You are on page 1of 14

Received: 4 June 2017 | Accepted: 7 November 2017

DOI: 10.1111/gcb.14011

PRIMARY RESEARCH ARTICLE

Who owns the Brazilian carbon?

Flavio L. M. Freitas1 | Oskar Englund2 | Gerd Sparovek3 € ran Berndes2


| Go |
Vinicius Guidotti4 € rtberg1
| Luıs F. G. Pinto4 | Ulla Mo

1
Department of Sustainable Development,
Environmental Science and Engineering, Abstract
KTH Royal Institute of Technology, Brazil is one of the major contributors to land-use change emissions, mostly driven
Stockholm, Sweden
2 by agricultural expansion for food, feed, and bioenergy feedstock. Policies to avoid
Department of Space, Earth and
Environment, Physical Resource Theory, deforestation related to private commitments, economic incentives, and other sup-
Chalmers University of Technology,
port schemes are expected to improve the effectiveness of current command and
Gothenburg, Sweden
3
Soil Science Department, University of S~ao control mechanisms increasingly. However, until recently, land tenure was unknown
Paulo, Piracicaba, Brazil for much of the Brazilian territory, which has undermined the governance of native
4
Instituto de Manejo e Certificacß~ao
vegetation and challenged support and incentive mechanisms for avoiding defor-
Florestal e Agrıcola—IMAFLORA,
Piracicaba, Brazil estation. We assess the total extent of public governance mechanisms protecting
aboveground carbon (AGC) stocks. We constructed a land tenure dataset for the
Correspondence
Flavio L. M. Freitas, Department of entire nation and modeled the effects and uncertainties of major land-use acts on
Sustainable Development, Environmental
protecting AGC stocks. Roughly 70% of the AGC stock in Brazil is estimated to be
Science and Engineering, KTH Royal
Institute of Technology, Stockholm, Sweden. under legal protection, and an additional 20% is expected to be protected after
Email: flmdf@kth.se
areas in the Amazon with currently undesignated land undergo a tenure regulariza-
Funding information tion. About 30% of the AGC stock is on private land, of which roughly two-thirds
Conselho Nacional de Desenvolvimento
are protected. The Cerrado, Amazon, and Caatinga biomes hold about 40%, 30%,
gico, Grant/Award
Cientıfico e Tecnolo
Number: 249404/2013-3 and 20% of the unprotected AGC, respectively. Effective conservation of protected
and unprotected carbon will depend on successful implementation of the Forest
Act, and regularization of land tenure in the Amazon. Policy development that prior-
itizes unprotected AGC stocks is warranted to promote conservation of native vege-
tation beyond the legal requirements. However, different biomes and land tenure
structures may require different policy settings considering local and regional speci-
fics. Finally, the fate of current AGC stocks relies upon effective implementation of
command and control mechanisms, considering that unprotected AGC in native veg-
etation on private land only accounts for 6.5% of the total AGC stock.

KEYWORDS
aboveground carbon, Brazil, climate change, ecosystem services, forest Act, land tenure, land-
use governance

1 | INTRODUCTION use and land-use change (LUC)—tropical deforestation, in particular


—account for roughly 10% of annual global GHG emissions (IPCC,
The Paris Agreement creates a framework for nations to mitigate cli- 2014). Brazil was the top CO2-equivalent emitter from forest land
mate change, adapt to its effects, and ratchet up ambitions for from 1990 to 2014 (FAO, 2014), due to land-use change mostly dri-
reductions of anthropogenic greenhouse gas (GHG) emissions (UN, ven by agricultural expansion for food, feed, and bioenergy feed-
2015). Native vegetation stores vast quantities of carbon, and land stock. As it remains the nation with the largest areas of preserved

Glob Change Biol. 2017;1–14. wileyonlinelibrary.com/journal/gcb © 2017 John Wiley & Sons Ltd | 1
2 | FREITAS ET AL.

tropical forest in the world (Zarin et al., 2016), addressing land use vegetation, and freshwater supply (Banks-Leite et al., 2014, Harris
and associated GHG emissions in Brazil is crucial for global policies et al., 2012; Lapola et al., 2014; Nazareno & Laurance, 2015).
and strategies to reduce emissions from tropical deforestation (Han- The revision of the Forest Act in 2012 entailed several changes
sen et al., 2013; Lapola et al., 2014; Nogueira, Yanai, Fonseca, & that resulted in less stringent requirements for restoration of native
Fearnside, 2015; Zarin, 2012; Zarin et al., 2016). vegetation on agricultural land (Metzger et al., 2010; Soares-Filho
Deforestation in the Brazilian Amazon, where more than 80% of et al., 2014; Sparovek, Berndes, Barretto, & Klug, 2012). Several stud-
the Brazilian AGC is stored, decreased by about 80% between 2005 ies of the 2012 Forest Act find negative impacts on the provision of
and 2012 (INPE, 2017) as a result of the expansion of protected valuable ecosystem services (Alarcon et al., 2015; De Sousa, Fernan-
areas, improved surveillance mechanisms, increased market-driven dez Piedade, & Candotti, 2011; Pinto et al., 2014; Soares-Filho et al.,
initiatives to avoid deforestation, and decreased commodities prices 2014; Sparovek et al., 2012), although the 2012 revision introduced
(Assuncß~ao, Gandour, & Rocha, 2015; Assuncß~ao, Gandour, & Rocha, elements intended to reduce illegal deforestation—land clearing
2017; Gibbs et al., 2015; Godar, Gardner, Tizado, & Pacheco, 2014; within protected areas—and make command and control regulations
Lapola et al., 2014; Nepstad et al., 2014). Recently, however, this more efficient (Silva Jaan et al., 2011; Soares-Filho et al., 2014; Spar-
deforestation rate has increased—a 30% increase from 2012 to ovek, Berndes, Klug, & Barretto, 2010; Stickler, Nepstad, Azevedo, &
2016 (INPE, 2017)—and Brazil remains the nation with the highest Mcgrath, 2013). Special mention should be made of one such ele-
annual losses of native tropical forest, and the largest carbon emitter ment, the Rural Environmental Registry (CAR, Portuguese acronym), a
from gross deforestation (Zarin et al., 2016). The pattern of defor- voluntary geo-referenced database of the boundaries, protected
estation in the Brazilian Amazon has changed recently, now occur- areas, and native vegetation occurrence in private rural properties.
ring in small patches, often in remote parts of the Amazon, with The CAR is a comprehensive nationwide geo-referenced information
increased relative participation of small-scale landholders who have source on private rural properties, a widely recognized tool for trans-
not responded to the existing conservation policies (Assuncßa~o, Gan- parency in the monitoring of legal compliance with environmental
dour, Pessoa, & Rocha, 2017; Godar et al., 2014). To reduce defor- legislation (Alarcon et al., 2015; Assuncß~ao, Gandour, Pessoa et al.,
estation, the scientific community has proposed that conservation 2017; Gibbs et al., 2015; Soares-Filho et al., 2014). CAR and other
policies be based on support and incentive mechanisms and tailored existing law enforcement tools—such as the DETER system, which
by the parties involved (Assuncß~ao, Gandour, Pessoa et al., 2017; provides real-time deforestation detection, and the PRODES project,
Azevedo et al., 2017; Godar et al., 2014). which monitors deforestation of the Amazon rainforest at a fine reso-
A challenge in stopping deforestation is the constant need for lution—are expected to make public governance of native vegetation
new areas of agricultural production. With an area of over 250 Mha on private land more effective (Sparovek et al., 2016).
of farmland, Brazil is among the top producers and exporters of In this new scenario of improved law enforcement and increased
major food, feed, and bioenergy crops, such as beef, maize, soybean, accountability for illegal deforestation, spatially explicit assessment
rice, coffee, and sugarcane (FAO, 2014). In response to the growing of the extent of public governance protection is of particular impor-
demand for these products, Brazil is expected to increase agricultural tance to orient public intervention and furnish the formulation of
outputs in the coming years (MAPA, 2016). The Brazilian govern- much needed policies aimed to preserve nature beyond the limits of
ment projects a 30% growth in cereal production up to 2026, to be existing public mechanisms of protection (Assuncß~ao, Gandour, Pes-
achieved through higher yields as well as expansion (7.4 Mha) of soa et al., 2017; Azevedo et al., 2017; Barlow et al., 2016; Godar
agricultural land (MAPA, 2016). The production of bioenergy prod- et al., 2014).
ucts from, for example, sugarcane—currently occupying an area of Despite the intense scientific debate about the 2012 Brazilian
more than 10 Mha—is projected to increase at the same rate Forest Act, only a few studies have provided nationwide quantita-
(MAPA, 2016). This development will increase the pressure on land tive analyses regarding the changes in protection for native vegeta-
and thus lead to further direct and indirect LUC. tion on private land (Soares-Filho et al., 2014; Sparovek et al.,
The Native Vegetation Protection Law (Brasil, 2012), henceforth 2012), focusing primarily on changes in the amount of land required
referred to as the “Brazilian Forest Act”, is the main legal framework to be protected. However, carbon storage in the diverse native veg-
to limit agricultural expansion into native vegetation on private land. etation varies greatly by location. Although previous studies provide
It regulates the two most important mechanisms for the protection rough estimates of carbon stored in native vegetation and soil
of native vegetation on private territory, the permanent preservation (Soares-Filho et al., 2014), they do not capture the spatial variability
areas (PPAs) and legal reserves (LRs). The PPAs consist of riparian of carbon in native vegetation nor the relationship to the various
zones along rivers, streams, and lakes, steep slopes, hilltops, and mechanisms of public governance. Using a geographically explicit
mangroves that must be protected from any disturbance. The LRs land-use model, Sparovek, Barretto, Matsumoto, and Berndes (2015)
are made up of variable shares of private rural properties (20%–80% demonstrated the need to assess the 2012 Forest Act in relation to
depending on the biome and vegetation type) to be set aside for other regulatory frameworks for public (Brasil, 1973, 2000) and pri-
nature protection. LRs are essential for the protection of natural vate lands (Brasil, 2006), highlighting the importance of tenure regu-
habitats and biodiversity, and for the provision of ecosystem ser- larization for the implementation of public protection mechanisms in
vices, for example, global climate regulation by carbon storage in the Amazon.
FREITAS ET AL. | 3

A significant portion of Brazil consists of undesignated land, 2 | MATERIALS AND METHODS


vast areas—primarily in the Amazon—with unclear land-use rights,
where private ownership is not recognized and public conservation The LUGA is a modeling framework developed to perform an inte-
is not institutionalized (Damasceno, Chiavar, & Lopes, 2017). These grated assessment of land-use protection mechanisms in Brazil. The
areas are often subject to conflicts over land possession, and high model was initially presented by Sparovek et al. (2010) and later
rates of illegal deforestation due to the lack of governance (Damas- applied and further developed in several other studies (Freitas, Sparo-
ceno, 2016; Lapola et al., 2014; Mueller, 2016; Reydon, Fernandes, vek, & Matsumoto, 2016; Freitas, Sparovek et al., 2017; Sparovek,
& Telles, 2015; Zarin et al., 2016). The scientific community and Barretto, Klug, Papp, & Lino, 2011; Sparovek et al., 2012, 2015). Here,
civil society have been pushing for tenure regularization as a mean we update the model version presented in Sparovek et al. (2015) to
to stop deforestation (Damasceno, 2016; Lapola et al., 2014; MMA, include a new component for the construction of a nationwide tenure
2017, Sparovek et al., 2015). To accelerate this process, the Brazil- dataset. The updated model is used to analyze the effects and uncer-
ian government has institutionalized the Terra Legal Programme tainty of tenure regularization in relation to major mechanisms of pub-
(Brasil, 2009), an initiative aimed at providing tenure security in the lic protection on private and public land. The role of these mechanisms
Amazon region by titling undesignated land legally appropriated by in the protection of AGC stocks is quantified. The data processing and
farmers, especially small ones, in order to prevent illegal land appro- modeling in this study use the Geographic Information System (GIS)
priation and deforestation, ensuring the preservation of natural software ArcGIS (Esri, 2010) with Python scripting integrated with a
ecosystems (L’roe, Rausch, Munger, & Gibbs, 2016). More recently, MySQL open source database.
the Brazilian government has enacted act 13.465/2017, which facil-
itates the tenure regularization of large untitled properties (Brasil,
2.1 | Integration of secondary tenure datasets
2017). This act is heavily criticized by civil society and by the oppo-
sition in parliament, who argue that the new regulation will enable We constructed a Brazilian land tenure dataset with national cover-
regularization of public land that has been grabbed illegally and may age by integrating publicly available geo-referenced datasets, includ-
promote deforestation in the region. The future of undesignated ing boundaries of indigenous reserves, conservation units, military
land in Brazil is uncertain, and studies providing a quantitative anal- land, rural settlements, and private properties (See Table S1 for a
ysis of the potential impacts of land tenure developments on the detailed list of datasets). Land tenure in the resulting dataset is cate-
preservation of carbon stocks are lacking. The absence of an inte- gorized as public land, private land, or undesignated land.
grated database of land tenure in Brazil presents a challenge. Exist- Among databases of private properties, we highlight the CAR data-
ing datasets are scattered among different institutions and are sets on property boundaries. The CAR datasets are based on farmers’
often inconsistent (Damasceno, 2016; Lapola et al., 2014; Reydon self-declaration of their property boundaries and native vegetation
et al., 2015). stocks. The declarations are made by either submitting existing maps
This study aimed to quantify the scope of different mecha- or manually drawing property boundaries on satellite images. This
nisms of public governance on public and private land regarding approach enables a relatively rapid database construction, but sub-
the protection of aboveground carbon (AGC) stocks. The main goal stantial numbers of conflicting edges indicate discrepancies between
is to assess whether regulatory mechanisms provide sufficient pro- reported and actual property boundaries. Registries that overlap many
tection for carbon stocks in Brazilian native vegetation. For this other properties were also found. We attempted to resolve these
purpose, we construct a national land tenure dataset by integrating issues by manually removing (i) polygons with boundary patterns that
existing tenure databases and performing complementary land are unlikely to represent real properties (such as triangles or circles)
tenure simulations. An updated version of the geographically expli- and (ii) polygons overlapping many smaller ones. These measures
cit land-use model presented by Sparovek et al. (2015), here called resulted in the removal of 1,404 polygons. Conflicting edges were
the land-use governance assessment (LUGA) model, was employed resolved by removing overlapping regions from the polygons intersect-
to identify protected areas and evaluate the effects of regularizing ing other polygons. In total, these procedures reduced the total area
the tenure of undesignated land on the implementation of major included in CAR datasets by roughly 70 Mha, half of it due to the
command and control mechanisms. Moreover, we combine the removal of inconsistent polygons (See Figure S3). This approach is not
geographically explicit information on land tenure and protected likely to influence the final results significantly, because possible errors
areas with high-resolution maps of AGC stocks (Englund et al., resulting from the removal of polygon intersections are mostly related
2017). Finally, we discuss the effectiveness of policy measures and to shifting boundaries of rural properties but have little effect on the
related sustainability implications. The outcomes of this study can extent of protected areas on private territory (model output). Even
elucidate the relationship between Brazilian land-use governance though changing the boundaries of a given rural property may
mechanisms and the protection of carbon stocks as well as identify decrease or increase the protected area, this change will be offset in
the distribution of unprotected AGC stocks among different rural the neighboring rural property. Other datasets of private properties
property sizes. This is strategic information for developing policies from the databases of INCRA (see Table S1)—based on geospatial field
to provide incentives for the preservation of native vegetation with data collection using centimeter-precision GPS receivers, following
high AGC. standard procedures to avoid overlapping with other rural properties
4 | FREITAS ET AL.

(INCRA, 2015)—are expected to have higher geographic precision, simulated private properties considered information about common
and thus, datasets from INCRA took precedence over datasets from boundaries of rural properties, that is, watercourses, roads, and
the CAR database. known boundaries to other properties. The algorithm used for the
On public land, there is overlap between datasets of conserva- geographic allocation was formulated to place properties in polygons
tion units and those of indigenous reserves. This overlap can result within the municipality that best fit the size of the rural property. If
from a lack of coordination among government agencies responsible the polygon was larger than the property-size range, we allocated
for demarcation of indigenous reserves and conservation units in several rural properties to the same polygon by subdividing the poly-
cases of conflict between agencies over the administration of the gon using a Voronoi diagram. The algorithm ensured that the distri-
land (Ricardo, 2004). However, it is also possible that the involved bution of properties per range size was as similar as possible to that
agencies share management of these overlapping zones. Although in the agricultural census database.
overlapping areas are traceable in the final land tenure dataset, such The allocation of small, medium, and large rural properties with
overlapping areas were here classified as indigenous reserves, unknown boundaries within the municipalities enabled the identifica-
because Brazilian legislation recognizes the sovereignty of indige- tion of the portion of the territory dominated by each farmer profile,
nous people within the national territory (Brasil, 1973). Therefore, but it created some uncertainty regarding the total area under LR
the total area of conservation units presented here will be smaller protection. Farm-size categorization is based on the fiscal module
(6.5 Mha) than in the reference dataset. unit, the official land unit used by the Brazilian Government for taxa-
Other conflicting zones include intersections between public land tion and land governance purposes. The fiscal module varies from 5
(indigenous reserves, conservation units, and military land) and pri- to 110 hectares, depending on the region (Figure S1) and the
vate property, due to disputes between farmers and indigenous peo- expected economic return on the predominant agricultural activities
ple or environmental agencies (Paixao et al., 2015). Such overlapping in the municipality.
areas were here classified as public land due to uncertainties regard- The Forest Act of 2012 exempted small-scale farmers (area smaller
ing the CAR registries, as described earlier, and existing legal barriers than four fiscal modules) from LR restoration but not medium- and
to modifying public land boundaries. This assumption was evaluated large-scale farmers. Therefore, if small private properties are allocated
in the sensitivity analysis. Rivers and other water bodies, roads, rail- to the part of the municipality territory most affected by humans
ways, and urban areas were grouped to a single layer. The final data- (where farmers often converted the native vegetation beyond the legal
set of undesignated land was composed of untitled land from the limits), the extent of native vegetation protected under LR will be less
Terra Legal Programme and public land without any designation by than if small properties are allocated to the part of the municipality
federal or state governments. The integration of all available datasets where most of the native vegetation is preserved. Attempting to control
related to land tenure resulted in the coverage of about 670 Mha or this uncertainty, we randomized the spatial allocation of simulated pri-
79% of the Brazilian territory. vate properties so that small and medium-to-large properties were allo-
cated to different regions within the municipality in each repetition.
This procedure was repeated ten times, generating, therefore, ten dif-
2.2 | Complementary land tenure simulation
ferent datasets of simulated boundaries of private properties. For each
For the remaining 21%—representing one-third of the private land— municipality, we adopted the simulated dataset in which the final out-
the land tenure was simulated. We assumed that this land is made put values for protected AGC stocks (the sum of all mechanisms of pub-
up of private rural properties not yet registered in INCRA or CAR lic protection) were closest to the average value of the ten repetitions.
databases. Using the land tenure dataset compilation and agricultural The land tenure dataset produced here was based on the compi-
census data (IBGE, 2006), we estimated the number and the size of lation of existing datasets and tabular data from the agricultural cen-
rural properties at a municipal level and used this information to cre- sus of 2006. For validation purposes, we compared the results of
ate simulated properties on land with unknown land tenure. this analysis with the tabular CAFIR database—the Rural Properties
The agricultural census database does not cover all rural private Registry—which is a mandatory registry of rural properties managed
property; the surveyed private property corresponds to about by the tax agency of Brazil. We compared the number of rural prop-
330 Mha of an estimated total area of about 510 Mha (i.e., the sum erties aggregated by size range and state as this aggregated informa-
of the already known and mapped area of private property and the tion was the only data from the CAFIR database publicly available
unknown territory used for the land tenure simulation). We assumed (Table S2). Additional methodological details can be found in Freitas,
that the unsurveyed territory had a similar distribution of rural prop- Guidotti, and Sparovek (2017). The final tenure dataset is publicly
erties per size range as the surveyed territory. Based on this assump- available for download (www.imaflora.org/atlasagropecuario).
tion, we estimated the universe of private properties per size range
in each municipality. We obtained the number of properties to be
2.3 | Simulation of tenure regularization of
simulated by subtracting, individually for each size range, the number
undesignated land
of private properties within municipalities that have known bound-
aries (based on the compilation of existing datasets), from the total The tenure regularization of undesignated land is an ongoing process
number of private properties. The geographic allocation of the that is crucial to tackling deforestation (MMA, 2017). However, the
FREITAS ET AL. | 5

extent and pace of the regularization processes for undesignated located close to land that is already protected. Such areas were iden-
land are very uncertain. For modeling purposes, we considered tified by determining neighborhood occurrence of legally protected
undesignated land assignable to either public conservation—in the areas or preserved native vegetation, using a search window of
form of indigenous reserves, conservation units, or military use—or 15 9 15 cells.
private use—through the creation of rural settlements or private We opted to model the allocation of LRs on the rural private
properties, as in the Terra Legal Programme. properties despite the existence of geo-referenced data on LRs for
We based our analysis on the assumption that the designation properties in the CAR database because the land-use map of native
correlates with the level of human influence on the native vegeta- vegetation does not necessarily match the LR shapefiles due to
tion. The outcomes of the regularization process were therefore set imprecisions in both the CAR database and the land-use map. This
based on the native vegetation coverage. Polygons with undesig- mismatch between LRs and native vegetation datasets could result
nated land were classified as private land if native vegetation cover- in an overestimation of LR deficits and unprotected native vegeta-
age was less than 95% and as public conservation land if coverage tion on private land. Additionally, other data sources of boundaries
was above 95%, which is the same threshold value used by Sparovek of private properties (from INCRA databases in Table S1) do not
et al. (2015). specify the LRs, which have to be simulated and allocated. For con-
sistency, we used a single methodological approach on all private
properties.
2.4 | Modeling the implementation of public
protection mechanism on private land
2.6 | Analyzing aboveground carbon stocks in
Implementation of the environmental legislation on private land was
relation to mechanisms of public protection
modeled using the mapping of land tenure described above and a
land use/land cover (LULC) map with 50 m spatial resolution (Sparo- To quantify the amount of carbon stored in vegetation depending
vek et al., 2015). This land-use map best represents the LULC in on land tenure, we used a newly produced aboveground carbon
2008, the expiration date for amnesty for illegal deforestation. Land compilation map for the entire Brazilian territory with a 50 m
was assigned so as to comply with the prevailing land-use legislation pixel resolution (Englund et al., 2017), created based on data from
by following the steps proposed by Sparovek et al. (2015) and existing carbon maps (Baccini et al., 2012; MCTI, 2016, Nogueira
updated by Freitas et al. (2016). In short, the total PPA was esti- et al., 2015; Saatchi, Houghton, Dos Santos Alvala, Soares, & Yu,
mated employing datasets on watercourses and surface water from 2007; Saatchi et al., 2011) and an up-to-date LULC map (Sparovek
the Brazilian Institute of Geography and Statistics (IBGE, 2015) to et al., 2015). This map represents the current LULC well and does
identify PPAs in riparian zones. On steep slopes, we employed a dig- not generally overestimate carbon levels on land with low carbon
ital elevation model (Farr et al., 2007). The LRs were quantified stocks, such as agricultural land, thus enabling farm-level
applying the rules described in Articles 12, 13, 15, and 67 of the assessment. Other carbon pools than aboveground biomass, that
Forest Act (Freitas et al., 2016). The layers containing water sur- is, belowground biomass and soil organic matter, were not
faces, urban areas, areas of inconsistent data, and infrastructure considered.
were masked out of the processing universe, leaving a total pro-
cessed area of 808 Mha. Conservation units, indigenous land, and
2.7 | Sensitivity analysis
land reserved for military purposes were included in land for public
conservation (Figure S2). We performed a global sensitivity analysis considering three major
sources of uncertainty in regularizing tenure and implementing the
mechanisms of public protection for native vegetation:
2.5 | Allocation of legal reserves
The Forest Act determines that the allocation of a LR within a rural 1. Zone conflicts: Public conservation or private land? We assumed
private property should consider several aspects, such as ecosystem that public conservation land always prevails if there is a zone
services conservation, the formation of ecological corridors, and con- conflict with private land. While this might be the most likely
nectivity to conservation units and PPAs (Brasil, 2012). While such scenario, there are several ongoing legal disputes in which land-
conservation factors are relevant, landholders, who ultimately sug- holders claim the right to land inside public conservation areas.
gest where on their land the LR should be located, may prioritize Additionally, several bills currently under congressional assess-
economic aspects and are likely to make decisions that ensure the ment aim to revoke or reduce indigenous reserves and conserva-
best financial returns. We based the LR allocation on the assumption tion units. Furthermore, when rural properties overlap with
that LRs will be allocated to those areas that are the least suitable conservation units, the Forest Act allows landholders to credit
for agricultural production. This land was identified using an agricul- these overlapping zones as LRs, thereby reducing the total area
tural suitability map produced by Barretto (2013), using soil, slope, protected as LR. To capture these possible variations in the sen-
and climate parameters. In areas where agricultural suitability is sitivity analysis, we let private land prevail when in conflict with
homogenous throughout the landscape, we assumed that LR will be public conservation land.
6 | FREITAS ET AL.

T A B L E 1 Summary of LUGA model design, including input data, data processing, and model outputs

• Indigenous territory datasets • Combination of secondary tenure data- • Geoexplicit tenure dataset
• Conservation units’ datasets sets • Area and AGC stock under each
Input • Boundaries of private rural Data • Complementary tenure simulation Output mechanism of public protection
data properties (CAR and INCRA processing • Simulation of tenure regularization of data (permanent protected areas,
datasets) undesignated land legal reserves, conservation
• Public forests datasets • Combination of tenure dataset with units, indigenous land, military
• Rural settlements 2008 land-use map of native vegetation land, Atlantic Forest Law)
• Military land • Simulation of the implementation of the • Geoexplicit information on
• Water courses and surfaces revised Forest Act 2012 unprotected native vegetation
• 2008 land-use map • Allocation of Legal Reserves and associated AGC stocks
• Agricultural census data of • Analysis of AGC map in relation to • Distribution of area and AGC
2006 mechanism of public protection and stocks among different profiles
• Federal and state roads and tenure information of rural private owners
railways datasets
• Urban areas dataset
• Digital elevation model
• Agricultural suitability dataset
Key parameters and assumptions Sensitivity analysis

• When combining secondary tenure datasets, we • What if private land instead prevails over public conservation. When boundaries conflict?
assumed that public conservation land prevails over
private land if boundaries conflict.
• To simulate regularization of tenure on undesignated • What if we instead let the threshold range from 85% to 99%?
land, we assign such land to public conservation when
native vegetation covers more than 95% of the poly-
gon
• Of 10 sets of complementary simulated tenure data, • What if we instead select the sets of data with the highest, or the lowest, protected AGC
we selected the one with protected AGC stock values stocks for each municipality?
closest to the average value for the 10 sets.

2. Regularization process for undesignated land: We based the assign- Protected land represents about half the Brazilian territory (418 Mha
ment of undesignated land to public or private use on the share in total), mostly located in the Amazon region (285 Mha)—home to
of native vegetation within polygons, considering a 95%-thresh- most of the public conservation land in Brazil (Figure 3, Figure 1c,
old value. This value is uncertain as other factors not addressed Table S4). Protected land holds about three-quarters (38 Gt) of the
in this study help decide the fate of the undesignated lands, for AGC stock in Brazil. The regularization of undesignated land adds
example, political priorities at federal, state, and municipal levels. 80 Mha of protected land with nearly 10 Gt of AGC stored in native
We tested how sensitive the model is to variations in this thresh- vegetation. Public governance protection would thus cover about
old value by varying it from 85% to 99% native vegetation. 60% of the total territory, on which more than 90% of the total
3. Simulation of land tenure: As described earlier, we adopted a data- AGC stock would be stored.
set of simulated boundaries for private properties, namely that About half of the AGC is on public conservation land, roughly
dataset for which the final output value for protected AGC equally distributed on indigenous reserves and conservation units—
stocks was close to the average value of 10 repetitions. To test full protection + sustainable use—(Figure 1b). About one percent of
the robustness of results based on these simulated boundaries, the AGC is found on military land. Expansion of public conservation
we reran the model using the repetitions that resulted in the on undesignated land is expected to cover an additional 70 Mha of
lowest and highest protection of AGC stocks. A summary of the land and 17% (8.9 Gt) of the total AGC stock. About 62% of Brazil-
LUGA model design is presented in Table 1. ian land is privately owned, holding about 30% (15.8 Gt) of the total
AGC. Tenure regularization may expand private land to 65% of the
territory and increase the relative share of AGC stock to 32.5%.
3 | RESULTS
Almost three-quarters of this carbon are protected in the form of
LRs (21.5%), permanent preservation areas (1.7%), or as native vege-
3.1 | Public protection of AGC stocks
tation protected by the Atlantic Forest Law (1.2%).
Figure 1 shows the aggregated land areas (a) and AGC stocks (b) in
relative terms, by land tenure and land use, for the different mecha-
3.2 | Unprotected AGC stocks
nisms of public protection on private and public land. The modeled
land tenure is shown on the map in Figure 1c. The figure also pre- Private land with no legal obligations comprises ca. 324 Mha (40% of
sents the relative distribution of land and associated AGC among pri- the processed territory). This is mainly used for agriculture and holds
vate property sizes (small, medium, and large) and LULC type. a relatively small share (8.5%) of the AGC stock. We estimate that the
FREITAS ET AL. | 7

(a) 13.0% Land tenure categories


34.9% Protected territory in public land
PC → Public conservation
IL
13.9% CUPI IL → Indigenous land
Large
20.0% 10.3% CUPI → Conservation unit of full protection
7.1%
CUSU CUSU → Conservation unit of sustainable use
2.0% M A → M ilitary land
MA UL to PC → Undesignated land assigned to public conservation
PRnoOB 0.4%
40.0% PC
5.5%
27% Protected territory in private land
Medium 808 UL to PC
7.9% 8.7% LR → Legal reserve
2.4%
PR Mha UL
PPA → Permanent protected areas
62% 11% UL to PR
2.6% AF → Atlantic Forest Law
PPA
Small AF 3.3%
12.1%
Private land
9.0% 1.0%
UL to PR → Undesignated land assigned to private use
LR PRnoOB→ Private land with no legal obligation
Small
20.3% 0.4% Small → Smallholder farmers (<4 fiscal modules)
7.1%
3.1% Medium 6.5% M edium → M edium farmers (4 – 15 fiscal modules)
Large
4.6% Large → Large farmers (>15 fiscal modules)
13.0%
3.7% 0.6%
1.4% 11.5%
0.9% Land use
0.4% 0.9% Agriculture land
(b) 1.1%
3.7% Native vegetation
0.7%
1.7%
0.1% Large 4.6%
Medium 1.5%
12.7% (c)
Small 2.4%
Large
12.8%
PRnoOB IL
8.5% 25.5%
LR
21.5% PR PC
3.5% Medium 50%
30%
3.5% 52
AF
1.2% Gt CUPI
UL 19.6%
Small
20%
8.0%
8.0%
CUSU
PPA 4.3%
1.7% MA
0.5%
UL to PR
2.5% UL to PC
17.1%

67.1%

F I G U R E 1 Distribution of area (a) and AGC stock (b), in relative terms, by land tenure and land use, and the geographic location (c) of
selected land tenure categories. The distribution of private land among small rural properties (0–4 fiscal modules), medium rural properties (4–
15 fiscal modules), and large rural properties (more than 15 fiscal modules) is also specified

biomass of about 223 Mha of crops stores approximately 2% (~1 Gt) The Amazon biome, the largest in Brazil, sustains a relatively
of the total AGC stock in Brazil. Native vegetation covers about small share of the unprotected native vegetation (~9 Mha). However,
101 Mha of private land with no legal obligations; this is unprotected this carbon-dense vegetation stores almost one-third of the unpro-
land that can legally be converted to agriculture. This vegetation holds tected AGC (~1 Gt). The Cerrado biome, holding more than 40%
about 6.5% (~3.4 Gt) of the Brazilian AGC stock (Figure 1b). (~1.4 Gt) of total unprotected AGC, is the largest unprotected AGC
8 | FREITAS ET AL.

No of rural properties (% of 6.5 Million)


Area (% of 522 Mha)
AGC stock (% of 15.8 Gt)

78%
Small Medium Large

50%
50%
23%

20%

16%
16%

10%
9%

6%
6%

6%

5%
5%

2%
Farm-size range
<1 1–2 2-4 4 – 15 >15
(Fiscal modules)

No of properties
per data source

Area of private land


per data source

INCRA databases CAR database Simulated properties

F I G U R E 2 Number of private rural properties, associated area, and AGC stock by farm size. The figure also presents the relative
importance of each data source used to subdivide private land into different farm sizes

pool among the Brazilian biomes (Figure 3, Table S5). The unpro- registration. For these, tenure simulation was especially important. In
tected AGC stock in native Cerrado vegetation is slightly greater contrast, existing datasets cover most of the medium and large-scale
than the unprotected AGC in the Amazon biome, but it is distributed properties (Figure 2). The final constructed dataset enabled spatially
over a much larger area of unprotected native vegetation (44 Mha, explicit identification of land ownership for the entire territory of
Figure 3). In Caatinga, about 35 Mha of native vegetation are unpro- Brazil. Aggregated numbers of individual private properties closely
tected, corresponding to a third of the unprotected native vegeta- match existing tabular databases, with a few exceptions such as in
tion, where an estimated 20% (~0.7 Gt) of AGC is stored (Figure 3, Bahia state, one of the largest Brazilian states, where the number of
Table S8). Together, the two smallest Brazilian biomes, Pantanal and small rural properties is half that in the existing tabular databases
Pampa, represent less than 12% of the unprotected native vegeta- (see Table S2).
tion and about 9% of the unprotected AGC stock (Figure 3, Two percent of the landholders own half of this territory and
Table S7, Table S9). half of the AGC stock on private land. At the other end, roughly
one-third of the private territory and AGC stock is owned by small-
holder farmers who represent 93% of the landholders in Brazil (Fig-
3.3 | Ownership of AGC stock on private land
ure 2).
Small rural properties are underrepresented in existing geographic On the national level, the AGC stock in unprotected native
databases due to the difficulties and costs associated with vegetation is roughly equally distributed between small-to-medium
FREITAS ET AL. | 9

4.5
AGC stocks in Unprotected NV (% of 3.4 Gt) Unprotected NV (% of 102 Mha) Landholders (% of 1.8 Million)
3.5
Large 0.4% 4.8% 14.5% 1.6% 26.0% 25.3% 0.1% 7.4% 7.3% 0.7% 15.9% 9.6% 0.3% 2.4% 0.5%
Private land

2.5
30% Medium 0.7% 1.6% 3.9% 3.1% 9.3% 8.7% 0.1% 0.6% 0.6% 1.8% 6.4% 3.7% 0.5% 1.0% 0.2%
1.5
Small 6.1% 2.9% 10.0% 29.9% 8.4% 8.6% 0.4% 0.2% 0.2% 50.5% 12.5% 6.7% 3.9% 0.8% 0.2%
20% 0.5

10%

0%
Amazon Cerrado Atlantic Forest Pantanal Caatinga Pampa

10%
6%
20%
Public conservation

4%
30%

2%
40%

0%
50%
Cerrado Atlantic Forest Pantanal Caatinga Pampa
–2%
60%
Protected NV (% 808 Mha)
70% AGC stock in protected NV (% of 52 Gt)
Unprotected NV (% of 808 Mha)
AGC stock in unprotected NV (% of 52 Gt)
80%

F I G U R E 3 Distribution of protected and unprotected land covered with native vegetation (NV) and the associated aboveground carbon
(AGC) stocks for each biome. The percentage of protected and unprotected NV represents the share of the total processed territory
(808 Mha) and associated AGC stocks (52 Gt). The red error bars represent the possible variation in our results considering the extreme values
from the sensitivity analysis. In the top of the figure, the colored circles represent the shares of landholders owning unprotected native
vegetation and the corresponding shares of the total area and of the AGC stock. As discussed earlier, all Atlantic Forest is protected by the
Atlantic Forest Law

rural properties and large ones (Figure 2). However, the distribu- unprotected AGC, while the 600,000 small-to-medium properties in
tion can vary considerably among different biomes (Figure 3, the Cerrado hold 17% (~0.6 Gt).
Tables S4–S9). Here, we highlight the Amazon, Cerrado, and Caa- The Caatinga biome is largely dominated by very small rural
tinga biomes, where most of the unprotected AGC stock is properties, compared with the other biomes. There are almost one
stored. million small-scale landowners in Caatinga that hold unprotected
In the Amazon biome, small-to-medium properties hold as AGC on their land, constituting half of all landowners in Brazil hold-
much unprotected AGC as the large ones. We estimate that the ing unprotected AGC (Figure 3).
7,000 large properties in the Amazon region hold almost 15%
(~0.5 Gt) of the national unprotected AGC, while the
3.4 | Sensitivity analysis
110,000 small properties in the same biome hold about 10%
(~0.34 Gt). Figure 3 shows the possible variation in the extent of public pro-
Large private properties dominate the Cerrado biome, where tection of native vegetation and associated AGC stocks based on
about 30,000 large properties hold a quarter of the national the sensitivity analysis. The full results from the sensitivity analysis
10 | FREITAS ET AL.

are presented aggregated at the national and biome level in


4.2 | Conservation policies for unprotected AGC
Table S3 thru Table S9. In general terms, the extent of public pro-
tection of AGC stocks—estimated to be over 90%—is robust and Greater compliance with the Forest Act can be expected among
not sensitive to uncertainties related to tenure simulation or tenure farmers, after tenure consolidation and full implementation of CAR
regularization, as whether assigned to public conservation (fully registration accompanied by improved satellite-based monitoring of
protected), or to private land (80% protected as LR in the Amazon deforestation in all Brazilian biomes (Sparovek et al., 2016). Full
region, where undesignated land is mostly located), the majority of compliance may still depend on further economic incentives and
the undesignated land will be protected after the tenure regulariza- market barriers (Azevedo et al., 2017). Also by considering a scenario
tion. However, the sensitivity tests reveal that the estimated of perfect compliance with the Forest Act, GHG emissions from LUC
unprotected AGC in the Amazon biome (1 Gt) may vary by a factor may remain high as a relevant share of the total AGC is stored in
of two, depending on the outcomes of tenure regularization. The unprotected vegetation on private land. Much of this vegetation is in
unprotected AGC stocks in the other biomes were not sensitive to remote areas or on land that is not suitable for agricultural explo-
tenure simulation or tenure regularization. At the national level, in ration, but this may change in the long run with infrastructure and
a worst-case scenario—considering the boundaries of our sensitiv- technological development. Policy mechanisms that have the poten-
ity tests—instead of 101 Mha storing 3.4 Gt, the unprotected tial to conserve the unprotected native vegetation on private land,
native vegetation could be 110 Mha, storing roughly 4.5 Gt of such as the Environmental Reserve Quotas—a mechanism that
AGC. enables the offsetting of legal deficits of native vegetation among
landholders (Bernasconi, Blumentrath, Barton, Rusch, & Romeiro,
2016; Freitas, Sparovek et al., 2017)—or Payment for Environmental
4 | DISCUSSION Services (PES) schemes, are struggling to be legally regulated, and
their scale and effectiveness are still uncertain (Bernasconi et al.,
The LUGA model presented in this paper enables spatially explicit 2016; Freitas, Sparovek et al., 2017).
information on land tenure and the effects of the current public pro- Our analysis indicates that the largest portion of unprotected
tection mechanisms (Table 1), allowing a pixel-level analysis of the AGC is stored in the Cerrado vegetation, where the mechanisms
ownership of AGC stocks in Brazil. The outcomes provide a detailed accountable for the deforestation slowdown in the Brazilian Amazon
view of the impacts of public governance on AGC stocks. (Assuncß~ao et al., 2015; Gibbs et al., 2015; Nepstad et al., 2014) are
not yet in place. This is a biodiversity hotspot where most of the
recent agricultural expansion into native vegetation has taken place,
4.1 | Tenure simulation and undesignated land
supported by infrastructure development and government incentives
The area classified as undesignated land was considerably smaller (Gibbs et al., 2015; Strassburg, Latawiec, & Balmford, 2016). Civil
than in previous estimates (Freitas et al., 2016; Sparovek et al., society and the scientific community have been pushing for an
2015), mainly due to the inclusion of the CAR dataset. However, expansion of market barriers to the Cerrado biome, referring to the
there are uncertainties associated with the use of a dataset based positive outcome in the Amazon biome (Azevedo et al., 2017; Gibbs
on farmers’ self-declarations. In addition to what has already been et al., 2015; Nepstad et al., 2014). Such an intervention may be
mentioned, there are concerns that the CAR registry may have been effective in reducing land-use-related emissions as ownership of
used for illegal appropriation of public land or other lands under dis- AGC stock in this biome is dominated by large-scale landholders,
pute among farmers and indigenous or traditional communities (Ciro who are particularly sensitive to trade restrictions and strongly influ-
Barros, 2016; L’roe et al., 2016). Therefore, there could be cases of enced by international markets (Assuncß~ao, Gandour, Pessoa et al.,
forged, or otherwise incorrect, CAR registries that have been used 2017; Godar et al., 2014). On the other hand, a third of the Brazilian
to determine land tenure for previously undesignated territory. Sup- AGC and a quarter of the unprotected AGC on private land are
porting this hypothesis, a relevant share (8.5 Mha, about 6% of the owned by smallholder farmers. These actors are less responsive to
private land in the Amazon biome) of the CAR registries in the Ama- the existing mechanisms for preventing deforestation and are
zon biome have more than 95% of the territory covered by native responsible for an increasing share of deforestation in the Amazon
vegetation, indicating a very low level of human disturbance in these (Assuncß~ao, Gandour, Pessoa et al., 2017; Godar et al., 2014;
properties. Public authorities are expected to verify and validate the Richards & Vanwey, 2015). Small-scale landholders dominate the
CAR registry, but the pace of this validation process is highly uncer- Caatinga biome. Although the carbon density of Caatinga vegetation
tain and may take several years. is lower than in, for example, the Amazon and Cerrado biomes, AGC
Despite the progress on designating undesignated land, this terri- stored in unprotected native vegetation is not negligible. Moreover,
tory stores over 20% of the Brazilian AGC stock. The tenure regular- the Caatinga biome hosts a large number of endemic and endan-
ization of this territory is a crucial step for the enforcement of public gered species (Lewinsohn, 2006). Recent infrastructure development
protection mechanisms and the means for further reductions in within the Caatinga—especially the inter-basin water transfer of the
emissions from deforestation in Brazil (Damasceno, 2016; Mueller, S~ao Francisco River—may bring changes, boost rural development,
2016; Zarin et al., 2016). and enable, for example, irrigated agriculture. While this could
FREITAS ET AL. | 11

contribute positively to local and regional economies, it may result in development of solutions that are specific to the Amazon region.
the displacement of native vegetation, with associated impacts on This development, however, is challenged by the lack of local human
important ecosystem services and biodiversity (Santos, Leal, capacities, incentives, and enabling regulatory frameworks (Nobre
Almeida-Cortez, Fernandes, & Tabarelli, 2011). Given the large share et al., 2016). The avoid and mitigation strategy depend on a novel
of unprotected native vegetation in this biome (Figure 3, Table S8), set of actions, as described here, because it is not likely that the cur-
such impacts could be severe and therefore deserve further investi- rent and follow-up government and parliament representatives will
gation. To date, land-use-related impacts of this infrastructure devel- repeat the strategies that reduced deforestation and increased envi-
opment have received little attention from the scientific community. ronmental protection from the past, by creating conservation units
and increasing surveillance. These measures depended on high public
investments and intervention strategy that in recent times are being
4.3 | AGC under public conservation
undermined by weakening environmental regulations on farmland
The demarcation of indigenous land and conservation units that has and reducing public investments in the social and environmental are-
occurred in the last decades is one of the most important elements nas.
of the Brazilian climate agenda, considering the associated carbon In this study, we focused on the Brazilian case; however,
stocks and the high effectiveness of this type of protection mecha- strengthening land-use governance in Brazil can result in negative
nism (Barber, Cochrane, Souza, & Laurance, 2014; Nepstad et al., LUC dynamics elsewhere (Aguiar et al., 2016), including other coun-
2006, 2014; Sparovek et al., 2010). tries in Latin America, Asia, or Africa, where emissions from defor-
Concerns regarding low environmental and social priorities have estation are still increasing and public governance of native
been raised in recent years (Fearnside, 2016; Ferreira et al., 2014) vegetation is weaker, thus offsetting the efforts made in Brazil to
and have intensified because of the recent political turbulence in prevent land-use-related emission (Zarin et al., 2016).
Brazil. Several bills have been formulated and moved forward to Our analysis did not include all carbon pools, such as below-
weaken environmental protection and slow the demarcation of ground biomass, and soil carbon. This results in a lower relative “im-
indigenous land, or even reduce or repeal existing conservation units portance” of regions with high carbon concentrations in the soil, for
and indigenous reserves (Fearnside, 2016). The purpose of initiatives example, areas in the Pantanal biome, where native aboveground
to weaken protection is to enable large infrastructure investments vegetation stores small amounts of carbon (Table S7), but large areas
and appropriation of land for agriculture production, mining, and of wetland and peatland store large amounts of belowground carbon
hydroelectric dams (Ferreira et al., 2014). Such political decisions can for which our analysis does not account. Drainage of these areas
result in a reduction of protected native vegetation and associated could trigger substantial carbon dioxide emissions (Joosten, 2009).
carbon stocks, especially in the Amazon. Furthermore, while results at the national level are robust to the
Conservation units and undesignated land host thousands of major sources of uncertainty related to data and modeling assump-
communities, which—despite improvements over the last decades— tions, several other sources of uncertainty were not considered in
are often subject to poverty and related social issues (Guedes et al., the sensitivity analysis, for example, uncertainties associated with
2012; Ioris, 2016). These socioeconomic challenges undermine the the LULC dataset and the AGC dataset. Moreover, ongoing negotia-
capacity of these communities to sustain their livelihoods based on tions in the parliament may affect the outcomes presented here. Bills
sustainable land-use practices. and provisional measures may revoke conservation units and indige-
There are reports of remunerations offered to communities in pro- nous land, while facilitating the titling of illegally appropriated land.
tected areas in exchange for illegal selective logging (Rist et al., 2012). Finally, if we detach the carbon agenda from other agendas
This can degrade the forest and lead to significant impacts on carbon related to the protection of native vegetation, for example, biodiver-
stocks and biodiversity (Asner et al., 2005; Barlow et al., 2016), a sity, water safety, and the scenic beauty of landscapes, the total
social dynamic that may escalate in a national scenario of economic amount of AGC is of higher importance than its location. Given the
crisis and political instability, expected to endure for several years. relatively small amounts of AGC stored in native vegetation on
Economic and social development in poor communities is a pre- unprotected private land (6.5% of total), conversion of such land to
requisite for the preservation of carbon stocks in the Amazon region agricultural land can be considered of minor importance from a car-
(Aguiar et al., 2016). The role of these communities in protecting the bon perspective, but it may bring severe impacts to other important
forest should be better acknowledged in public and private gover- ecosystem services and biodiversity.
nance. Schemes of payment for ecosystem services may be an Depending on the degree of impacts on biodiversity and other
important mechanism for financing traditional communities in the ecosystem services, agricultural expansion on highly productive sites
short term. Unfortunately, the proliferation of such schemes has covered by native vegetation may be considered acceptable if this
been very slow. Nobre et al. (2016) argued that new ways of think- development brings positive economic and social returns. Unpro-
ing about sustainable development of the Amazon are needed, tected native vegetation on farmland under surveillance, or of major
focusing on developing and scaling up innovative high-value prod- concern, could be reduced if only considering land with high conser-
ucts and services based on a sustainable use of the Amazon rainfor- vation value. This would also allow efforts to be focused where they
est. Achieving this may require public-private partnerships for the are most needed. At the same time, high priority areas might shift in
12 | FREITAS ET AL.

the future due to climate change, which could motivate the protec- the Brazilian Amazon. Global Change Biology, 22, 1821–1840.
tion of additional areas as a safety net for adaptive management. https://doi.org/10.1111/gcb.13134
Alarcon, G. G., Ayanu, Y., Fantini, A. C., Farley, J., Filho, A. S., & Koellner,
Regarding carbon, major immediate efforts should be focused on
T. (2015). Weakening the Brazilian legislation for forest conservation
legal protection, more specifically on (i) ensuring full regularization of has severe impacts for ecosystem services in the Atlantic Southern
undesignated land and strong enforcement of the Forest Act; (ii) Forest. Land Use Policy, 47, 1–11. https://doi.org/10.1016/j.landuse
reverting the current increase in illegal deforestation on conservation pol.2015.03.011
Asner, G. P., Knapp, D. E., Broadbent, E. N., Oliveira, P. J. C., Keller, M.,
units and undesignated land; and, most importantly, (iii) keeping or
& Silva, J. N. (2005). Selective logging in the Brazilian Amazon.
expanding the current area of conservation units and indigenous Science, 310, 480–482. https://doi.org/10.1126/science.1118051
reserves, which have both recently been challenged by government Assuncß~ao, J., Gandour, C., Pessoa, P., & Rocha, R. (2017). Property-level
actions. All these are related to the government and the parliament assessment of change in forest clearing patterns: The need for tailor-
ing policy in the Amazon. Land Use Policy, 66, 18–27. https://doi.org/
in Brazil, which are both affected by political and economic chal-
10.1016/j.landusepol.2017.04.022
lenges. Assuncß~ao, J., Gandour, C., & Rocha, R. (2015). Deforestation slowdown
Because of the earlier effectiveness of parliament—and govern- in the Brazilian Amazon: Prices or policies? Environment and Develop-
ment—driven reduction of deforestation in the Amazon (Nepstad ment Economics, 20, 697–722. https://doi.org/10.1017/S1355770X
15000078
et al., 2014), revision of the Forest Act, and establishment of vast
Assuncß~ao, J., Gandour, C., & Rocha, R. (2017). DETERring deforestation
areas of conservation units, the environmental agenda shifted to the
in the Amazon: environmental monitoring and law enforcement. In:
private agricultural sector (e.g., the soy moratorium, sectorial commit- Nucleo de Avaliacß~ ao de Polıticas Climaticas. (ed Initiative CP),
ments on Zero Deforestation). As our analysis shows, the focus should PUC-RIO.
immediately return to the political landscape. This article can hopefully Azevedo, A. A., Raj~ao, R., Costa, M. A., Stabile, M. C., Macedo, M. N., dos
Reis, T. N., . . . Pacheco, R. (2017). Limits of Brazil’s forest code as a
contribute to the development of new policies that aim at maximizing
means to end illegal deforestation. Proceedings of the National Acad-
the effects of land preservation on AGC stocks—such as targeted pro- emy of Sciences, 114, 7653–7658. https://doi.org/10.1073/pnas.
tection—and guide PES investments to where they provide the best 1604768114
outcomes. The modeling framework presented here can be further Baccini, A. G. S. J., Goetz, S. J., Walker, W. S., Laporte, N. T., Sun, M.,
Sulla-Menashe, D., . . . Samanta, S. (2012). Estimated carbon dioxide
developed to consider additional carbon pools, as well as other ecosys-
emissions from tropical deforestation improved by carbon-density
tem services, such as freshwater supply and maintenance of soil fertil- maps. Nature Climate Change, 2, 182–185. https://doi.org/10.1038/
ity, and biodiversity conservation. In this way, a sustainable balance nclimate1354
between producing food, fiber and bioenergy feedstock, and other Banks-Leite, C., Pardini, R., Tambosi, L. R., Pearse, W. D., Bueno, A. A.,
Bruscagin, R. T., . . . Metzger, J. P. (2014). Using ecological thresholds
ecosystem services, as well as biodiversity, can be achieved.
to evaluate the costs and benefits of set-asides in a biodiversity hot-
spot. Science, 345, 1041–1045.
Barber, C. P., Cochrane, M. A., Souza, C. M. Jr, & Laurance, W. F. (2014).
ACKNOWLEDGEMENTS
Roads, deforestation, and the mitigating effect of protected areas in
the Amazon. Biological Conservation, 177, 203–209. https://doi.org/
We would like to express our appreciation to the anonymous
10.1016/j.biocon.2014.07.004
reviewers whose suggestions and comments were very valuable and Barlow, J., Lennox, G. D., Ferreira, J., Berenguer, E., Lees, A. C., Mac
helped us improving the article. We are also grateful to the Brazilian Nally, R., . . . Parry, L. (2016). Anthropogenic disturbance in tropical
National Council for Scientific and Technological Development forests can double biodiversity loss from deforestation. Nature, 535,
144–147. https://doi.org/10.1038/nature18326
(249404/2013-3) for financing this research. The authors of this
Barretto, A. G. D. O. P. (2013). Agricultural land-use expansion dynamics in
manuscript have no conflict of interest to declare.
Brazil. Piracicaba, Brazil: Unpublished PhD University of S~ao Paulo.
https://doi.org/10.11606/T.11.2013.tde-28032013-110939
Bernasconi, P., Blumentrath, S., Barton, D. N., Rusch, G. M., & Romeiro, A.
ORCID R. (2016). Constraining forest certificate’s market to improve cost-
effectiveness of biodiversity conservation in S~ao Paulo State, Brazil. PLoS
Flavio L. M. Freitas http://orcid.org/0000-0002-8313-5845
ONE, 11, e0164850. https://doi.org/10.1371/journal.pone.0164850
Oskar Englund http://orcid.org/0000-0002-1662-6951 Brasil(1973). Indigenous People Statute “Estatuto do Indio”. In: Law no
Gerd Sparovek http://orcid.org/0000-0001-8301-8529 6001, 1973-12-19. Brasilia, Brazil.
€ran Berndes
Go http://orcid.org/0000-0003-1126-6835 Brasil (2000). National System of Conservation Units “Sistema Nacional de
Unidades de Conservacßa ~o da Natureza”. In: Law n° 9.985, 2000-07-18.
Vinicius Guidotti http://orcid.org/0000-0002-3639-2101
Brasilia, Brazil.
€rtberg
Ulla Mo http://orcid.org/0000-0002-1640-8946 Brasil (2006). Atlantic Forest Law “Lei de Mata Atla ^ntica”. In: Law N°
11.428, 2006-12-22. Brasil.
Brasil (2009). Terra Legal Programme “Programa Terra Legal”. In: Law N°
11.952. Brasilia, Brazil.
REFERENCES
Brasil (2012). Brazilian Native Vegetation Protection Law. In: Law N°
Aguiar, A. P. D., Vieira, I. C. G., Assis, T. O., Dalla-Nora, E. L., Toledo, P. 12.651, 25-May-2012. Brasilia, Brazil.
M., Oliveira Santos-Junior, R. A., . . . Nobre, C. A. (2016). Land use Brasil (2017). Act for tenure regularization of rural and urban land. In: Law
change emission scenarios: Anticipating a forest transition process in N° 13.465, 2017-07-11.
FREITAS ET AL. | 13

Ciro Barros, L. B. (2016). Crime e grilagem com uso do CAR. In T. Dome- INPE (2017). Satellite-based monitoring of deforestation of the Amazon for-
nici (Ed.), Ag^encia De Reportagem E Jornalismo Investigativo. Sao Paulo, est/PRODES “Monitoramento da floresta amaz^ onica brasileira por
Brazil: Age ^ncia Publica. satelite”.
Damasceno, R. (2016). Insecure land rights in Brazil: Consequences for rural Ioris, A. A. R. (2016). The paradox of poverty in rich ecosystems: Impov-
areas and challenges for improvement. In: Nu cleo de Avaliacß~ao de erishment and development in the Amazon of Brazil and Bolivia. The
Polıticas Climaticas. PUC-Rio, Climate Policy Initiative. Geographical Journal, 182, 178–189. https://doi.org/10.1111/geoj.
Damasceno, R., Chiavar, J., & Lopes, C. L.. (2017). Evolution of land rights 12124
in Brazil. In: Nu cleo de Avaliacß~ao de Polıticas Climaticas. PUC-Rio, Cli- IPCC (2014) Climate change 2014: Synthesis report. Contribution of Work-
mate Policy Initiative. ing Groups I, II and III to the Fifth Assessment Report of the Inter-
De Sousa, P. T., Fernandez Piedade, M. T., & Candotti, E. (2011). Ecologi- governmental Panel on Climate Change [Code Writing, R. K. Pachauri
cal oversight: Brazil’s forest code puts wetlands at risk. Nature, 478, and L. A. Meyer (Eds.)]Geneva, Switzerland: IPCC.
458. https://doi.org/10.1038/478458b Joosten, H. (2009). The Global Peatland CO2 Picture: Peatland status and
Englund, O., Sparovek, G., Berndes, G., Freitas, F., Ometto, J. P., Oliveira, drainage related emissions in all countries of the world. Wageningen,
P. V. D. C. E., . . . Lapola, D. (2017). A new high-resolution nationwide The Netherlands: Wetlands International.
aboveground carbon map for Brazil. Geo: Geography and Environment, Lapola, D. M., Martinelli, L. A., Peres, C. A., Ometto, J. P., Ferreira, M. E.,
4, e00045. Nobre, C. A., . . . Joly, C. A. (2014). Pervasive transition of the Brazil-
Esri. (2010). ArcGIS Version 10.3.1 [GIS Application]. Redlands, CA: Envi- ian land-use system. Nature Climate Change, 4, 27–35.
ronmental Systems Research Institute, Inc. Lewinsohn, T. M. O. (2006). Avaliacßa ~o do estado do conhecimento da bio-
FAO. (2014). FAOSTAT statistics database. Rome, Italy: Food Agriculture diversidade brasileira - Volume I e II, Brasılia, Brazil: MMA.
Organization of the United Nations. L’roe, J., Rausch, L., Munger, J., & Gibbs, H. K. (2016). Mapping proper-
Farr, T. G., Rosen, P. A., Caro, E., Crippen, R., Duren, R., Hensley, S., . . . ties to monitor forests: Landholder response to a large environmental
Seal, D. (2007). The shuttle radar topography mission. Reviews of registration program in the Brazilian Amazon. Land Use Policy, 57,
Geophysics, 45, https://doi.org/10.1029/2005RG000183. 193–203. https://doi.org/10.1016/j.landusepol.2016.05.029
Fearnside, P. M. (2016). Brazilian politics threaten environmental policies. MAPA (2016). Brasil Projecß~ oes do agronegocio 2015/2016 a 2025/2026.
Science, 353, 746–748. https://doi.org/10.1126/science.aag0254 Brasılia-DF, Brazil: Ministe rio da Agricultura, Pecu aria e Abasteci-
Ferreira, J., Arag~ao, L. E. O. C., Barlow, J., Barreto, P., Berenguer, E., Busta- mento.
mante, M., . . . Pardini, R. (2014). Brazil’s environmental leadership at MCTI (2016). Third national communication of Brazil to the United Nations
risk. Science, 346, 706–707. https://doi.org/10.1126/science.1260194 framework convention on climate change (ed Ministry of Science TaI).
Freitas, F. L. M., Guidotti, V., & Sparovek, G. (2017). Technical note: Land Brasilia, Brazil.
tenure map of Brazil, v.170321. In: Atlas - A Geografia da Agrope- Metzger, J. P., Lewinsohn, T. M., Joly, C. A., Verdade, L. M., Martinelli, L. A.,
cuaria Brasileira, 2017. Piracicaba. & Rodrigues, R. R. (2010). Brazilian law: Full speed in reverse? Science,
Freitas, F. L. M., Sparovek, G., & Matsumoto, M. (2016). A adicionalidade 329, 276–277. https://doi.org/10.1126/science.329.5989.276-b
do mecanismo de compensacß~ao de Reserva Legal da Lei 12.651/2012: MMA (2017). Brazil’s forest reference emission level for reducing emissions
Uma ana lise da oferta e demanda de Cotas de Reserva Ambiental. In: from deforestation in the cerrado biome for results-based payments for
As Mudancßas no Co  digo Florestal Brasileiro: desafios para a imple- REDD+ under the United Nations Framework Convention on Climate
mentacß~ ao da nova lei. (ed Ipea). Brasılia, IPEA. Change. (ed Environment MOT). Brasilia.
Freitas, F. L. M. D., Sparovek, G., Mo € rtberg, U., Silveira, S., Klug, I., & Mueller, B. (2016). Key issues for property rights in Brazil: Implications for
Berndes, G. (2017). Offsetting legal deficits of native vegetation the forest code. In: Nu cleo de Avaliacß~ao de Polıticas Clim aticas. (ed
among Brazilian landholders: Effects on nature protection and socioe- Initiative CP). PUC-Rio.
conomic development. Land Use Policy, 68, 189–199. https://doi.org/ Nazareno, A. G., & Laurance, W. F. (2015). Brazil’s drought: Beware
10.1016/j.landusepol.2017.07.014 deforestation. Science, 347, 1427–1427.
Gibbs, H. K., Rausch, L., Munger, J., Schelly, I., Morton, D. C., Noojipady, Nepstad, D., McGrath, D., Stickler, C., Alencar, A., Azevedo, A., Swette,
P., . . . Walker, N. F. (2015). Brazil’s Soy Moratorium. Science, 347, B., . . . Armijo, E. (2014). Slowing Amazon deforestation through pub-
377–378. https://doi.org/10.1126/science.aaa0181 lic policy and interventions in beef and soy supply chains. Science,
Godar, J., Gardner, T. A., Tizado, E. J., & Pacheco, P. (2014). Actor-speci- 344, 1118–1123. https://doi.org/10.1126/science.1248525
fic contributions to the deforestation slowdown in the Brazilian Ama- Nepstad, D., Schwartzman, S., Bamberger, B., Santilli, M., Ray, D., Sch-
zon. Proceedings of the National Academy of Sciences, 111, 15591– lesinger, P., . . . Rolla, A. (2006). Inhibition of Amazon deforestation
15596. https://doi.org/10.1073/pnas.1322825111 and fire by parks and indigenous lands. Conservation Biology, 20(1),
Guedes, G. R., Brondızio, E. S., Barbieri, A. F., Anne, R., Penna-Firme, R., 65–73. https://doi.org/10.1111/j.1523-1739.2006.00351.x
& D’antona A.  O. (2012). Poverty and inequality in the Rural Brazilian Nobre, C. A., Sampaio, G., Borma, L. S., Castilla-Rubio, J. C., Silva, J. S., &
Amazon: A multidimensional approach. Human Ecology: An Interdisci- Cardoso, M. (2016). Land-use and climate change risks in the Amazon
plinary Journal, 40, 41–57. https://doi.org/10.1007/s10745-011- and the need of a novel sustainable development paradigm. Proceed-
9444-5 ings of the National Academy of Sciences, 113, 10759–10768.
Hansen, M. C., Potapov, P. V., Moore, R., Hancher, M., Turubanova, S., https://doi.org/10.1073/pnas.1605516113
Tyukavina, A., . . . Kommareddy, A. (2013). High-resolution global Nogueira, E. M., Yanai, A. M., Fonseca, F. O. R., & Fearnside, P. M.
maps of 21st-century forest cover change. Science, 342, 850–853. (2015). Carbon stock loss from deforestation through 2013 in Brazil-
https://doi.org/10.1126/science.1244693 ian Amazonia. Global Change Biology, 21, 1271–1292. https://doi.org/
Harris, N. L., Brown, S., Hagen, S. C., Saatchi, S. S., Petrova, S., Salas, W., 10.1111/gcb.12798
. . . Lotsch, A. (2012). Baseline map of carbon emissions from defor- Paixao, S., Hespanha, J. P., Ghawana, T., Carneiro, A. F. T., Zevenbergen,
estation in tropical regions. Science, 336, 1573–1576. J., & Frederico, L. N. (2015). Modeling indigenous tribes’ land rights
IBGE (2006). Agricultural census. with ISO 19152 LADM: A case from Brazil. Land Use Policy, 49, 587–
IBGE (2015). Digital Cartographic Database. 597. https://doi.org/10.1016/j.landusepol.2014.12.001
INCRA (2015). Land tenure database. National Institute of Colonization Pinto, S., Melo, F., Tabarelli, M., Padovesi, A., Mesquita, C. A., de Mattos
and Agrarian Reform “Instituto Nacional de Colonizacß~ao e Reforma Scaramuzza, C. A., . . . Ce sar, R. G. (2014). Governing and delivering a
Agraria” biome-wide restoration initiative: The case of Atlantic Forest
14 | FREITAS ET AL.

restoration pact in Brazil. Forests, 5, 2212–2229. https://doi.org/10. and conservation in Brazil. Environmental Science & Technology, 49,
3390/f5092212 10285–10293. https://doi.org/10.1021/acs.est.5b01300
Reydon, B. P., Fernandes, V. B., & Telles, T. S. (2015). Land tenure in Bra- Sparovek, G., Berndes, G., Barretto, A. G. D. O. P., & Klug, I. L. F. (2012).
zil: The question of regulation and governance. Land Use Policy, 42, The revision of the Brazilian Forest Act: Increased deforestation or a
509–516. https://doi.org/10.1016/j.landusepol.2014.09.007 historic step towards balancing agricultural development and nature
Ricardo, F. (2004). Terras Indıgenas & Unidades de Conservacßa ~o da natur- conservation? Environmental Science & Policy, 16, 65–72. https://doi.
eza : o desafio das sobreposicßo ~es. Sao Paulo, Brazil: Instituto Socioam- org/10.1016/j.envsci.2011.10.008
biental. Sparovek, G., Berndes, G., Klug, I. L. F., & Barretto, A. G. O. P. (2010).
Richards, P. D., & Vanwey, L. (2015). Farm-scale distribution of defor- Brazilian agriculture and environmental legislation: Status and future
estation and remaining forest cover in Mato Grosso. Nature Climate challenges. Environmental Science & Technology, 44, 6046–6053.
Change, 6, 418. https://doi.org/10.1021/es1007824
Rist, L., Shanley, P., Sunderland, T., Sheil, D., Ndoye, O., Liswanti, N., & Stickler, C. M., Nepstad, D. C., Azevedo, A. A., & Mcgrath, D. G. (2013).
Tieguhong, J. (2012). The impacts of selective logging on non-timber Defending public interests in private lands: Compliance, costs and poten-
forest products of livelihood importance. Forest Ecology and Manage- tial environmental consequences of the Brazilian Forest Code in Mato
ment, 268, 57–69. https://doi.org/10.1016/j.foreco.2011.04.037 Grosso (p. 368). Biological Sciences: Philosophical Transactions of the
Saatchi, S. S., Harris, N. L., Brown, S., Lefsky, M., Mitchard, E. T., Salas, Royal Society B.
W., . . . Petrova, S. (2011). Benchmark map of forest carbon stocks in Strassburg, B. B. N., Latawiec, A., & Balmford, A. (2016). Brazil: Urgent
tropical regions across three continents. Proceedings of the National action on Cerrado extinctions. Nature, 540, 199. https://doi.org/10.
Academy of Sciences, 108, 9899–9904. https://doi.org/10.1073/pnas. 1038/540199a
1019576108 UN (2015). Paris agreement. United Nations.
Saatchi, S. S., Houghton, R. A., Dos Santos Alvala, R. C., Soares, J. V., & Zarin, D. J. (2012). Carbon from tropical deforestation. Science, 336,
Yu, Y. (2007). Distribution of aboveground live biomass in the Ama- 1518–1519. https://doi.org/10.1126/science.1223251
zon basin. Global Change Biology, 13, 816–837. https://doi.org/10. Zarin, D. J., Harris, N. L., Baccini, A., Aksenov, D., Hansen, M. C., Aze-
1111/j.1365-2486.2007.01323.x vedo-Ramos, C., . . . Allegretti, A. (2016). Can carbon emissions from
Santos, J. C., Leal, I. R., Almeida-Cortez, J. S., Fernandes, G. W., & Tabar- tropical deforestation drop by 50% in 5 years? Global Change Biology,
elli, M. (2011). Caatinga: The scientific negligence experienced by a 22, 1336–1347. https://doi.org/10.1111/gcb.13153
dry tropical forest. Tropical Conservation Science, 4, 276–286.
https://doi.org/10.1177/194008291100400306
Silva Jaan, A. D., Manzatto, C. V., Joly, C. A., Rodrigues, R. R., Skorupa, L.
A., Nobre, C. A., . . . Rech Filho, E. L. (2011). O Co digo Florestal e a SUPPORTING INFORMATION
Ci^encia. S~ao Paulo, Brazil: SBPC/ABC.
Additional Supporting Information may be found online in the sup-
Soares-Filho, B., Raj~ ao, R., Macedo, M., Carneiro, A., Costa, W., Coe, M.,
. . . Alencar, A. (2014). Cracking Brazil’s forest code. Science, 344, porting information tab for this article.
363–364. https://doi.org/10.1126/science.1246663
Sparovek, G., Antoniazzi, L. B., Barretto, A., Barros, A. C., Benevides, M.,
Berndes, G., . . . Nogueira, M. P. (2016). Sustainable bioproducts in
Brazil: Disputes and agreements on a common ground agenda for How to cite this article: Freitas FLM, Englund O, Sparovek G,
agriculture and nature protection. Biofuels, Bioproducts and Biorefining, et al. Who owns the Brazilian carbon? Glob Change Biol.
10, 204–221. https://doi.org/10.1002/bbb.1636 2017;00:1–14. https://doi.org/10.1111/gcb.14011
Sparovek, G., Barretto, A., Klug, I., Papp, L., & Lino, J. (2011). A revis~ ao
do Co  digo Florestal brasileiro. Novos Estudos - CEBRAP, 89, 111–135.
https://doi.org/10.1590/S0101-33002011000100007
Sparovek, G., Barretto, A. G. D. O. P., Matsumoto, M., & Berndes, G.
(2015). Effects of governance on availability of land for agriculture

You might also like