Content uploaded by Raoni Rajão
Author content
All content in this area was uploaded by Raoni Rajão on Jan 17, 2017
Content may be subject to copyright.
www.sciencemag.org SCIENCE VOL 344 25 APRIL 2014 363
POLICYFORUM
Roughly 53% of Brazil’s native vege-
tation occurs on private properties.
Native forests and savannahs on these
lands store 105 ± 21 GtCO2e (billion tons
of CO2 equivalents) and play a vital role in
maintaining a broad range of ecosystem ser-
vices ( 1). Sound management of these private
landscapes is critical if global efforts to mit-
igate climate change are to succeed. Recent
approval of controversial revisions to Bra-
zil’s Forest Code (FC)—the central piece of
legislation regulating land use and manage-
ment on private properties—may therefore
have global consequences. Here, we quantify
changes resulting from the FC revisions in
terms of environmental obligations and rights
granted to land-owners. We then discuss con-
servation opportunities arising from new pol-
icy mechanisms in the FC and challenges for
its implementation.
Created in 1965, the FC was transformed
during the 1990s into a de facto environmen-
tal law via a series of presidential decrees. As
of 2001, the FC required landowners to con-
serve native vegetation on their rural proper-
ties, setting aside a Legal Reserve (LR) that
occupies 80% of the property area in the Ama-
zon and 20% in other biomes [supplementary
material (SM), fig. S1, and table S1]. The
law also designated environmentally sensi-
tive areas as Areas of Permanent Preservation
(APPs), aiming to conserve water resources
and prevent soil erosion. APPs include both
Riparian Preservation Areas (RPAs) that pro-
tect riverside forest buffers, and Hilltop Pres-
ervation Areas (HPAs) at hilltops, high eleva-
tions, and steep slopes.
The FC severely restricted deforestation
on private properties but proved challeng-
ing to enforce, particularly in the Amazon.
As deforestation rates rose in the early 2000s,
efforts to strengthen enforcement increased
pressure on the farming sector, which trig-
gered a backlash against the FC. The agri-
business lobby took advantage of a favorable
political moment, related to a substantial drop
in deforestation rates in the Brazilian Ama-
zon, to propose creation of a new FC, which
was approved in late 2012 ( 2). Some criticize
the legislation for being too lenient on land-
owners; others maintain that it is a barrier to
agricultural development. Regulations detail-
ing key implementation mechanisms of the
revised FC are still under negotiation.
Amnesty for Illegal Deforestation
The 2012 FC maintains conservation require-
ments for LRs and RPAs —i.e., land that
may not be deforested (table S1). These two
requirements protect 193 ± 5 Mha of native
vegetation containing 87 ± 17 GtCO2e (see
the map). Changes in the defi nition of HPAs
reduced their total area by 87% (table S8).
Because the new law differentiates
between conservation and restoration require-
ments, the 2012 FC reduced by 58% Brazil’s
“environmental debt”—i.e., areas of LR and
RPA deforested illegally before 2008 that,
under the previous FC, would have required
restoration at the landowner’s expense (fi g.
S2). This was accomplished by forgiving the
LR debt of “small” properties, ranging in size
from 20 ha in southern Brazil to 440 ha in the
Amazon. Under these new rules, 90% of Bra-
zilian rural properties qualify for amnesty.
Further reductions resulted from including
RPAs in the calculation of the LR area, reduc-
ing the LR restoration requirement to 50% in
Amazonian municipalities occupied predomi-
nantly by protected areas, and relaxing RPA
restoration requirements on small properties
(table S1).
Together, these measures decreased
the total area to be restored from 50 ± 6 to
21 ± 1 Mha, of which 78% encompasses LRs
and 22% RPAs (tables S2 and S3). Reduc-
tions in the environmental debt were uneven
across states and biomes, affecting mainly the
Amazon, Atlantic Forest, and Cerrado (fi g.
S2). These losses may have a large impact on
Cracking Brazil’s Forest Code
LAND USE
Britaldo Soares-Filho,
1
* Raoni Rajão,
1 Marcia Macedo,
1, 2 Arnaldo Carneiro,
3 William Costa,
1
Michael Coe,
2 Hermann Rodrigues,
1 Ane Alencar
4
Brazil’s controversial new Forest Code grants
amnesty to illegal deforesters, but creates new
mechanisms for forest conservation.
Alagoas
Bahia
Espírito Santo
Paraíba
Pernambuco
Rio de Janeiro
Rio Grande do Norte
Sergipe
São Paulo
Acre
Amazonas
Amapá
Ceará
Distrito
Federal
Gisáo
Maranhão
Minas
Gerais
do Sul
Mato
Grosso
Mato
Grosso
Pará
Piauí
Paraná
Rondônia
Roraima
Ri o
Grande do Sul
Santa Catarina
Tocantins
AM
PT
CE
PA
CA
AF
Biomes
States
10° S
60° W
50° W
70° W
0° 0°
60° W
70° W 10° S
20° S
20° S
30° S 30° S
50001000 km
40° W
Main rivers
40° W
Compliance levels
> 300%
70% — 300%
50% —
70%
20% — 50%
0 —20%
–20% — 0
–50% — –20%
–70% — –50%
–90% — –70%
< –90%
FC inapplicable
Atlantic
Forest
(AF)
Pampas
(PA)
Amazon
(A M)
Cerrado
(CE)
Caatinga
(CA)
Pantanal
(PT)
Compliance levels under Brazil’s 2012 FC. Percent difference between the remaining area of native vegeta-
tion and the area required to comply with the 2012 FC. Positive values indicate forest surpluses or land that may
be legally deforested. Negative values indicate forest debts or land that requires restoration. See SM for details.
1Universidade Federal de Minas Gerais, Belo Horizonte,
MG 31270-901, Brazil. 2Woods Hole Research Center, Fal-
mouth, MA 02540, USA. 3Secretaria de Assuntos Estraté-
gicos da Presidência da República, Brasília, DF 70052-
900, Brazil. 4Instituto de Pesquisa Ambiental da Amazônia,
Brasília, DF 71.503-505, Brazil. *Corresponding author:
britaldo@csr.ufmg.br
Published by AAAS
25 APRIL 2014 VOL 344 SCIENCE www.sciencemag.org
364
POLICYFORUM
biodiversity conservation ( 3) and forest resto-
ration programs ( 4), especially in the Atlantic
Forest, where only 12 to 16% of the original
forest cover remains ( 5).
Furthermore, both old and new FCs allow
an additional 88 ± 6 Mha of legal deforesta-
tion on private properties (table S4 and the fi g-
ure). This area of native vegetation, exceeding
LR and RPA conservation requirements, con-
stitutes an “environmental surplus” with the
potential to emit 18 ± 4 GtCO2e (SM, §2.1).
New Mechanisms for New Markets
Although the 2012 FC reduces restoration
requirements, it introduces new mechanisms
to address fi re management, forest carbon,
and payments for ecosystem services, which
could reduce deforestation and bring envi-
ronmental benefi ts. Perhaps the most impor-
tant mechanism is the Environmental Reserve
Quota (Portuguese acronym, CRA), a tradable
legal title to areas with intact or regenerating
native vegetation exceeding the FC require-
ments. The CRA (surplus) on one property
may be used to offset a LR debt on another
property within the same biome and, prefer-
ably, the same state. Implementating the CRA
could create a trading market for forested
lands, adding monetary value to native veg-
etation. This CRA market could potentially
abate 56% of the LR debt (fi g. S3). Given the
high costs of forest restoration ( 6), exchange
of CRAs could become a cost-effective way
to facilitate compliance, meanwhile protect-
ing forest surpluses that might otherwise be
legally deforested. A balanced use of CRAs
should focus on improving functional and
ecological attributes of forested landscapes,
e.g., habitat integrity (and thus biodiversity),
carbon stocks, and water balance regulation,
crucial for maintaining hydroelectric power
generation in Brazil ( 7).
One of the strongest arguments of the agri-
business lobby is that forest restoration con-
fl icts with agricultural production. Our results
suggest that, with respect to land availability,
this concern is unfounded. Of the 4.5 ± 1 Mha
of RPAs slated for restoration, only 0.6 ± 0.35
Mha are currently occupied by crops, repre-
senting less than 1% of all croplands nation-
wide. Moreover, if restoration of the remain-
ing LR debt (after compensation via CRAs)
occurred exclusively in pasturelands unsuit-
able for agriculture, as few as ≈ 550,000 ha
of required restoration would remain in ara-
ble lands (SM §§2.2 and 2.3 and fi gs. S3 to
S5). Such a large-scale transition from cattle
ranching to agriculture would require sub-
stantial increases in stocking densities to sus-
tain current levels of meat production and
allow for forest restoration. To this end, Bra-
zil has created a national Low-Carbon Agri-
culture (ABC) program that provides ~U.S. $
1.5 billion in annual subsidized loans aimed
at increasing agricultural productivity while
reducing associated carbon emissions and
supporting forest restoration (table S5).
Key to success of the FC is the Rural Envi-
ronmental Registry System (SICAR), a geo-
referenced Web system that will enable docu-
mentation of over 5 million rural properties,
improving transparency and providing a path-
way to environmental compliance. SICAR
could facilitate the market for CRAs and pay-
ments for ecosystem services [for example,
( 8)], which will be critical to offset the often-
prohibitive costs of forest restoration, espe-
cially for small landowners. We estimate that
elimination of the FC debt via forest resto-
ration would sequester up to 9 ± 2 GtCO2e
(SM, §2.1).
Enforcement and Private Initiatives
Effective implementation of Brazil’s 2012
FC will be enormously challenging. The
fi rst crucial challenge is to convince the agri-
business sector of the potential gains from
the new FC. Even though law enforcement
activities have intensifi ed in recent years, the
agribusiness constituency has historically
taken advantage of the government’s rela-
tively weak enforcement of environmental
laws. Amnesty afforded by the new FC could
lead to the perception that illegal deforesters
are unlikely to be prosecuted and may even
be exonerated in future law reforms. To meet
this challenge, Brazil must continue to invest
in its monitoring and enforcement capabili-
ties. Satellite-based deforestation monitor-
ing systems maintained by the National
Institute for Space Research (INPE) need to
be expanded to other Brazilian biomes and
adapted to detect subtler land-use changes,
including forest degradation and deforesta-
tion in savannahs, riparian forests, and small
remnants of the Atlantic Forest.
More important, there is a need to
strengthen and integrate efforts across the
myriad state and federal agencies responsible
for implementing the FC, establishing clear
land tenure, granting environmental licenses,
and supporting agricultural production. This
integrated system must be transparent and
harnessed to economic incentives for conser-
vation; otherwise, it might only exhort land-
owners to exercise rights to deforest ( 9).
Fortunately, private initiatives are align-
ing to assist landowners in attaining compli-
ance. These include international certifi ca-
tion standards, commodity roundtables, and
boycotts of agricultural products grown in
recently deforested or high-biodiversity areas.
Increasingly, farmers and ranchers are adher-
ing to voluntary registries that require com-
mitments to improving social and environ-
mental performance [for example, ( 10, 11)].
Both certifi cation schemes and voluntary reg-
istries may eventually enable access to special
markets that provide fi nancial incentives to
participating producers. These mechanisms
are particularly important in the Cerrado, the
most coveted biome for agribusiness expan-
sion, given its 40 ± 3 Mha of environmental
surplus that could be legally deforested (table
S4). Moreover, conservation efforts must aim
at expanding protected areas outside the Ama-
zon. Whereas these areas cover 46% of the
Brazilian Amazon, the level of protection in
other major biomes (7% of the Cerrado and
2.6% of the Atlantic Forest) is well below the
17% recommended by the 10th Convention
on Biological Diversity. Conservation initia-
tives will be vital to protect large expanses of
native vegetation, particularly in the Cerrado
and Caatinga, where additional protection by
land-use zoning is low.
Brazil has achieved an unprecedented suc-
cess in reducing deforestation in the Ama-
zon. However, this gain is not yet secured.
Recently, deforestation rates ceased to decline
in the Amazon and Atlantic Forest, and surged
in the Cerrado (fi g. S6). Our analysis suggests
that the FC will allow additional deforesta-
tion, especially in the Cerrado and Caatinga.
Economic incentives for conserving forests,
including the Warsaw Framework for Reduc-
ing Emissions from Deforestation and Forest
Degradation as (REDD+), will be essential to
help implement the FC and to enable Brazil to
better reconcile environmental conservation
with agricultural development.
References and Notes
1. Millennium Ecosystem Assessment, Ecosystem and Human
Well-Being: Synthesis (Island Press, Washington, DC,
2005); www.millenniumassessment.org/en/index.aspx.
2. Federal Law 12.727, 17 October 2012; www.planalto.
gov.br/ccivil_03/_Ato2011-2014/2012/Lei/L12727.htm.
3. A. C. Lees, C. A. Peres, Conserv. Biol. 22, 439 (2008).
4. R. Rodrigues et al., For. Ecol. Manage. 261, 1605 (2011).
5. M. Ribeiro et al., Biol. Conserv. 142, 1141 (2009).
6. R. Cury, O. Carvalho, “Manual para restauro fl orestal:
Florestas de transição” (IPAM, Canarana, 2011).
7. C. M. Stickler et al., Proc. Natl. Acad. Sci. U.S.A. 110,
9601 (2013).
8. Project Oasis, Fundação Grupo Boticário; www.funda-
caogrupoboticario.org.br/en/what%20we%20do/oasis/
pages/default.aspx.
9. R. Rajão et al., Public Adm. Dev. 32, 229 (2012).
10. Aliança da Terra, aliancadaterra.org.br.
11. Round Table on Responsible Soy Association, responsible-
soy.org.
Acknowledgments: See the supplementary materials for
funding services and acknowledgments.
10.1126/science.1246663
Supplementary Materials
www.sciencemag.org/content/344/6182/363/suppl/DC1
Published by AAAS
1
www.sciencemag.org/content/344/6182/363/suppl/DC1
Supplementary Materials for
Cracking Brazil’s Forest Code
Britaldo Soares-Filho, Raoni Rajão, Marcia Macedo, Arnaldo Carneiro, William Costa,
Michael Coe, Hermann Rodrigues, Ane Alencar
*Corresponding author. E-mail: britaldo@csr.ufmg.br
Published 25 April 2014, Science 344, 363 (2014)
DOI: 10.1126/science.124663
This PDF file includes
Materials and Methods
Supplementary Text
Figs. S1 to S16
Tables S1 to S10
References
Other Supplementary Material for this manuscript includes the following:
Input data, models, and main outputs of the Forest Code analyses available at
www.csr.ufmg.br/forestcode.
2
Materials and Methods
S§1. Quantifying the impacts of revision to the Forest Code
Here, we evaluate how recent modifications (Table S1) to the Forest Code (FC)
(Law No. 12,727, 10/17/2012) changed the spatial distribution and area of land
designated for conservation or restoration of native ecosystems (Figs. S1 and S2),
compared to the former version [Law No. 4.771, 09/15/1965, modified by provisional
measures (MPs) no. 1.1511, 06/25/1996 and 2.166-67, 08/24/2001]. In doing so, we
provide estimates of these new requirements, as well as the level of uncertainty resulting
from the methods and data sets used. The study aims to provide decision-makers with a
spatially explicit understanding of the impacts of the FC revision on forest conservation
in Brazil and the magnitude of the effort required to fully implement the new law.
S§2. Methods
Quantifying the conservation and restoration requirements of both versions of the
FC is not a trivial task because of Brazil’s continental scale. There are many data
limitations, including the absence of a unified national land-registry database that
integrates information on the approximately five million rural properties; a lack of fine-
scale maps of the drainage network and river widths; and inconsistent information on the
remaining native vegetation for all biomes. For example, monitoring the 30-m-minimum
RPA (Riparian Preservation Area) width for conservation would require an integrated
database spanning the entire national territory, with a cartographic accuracy equal to or
higher than 15 m (that is, a scale of 1:50,000 or higher). The absence of an ideal,
integrated database does not impede the development of national-level estimates, as
attempted in previous analyses (12). Several mapping projects of vegetation remnants and
cartographic databases are available for Brazil at various scales (Table S6). These data
sets enable estimation of the FC balance (debts and surpluses—i.e., areas that must be
reforested and areas that may be legally deforested) with reasonable accuracy, given that
the aggregated uncertainty is sufficiently large to encompass the cartographic uncertainty
arising from combining data sets with different spatial scales. Moreover, improvements
in computing capacity have enabled increasingly fine-scale reanalysis of these massive
databases, making it feasible to assess the FC balance throughout the Brazilian territory at
the microwatershed scale.
In this study, we developed a unified cartographic database with a grid-cell
resolution of 60 m × 60 m. This resolution allows quantification of 30-m RPA, combined
on either side of a watercourse and is compatible with the cartographic scale of drainage
maps used (Table S6). For each input data layer of Table S6, we generated a cartographic
raster (matrix) with 71,000 columns and 73,000 rows in the Albers Conical Equal-Area
projection, which minimizes distortion of area. Our analytical models were implemented
using Dinamica EGO freeware (13) (csr.ufmg.br/dinamica) and Excel 2103 64 bits, and
all processing was performed using the computing resources of the Center for Remote
Sensing (www.csr.ufmg.br) of the Federal University of Minas Gerais (Belo Horizonte,
Brazil). All input data, analytical models, and main outputs are available for download at
csr.ufmg.br/forestcode. The user-friendly graphical interface of Dinamica EGO allows
for designing a model as a diagram, whose graph of operators establishes a visual data
3
flow (Fig. S7). Users can add comments to and name operators and submodels with
aliases to make the model diagram self-explanatory. By opening the FC models on the
Dinamica EGO graphical interface, users will be able to follow the model diagrams and
understand the algorithms employed in each model step. All calculation, logical or
arithmetical, is stored within the operator and can be accessed by opening its calculation
window. In addition, models can also be viewed and edited as a script language by using
a text editor.
In the absence of a unified land registry for the entire country of Brazil, we chose
to use 12th-order watersheds (Ottobacias) provided by ANA (Brazil’s National Water
Agency) as a proxy for rural properties (Table S6, watersheds_otto_12_reclass.tif in
Inputs at csr.ufmg.br/forestcode). These watersheds constitute 166,000 units with a mean
area of 3683 ha (considering their portions where the FC is applicable). Our analysis
indicates that, when estimating the FC balance, the uncertainty associated with using
microwatersheds to represent rural properties is inversely proportional to the number of
properties contained within that microwatershed (section S§2.6) and directly proportional
to the microwatershed size. That is, the smaller the watersheds and larger the number of
properties contained therein, the lower the uncertainty. We therefore estimated the
uncertainty associated with each microwatershed and calculated a total value by adding it
to the uncertainty estimate for the FC balance derived from estimation of RPA width as
calculated below. In addition, we validated our analysis of the FC balance by comparing
figures obtained by using the microwatershed methodology with those obtained by using
the INCRA (National Institute for Colonization and Agrarian Reform) data set (Table S6)
of rural properties (section S§2.5).
The models applied to our analyses are grouped into three sets: Preparatory, Main,
and Ancillary models (see tables at csr.ufmg.br/forestcode). Models that are dependent on
other models’ outputs are numbered in order to provide the sequence of execution.
To compute the remaining areas of native vegetation (remnants.tif in Inputs at
csr.ufmg.br/forestcode), we integrated the most comprehensive national database of maps
from PRODES (Project of Monitoring Deforestation in Legal Amazon), SOS Mata
Atlântica, and PMDBBS (Project for Satellite-based Monitoring of Deforestation in the
Brazilian Biomes) (Table S6). Models applied to obtain this integrated map of remnants
include 1_campos_em_MA_ibama.ego and 2_add_remnants.ego―model names are
written in italic (Preparatory models at csr.ufmg.br/forestcode).
For the calculation of the FC balance (surplus and debts), we quantified for each
microwatershed the total area where the FC is applicable, named hereafter accountable
areas (i.e., the total area occupied by rural properties). To map these areas
(accountable_and_non_areas.tif in Inputs at csr.ufmg.br/forestcode), we subtracted urban
areas, water bodies, conservation reserves, indigenous lands, and 30-m buffers along
roads and railroads from the microwatershed’s total area (accountable_areas.ego in
Preparatory models at csr.ufmg.br/forestcode).
As established by both FC laws, the required Legal Reserve (LR) area is a
proportion of the property area―i.e., in our analyses, the microwatershed accountable
area. As this proportion varies across Brazil (Table S1), we combined maps of the Legal
Amazon and vegetation physiognomy (Table S6) to calculate LR percentage for both
4
conservation and restoration rules (legal_reserve_percent_ativo.ego,
legal_reserve_percent_passivo.ego in Preparatory models at csr.ufmg.br/forestcode).
When a microwatershed is split between regions with different requirements (i.e., the
Cerrado and Amazon), the appropriate rules were weighted according to the proportion of
the microwatershed in each region and summed to arrive at a final estimate for the
microwatershed.
The main sequence of models to obtain both the new and old FC balances is
depicted in Fig. S8. At the top, a set of preparatory models output the main spatial inputs
for a central model named 1_1_forest_code.ego in Main at csr.ufmg.br/forestcode, which
are (i) remnants.tif, (ii) rivers.tif, (iii) watersheds_otto_12_reclass.tif, (iv)
accountable_and_non_areas.tif, (v) legal_reserve_percent.tif, (vi)
legal_reserve_percentpass.tif (the latter depicts LR percentage for restoration purpose)
(vii) apps_non_hierarchical.tif, (viii) apps_hierarchical.tif, and (ix)
apps_reconstituicao.tif. The last three maps consist of buffer zones along river and water
streams and bodies that represent RPA width requirements.
A major source of uncertainty inherent in mapping RPA buffers stems from the
lack of information about stream widths, which determine buffer-width requirements
(Table S1). The drainage basin data layer from ANA (Table S6) includes information on
watershed hierarchy (the first order representing the longest river) but not stream width.
We therefore assigned a hypothetical width based on the river order within the drainage
basin, as specified in Table S7 to produce maps vii and viii (APP, Portuguese acronym
for “Área de Proteção Permanente”―Area of Permanent Preservation). We developed
two databases of RPAs for conservation purpose, the first considering fixed 30-m buffers
along all watercourses (map vii) and the second (map viii) using the hierarchy specified
in Table S7. Sparovek et al. (12) used a similar method, but our analysis was conducted
at a finer spatial scale than that established by those authors. Taking the mean value of
RPA width for each microwatershed, we defined the uncertainty as 75% of the difference
relative to the maximum value, because the chance of extreme values is low. Last, the
map apps_reconstituicao.tif contains river buffer widths according to Table S1, Article
no. 61 §6 and Article no. 61-B of the new FC. Note that for this particular case, the
minimum RPA width is half of the cell resolution, thus areas of RPA widths <30 m are
calculated using a discount factor within the model 1_Forestcode_balance.xlsx in Main
models at csr.ufmg.br/forestcode.
In addition to the input maps, another set of preparatory models produce the
spatial representation of the fiscal modules, which vary in size throughout Brazilian
municipalities and, thus, within each microwatershed. The size of fiscal modules and
information on average size of rural properties, including the number of properties and
their extent, are obtained from the 2006 national agricultural census (14). Because these
data are organized according to municipal units, the first step in our spatial analysis
involved converting the municipality data into units of microwatershed by calculating the
proportion of municipal areas relative to the area of each microwatershed. These models
then generate the areal percentage of properties per fiscal module and convert the
resulting municipality data into microwatershed representation
(1_calcPercent_module.ego, 2_muni_to_water_modulo_fiscais&MF.ego). Next, the
models 1_calc_muni_state_PApercent.ego and 2_muni_to_water_PApercent.ego
5
calculate the municipality percentage covered with protected areas (Article no. 12 §§ 4
and 5, Table S1) and convert these values into microwatershed representation
(Preparatory models at csr.ufmg.br/forestcode). Both sets of models’ output tables are in
CSV (Comma-Separated Value) format.
In turn, model 1_1_forest_code.ego calculates the basic areal information per
each microwatershed, outputting a table (forestcode.csv) containing the following
variables: (i) microwatershed area, (ii) total remnant forest area, (iii) accountable area,
(iv) remnants in accountable area, (v) area of hierarchical RPA requirement, (vi) area of
nonhierarchical RPA requirement, (vii) river width, (viii) percent of LR requirement for
conservation, (ix) percent of LR requirement for restoration, and (x) area of RPA
requirement for restoration.
Because none of the land-cover maps have sufficient spatial accuracy to detect
remnant vegetation along RPAs, we calculated the combined area of LR and RPA within
each microwatershed unit, adding the RPA requirement area to the LR requirement area
in each microwatershed. Area of RPA to be restored was estimated indirectly, see end of
this section and S§2.6. To evaluate the balance (i.e., compliance level) of the FC, we
subtracted the total area required for RPAs and LRs from the remaining areas of native
vegetation in accountable areas of each microwatershed. We defined a positive result as
an environmental surplus and a negative result as an environmental debt. To estimate the
environmental debt, we evaluated the impact of reducing the LR from 80 to 50% of
properties in the Legal Amazon (Table S1), for the case of restoration in areas indicated
for agricultural consolidation (deemed suitable for agriculture or cattle production) by the
state-level Ecological and Economic Zoning programs (ZEE). To define those areas, we
used the ZEE suggested by the Ministry of the Environment (MMA, Table S6), given the
variability in state-level ZEE planning processes. In Legal Amazon, we maintained 80%
for LR in Areas for Environmental Protection (APAs)―a land-use zone that allows
private properties within it.
Although the rules governing forest conservation did not change under the revised
FC, save for the definition of Hilltop Preservation Areas―HPAs (Table S1), those
governing restoration became far more complex. First, the restoration of the RPA debt
along watercourses is now regulated by a rule called the “escadinha” (little staircase),
which specifies the buffer to be restored according to the property size (defined by the
number of fiscal modules) and width of the river. Small properties (up to four fiscal
modules) are now exempt from restoring their LR in areas that are already deforested and
in production (i.e., consolidated—defined as rural properties with human occupation
predating 22 July 2008). Additionally, LRs in the Amazon may be reduced to 50% when
the municipality has more than 50% of its area occupied by public conservation areas and
indigenous reserves or when the state has an approved ZEE and more than 65% of its
territory occupied by public conservation areas and indigenous reserves. Finally, RPAs
are now included when calculating the total area of LR, as long as the property is
registered in the Rural Environmental Registry (CAR, Table S1). Because all properties
are legally obligated to enter the system, we assume that 100% of rural properties in
Brazil will obtain a CAR registration.
To calculate the FC debt, we did not consider HPAs, because the new FC does not
stipulate the restoration of this type of APP in consolidated areas. Furthermore, the
6
calculation of the FC surplus―i.e., land that exceeds the conservation requirements of
the FC and thus can be legally deforested―does not account for the fact that a small
fraction (≤3%) of this total may be located in HPAs, where deforestation is not allowed.
Nonetheless, HPA areas are implicitly represented in our analysis because LR are more
likely to be located in these areas, which may be less suitable for agricultural production.
The table forestcode.csv plus the tables PA_water_percent.csv,
water_mf_mf_percent.csv, and waterModulos_fiscais.csv output from the preparatory
models together with the output from 1_2_biomass_average.ego are input for the model
1_Forestcode_balance.xlsx. One needs to copy the tables and paste them on the right
space of the spreadsheet as identified by the tabs with the corresponding names. The
Excel model incorporates the complex set of rules (cell formulas can be traced back in
order to map the rules) that quantifies for each microwatershed the FC balance in terms
of debt and surplus, as well as the changes to these rules under the new FC (Table S1).
The results in terms of surplus, debts, and their respective carbon stock and sequestration
potential for the old and new FC versions appear on the tab “forestcode” of
1_Forestcode_balance.xlsx. We then produced aggregate estimates of debts and surplus
for each biome and state by combining microwatershed data on the FC balance with
spatial information on municipal, biome, and microwatershed boundaries. In order to do
so, these data are extracted and ported to tables named output_##.csv, where # symbol is
a placeholder for a sequential number (e.g., 01, 02). These CSV tables must have only
two columns, the first representing the microwatershed code and the second the
associated data. Then the models 2_forest_code_water_muni.ego and
3_sum_state_biome.ego (Main models at www.csr.ufmg.br/forestcode) are used, first, to
convert look-up tables indexed by microwatershed to look-up tables indexed by
munibiomes (spatial unit consisting of a unique combination of municipality and biome
output from cross_muni_biomes.ego in Preparatory models) and, second, to totalize data
per state and biome (Tables S2 to S4). Because large blocks of forests in the state of
Amazonas remain undesignated public land to date (15), we ignored the surplus of this
state in summing the grand total of the FC surplus (Table S4).
Finally, we estimated both the extent of RPAs to be restored along watercourses
and the likelihood of their being occupied by agriculture using an indirect method, which
we refer to as the spatial Bootstrap (see section S§2.6). This calculation is performed
using the models 1_muni_to_water_PAM.ego and 2_Calc_APP_andUncertainties.xlsx in
Ancillary models at csr.ufmg.br/forestcode. For this calculation, we assumed that at least
90% (this figure is conservative; it could be even higher) of the 24 Mha of soybean
croplands plus 7.6 Mha of single-cropped cornfields in Brazil occur outside of RPAs,
given that mechanized agriculture (e.g., soybeans and corn) cannot operate in riparian
areas because of the high water table. The resulting 28.5 Mha (0.9*31.6) represents 40%
of all croplands (±70 Mha). Next, we used the spatial bootstrap simulation to estimate the
probability of any cropland (mechanized or not) occurring within RPAs. Finally, to
estimate the area of RPAs realistically occupied by croplands, we multiplied the
probability of any cropland occurring in RPAs (calculated from the spatial bootstrap) by
0.6 (1 minus 0.4—or the probability of nonmechanized agriculture occurring in RPAs).
This combined calculation yields the total area of cropland that is likely to fall within
RPAs.
7
S§2.1. Estimates of potential CO2 emissions and sequestration
For each microwatershed, we multiplied the areas of surpluses and debts by the
mean potential biomass of native vegetation to estimate the potential for carbon
sequestration via restoration projects, as well as the potential CO2 emissions from future
(legal) deforestation of environmental surplus areas. The potential biomass map
reconstructs the biomass of the original vegetation present in the Brazilian biomes (16).
We added 20% to the overall uncertainty estimate to account for the inherent uncertainty
in the biomass map. We assumed that carbon content is 50% of woody biomass (17) and
that 85% of the carbon contained in trees is released to the atmosphere after deforestation
(18).
S§2.2. Potential market for CRA
Article 44 of Law no. 12.651, 05/25/2012 specifies that the “Cota de Reserva
Ambiental” (Portuguese acronym, CRA―Environmental Reserve Quota) is a tradable
legal title to areas with intact or regenerating native vegetation exceeding FC
requirements. The CRA (surplus) on one property may be used to offset an LR debt on
another property within the same biome and, preferably, the same state. Paragraph §4
stipulates that the LRs of small landholders (up to four fiscal modules) can also constitute
CRA titles.
To quantify the potential market for CRAs, we used the model
1_Compensation_with_CRAs.xlsx (Ancillary models at csr.ufmg.br/forestcode) first to
calculate the extent of native vegetation exceeding FC requirements in microwatersheds
within unique combinations of biomes and states. We then compared these quantities
with the total LR debt within the same territorial units. If the total LR debt was less than
the FC surplus, we deducted this difference from the total amount of FC surplus to
estimate the potential regional (biome and state unit) market for CRA. In other words, the
potential CRA market depends on both availability of CRAs and LR debts within the
same biome and state. Regulation of the CRA under paragraph §4 still depends on
implementation of specific legislation by each Brazilian state (19). Because our estimates
did not account for these criteria, our figures for the potential market for CRA may be
conservative.
S§2.3. Pasturelands suitable for growing crops
We estimated the extent of suitable pasturelands for growing crops by applying
the following sequence of models: 1_deviance_slope.ego and 2_suitability.ego in
Preparatory models, and 1_1_calc_aptitute.ego, 2_sum_state_biome.ego
1_Apt_per_municipalities.xlsx and 2_Compesantion_after_CRAs_in_inapts.xlsx in
Ancillary models at csr.ufmg.br/forestcode.
Our calculation of suitable areas for cropland included areas with slopes less than
15%―appropriate for use of the heavy machinery required by agribusiness (20)―and
eliminated areas with soils that are highly unsuitable for agriculture. As in Nepstad et al.
(21), our soil criteria excluded soils with strong edaphic restrictions (e.g., ultisols,
lithosols, dysthropic podzols, sands, and hydromorphic soils). Comparing our suitability
map to soy and sugarcane croplands identified by the CANASAT project (22), we found
8
that 90% of existing croplands fell within areas classified as suitable. To calculate the
amount of pasturelands suitable for agriculture—and hence the extent of unsuitable
pasturelands that could be used for forest restoration—we first deducted the ~70 Mha of
existing croplands estimated by the Instituto Brasileiro de Geografia e Estatística (IBGE)
(23) from the roughly 290 Mha of land currently in production in Brazil. Of the 220 ± 10
Mha of pastures in various stages of occupation and productivity, ~60% could be utilized
for crops, if one assumes no climatic restrictions (Figs. S4 and S9). The uncertainty
bounds estimated for pastureland extent arise from uncertainties in the scales of maps of
remaining vegetation in Brazil, particularly for biomes other than the Amazon (Table S6).
S§2.4. Mapping Hilltop Preservation Areas
The models 1_brasil_hill_top_old_code_per_watershed_5.ego,
2_brasil_hill_top_new_code_per_watershed_5.ego in Ancillary models at
csr.ufmg.br/forestcode) were designed for calculating HPAs according to the old and new
definitions of the FC.
They comprise the following steps (Fig. S10): First, the elevation map is
quantized into hypsometric slices. Contiguous slices are labeled as individual elevation
patches. For each elevation patch, the algorithm calculates the mean elevation, slope,
area, and neighboring patches. Each patch consists of a node that is inserted into a
computer graph algorithm, which orders all nodes from local minima to local maxima,
forming a tree graph. The algorithm then uses the CalcHillTop operator (Fig. S11) to
analyze this graph and outputs tables of hilltops, hill heights, hill slopes, plateaus, and
local minima and maxima. The CalcHillTop operator was designed by Leandro Lima for
use in the Dinamica EGO software platform. Visual inspections of the three-dimensional
(3D) digital terrain models (Fig. S12) indicated an accuracy of 80% for the mapping
algorithm. Because of the high complexity of this algorithm and its sensitivity to spatial
resolution and slicing threshold of the elevation map, we only calculated the relative
impact (percent of reduction) of the new definition of HPAs (Table S8).
S§2.5 Validation of the FC balance with INCRA rural properties
We applied a data set from INCRA (Institute for Agrarian Reform) containing
62,897 rural properties distributed throughout Brazil to validate our analysis of the FC
balance. (See sequence of models employed for validation in Validation at
csr.ufmg.br/forestcode.) Table S9 shows a comparison between the INCRA data set of
rural properties and ANA microwatersheds. The accountable area of the latter is 6.72
larger than that of the former. We multiplied by this ratio the total figures for the FC
surplus and debts (before and after revision) obtained by using the INCRA data set to
compare with the figures obtained by using ANA microwatersheds. After scaling up, we
found a deviance of 11% between figures for the FC surplus and 7% between figures for
the old FC debt, which conform to the uncertainty bounds of our analysis. For the new
FC debt, we initially found a large difference between figures because of difference in
property size distribution from INCRA sample to IBGE census data (Table S9). We fixed
this difference by applying the same rules that govern the requirements for restoration as
if INCRA sample had the same property size distribution of IBGE census data. After this
adjustment and scaling up by 6.72, we found a deviance of only 6% (Table S10). We also
9
compared the spatial matching between maps of percentage of municipality in
compliance with the FC obtained from both analyses (Fig. S13). We applied two pairwise
tests. The Contingency Coefficient test (24) with 10 categorical intervals
(2_determinie_correlation10_intervals.ego) yielded a correlation of 74% and the
Reciprocal Similarity method (13) with 4 categorical intervals
(3_pattern_maching_4_intervals.ego) resulted in a spatial matching of 74%.
S§2.6 Spatial bootstrap of the RPA calculation, area of RPA occupied by croplands,
and model uncertainties
Bootstrapping is a statistical technique used to estimate parameters for small
sample sizes, including the distribution of the mean and its variance. The technique uses
repetitive sampling, with substitution of the samples selected from the data, using a
Monte Carlo simulation with a large number of iterations (1000 to 10,000, for example).
In this study, we adapted bootstrap methods to estimate the RPA debt (area to be
restored), RPA area potentially occupied by crops, and uncertainty related to the use of
ANA microwatersheds as a proxy for properties. To do so, we first created a map with
100 cells (10 × 10) to represent a microwatershed. We then used Dinamica EGO to run
spatial simulations that varied the extents and locations of vegetation remnants, LRs,
RPAs, crop area, and the number of properties within a watershed.
A set of simulations was carried out by varying the extent of vegetation remnants
from 20 to 80% and the width of the RPA from 10 to 50% of the microwatershed. The
model (1_Uncertainty_of_app&legal_reserve.ego in Ancillary models at
csr.ufmg.br/forestcode) randomly allocates remnant vegetation and RPAs to cells, using a
suite of 10,000 repetitions. For each iteration, we superimposed the simulated vegetation
remnants on the simulated RPAs and compared the results to the calculated debt (Fig.
S14). Based on the convergence of results, we concluded that the mean RPA debt,
expressed as a percentage of the accountable area of each microwatershed, could be
approximated using the following equation:
/RPAd RPAa RPAa RVa MWa
(Eq. S1)
Where RPAd is the RPA debt, RPAa is the RPA area, RVa is the remnant vegetation area,
and MWa is the microwatershed area.
In the same manner, we estimated the uncertainty in the calculation of the RPA
debt at one standard deviation (Sdd) as follows:
2
1/Sdd RPAd RPAa
(Eq. S2)
By extension, we used the equations below to infer the area of RPA debt that is
potentially occupied by croplands, as well as its respective uncertainty at one standard
deviation (Sdc):
/RPAc RPAd CRa WCa
(Eq. S3)
Where RPAc is the RPA area potentially occupied by croplands, CRa is the total cropped
area within the microwatershed, and WCa is the microwatershed area converted to
croplands.
10
2
1/Sdd RPAc RPAa
(Eq. S4)
Another set of simulations (2_Uncertainty_from_propertysize.ego in Ancillary
models at csr.ufmg.br/forestcode) demonstrated that, given the large number of
microwatersheds used in the analysis, the uncertainty in the balance of the LR depends
on: 1) the number of properties within a microwatershed; 2) the stipulated percentage of
the LR in a given microwatershed; and 3) the percentage of debt or surplus relative to the
accountable area of the microwatershed (Fig. S15). To simplify, we derived a mean
estimate of this parameter for each microwatershed such that (Fig. S16):
1.11
2.542U NPw
(Eq. S5)
Where U of the mean uncertainty and NPw is the number of properties within each
microwatershed.
Finally, we modified the above equation to account for the fact that the
uncertainty tends to zero when the number of properties is equal to 1 or the area of
vegetation remnants approaches 0% or 100% of the accountable microwatershed area.
We therefore applied the following rule to estimate the uncertainty of the forest code
balance:
If NPw = 1 or RVa <5% or RVa > 95%, then U = 0, else
1.11
2.542U NPw
(Eq. S6)
On the basis of the above calculations, we estimated that the total uncertainty arising
from using microwatersheds as a proxy for rural properties (considering two standard
deviations) was 0.6 Mha—representing just 3% of the total environmental debt under the
revised FC. This suggests that the methodology employed in this study is robust and
provides a reasonably good approximation of the FC balance.
Supplementary Text
Acknowledgments: Supported by Secretaria de Assuntos Estratégicos da Presidência da
República, the Climate and Land Use Alliance, Fundação de Amparo à Pesquisa do
Estado de Minas Gerais, Conselho Nacional de Desenvolvimento Científico e
Tecnológico, Deutsche Gesellschaft für Internationale Zusammenarbeit, SERVAMB
(Serviços Ambientais da Amazônia), the Gordon and Betty Moore Foundation, the
National Aeronautics and Space Administration (NNX11AE56G), and the National
Science Foundation (DEB 0949996 and DEB 0743703). F. Merry provided valuable
comments and insights.
11
Fig. S1.
Fig. S1. Spatial representation of the main requirements of the FC on a Google Earth 3D
view. Forest surplus represents extent of native vegetation that exceeds FC requirements.
12
Fig. S2.
Fig. S2. Reductions in the environmental debt resulting from revisions to the FC. “Forest
Code Inapplicable” refers to areas where other legislation (e.g., protected areas)
supersedes the FC. AC, Acre; AM, Amazonas; AP, Amapá; BA, Bahia; CE, Ceará; GO, Goiás; MA,
Maranhão; MG, Minas Gerais; MS, Mato Grosso do Sul; MT, Mato Grosso; PA, Pará; PI, Piauí; PR,
Paraná; RO, Rondônia; RR, Roraima; RS, Rio Grande do Sul; SP, São Paulo; SC, Santa Catarina; TO,
Tocantins. RN, Rio Grande do Norte; PB, Paraíba; PE, Pernambuco; AL, Alagoas; SE, Sergipe; ES,
Espírito Santo; RJ, Rio de Janeiro. Biomes: AM, Amazon; CE, Cerrado; CA, Caatinga; AF, Atlantic Forest;
PN, Pantanal; PP, Pampas.
13
Fig. S3.
Fig. S3. Potential for forest compensation of LR debts via the Environmental Reserve
Quotas (CRAs) within the same biome (colors) and state (horizontal axis). Positive
numbers indicate a reduction (offset) in the LR debt using CRAs and negative numbers
indicate the remaining debt after offset.
14
Fig. S4.
Fig. S4. Pasturelands suitable for agriculture, without considering climatic or land-use
zoning restrictions, per biome (colors) and state (horizontal axis).
15
Fig. S5.
Fig. S5. Potential for restoration of the LR debt (after compensation via the CRA, Fig.
S2) on pasturelands unsuitable for mechanized agricultural, per biome (colors) and state
(horizontal axis). Positive numbers indicate the area restored (debt reduction) and
negative numbers indicate the remaining LR debt that must be restored on arable lands.
16
Fig. S6.
Fig. S6. Deforestation trajectories (km2·year–1) in the three major Brazilian biomes. Note
that deforestation rates for the Atlantic Forest are depicted on a different scale than that
of the Amazon and Cerrado. Data for the Amazon come from INPE (25), for Cerrado
from LAPIG (26), and for the Atlantic Forest from Fundação SOS Mata Atlântica (27).
17
Fig. S7
Fig. S7. Dinamica EGO graphical interface showing a model diagram. Operators are
connected with arrows to establish a visual data flow. Each operator can be edited and its
output viewed.
18
Fig. S8.
accounta
ble_areas
.ego
legal_reserv
e_percent_a
tivo.ego
legal_reserv
e_percent_p
assivo.ego
cross_mu
ni_biome
s.ego
1_campo
s_em_M
A_ibama.
ego
2_add_re
mnants.e
go
legal_res
erve_per
cent.tif
1_calcPer
cent_mo
dule.ego
2_muni_to_
water_modu
lo_fiscais&M
F.ego
1_calc_m
uni_state
_PAperce
nt.ego
2_muni_t
o_water_
PApercen
t.ego
1_1_forest_code.ego
2_forest_
code_wat
er_muni.
ego
3_sum_st
ate_biom
e.ego
1_Forestcode_balance.
xlsx
legal_res
erve_per
centpass.
tif
muni_bi
omes.tif
apps_hie
rarchical.
tifve_per
cent_a
apps_no
n_hierar
chical.tif
accounta
ble_and_
non_area
s.tif
remnan
ts.tif
watershe
ds_otto_
12_reclas
s.tif
rivers.tif
water_mf_mf
_percent##.cs
v
watwershedP
A_percent##.
csv
forestcode.cs
v
Surplus
(output_01.cs
v)
Old Deficit
(output_02.cs
v)
New Deficit
(output_03.cs
v)
(Surplus)
1_biome_stat
es.csv
(Old Deficit)
2_biome_stat
es.csv
(New Deficit)
3_biome_stat
es.csv
New Deficit
(muni03.csv)
Old Deficit
(muni02.csv
Surplus
(muni01.csv)
apps_re
constitu
icao.tif
Fig. S8. Flowchart of FC main models.
19
Fig. S9.
Fig. S9. Land use map of Brazil, showing pasturelands with varying suitability for
croplands.
20
Fig. S10.
Fig. S10. Flowchart of the hilltop-mapping algorithm.
21
Fig. S11.
Fig. S11. Graphical interface of input and output ports of CalcHillTop operator.
22
Fig. S12.
Fig. S12. 3D Google Earth diagrams overlaid with HPAs mapped according to the old
and new FC definitions. Note the reduction of these areas after the revisions to the FC.
23
Fig. S13.
Fig. S13. Percentage of municipality in compliance with the new FC. Maps (a) estimated
by watershed balance, (b) estimated by using INCRA-certified private rural properties.
24
Fig. S14.
Fig. S14. Simulations with random allocation of RPA of 20 cells (black) and vegetation
remnant of 50 cells (light green) within a 100-cell watershed. Dark green corresponds to
the RPA with vegetation remnant and white to the absence of RPA and vegetation
remnant. On average, the RPA debt is approximately 10 cells.
25
Fig. S15.
Fig. S15. Uncertainty (in percentage of the microwatershed accountable area) in the
calculation of the “forest balance” relative to the number of properties within a
microwatershed and the fraction occupied by the environmental debt or surplus. (above)
80% of legal reserve, (below) 20% of legal reserve. The number of properties varies from
2 to 25 (color lines).
26
Fig. S16.
Fig. S16. Relation between the uncertainty (in percentage of the microwatershed
accountable area) of forest balance estimates per microwatershed and the number of
properties within it.
27
Table S1. Principal differences between the old and new FCs.
Old FC (Law no. 4.771, 09/15/1965
modified by MPs no. 1.1511, 06/25/1996
and 2.166-67, 08/24/2001)
New FC (Law no. 12.651, 05/25/2012,
modified by Law no. 12.727 and
Decree 7.830, 10/17/2012)
Reduction
due to
article
Legal Reserve (LR)
Conservation measures
Article no. 16
I. Located in the Legal Amazon:
a. 80% of properties located in forested
areas;
b. 35% of properties located in Cerrado
areas;
c. 20% of properties located in grassland
areas;
II. Located in other regions of Brazil:
a. 20% of the property
Article no. 12
The same.
̶
Restoration measures*
Article no. 16 §5
50% of properties, exclusively for the
purpose of regularization (coming into
compliance), where designated as a
consolidation zone by the Ecological-
Economic Zoning – ZEE.
Article no. 13
The same.
̶
Otherwise Article nº 16 § nº 5, same as
conservation.
Article no. 15
RPAs now count toward the required
percentage of a property’s Legal
Reserve, as long as the property
adheres to the Rural Environmental
Registry (CAR).
≈4 Mha
(Million
hectares)
Article no. 67
In properties with up to 4 fiscal
modules, the Legal Reserve consists
of the area occupied by native
vegetation as of 22 July 2008.
≈17 Mha
Article no. 12 §4
50% in Amazonian municipalities with
more than 50% of their areas
occupied by public conservation
areas or indigenous lands.
≈1 Mha (both
articles)
Article no. 12 §5
50% in Amazonian municipalities
when the state has an approved
Ecological-Economic Zoning Plan
and more than 65% of its territory
occupied by public conservation
areas or indigenous lands.
28
Riparian Preservation Areas (RPAs)
Conservation measures
Article no. 2
Buffers:
30 m, for watercourses less than 10 m wide;
50 m, for watercourses from 10 m to 50 m
wide;
100 m, for watercourses from 50 m to 200 m
wide;
200 m, for watercourses from 200 m to 600
m wide;
500 m, for watercourses wider than 600 m;
Areas surrounding natural lakes and ponds,
with minimum width of:
100 m, in rural areas, except for water bodies
with surface areas of up to 20 ha, whose
buffer width shall be 50 m.
Article no. 4
The same.
̶
Restoration measures*
Same as conservation.
Article no. 61 §6
≤ 1 fiscal module: restoration of 5-m
riparian buffer from the edge of the
regular stream channel, regardless of
the width of the watercourse.
From 1 to 2 fiscal modules: 8 m.
From 2 to 4 fiscal modules: 15 m.
From 4 to 10 fiscal modules: 20 m for
watercourses up to 10 m wide. For
wider streams, the rules for
properties larger than 10 fiscal
modules apply.
≥10 fiscal modules: half the width of
the stream channel, observing a
minimum of 30 m and maximum of
100-m buffers from the edge of the
regular stream channel.
For lakes, idem up to 4 modules, 30 m
for larger modules.
≈8 Mha
Article no. 61-B
<2 fiscal modules: RPA not to exceed
10% of the total area of the property
or
2 to 4 fiscal modules: RPA not to
exceed 20% of the total area of the
property.
≈0 Mha
29
Hill Top Preservation Areas (HPAs)
Conservation measures
CONAMA resolution 303, 03/20/2002
Hilltops with a minimum height of 50 m
(measured from the base), maximum
height of 300 m, and mean slope ≥17%.
Hilltops are defined as areas situated above
two-thirds of the total height.
Elevation higher than 1800 m.
Areas situated above two-thirds of the
height of hills and ridges, with height
>300 m, mean slope ≥30%, and hilltops
<500 m away.
Mesas with more than 10 ha and slope
<10%, characterizing plateaus with
elevation >600 m.
Mesa escarpment with min. horizontal
width of 100 m.
Baseline defined as the horizontal surface
of the adjacent plain or water surface, or
upon the nearest saddle point in
undulated terrain.
Article no. 4
Hilltops and ridges with minimum
height of 100 m and mean slope
≥25%. Does not specify maximum
height.
Hilltops defined as areas situated above
two-thirds of the total height.
Elevation higher than 1800 m.
All areas with slope ≥45%.
Mesa escarpment with min. horizontal
width of 100 m.
Baseline defined as the horizontal
surface of the adjacent plain or water
surface, or on the nearest saddle
point in undulated terrain.
87%
Restoration measures
Same as conservation.
Absent.
Not
quantified
*Restoration applies to the consolidated rural area―that is, the portion of the rural
property with deforestation (anthropic occupation) predating 22 July 2008.
30
Table S2.Environmental debt in LRs and reduction in their areas (ha) under the revised FC, summarized by biome and state.
Values in brackets correspond to the percent reduction relative to requirements under the former FC.
States\Biomes
Amazon
Atlantic Forest
Cerrado
Caatinga
Pampas
Pantanal
State totals
Amapá
- (100%)
-
-
-
-
-
- (100%)
Roraima
- (100%)
-
-
-
-
-
- (100%)
Piauí
-
-
- (100%)
- (100%)
-
-
- (100%)
Amazonas
- (100%)
-
-
-
-
-
- (100%)
Distrito Federal
-
-
11.1E+3 (56%)
-
-
-
11.1E+3 (56%)
Rio Grande do Norte
-
6.9E+3 (70%)
-
11.2E+3 (53%)
-
-
18.1E+3 (61%)
Ceará
-
-
-
20.2E+3 (57%)
-
-
20.2E+3 (57%)
Paraíba
-
23.3E+3 (56%)
-
- (100%)
-
-
23.3E+3 (67%)
Sergipe
-
25.1E+3 (79%)
-
15.7E+3 (76%)
-
-
40.8E+3 (78%)
Santa Catarina
-
46.5E+3 (88%)
-
-
16.0E+0 (81%)
-
46.5E+3 (88%)
Acre
57.8E+3 (90%)
-
-
-
-
-
57.8E+3 (90%)
Pernambuco
-
39.2E+3 (74%)
-
27.1E+3 (84%)
-
-
66.3E+3 (79%)
Alagoas
-
71.2E+3 (55%)
-
24.5E+3 (75%)
-
-
95.7E+3 (62%)
Rio de Janeiro
-
120.8E+3 (57%)
-
-
-
-
120.8E+3 (57%)
Espírito Santo
-
179.0E+3 (64%)
-
-
-
-
179.0E+3 (64%)
Rondônia
240.6E+3 (85%)
-
102.0E+0 (56%)
-
-
-
240.7E+3 (85%)
Rio Grande do Sul
-
217.7E+3 (77%)
-
-
287.3E+3 (49%)
-
505.0E+3 (67%)
Goiás
-
81.6E+3 (39%)
431.6E+3 (54%)
-
-
-
513.2E+3 (52%)
Bahia
-
563.6E+3 (52%)
22.5E+3 (26%)
233.3E+3 (71%)
-
-
819.4E+3 (60%)
Tocantins
604.2E+3 (31%)
-
238.3E+3 (53%)
-
-
-
842.4E+3 (39%)
Mato Grosso do Sul
-
433.0E+3 (28%)
559.8E+3 (31%)
-
-
- (100%)
992.8E+3 (28%)
Minas Gerais
-
764.2E+3 (68%)
233.8E+3 (62%)
- (100%)
-
-
998.0E+3 (67%)
Maranhão
1.1E+6 (61%)
-
27.1E+3 (91%)
- (100%)
-
-
1.1E+6 (64%)
Paraná
-
1.2E+6 (48%)
13.6E+3 (40%)
-
-
-
1.2E+6 (48%)
Pará
1.3E+6 (68%)
-
- (100%)
-
-
-
1.3E+6 (68%)
São Paulo
-
1.0E+6 (43%)
522.6E+3 (39%)
-
-
-
1.5E+6 (42%)
Mato Grosso
3.9E+6 (41%)
-
1.6E+6 (34%)
-
-
37.7E+3 (51%)
5.6E+6 (40%)
Totals
7.2E+6 (59%)
4.8E+6 (57%)
3.7E+6 (44%)
332.0E+3 (73%)
287.3E+3 (49%)
37.7E+3 (3%)
16.3±1E+6
(56%)
31
Table S3. Environmental debt in RPAs and reduction in their areas (ha) under the revised FC, summarized by biome and
state. Values in brackets correspond to the percent reduction relative to the former FC.
States\Biomes
Cerrado
Atlantic Forest
Amazon
Caatinga
Pampas
Pantanal
State totals
Distrito Federal
3.6E+3 (46%)
-
-
-
-
-
3.6E+3 (46%)
Amapá
-
-
6.8E+3 (66%)
-
-
-
6.8E+3 (66%)
Roraima
-
-
11.5E+3 (57%)
-
-
-
11.5E+3 (57%)
Acre
-
-
17.4E+3 (66%)
-
-
-
17.4E+3 (66%)
Sergipe
-
15.4E+3 (82%)
-
6.5E+3 (88%)
-
-
21.9E+3 (84%)
Alagoas
-
26.7E+3 (80%)
-
8.7E+3 (86%)
-
-
35.4E+3 (82%)
Paraíba
-
8.5E+3 (82%)
-
30.5E+3 (82%)
-
-
39.0E+3 (82%)
Rio Grande do Norte
-
4.8E+3 (82%)
-
46.5E+3 (81%)
-
-
51.3E+3 (81%)
Espírito Santo
-
59.5E+3 (82%)
-
-
-
-
59.5E+3 (82%)
Rio de Janeiro
-
59.9E+3 (74%)
-
-
-
-
59.9E+3 (74%)
Pernambuco
-
20.1E+3 (84%)
-
46.3E+3 (81%)
-
-
66.3E+3 (82%)
Piauí
19.5E+3 (69%)
-
-
47.1E+3 (70%)
-
-
66.7E+3 (70%)
Santa Catarina
-
67.2E+3 (82%)
-
-
21.4E+0 (83%)
-
67.2E+3 (82%)
Rondônia
31.2E+0 (35%)
-
75.4E+3 (48%)
-
-
-
75.4E+3 (48%)
Ceara
-
-
-
76.5E+3 (81%)
-
-
76.5E+3 (81%)
Amazonas
-
-
120.7E+3 (52%)
-
-
-
120.7E+3 (52%)
Tocantins
104.1E+3 (52%)
-
27.8E+3 (73%)
-
-
-
131.8E+3 (59%)
Maranhão
58.8E+3 (77%)
-
74.1E+3 (76%)
2.3E+3 (77%)
-
-
135.1E+3 (76%)
Paraná
4.0E+3 (19%)
227.8E+3 (47%)
-
-
-
-
231.7E+3 (47%)
Para
196.9E+0 (33%)
-
307.6E+3 (66%)
-
-
-
307.8E+3 (66%)
Rio Grande do Sul
-
94.6E+3 (81%)
-
-
213.8E+3 (63%)
-
308.4E+3 (71%)
Mato Grosso do Sul
241.4E+3 (34%)
79.1E+3 (30%)
-
-
-
22.2E+3 (50%)
342.7E+3 (34%)
Bahia
65.0E+3 (60%)
154.3E+3 (80%)
-
144.7E+3 (82%)
-
-
363.9E+3 (79%)
São Paulo
132.0E+3 (19%)
248.7E+3 (45%)
-
-
-
-
380.7E+3 (38%)
Goiás
373.9E+3 (46%)
31.4E+3 (26%)
-
-
-
-
405.3E+3 (44%)
Mato Grosso
227.7E+3 (46%)
-
257.9E+3 (43%)
-
-
20.0E+3 (49%)
505.6E+3 (45%)
Minas Gerais
322.0E+3 (61%)
290.0E+3 (71%)
-
8.7E+3 (64%)
-
-
620.7E+3 (66%)
Totals
1.6E+6 (51%)
1.4E+6 (70%)
899.2E+3 (60%)
417.7E+3(81%)
213.8E+3 (63%)
42.2E+3 (49%)
4.5±1E+6
(65%)
32
Table S4. Environmental surplus or area of land (ha) that is legally available for
conversion (deforestation) from native vegetation to other uses, summarized by
biome and state.
States/Biomes
Cerrado
Caatinga
Amazon
Pantanal
Atlantic
Forest
Pampas
Totals
Espírito Santo
-
-
-
-
44.6E+3
-
44.6E+3
Distrito Federal
44.6E+3
-
-
-
-
-
44.6E+3
Alagoas
-
51.5E+3
-
-
6.0E+3
-
57.5E+3
Rio de Janeiro
-
-
-
-
127.7E+3
-
127.7E+3
Sergipe
-
139.9E+3
-
-
9.0E+3
-
148.9E+3
Rondônia
404.1E+0
-
310.8E+3
-
-
-
311.2E+3
Paraná
23.5E+3
-
-
-
412.5E+3
-
435.9E+3
São Paulo
31.6E+3
-
-
-
510.2E+3
-
541.8E+3
Acre
-
-
831.2E+3
-
-
-
831.2E+3
Amapá
-
-
913.0E+3
-
-
-
913.0E+3
Santa Catarina
-
-
-
-
1.1E+6
48.3E+0
1.1E+6
Rio Grande do Norte
-
1.4E+6
-
-
5.6E+3
-
1.4E+6
Paraíba
-
1.5E+6
-
-
2.0E+3
-
1.5E+6
Roraima
-
-
1.9E+6
-
-
-
1.9E+6
Pernambuco
-
2.0E+6
-
-
16.7E+3
-
2.0E+6
Pará
1.2E+3
-
3.0E+6
-
-
-
3.0E+6
Rio Grande do Sul
-
-
-
-
664.0E+3
3.0E+6
3.7E+6
Goiás
4.5E+6
-
-
-
1.1E+3
-
4.5E+6
Ceará
-
5.1E+6
-
-
-
-
5.1E+6
Mato Grosso do Sul
1.1E+6
-
-
5.3E+6
21.6E+3
-
6.4E+6
Tocantins
6.4E+6
-
8.5E+3
-
-
-
6.4E+6
Maranhão
6.7E+6
154.7E+3
77.1E+3
-
-
-
6.9E+6
Minas Gerais
6.4E+6
228.2E+3
-
-
281.8E+3
-
6.9E+6
Mato Grosso
4.0E+6
-
1.6E+6
2.0E+6
-
-
7.6E+6
Amazonas
-
-
10.5E+6
-
-
-
10.5E+6
Piauí
4.9E+6
7.1E+6
-
-
-
-
12.0E+6
Bahia
5.9E+6
8.1E+6
-
-
176.8E+3
-
14.1E+6
Totals
39.9E+6
25.8E+6
8.6E+6
7.3E+6
3.4E+6
3.0E+6
88±6E+6
The grand total of the FC surplus does not include the surplus of the state of Amazonas
(10.5 Mha), because large blocks of forests in this state remain undesignated public land
to date (15).
33
Table S5. Main activities supported by the ABC (low-carbon agriculture)
investment program. ABC provides US$ 1.5 billion in annual subsidized loans to the
above activities in order to reduce Green House Gas emissions from agriculture (28).
Activities
Recovery of degraded pasture
Crop, livestock, and forestry integration
FC compliance including forest restoration
Treatment of agriculture residues for bioenergy generation
Organic agriculture
Direct seeding (i.e., no-till agriculture)
Improved forestry management, especially for charcoal production
Biological nitrogen fixation
Oil palm (Dedenzeiro) plantation on degraded pastureland
34
Table S6. Data sets used in the FC analysis. PNLT: National Plan for Logistics and Transport, IBGE:
Brazilian Institute for Geography and Statistics, SFB: Serviço Florestal Brasileiro, IPAM: Instituto de Pesquisas
Ambientais da Amazônia, MMA: Ministry of the Environment, INCRA: Institute for Agrarian Reform, CSR: Center
for Remote Sensing, UFMG, ANA: Brazilian National Water Agency, NASA: National Aeronautics and Space
Administration, ICMBio: Instituto Chico Mendes de Conservação da Biodiversidade, PRODES: Project for Monitoring
Deforestation in the Amazon (25), PMDBBS: Project for Monitoring Deforestation in the Brazilian Biomes via
Satellite (29) Terra Class (30), SOS: Fundação SOS Mata Atlântica (27). Leite et al. 2012 (16). PAM: Municipal
Agricultural Production.
Theme
Map
Source
Date
Scale
Infrastructure
Railroad network
PNLT
2009
1:1,000,000
Road network
PNLT
2009
1:1,000,000
Demography and
administrative limits
Urban areas within Brazilian census
tracts
IBGE
2010
1:100,000
Municipalities of Brazil
IBGE
2010
1:100,000
States of Brazil
IBGE
2010
1:100,000
Brazilian Legal Amazon
MMA
2011
1:5,000,000
Certified rural properties
INCRA
2013
1:100,000
Protected Areas
Protected Areas including indigenous
reserves, sustainable use areas, and
strict protected areas
CSR
2012
1:100,000
(APA) Areas of Environmental
Preservation
CSR
2012
1:100.000
ZEE
Ecological and economic zoning
MMA
2007
1:5.000.000
Hydrography
Ottobacias (watersheds) with order up
to 12
ANA
2010
1:100.000
Hydrographic network
ANA
2010
1:100.000
Perennial rivers with two margins,
lakes, and reservoirs
IBGE
2006
1:1.000.000
Physiography
Principal vegetation classes in Brazil
IBGE
2002
1:5,000,000
Biomes of Brazil
IBGE
2011
1:5,000,000
Potential biomass of the original
vegetation
Leite et al
(16)
2012
1:5,000,000
Brazilian Soils Map, classified
according to the Brazilian System for
Soil Classification developed by
EMBRAPA
IBGE
1999
1:5,000,000
Topography from Space Shuttle Radar
Topographic Mission
NASA
2011
1:250,000
Remaining native
vegetation
Remnants in the Cerrado biome
PMDBBS
1:250,000
Remnants in the Pampas biome
PMDBBS
2009
1:250,000
Remnants in the Caatinga biome
PMDBBS
2009
1:250,000
Remnants in the Pantanal biome
PMDBBS
2009
1:250,000
Remnants in the Amazon biome
PRODES
2011
1:250,000
Remnants in the Atlantic Forest biome
SOS
2009
1:250,000
Secondary vegetation in Amazonia
TERRACLAS
2012
1:100,000
Cerrado Deforestation from 2009 to
2010
PMDBBS
2011
1:250,000
Census data
Agricultural Census
IBGE
2006
1:100,000
PAM
IBGE
2011
1:100,000
35
Table S7. RPA widths associated with ANA watershed hierarchy.
Order
Width (m)
1
240
2
180
3
90
4
60
5
60
6
30
7
30
8
30
9
30
≥10
30
36
Table S8. Absolute and relative changes in areas of HPAs from the old to the new
FC.
State
Old definition
(ha)
New definition
(ha)
Relative change
(%)
Acre
120.6E+3
4.0E+3
–97
Alagoas
332.5E+3
29.6E+3
–91
Amazonas
1.5E+6
159.0E+3
–89
Amapá
931.2E+3
42.7E+3
–95
Bahia
2.9E+6
403.4E+3
–86
Ceará
762.5E+3
170.1E+3
–78
Brasília - DF
19.4E+3
116.6E+0
–99
Espirito Santo
965.8E+3
228.4E+3
–76
Goiás
1.5E+6
206.8E+3
–86
Maranhão
2.1E+6
263.2E+3
–88
Minas Gerais
5.9E+6
945.2E+3
–84
Mato Grosso do Sul
145.2E+3
94.9E+3
–35
Mato Grosso
2.1E+6
306.7E+3
–86
Pará
7.3E+6
483.2E+3
–93
Paraíba
343.2E+3
43.6E+3
–87
Pernambuco
620.8E+3
92.6E+3
–85
Piauí
1.3E+6
238.0E+3
–82
Paraná
1.5E+6
141.6E+3
–90
Rio de Janeiro
767.0E+3
206.1E+3
–73
Rio Grande do Norte
186.3E+3
36.4E+3
–80
Rondônia
578.5E+3
65.2E+3
–89
Roraima
1.3E+6
194.6E+3
–85
Rio Grande do Sul
1.2E+6
181.7E+3
–85
Santa Catarina
1.2E+6
207.4E+3
–82
Sergipe
88.5E+3
2.5E+3
–97
São Paulo
1.3E+6
178.5E+3
–87
Tocantins
917.9E+3
162.5E+3
–82
Totals
38.0E+6
5.1E+6
–87
37
Table S9. Comparison between data of ANA 12 watersheds and INCRA private
properties.
FC
applicable
area (ha)
No. of
units
Average
size
Distribution by Fiscal Modules in terms of area (%)
≤1
≤2
≤3
≤4
>4
& ≤10
>10
ANA 12
watersheds
612,928,419
166,443
3,683
14
24
29
33
16
51
INCRA
properties
91,187,213
62,897
1,595
7
12
15
18
17
65
Ratio
6.72
2.65
2.31
38
Table S10. FC balance using ANA 12 watersheds and INCRA private properties
and relative deviance.
After area adjustment
After agrarian
adjustment
(Unit in ha)
Surplus
Old
deficit
New
deficit
Surplus
Old
deficit
New
deficit
New
deficit
ANA 12
Watersheds
98.6E+6
49.7E+6
20.7E+6
98.6E+6
49.7E+6
20.7E+6
20.7E+6
INCRA
properties
16.4E+6
8.0E+6
6.5E+6
110.3E+6
53.6E+6
44.0E+6
2.9E+6
19.5E+6
–11%
–7%
6%
39
References and notes
1. Millennium Ecosystem Assessment, Ecosystem and Human Well-Being: Synthesis
(Island Press, Washington, DC, 2005);;
www.millenniumassessment.org/en/index.aspx).
2. Federal Law 12.727, 17 October 2012; www.planalto.gov.br/ccivil_03/_Ato2011-
2014/2012/Lei/L12727.htm
3. A. Lees, C. Peres, Conservation value of remnant riparian forest corridors of
varying quality for Amazonian Birds and mammals. Conserv. Biol. 22, 439
(2008).
4. R. Rodrigues et al., Large-scale ecological restoration of high-diversity tropical
forests in SE Brazil. For. Ecol. Manag. 261, 1605 (2011).
5. M. Ribeiro et al., The Brazilian Atlantic Forest: How much is left, and how is the
remaining forest distributed? Implications for conservation. Biol. Conserv. 142,
1141 (2009).
6. R. Cury, O. Carvalho, “Manual para restauro florestal: florestas de transição”
(Série Boas Práticas, 5, IPAM, Canarana, 2011);
www.ipam.org.br/download/livro/Manual-para-Restauracao-Florestal/591).
7. C. Stickler et al., Dependence of hydropower energy generation on forests in the
Amazon Basin at local and regional scales. Proc. Natl. Acad. Sci. U.S.A. (2013),
doi: 10.1073/pnas.1215331110.
8. Project Oasis, Fundação Grupo Boticário.
www.fundacaogrupoboticario.org.br/en/what%20we%20do/oasis/pages/default.
aspx.
9. R. Rajão, A. Azevedo, M. C. C. Stabile, Institutional subversion and
deforestation: learning lessons from the system for the environmental licencing of
rural properties in Mato Grosso. Public Admin. Dev. 32, 229 (2012).
10. Aliança da Terra, aliancadaterra.org.br.
11. Round Table on Responsible Soy Association, responsiblesoy.org.
12. G. Sparovek, G. Berndes, I. Klug, A. Barretto, Brazilian agriculture and
environmental legislation: status and future challenges. Environ. Sci. Technol. 44,
6046-6053 (2010).
13. B. S. Soares-Filho, H. Rodrigues, M. Follador, A hybrid analytical-heuristic
method for calibrating land-use change models. Environ. Modell. Software 43, 80-
87 (2013).
14. Instituto Brasileiro de Geografia e Estatística, Censo Agropecuário de 2006
(IBGE Publication, Rio de Janeiro, 2006);
http://biblioteca.ibge.gov.br/visualizacao/periodicos/51/agro_2006.pdf.
15. Serviço Florestal Brasileiro, Instituto de Pesquisa Ambiental da Amazônia,
Florestas nativas de Produção Brasileiras (Report, Brasília, IPAM 2011);
www.florestal.gov.br/index.php?option=com_k2&view=item&task=download&i
d=121.
16. C. Leite, M. Costa, B. S. Soares-Filho, L. B. V. Hissa, Historical land use change
and associated carbon emissions in Brazil from 1940 to 1995. Global
Biogeochem. Cycles 26, 2011-2029 (2012).
40
17. R. A. Houghton, K. T. Lawrence, J. Hackler, L. S. Brown, The spatial distribution
of forest biomass in the Brazilian Amazon: A comparison of estimates. Global
Change Biol. 7, 731–746 (2001).
18. R. A. Houghton, et al. Annual fluxes of carbon from deforestation and regrowth
in the Brazilian Amazon. Nature 403, 301–304 (2000).
19. L. A Carvalho, O novo código florestal comentado: Artigo por artigo, com as
alterações trazidas pela lei 12.272, de 17/10/2012 e referências ao Decreto 7830,
de 17/10/2012 (Juruá, Curitiba, 2013).
20. C. Gouvello et al, Brazil Low-carbon Country Case Study (World Bank,
Washington, DC, 2010;
http://siteresources.worldbank.org/BRAZILEXTN/Resources/Brazil_LowcarbonS
tudy.pdf).
21. D. C. Nepstad, C. M. Stickler, B. S. Soares-Filho, F. Merry, Interactions among
Amazon land use, forests and climate: prospects for a near-term forest tipping
point. Philos. Trans. R. Soc. London B 363, 1737–1746 (2008).
22. “Sugarcane monitoring through satellite images” (CANASAT Project, INPE;
http://www.dsr.inpe.br/laf/canasat).
23. Instituto Brasileiro de Geografia e Estatística, Produção Agrícola Municipal:
culturas temporárias e permanentes (IBGE Publication, Rio de Janeiro, 2011;
http://www.ibge.gov.br/home/estatistica/economia/pam/2011/).
24. G. F. Bonham-Carter, Geographic Information Systems for Geoscientists:
Modeling with GIS (Pergamon, New York, 1994).
25. Instituto Nacional de Pesquisas Espaciais, Projeto PRODES – monitoramento da
floresta amazônica brasileira por satélite (INPE, São Paulo, 2013;
http://www.obt.inpe.br/prodes/index.php).
26. LAPIG, Laboratório de Processamento de Imagens e Geoprocessamento, Dados
Vetoriais de alertas de desmatamento no período de 2002 a 2012 (Universidade
Federal de Goiás, Goiânia, 2013;
http://www.lapig.iesa.ufg.br/lapig/index.php/produtos/dados-vetoriais).
27. Fundação SOS Mata Atlântica, Instituto Nacional de Pesquisas Espaciais, Atlas
dos remanescentes florestais da Mata Atlântica - período 2011-2012 (São Paulo,
2013; http://www.sosma.org.br/wp-content/uploads/2013/06/atlas_2011-
2012_relatorio_tecnico_2013final.pdf).
28. Banco Nacional de Desenvolvimento Econômico e Social - BNDES, Programa
para Redução da Emissão de Gases de Efeito Estufa na Agricultura – Programa
ABC (BNDES, Rio de Janeiro, 2013;
http://www.bndes.gov.br/SiteBNDES/bndes/bndes_pt/Institucional/Apoio_Financ
eiro/Programas_e_Fundos/abc.html).
29. IBAMA, Project for Satellite-based Monitoring of Deforestation in the Brazilian
Biomes (PMDBBS, 2012; http://siscom.ibama.gov.br/monitorabiomas/).
