la régulation environnementale par le numérique”, Études rurales, No.209, 40‑60.

Castree N., 2008, Neoliberalising Nature: The Logics of Deregulation and Reregulation, Environment and Planning A: Economy and Space, Vol.40, No.1, 131-152.

Coelho-Junior M. G., Valdiones A. P., Shimbo J. Z., Silgueiro V., Rosa M., Del Lama Marques C., et al., 2022, “Unmasking the impunity of illegal deforestation in the Brazilian Amazon: a call for enforcement and accountability”, Environmental Research Letters, Vol.17, No.041001.

Comisão Pastoral da Terra, 2023, “Conflitos no Campo Brasil 2022”, Goiânia.

Da Silva P., Sauer S., 2022, “Desmantelamento e desregulação de políticas ambientais e apropriação da terra e de bens naturais no Cerrado”, Raízes: Revista de Ciências Sociais e Econômicas, Vol.42, No.2, 298‑315.

Da Silva R. F. B., De Castro Victoria D., Ávila Nossack F., Viña A., Millington J. D. A., Vieira S. A., et al., 2023, “Slow-down of Deforestation Following a Brazilian Forest Policy Was Less Effective on Private Lands than in All Conservation Areas”, Communications Earth & Environment, Vol.4, No.1, 111.

Daugeard M., Cunha P. R., 2022, “Entre exploitation et protection : la question forestière au cœur d’une ambivalence historique au Brésil”, Caravelle, No.119, 51‑70.

Dauguet B., 2015, “Biodiversity Offsetting as a Commodification Process: A French Case Study as a Concrete Example”, Biological Conservation, Vol.192, 533‑540.

Devictor V., 2015, “When Conservation Challenges Biodiversity Offsetting”, Biological Conservation, Vol.192, 483‑484.

Devictor V., 2018, Dossier : La fabrique de la compensation écologique : controverses et pratiques – La compensation écologique : fondements épistémiques et reconfigurations technoscientifiques, Natures Sciences Sociétés, Vol.26, No.2, 136-149.

Droste N., Alkan Olsson J., Hanson H., Knaggård Å., Lima G., Lundmark L., et al., 2022, “A Global Overview of Biodiversity Offsetting Governance”, Journal of Environmental Management, Vol.316, 115231.

Eloy L., Aubertin C., Toni F., Laine Borges Lúcio S., Bosgiraud M., 2016, “On the Margins of Soy Farms: Traditional Populations and Selective Environmental Policies in the Brazilian Cerrado”, The Journal of Peasant Studies, Vol.43, No.2, 494‑516.

Eloy L., Da Silva A. L., Coelho Filho O., Ghiotti S., 2023, “The Water Frontier: Agribusiness vs. Smallholder Communities in the Brazilian Cerrado”, Water Alternatives, Vol.16, No.3, 869‑891.

Fairhead J., Leach M., Scoones I., 2012, Green Grabbing: a new appropriation of nature?, The Journal of Peasant Studies, Vol.39, No.2, 237-261.

Ferraz da Fonseca I., Bursztyn M., Pereira Lindoso D., Daugeard M., 2023, “A desconstrução organizada da política florestal no Brasil: estratégias de desmantelamento e de resistência”, in: Gomide A. de A., Silva M. M de Sá e, Leopoldi M. A. (éd.), Desmonte e reconfiguração de políticas públicas (2016-2022), Brasília, IPEA, INCT/PPED, 125‑155.

Frederico S., Albuquerque B. H., Castro De Almeida M., 2024, “Fronteira agrícola e green grabbing: apropriação digital de terras nos Cerrados Piauienses/Agricultural frontier and green grabbing: digital land grabbing in the Cerrados of Piauí/Frontière agricole et appropriation verte : l’appropriation numérique des terres dans les Cerrados de Piauí”, REVISTA NERA, Vol.27, No.4.

Freitas F. L. M., Sparovek G., Berndes G., Persson U. M., Englund O., Barretto A., Mörtberg U., 2018, “Potential Increase of Legal Deforestation in Brazilian Amazon after Forest Act Revision”, Nature Sustainability, Vol.1, No.11, 665‑670.

Freitas F. L. M. de, Sparovek G., Mörtberg U., Silveira S., Klug I., Berndes G., 2017, “Offsetting Legal Deficits of Native Vegetation among Brazilian Landholders: Effects on Nature Protection and Socioeconomic Development”, Land Use Policy, Vol.68, 189‑199.

Gaspar, M. T. P., Campos J. E. G., 2007, “O sistema aqüífero Urucuia”, Revista Brasileira de Geociências, Vol.37, No.4, 1068‑1078.

Gautreau P., Gisclard M., Langbehn L., Marquis-Dupont G., 2016, “Regular as fronteiras agrícolas sul-americanas? Experiências e negociações ambientais no Chaco argentino, no Uruguai e no Rio Grande do Sul – Brasil”, in: Bühler E. A., Guibert M., de Oliveira V. L. (éd.), Agriculturas empresariais e espaços rurais na globalização, Porto Alegre, Brasil, UFRGS Editora, 171‑190.

Gelot S., Bigard C., 2021, “Challenges to Developing Mitigation Hierarchy Policy: Findings from a Nationwide Database Analysis in France”, Biological Conservation, Vol.263, 109343.

Gibbs H. K., Rausch L., Munger J., Schelly I., Morton D. C., Noojipady P., et al., 2015, “Brazil’s Soy Moratorium”, Science, Vol.347, No.6220, 377‑378.

INPE, 2023, Terra Brasilis. PRODES Desmatamento. http://terrabrasilis.dpi.inpe.br/app/dashboard/deforestation/biomes/cerrado/increments.

L’Roe J., Rausch L., Munger J., Gibbs H. K., 2016, “Mapping Properties to Monitor Forests: Landholder Response to a Large Environmental Registration Program in the Brazilian Amazon”, Land Use Policy, Vol.57, 193‑203.

Levrel H., 2020, Les compensations écologiques, Paris, La Découverte.

Luiz C. H. P., Steinke V. A., 2022, “Recent Environmental Legislation in Brazil and the Impact on Cerrado Deforestation Rates”, Sustainability, Vol.14, No.13, 8096.

McAfee K., Shapiro E. N., 2010, “Payments for Ecosystem Services in Mexico: Nature, Neoliberalism, Social Movements, and the State”, Annals of the Association of American Geographers, Vol.100, No.3, 579‑599.

Maron M., Von Hase A., Quétier F., Sonter L. J., Theis S., Zu Ermgassen S. O. S. E., 2025, “Biodiversity Offsets, Their Effectiveness and Their Role in a Nature Positive Future”, Nature Reviews Biodiversity, Vol.1, No.3, 183‑196.

Martinez-Alier J., Munda G., O’Neill J., 1998, “Weak Comparability of Values as a Foundation for Ecological Economics”, Ecological Economics, Vol.26, No.3, 277‑286.

Mello K. de, Fendrich A. N., Sparovek G., Simmonds J. S., Maron M., Tavares P. A., et al., 2021, “Achieving Private Conservation Targets in Brazil through Restoration and Compensation Schemes without Impairing Productive Lands”, Environmental Science & Policy, Vol.120, 1‑10.

Moreno-Mateos D., Maris V., Béchet A., Curran M., 2015, “The True Loss Caused by Biodiversity Offsets”, Biological Conservation, Vol.192, 552‑559.

Padilla B., Gelot S., Guette A., Carruthers-Jones J., 2024, “La compensation écologique permet-elle vraiment de tendre vers l’absence de perte nette de biodiversité ?”, Cybergeo. Environnement, nature, paysage, No.1060.

Pinto L. F. G., Guidotti V., Cerignoni F., Freitas F. L. M., Sparovek G., et al., 2018, “Código Florestal: a abrangência e os vazios do CAR – quanto e quem falta”, Sustentabilidade em Debate, Vol.8, 14.

Porto-Gonçalves C. W., Britto Chagas S., 2019, “Os pivôs da discórdia e a digna raiva: uma análise dos conflitos por terra, água e território em Correntina-BA”, Bom Jesus da Lapa, Editora Bom Jesus.

Rajão R., Azevedo A, Stabile M. C., 2012, “Institutional subversion and deforestation: Learning lessons from the system for the environmental licensing of rural properties in Mato Grosso”, Public Administration and Development, Vol.32, 229-244.

Rajão R., Giudice R. del, Van der Hoff R., Bergo de Carvalho E., 2021, Uma breve história da legislação florestal brasileira: contém a Lei no 12.651, de 2012, com comentários críticos acerca da aplicação de seus artigos, Florianópolis, SC, Expressáo.

Rajão R., Loconto A., Gautreau P., 2020, “The limits of private certifications and the potential of state-led spatial data infrastructure in South America as to ensure deforestation-free exports to the EU”, European Commission feedback. https://hal.inrae.fr/hal-02912500/document.

Rasmussen L. V., Jung S., Dantas Brites A., Watkins C., Agrawal A., 2016, “Understanding smallholders’ intended deforestation behavior in the Brazilian Cerrado following environmental registry”, Environmental Research Letters, Vol.12, No.9, 094001.

Reid J., Bruner A., Chow J., Malky A., Rubio J. C., Vallejos C., 2015, “Ecological Compensation to Address Environmental Externalities: Lessons from South American Case Studies”, Journal of Sustainable Forestry, Vol.34, Nos.6‑7, 605‑622.

Rundcrantz K., Skärbäck E., 2003, “Environmental Compensation in Planning: A Review of Five Different Countries with Major Emphasis on the German System”, European Environment, Vol.13, No.4, 204‑226.

Sabourin E., Grisa C., Niederle P., Pereira Leite S., Milhorance C., Damasceno Ferreira A., et al., 2020, “Le démantèlement des politiques publiques rurales et environnementales au Brésil”, Cahiers Agricultures, Vol.29, No.31.

Salmona Y. B., Trondoli Matricardi E. A., Skole D. L., Andrade Silva J.F, Coelho Filho O. de A., Pedlowski M. A., et al., 2023, “A Worrying Future for River Flows in the Brazilian Cerrado Provoked by Land Use and Climate Changes”, Sustainability, Vol.15, No.5, 4251.

Sauer S., Pereira Leite S., 2012, “Agrarian Structure, Foreign Investment in Land, and Land Prices in Brazil”, The Journal of Peasant Studies, Vol.39, Nos.3‑4, 873‑898.

Seehusen S. E., Kieling D., Bruner A., Vilela T., 2017, Os instrumentos de mercado podem contribuir para o alcance de objetivos ambientais a menores custos? Cenários de implementação das cotas de reserva ambiental na Bahia, Série técnica, Conservation Strategy Fund. https://www.conservation-strategy.org/sites/default/files/field-file/PORT_Serie_Tecnica_CRAs_Bahia_CSF_Jan.2017.pdf.

Scott J.C., 1998, Seeing like a state : how certain schemes to improve the human condition have failed, New Haven, Yale University Press.

Silva J. S. da, Ranieri V. E. L., 2014, “The legal reserve areas compensation mechanism and its economic and environmental implications”, Ambiente & Sociedade, Vol.17, No.1, 115‑132.

Soares-Filho B., Rajao R., Macedo M., Carneiro A., Costa W., Coe M., et al., 2014, “Cracking Brazil’s Forest Code”, Science, Vol.344, No.6182, 363‑364.

Sparovek G., 2012, “Caminhos e escolhas na revisão do Código Florestal: quando a compensação compensa?”, Visão agrícola, No.10, 25‑28. https://www.esalq.usp.br/visaoagricola/sites/default/files/VA10-visao-tecnica04.pdf.

Stabile M. C. C., Guimarães A. L., Silva D. S., Ribeiro V., Macedo M. N., Coe M. T., et al., 2020, “Solving Brazil’s Land Use Puzzle: Increasing Production and Slowing Amazon Deforestation”, Land Use Policy, Vol.91, 104362.

Strassburg B. B. N., Brooks T., Feltran-Barbieri R., Iribarrem A., Crouzeilles R., Loyola R., et al., 2017, “Moment of Truth for the Cerrado Hotspot”, Nature Ecology & Evolution, Vol.1, No.4, 0099.

Wang L., Lv T., Zhang X., Hu H., Cai X., 2022, “Global Research Trends and Gaps in Ecological Compensation Studies from 1990 to 2020: A Scientometric Review”, Journal for Nature Conservation, Vol.65, 126097.

Haut de page

Annexe

Supporting Information

S1. Description of the GIS methodology deployed in the paper

Precisions about the spatial scope of the data

As the objective of the paper is to study the dynamics of agricultural frontiers in western Bahia and the role that Legal Reserve compensation plays in theses dynamics, we chose data that covered the entire Cerrado biome within the state of Bahia. Law 12.651 on the protection of Native Forests establishes in Article 46, paragraph 2, that compensation is authorized only within the same biome of the same state (with some exception to this second rule when compensation is made over priority conservation areas defined at federal level). The Environmental Reserve Quota (CRA) can only be used to offset the Legal Reserve of a rural property located in the same biome as the area to which the title is linked. This means that if a property owner needs to compensate for a deficient Legal Reserve on their rural property, they must acquire the CRA from an area within the same biome. The CRA serves as a mechanism to ensure that the compensation of Legal Reserves contributes to the preservation and conservation of the specific biome’s biodiversity and ecological balance. By requiring the CRA to be sourced within the same biome, the legislation aims to promote the conservation of the unique environmental characteristics and ecosystems associated with each specific region (Federal Law No 12.651/2012).

The data used in this study were obtained from the SICAR website (https://www.car.gov.br/​publico/​imoveis/​index), where they are accessible as open data. The primary analysis was conducted using data from September 2021. In order to examine the spatial evolution of Legal Reserves, additional SICAR files were downloaded from November 2016, February 2019, and April 2023. By incorporating data from multiple time points, this study aims to provide insight into the changes and trends in Legal Reserves over time, allowing for a comprehensive understanding of their spatial dynamics.

Data processing and analysis in detail

The combination of self-declared data by rural landowners and variable interpretation during the mapping and validation process can result in a significant amount of overlapping data in the Rural Environmental Registry (CAR). This combination of factors contributes to a certain level of data inaccuracy. It is worth noting that the state of Bahia demonstrates a higher quality of CAR data than other states in Brazil. However, overlapping registrations can still be observed, highlighting the ongoing need for improvement and monitoring in order to ensure the accuracy and reliability of the recorded information.

In our analysis, we converted our data layers into centroids using the “realcentroid” extension in QGIS. This transformation ensures that the centroid is located inside a polygon, even when dealing with complex geometries. It is worth noting that conventional centroid tools may generate centroids that fall outside complex features. Since the conversion of polygons into points was done systematically, we do not explicitly mention it in the subsequent paragraphs.

All cartographic results presented with a grid were generated by aggregating polygon areas within a 20 km-width hexagonal cell grid. This grid allows for a more detailed analysis of spatial variations of phenomena across municipalities with a west–east length of nearly 200 km. By utilizing this hexagonal cell grid, the analysis captures a finer level of geographic representation, enhancing the understanding of spatial patterns and trends within the study area.

Section 1. Data preparation

The municipalities within the Cerrado biome in the state of Bahia were downloaded for this analysis. The downloaded data were extracted and merged to create a single unified dataset. Although the data contained various layers, only some of the data layers were utilized for this analysis. All the data layers underwent geometry repair, and any duplicate geometries were eliminated. It is important to note that the presence of duplicated geometries was minimal, representing less than 1% of all entities within each data layer.

From the official SICAR website, all the data layer files of all the municipalities covering the Cerrado biome in the state of Bahia were downloaded. We retained five data layers: AREA_IMOVEL, which represents the boundaries of farm parcels; RESERVA_LEGAL, indicating the limits of declared Legal Reserves; APP, delineating the compulsory protected areas near rivers, water sources, and high slopes; AREA_CONSOLIDADA, representing the area occupied by agriculture within the farms prior to 22 July 2008, which cannot be considered to have been illegally deforested; and VEGETACAO_NATIVA, which denotes remnants of spontaneous vegetation outside of the declared area of Legal Reserve.

Section 2. Detection and elimination of overlapping polygons in the IMO (farm boundaries) data layer

The overlapping of polygons posed significant challenges in the subsequent steps of our analysis, especially when performing spatial joins to attribute data from other layers. To mitigate potential errors resulting from these spatial joins, it became necessary to address the issue of overlapping polygons in the IMO data layer (the farm boundaries layer). However, it should be noted that, due to the inherent inaccuracies in the self-declared SICAR database, even minor overlapping issues affected a large portion of the polygons. In order to retain a substantial amount of data without compromising the integrity of the layer, we established a maximum threshold for overlapping. This threshold helped us determine which polygons should be removed and which ones should be retained, striking a balance between data preservation and error reduction.

We conducted an analysis of overlapping polygons in the IMO data layer to determine the extent to which each polygon’s surface was overlapped by other polygons within the same file.

Figure A: Area covered by overlapping farms that were eliminated from the database (overlapping threshold: ≥ 5%)

Figure A: Area covered by overlapping farms that were eliminated from the database (overlapping threshold: ≥ 5%)

Source: authors, CAR database (2021)

We tried to optimize the overlapping threshold in order to reduce global overlapping in the database while retaining the highest number of farms. Since our research focuses on agribusiness strategies in the western municipalities of Bahia, we calculated the percentage of farms eliminated in the eight municipalities along the western border for threshold values of 15% (2,29% of farms), 10% (2,39%), and 5% (4,70%). We ultimately selected a threshold of 5% for overlapping as it provided a high level of accuracy in our analysis while retaining over 95% of the farms in the database. Figure a illustrates the area covered by the farms that were eliminated from the database.

Section 3. Establishing a typology of farms according to Legal Reserve compensation strategy

For this step, the objective was to classify the farms from the IMO data layer into four types (as illustrated in Document 1 of these Supporting Information):

  • The farms that have no Legal Reserve inside their perimeter are “full compensators” (type 1).

  • Farms that have a portion of their Legal Reserve, which does not reach 20% of their total area, and compensate for the remaining area by acquiring a compensated Legal Reserve outside their perimeter, are categorized as “partial compensators” (type 2).

  • The farms with one Legal Reserve covering 20% of their area correspond to the “normal situation” (type 3).

  • Farms that not only have their own Legal Reserve but also receive compensated Legal Reserve from other farms are referred to as “compensation receivers” (type 4).

For this step, we crossed the farms data layer (IMO) and the Legal Reserve layer (RL). The typology was obtained after three operations of spatial join within the IMO data layer, calculating several dimensions of the Legal Reserves that fall inside them (see following steps).

3.1. Preparation of the LR and IMO (farms) data layers

The Legal Reserve (LR) data layer is made of “multipart” polygons, meaning that multiple parts or geometries can be described in the database as a single polygon entry in the attribute table. To illustrate this, let’s consider a “normal” farm (type 3) where the Legal Reserve is non-continuous. In this case, the Legal Reserve can be represented by a single polygon entry in the attribute table but is comprised of six geometries.

In the case of “partial compensator farms” (type 2), their corresponding Legal Reserve includes one geometry within the farm’s perimeter and one or more additional geometries located on other farms. However, all these geometries are encoded as a single polygon entry in the attribute table.

We conducted an analysis of each polygon within the LR data layer, calculating the number of its parts and adding this information to a new column in the attribute table called ‘nb_parts’. This calculation was performed using the expression ‘num_geometries($geometry)’ in the field calculator within QGIS.

In order to convert the LR data layer from a “multi-part” format to a more manageable “single-part” format, we conducted a splitting operation. This process involved transforming Legal Reserve polygons, which originally consisted of multiple parts, into separate autonomous polygons. For instance, if a Legal Reserve polygon initially had five parts, it would be split into five individual polygons. Each of these newly formed polygons inherited the identifier of the original polygon and preserved the record of the number of parts it originated from, which was stored in the ‘nb_parts’ column.

By spatially joining the LR layer to the IMO (farms) layer, we added the following fields to the IMO layer:

  • RL_nb: the number of LR pieces that fall inside each farm.

  • unique_rl: the number of unique LR identifiers (ID) for each farm.

  • sum_nb_parts: the sum of the number of parts that every LR piece was coded with.

Figure B: Diagram of step 3

Figure B: Diagram of step 3

Source: authors

3.2. Farm typology processing

At the end of step 3.1, the IMO data layer gained three new columns or fields: ‘RL_nb’/‘unique_rl’/‘sum_nb_parts’.

The second step of the typology processing involved filtering the IMO data layer to achieve an initial categorization of farms based on specific criteria. This step included the following categorizations:

  1. Type 1 farms (integral compensators): by applying the expression ‘unique_rl = 0’ to the IMO data layer, we directly obtained farms that have no unique LR identifier associated with them. These farms are classified as integral compensators, indicating that they do not have any Legal Reserve within their perimeter.

  2. Type 4 farms (compensation receivers): applying the expression ‘unique_rl ≥ 2’ to the IMO data layer allowed us to directly identify farms that have two or more unique LR identifiers associated with them. These farms are classified as compensation receivers, indicating that they not only have their own Legal Reserve but also receive compensated Legal Reserve from other farms.

At the end of this second step, we had a third group of farms that includes both types 2 and 3. To separate them out, we segregated the unidentified group of farms and applied a filter using the expression ‘unique_rl = 1 AND (sum_nb_part > (RL_nb * RL_nb))’. By executing this step, we successfully obtain the type 2 classification for these farms. The type 3 farms were obtained by identifying the farms that do not belong to types 1, 2, or 4.

Section 4. Compensation link creation

The objective of this step was to draw a line between the farms that compensate (types 1 and 2) with the polygons of their compensated Legal Reserve. This step was developed in order to tackle a lack of information in the open database SICAR. For inexplicit reasons, the identifier that could link a farm (imóvel) with its Legal Reserve, the “IDF” code, is missing in the farm data layer (AREA_IMOVEL). This is problematic when the Legal Reserve is not located on its “own” farm but it is compensated for outside the farm: without an IDF code in the farm data layer, it becomes impossible to link compensated Legal Reserve with the distant farm it belongs to. If the IDF code were given in the farms data layer, step 4.1 would be unnecessary.

4.1. Attribution of IDF codes to the farms data layer

The process of assigning IDF codes to farm polygons is based on the principle that IDF codes are present in all data layers except farms, including Legal Reserve, APP, AREAc, and VEGnat. To link a compensated Legal Reserve polygon to its original farm, we relied on these three additional data layers, which represent features that cannot exist outside the farm boundaries. For example, APP areas are mandatory conservation zones located alongside watercourses, and they are always situated within the farm limits.

Each data layer (APP, AREAc, and VEGnat) was first divided into individual features (from multi-part to single-part) and then converted into points using the “realcentroid” extension in QGIS. After a series of three spatial join operations, we assigned IDF codes to the farm data layer for all farm polygons that contained any of these features (an APP, a portion of natural vegetation, or an agricultural “consolidated” area).

It is important to emphasize that only 35% of the farms could be assigned an IDF code. This is because a significant portion of the farms in the region, particularly those located in deforested areas in the central plateau without watercourses, do not have APP data or other complementary features.

4.2. Identifying Legal Reserve as “normal” or “compensated”

Nota bene: this step only concerns farms of type 4 and the Legal Reserve polygons that fall into them.

The objective of this step was to identify, for every farm belonging to type 4 (compensation receivers), which Legal Reserves come from other farms and which is the Legal Reserve of the type 4 farm in question. (Even if a farm receives compensated Legal Reserves from other farms, it must also have its own Legal Reserve to comply with the Forest Code.)

Thanks to step 4.1, we could compare the IDF codes of the farms and the IDF codes of the Legal Reserves. For that, we followed the next steps:

  • By spatial join, we added the IDF code of the type 4 farms (compensation receivers) to the LR data layer whenever a LR polygons fell inside it. A new column was created with name “IDF2.”

  • In the LR data layer, we created a new column, “compared IDF,” where we calculated the operation “IDF/IDF2.” When the result was 1, that meant that the Legal Reserve polygon was over its own farm. When the result was other than 1, that meant that the Legal Reserve was compensated (its IDF did not match the IDF of the farm polygon it fell inside).

To correctly classify all the Legal Reserve of the LR data layer, we joined a field to this layer indicating the type of the farm it fell inside (a ‘typology’ field). With this new field, we performed the filter [typology = 2 OR typology = 3 OR (typology = 4 AND compared_idf = 1)] to find the normal Legal Reserve and, by differentiation, the compensated ones.

Figure C: Diagram of step 4

Figure C: Diagram of step 4

Source: authors

4.3. Creation of lines between compensation farms (types 1 and 2) and compensated Legal Reserve

A specific LR data layer was created, containing only the compensated Legal Reserve polygons (see step 4.2). Lines were then drawn between the compensated LR data layer and the farm data layer (but only for the farm polygons that could be given an IDF).

Section 5. Quantification of Legal Reserve loss or gain

To study the spatial evolution of Legal Reserves over time, we utilized data layers obtained from the SICAR website in November 2016, February 2019, September 2021, and April 2023. In order to analyze the changes in the spatial data layers of Legal Reserves, we calculated the spatial differences for three specific periods: 2016-2019 (27 months), 2019-2021 (31 months), and 2021-2023 (19 months). Union, Difference and Intersection geoprocessing tools from QGis were used to identify these changes. For each period, we then calculated the new areas and the areas that were lost in comparison to the initial area covered by Legal Reserves (that is, 2016 for the 2016-2019 period; 2019 for the 2019-2021 period, and so on). These calculations allowed us to determine a decrease rate for each period. Additionally, we computed a monthly decrease rate to examine the trend over the entire duration of the study.

Section 6. Evaluation of land-use after declassification of Legal Reserves

To assess the possible transformation of Legal Reserves into agricultural lands, we compared the LR surfaces lost during the studied period (2016-2023) with data provided by the Brazilian Space Agency (INPE) through its platform PRODES-Cerrado for the period from August 2017 to August 2022 (data downloaded on 12 May 2023). The declassified LR area that overlaps with deforested area represents only 17,977.25 ha. This area represents 4,1% of the deforestation that occurred during this period (435,527.52 ha) and 5% of the LR area that disappeared (352,691.56 ha) during the same period.

S2. Area covered by vegetation on compensator farms in 2008

For every farm (“imóvel” in the SICAR database) that uses compensation, totally or partially, we calculated the percentage of its area covered with natural vegetation in the land cover MapBiomas dataset for the year 2009. 2009 was the closest dataset available to July 2008, the temporal limit established by the Forest Code of 2012 to allow compensation. Only the farms that had less than 20% of their area covered with natural vegetation on 22 July 2008 were allowed to compensate for their Legal Reserve. The farms that deforested after July 2008 only could restore their Legal Reserve. Farms in pink and red in the map are probably infringing on the Forest Code.

Figure D: Percentage of farm area covered with vegetation in 2008

Figure D: Percentage of farm area covered with vegetation in 2008

Sources: SICAR 2021; Projeto MapBiomas – Collection 8, Land cover and usage in Brazil. Map downloaded: 2009. Retrieved 24.09.2023, via link: https://brasil.mapbiomas.org/​.

S3. Global balance of compensation in the Bahia Cerrado (municipalities studied in this paper)

Area in ha

Calculation detail

Total area of compensator farms (full and partial compensation)

1,513,185.22 ha

A. Total area of compensated LR

316 657

B. Total LR deficit among full compensator farms

302 637

B = 20% of 1,513,185.22 ha

C. Total LR deficit among partial compensator farms

58 051

C = b – c

a. Total area of partial compensator farms: 643,504.55 ha

b. 20% of that = 128,700.91 ha

c. Total existing LR surface on partial compensator farms: 70,649.56 ha

D. Total deficit of LR in BA Cerrado

360 688

D = B + C

E. Unexplained deficit of LR

44 030

E = D – A

E = 12,2% of A

Source: authors analysis based on SICAR 2021

Haut de page

Notes

1 For savannahs located inside the administrative region called “Legal Amazon”, the Legal Reserve is 35% of the farm.

2 In practice, restoration in Brazilian farms is reached mainly by enclosure of the Legal Reserve area, to allow spontaneous reforestation. Very few active reforestation is practiced, due to its cost.

3 One of the main mechanisms used to force farmers to register in the CAR database was to block their access to public bank financing. As very large farmers are structurally dependent on this financing, they were the first to register collectively and massively in the CAR. Farmers who were less dependent on bank financing joined the scheme later and less systematically.

4 https://consultapublica.car.gov.br/publico/imoveis/index

5 https://brasil.mapbiomas.org/

6 Organization of the Brazilian Catholic Church born under the dictatorship (1964-1985), which intervenes in the territories in favor of rural workers, in defense of their working conditions and to record and denounce the violence of which they are victims.

7 Instituto do Meio Ambiente e Recursos Hídricos.

8 This situation changed in 2025, 13 years after the first release of the CAR.

9 Information collected through fieldwork interviews.

10 The conclusions of this article are valid for western Bahia and cannot be extrapolated to the whole of Brazil. The scale and systematic nature of the compensation practices we have identified in this region have not yet been observed elsewhere in the country. Future analyses will be needed to determine whether western Bahia is an exceptional case.

11 This information was only released after the submission of this paper, 13 years after the 2012 reform of the Forest Code.

Haut de page

Table des illustrations

Titre Figure 1: Typology of farms according to Legal Reserve compensation strategy
Légende Legend: Black polygons represent farms; green polygons represent legal reserves. Arrows represent compensation direction.
Crédits Source: authors
URL http://journals.openedition.org/cybergeo/docannexe/image/42687/img-1.jpg
Fichier image/jpeg, 272k
Titre Figure 2: Typology of farms according to their compensation strategy
Légende One third of compensations declared to CAR do not meet the Forest Code requirements.
Crédits Source: authors, based on CAR dataset (2021)
URL http://journals.openedition.org/cybergeo/docannexe/image/42687/img-2.jpg
Fichier image/jpeg, 692k
Titre Figure 3: Spatial concentration of emission or reception of compensated areas
Crédits Source: authors, based on CAR dataset (2021)
URL http://journals.openedition.org/cybergeo/docannexe/image/42687/img-3.jpg
Fichier image/jpeg, 392k
Titre Figure 4: Links between farms that emit and farms that receive compensated areas
Crédits Source: authors, based on CAR dataset (2021)
URL http://journals.openedition.org/cybergeo/docannexe/image/42687/img-4.jpg
Fichier image/jpeg, 280k
Titre Figure 5: Overlapping of compensated Legal Reserves and territories claimed by traditional peasant communities in the Rio Corrente hydrographical basin
Crédits Source: authors, based on CAR dataset (2021) and Comisão Pastoral da Terra (Fundos e fechos de pasto limits)
URL http://journals.openedition.org/cybergeo/docannexe/image/42687/img-5.jpg
Fichier image/jpeg, 604k
Titre Figure 6: Evolution of Legal Reserve declared in the CAR between 2016 and 2023
Crédits Source: authors, based on CAR datasets (2016 and 2021)
URL http://journals.openedition.org/cybergeo/docannexe/image/42687/img-6.jpg
Fichier image/jpeg, 1,0M
Titre Figure A: Area covered by overlapping farms that were eliminated from the database (overlapping threshold: ≥ 5%)
Crédits Source: authors, CAR database (2021)
URL http://journals.openedition.org/cybergeo/docannexe/image/42687/img-7.jpg
Fichier image/jpeg, 276k
Titre Figure B: Diagram of step 3
Crédits Source: authors
URL http://journals.openedition.org/cybergeo/docannexe/image/42687/img-8.jpg
Fichier image/jpeg, 280k
Titre Figure C: Diagram of step 4
Crédits Source: authors
URL http://journals.openedition.org/cybergeo/docannexe/image/42687/img-9.jpg
Fichier image/jpeg, 220k
Titre Figure D: Percentage of farm area covered with vegetation in 2008
Crédits Sources: SICAR 2021; Projeto MapBiomas – Collection 8, Land cover and usage in Brazil. Map downloaded: 2009. Retrieved 24.09.2023, via link: https://brasil.mapbiomas.org/​.
URL http://journals.openedition.org/cybergeo/docannexe/image/42687/img-10.jpg
Fichier image/jpeg, 534k
Haut de page

Pour citer cet article

Référence électronique

Pierre Gautreau, Eve Anne Bühler, Ludivine Eloy et Sylvain Théry, « Making environmental compensation visible. Biased data transparency and agrobusiness strategies in the Bahianese agricultural frontier of the Cerrado », Cybergeo: European Journal of Geography [En ligne], Politique, Culture, Représentations, document 1103, mis en ligne le 03 février 2026, consulté le 14 février 2026. URL : http://journals.openedition.org/cybergeo/42687 ; DOI : https://doi.org/10.4000/15ly4

Haut de page

Auteurs

Pierre Gautreau

Université Paris 1 Panthéon Sorbonne, Laboratoire PRODIG UMR 8586, France.
Contribution: research design, GIS data processing protocol, research and data analysis, writing.
pierre.gautreau@univ-paris1.fr

Articles du même auteur

Eve Anne Bühler

Geography Department/IGEO, Federal University of Rio de Janeiro, Brazil.
Contribution: research design, research and data analysis, writing.
eve.buhler@gmail.com

Ludivine Eloy

Laboratoire ART-Dev UMR 5281, Univ Montpellier, CIRAD, CNRS, Université Paul Valéry Montpellier 3, Université Perpignan Via Domitia, Montpellier, France.
Contribution: research and data analysis, writing.
ludivine.eloy@univ-montp3.fr

Articles du même auteur

Sylvain Théry

Laboratoire ART-Dev UMR 5281, Univ Montpellier, CIRAD, CNRS, Université Paul Valéry Montpellier 3, Université Perpignan Via Domitia, Montpellier, France.
Contribution: data analysis.
sylvain.thery@cnrs.fr

Haut de page

Droits d’auteur

CC-BY-4.0

Le texte seul est utilisable sous licence CC BY 4.0. Les autres éléments (illustrations, fichiers annexes importés) sont susceptibles d’être soumis à des autorisations d’usage spécifiques.

Haut de page
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search