jose guzman possible causes and reducers of … causes...subsidies for amazonian development,...
TRANSCRIPT
Jose Guzman_Possible Causes and Reducers of Deforestation in the Brazilian Legal Amazon:
The Role of Environmental Policies from 2005 to 2012_ PA 388K-59935_Fall_2015
Page 1
Possible Causes and Reducers of Deforestation in the Brazilian Legal Amazon: The Role of
Environmental Policies from 2005 to 2012
Jose Guzman
University of Texas at Austin
December 9, 2015
Introductory GIS
Jose Guzman_Possible Causes and Reducers of Deforestation in the Brazilian Legal Amazon:
The Role of Environmental Policies from 2005 to 2012_ PA 388K-59935_Fall_2015
Page 2
Executive Summary
Deforestation in the Brazilian Legal Amazon has declined by up to 82 percent in the last decade,
compared to previous years and the 2004 peak in deforestation levels.i The report will spatially
analyze the possible causes of this reversal. In general, the report expands on previous studies
analyzing the effect of large scale cattle ranching, population change and density on increased
deforestation; and the effects of environmental policies, such as priority municipalities, increased
monitoring mechanisms, protected areas and indigenous lands on decreasing deforestation rates
from 205 to 2012.
The report provides descriptive maps on deforestation for the years 2005, 2007, 2010, and 2012;
population and cattle density for the year 2012; population and number of cattle changes from
2005 to 2012; and a map of priority municipalities, and embargoes/fines from 2005 to 2012. The
report also makes use og spatial analysis methods such as Ordinary Least Squares (OLS), and
Geographically Weighted Regression (GWR), which use variables that cause (cattle, roads, and
population variables), and reduce deforestation (embargoes/fines, protected areas, and
indigenous lands).
The spatial analysis results show a positive correlation between cattle density for 2012,
population changes from 2005 to 2012, and priority areas, and aggregate deforestation
percentages for the years studied. The results also show a negative correlation between protected
areas and indigenous lands, and aggregate deforestation percentage levels, indicating that these
areas significantly reduce deforestation in the Amazon. Recommendations for further policy
based on the results include further analysis on the effectiveness of illegal deforestation
monitoring mechanisms, and enhancement or decentralization of management and administration
efforts for protected areas in indigenous lands.
Jose Guzman_Possible Causes and Reducers of Deforestation in the Brazilian Legal Amazon:
The Role of Environmental Policies from 2005 to 2012_ PA 388K-59935_Fall_2015
Page 3
I. Introduction and Literature Review
The Amazon biome occupies almost 50 percent of the Brazil distributed over the Acre, Amapá,
Amazonas, Pará, and Roraima States, 34 percent of the Maranhão and 9 percent Tocantins
States. The biome is known for its hot and humid climates, the predominance of tropical forests,
and its biological diversity. The biome harbors more than 30 thousand species of plants, 1.8
thousand species of continental fish, 1.3 thousand species of birds, 311 species of mammals, and
164 species of amphibians. The Brazilian government created the Amazonia Legal, one of the
three socio-geographic administrative units in the country through Federal Law No. 5.173 and
later alterations that integrated Mato Grosso with Art. 45 of the Complementary Federal Law No.
3111 of 1977, and the state of Tocantins and Transitory Dispositions of the Federal Constitution
(Art. 13)12 of 1988. The Amazonia Legal region is made up of about 5,016,136.3 square
kilometers and encompasses the states of Acre, Amapá, Amazonas, Pará, Rondônia, Roraima,
Tocantins, Mato Grosso, and part of Maranhão. The area represents 59 percent of the Brazilian
territory and has more than 24 million inhabitants. More than 300 thousand Brazilian Indians
who belong to 170 indigenous groups inhabit an area of 1,085,890 square kilometers, or 21.7
percent of Amazonia Legal (Socioenvironmental Institute).
Deforestation of the Brazilian Amazon began in the 1960s through government incentives to
expand and develop in the Amazon area, as well as the creation transportation projects such as
the Trans-Amazonian Highwayii. During the 1970s and 1980s the government increased
subsidies for Amazonian development, invested in road infrastructure, created unclear land
tenure, and created policies that promoted speculation by rewarding deforesters with land titles.iii
As of 2013, 3,341,908 square kilometers of the Brazilian Amazon forest covers remain out of the
4,100,000 square kilometers estimated in the 1970s, representing 81.5 percent of the remaining
forest cover and a total of 758,092 square miles of forest loss since 1970. Deforestation rates
have fluctuated since the 1970s and annual forest loss reached a peak of 27,772 square miles of
forest loss in 2004.iv
The causes of deforestation in the Amazon vary according to different studies, but most point out
to both small and large cattle ranching and farming as the main drivers of deforestation.
Fearnside states, for example, that the number of properties censused in each size class explains
74% of the variation in deforestation rate among the nine Amazonian states, further indicating
through multiple regressions that 30% of the clearing in 1991 can be attributed to small farmers
(properties < 100 hectares in area), and the remaining 70% to either medium or large ranchers.v
Bouchardet and Porsse use spatial analysis to study the spatial pattern of deforestation in a cross-
section and longitudinal perspective using global and local analysis estimating Moran’s I statistic
and Local Indicator of Spatial Association clusters maps. The authors applied statistics to three
outcome variables: absolute deforestation, fraction of the total area deforested in each year and
the cumulative sum of the second variable. The studied resulted in a presence of high spatial
correlation dependency, and persistent spatial concentration of deforestation. The authors
conclude that policy control mechanisms have achieved relative success to reduce the average
deforestation but have not influenced its spatial dispersion.vi
Jose Guzman_Possible Causes and Reducers of Deforestation in the Brazilian Legal Amazon:
The Role of Environmental Policies from 2005 to 2012_ PA 388K-59935_Fall_2015
Page 4
Sydenstricker-Neto utilizes a mixed methodology to investigate the effect of population on
deforestation in the Brazilian Amazon. The GIS analysis section of the study uses among
participatory mapping methods, analysis of land use change. The final land-cover classification
scheme of the analysis included seven categories: primary forest, secondary forest, transition
(i.e., areas recently cleared burned or not, and not in use), pasture, crops, bare soil, and water.
The author concludes that population does affect deforestation. However, population alone is not
the main factor; it has to be conceptualized as being mediated by economic, social, cultural, and
institutional factors.
Other factors contributing to deforestation include soil quality and vegetation density, and factors
affecting transport costs (such as distance to major markets and both own- and neighboring-
county roads) and development projects.vii Kaimowitz et al drew similar conclusions by applying
a spatial regression model to analyze deforestation in Santa Cruz, Bolivia. The authors used as
the dependent variable the probability that a given location covered with forest in 1989 still had
forest in 1994; distance to roads, railroads, trails, and markets, “land use potential” (soils and
topography), rainfall, among others were used as independent variables. The study found that the
forests in Santa Cruz that existed in 1989 were more likely to be deforested over the next five
years if they were closer to roads, trails, railways, markets, and areas that had already been
deforested in 1989.viii
Logging is another factor that contributes to deforestation in the region by increasing
susceptibility to forest fires. Climatic changes or events such as dry seasons or events such as El
Niño also compound to deforestation rates by increasing the chances of forest fires.ix
Deforestation is associated to different impacts in the Brazilian Amazon. Deforestation affects
soil erosion, nutrient depletion, and soil compaction, which in turn affects land productivity.
Deforestation also leads to changes in hydrological patterns, such as watershed functions and
precipitation patterns. Biodiversity loss is also a negative outcome of deforestation by destroying
the natural habitats of highly endemic species. Furthermore, deforestation caused by fires
releases trapped carbon dioxide in trees, with negative implications for world climate change.x
Although Brazil has had one of the fastest rates of deforestation in the world (Indonesia
surpassed Brazil in 2014), the country has experienced a decrease in deforestation rates. When
looking at deforestation data from Brazil's National Institute for Space Research (INPE), which
has monitored deforestation rates since the 1970s, deforestation rates have decreased by more
than 70 percent since 2004.
Similar to the causes of deforestation, several factors appear influence current trends towards
decreased deforestation levels, protected areas and indigenous territories being some of the main
factors.xi Article 68 of the new Brazilian Constitution assured lands to indigenous Brazilians,
recognizing land claims in forested areas and ratifying traditional forms of natural resource
management and ownership. The constitutional article also expanded concepts of inhabited
conservation areas. This concept can be contrasted to the notion of uninhabited, pristine
landscapes and protected areas in the United States. These changes have resulted in an increase
Jose Guzman_Possible Causes and Reducers of Deforestation in the Brazilian Legal Amazon:
The Role of Environmental Policies from 2005 to 2012_ PA 388K-59935_Fall_2015
Page 5
in conservation areas such as protected areas and recognized indigenous areas, which now
account for more than 50 percent of the remaining forests of the Brazilian Amazon.
Apart from increases in protected areas, Nepstad et al developed and examined other
hypothetical causes for the decline in deforestation rates. The authors include higher risks for
landowners of reduced access to markets and finance, increased and harsher legal restrictions and
fines on illegal activities, benefits through payments for ecosystem services, price premiums
from certification, and access to new credit lines by avoiding deforestation, stalled highway
projects, decline of profitability of soy production, increased beef production on already cleared
lands, and a reduction in cattle herd sizes.xii Macedo et al provide further evidence that policies
promoting efficient production of already cleared lands while restricting deforestation can result
in increased soy productivity that does not affect deforestation.xiii
Assunção and Gandour analyzed possible policies that led to the reduction of annual
deforestation levels starting in 2004. The authors single out Presidential Decree 6.321/2007 as
one of the policies that has helped reduce deforestation levels. The decree established
municipalities in the Legal Amazon classified as needing priority action to prevent, monitor, and
combat illegal deforestation. Priority municipalities were chosen following three criteria: (i) total
deforested area; (ii) total area deforested in the past three years; and (iii) increase in deforestation
rate in at least three of the past five years. Priority municipalities have experienced stricter
command and control policies and monitoring of irregular activities. Monitoring measures
include harsher registration, licensing requirements, and the revision of private land titles to
avoid fraudulent documents and illegal occupation. Assunção and Gandour concluded that
“changes to conservation policies implemented beginning in 2004 and 2008 significantly
contributed to the curbing of deforestation rates, even after controlling for different sorts of price
effects”.xiv
II. Research Questions
This report will address the following questions:
What factors help explain the decrease in deforestation rates in the Brazilian Amazon in the past
decade? How do demographic trends, herd sizes, and environmental policies affect
deforestation?
Are monitoring efforts effectively allocated in the Brazilian Amazon’s municipalities?
This report is intended to demonstrate the positive impact that environmental policies have in
reducing deforestation rates in the Brazilian Amazon.
III. Methods and Data collection
1. PRODES –
The PRODES project, operated by Brazil’s National Institute for Space Research (INPE),
monitors annual rates of deforestation in the Brazilian Amazon through satellite imagery. The
project measures annual rates based on increments in deforestation using Landsat satellite images
(20 to 30 meter spatial resolution available every 16 days) in a combination that seeks to
Jose Guzman_Possible Causes and Reducers of Deforestation in the Brazilian Legal Amazon:
The Role of Environmental Policies from 2005 to 2012_ PA 388K-59935_Fall_2015
Page 6
minimize the problem of cloud cover. Environmental organizations, such as Imazon, publish
alternative results through their Deforestation Alert System (SAD), based on MODIS and
Landsat imagery. National and international scientists nevertheless consider PRODES estimates
as reliable for analysis.xvDeforestation analysis for the report will thus be based only on
PRODES deforestation data to retain consistency in the deforestation rates analyzed in each year
of study.
2. The Brazilian Institute on Geography and Statistics (IBGE) –
IBGE provides a rich data set, including the Legal Amazon’s municipality shapefiles, and
population data for each municipality, which this report will include to calculate population
density per municipality for the years 2005 and 2012 (measured in squared kilometers), and
population percent change from 2005 to 2012. IBGE also provides cattle numbers per
municipality over the years 2005, 2007, 2010, and 2012, which the analysis will include to
measure number of cattle density per municipality, and number of cattle percentage change from
2005 to 2012. This data on cattle numbers will be used as a proxy measurement to cattle
production intensity for the years studied. Other data used in the report are Brazil vegetation
types, and data on urban centers.
3. Brazil Environmental Ministry (MMA)
MMA provides the list of municipalities that the Federal government established as priority
municipalities in the Legal Amazon for increased monitoring. This data will be analyzed to see
whether deforestation prevention efforts are effectively allocated in the most deforested
municipalities.
4. Brazilian Institute of Environment and Renewable Natural Resources (IBAMA)
IBAMA provides data on legal enforcement mechanisms such as number of land embargoes per
municipalities, and fines related to illegal deforestation. The report will classify this data
according to the years studied and will be transformed into point shapefiles indicating a count of
an instance of a land embargo or fine.
5. World Database on Protected Areas (WDPA)
WDPA provides shapefile data on Brazil’s protected areas. The shapefiles include protected
areas by categories laid out by the International Union for Conservation of Nature (IUNC). The
categories classified for the descriptive and analysis maps are the following:
A. Ia – Strict Nature Reserves:
Strictly protected areas set aside to protect biodiversity and also possibly
geological/geomorphological features, where human visitation, use and impacts are strictly
controlled and limited to ensure protection of the conservation values.
B. II – National Park
Protected areas are large natural or near natural areas set aside to protect large-scale
ecological processes, along with the complement of species and ecosystems characteristic of
Jose Guzman_Possible Causes and Reducers of Deforestation in the Brazilian Legal Amazon:
The Role of Environmental Policies from 2005 to 2012_ PA 388K-59935_Fall_2015
Page 7
the area, which also provide a foundation for environmentally and culturally compatible,
spiritual, scientific, educational, recreational, and visitor opportunities.
C. V – Protected Landscape
To maintain a balanced interaction of nature and culture through the protection of landscape
and/or seascape and associated traditional management approaches, societies, cultures and
spiritual values.
The categories above where chosen apart from other categories, such as sustainable
development, or indigenous areas, in order to narrow down the scope of protected areas into ones
with strict conservation criteria and ecological value, which the report assumes would be areas
where deforestation will be better managed or prevented.
6. World Resources Institute (WRI)
As part of WRI’s Global Forest Watch forest monitoring alert system, the organization provides
data on indigenous lands in Brazil. The data includes the tribe names of each land/area, type of
indigenous land (Reserve/Territory), and land area.
IV. Methods and Process
All gathered shapefile data was projected to the South America Albers Equal Area Conic
projected coordinate system.
1. Context Maps – Four descriptive maps showing annual deforestation for the years 2005,
2007, 2010, and 2012.
a. Reference Map
- Added Projected country borders of South America shapefile SouthAmerica_Project.shp
- Added Brazil states shapefile Brazil_admin_project.
- Added state municipality shapefile for the Legal Amazon states of Acre, Amapá,
Amazonas, Pará, Rondônia, Roraima, Tocantins, Mato Grosso, and Maranhão. Merged
shapeflies.
- Selected all municipalities in the merged file except those in the state of Maranhão that
are not part of the Legal Amazon. Created new shapefile from the selected features,
Legal_Amazon_Muni.shp.
- Added field to attribute table shape_area. Calculated geometry to measure area of
municipalities in Km Sq
b. Annual Deforestation Maps
- Added SouthAmerica_Project.shp, Brazil_admin_Project, and Legal_Amazon_Muni,
shapefile.
Jose Guzman_Possible Causes and Reducers of Deforestation in the Brazilian Legal Amazon:
The Role of Environmental Policies from 2005 to 2012_ PA 388K-59935_Fall_2015
Page 8
- Added Brazil river network shapefile. Clipped it to the Legal_Amazon_Muni.shp to
create Amazon_River_Network.shp file.
- Added Shapefile of Brazil Vegetation cover types Radam_Vegetacao_SIRGAS.shp.
Clipped it to Legal_Amazon_Muni to create Vegetation_Legal_Amazon.shp.
- Based on IBGE data, selected by attributes types of vegetation that falls under dense
rainforest vegetation. Created DenseOmbophylousForest_1.shp.
- Based on IBGE data, selected by attributes types of vegetation that falls under open
rainforest vegetation. Created OpenOmbrophylousForest_1.shp.
- Based on IBGE data, selected by attributes types of vegetation that falls under mixed
forest.
- Added annual deforestation shapefile data for years 2005, 2007, 2010, and 2012 into
separate maps.
2. Descriptive Maps
a. 2012 number of cattle and population density per municipality (Km Sq)
- Added Legal_Amazon_Muni.shp.
- Added Amazon_River_Network.shp.
- Added shapefile for urban centers for with populations above 100 thousand.
- Joined population data for Legal Amazon municipalities for the years 2005 and 2012.
- Joined number of cattle data for the years 2005, 2007, 2010, and 2012.
- Added field PopDen2012. Used field calculator to divide Pop2012 field by the
shape_area field to calculate population density for 2012.
- Added field Cowden2012. Used field calculator to divide Cow2012 (Number of cattle
for the year 2012) field by the shape_area field to calculate number of cattle density for
the year 2012.
- Classified population density for 2012 into four classes. First, classified by Jenks
(Natural Breaks) method and then manually classified density ranges based on Jenks for
easier interpretation.
- Classified number of cattle density for 2012 into four classes. First, classified by Jenks
method and then manually classified density ranges based on Jenks for easier
interpretation.
- Labeled some of the major municipalities (capitals) for each of the Legal Amazon states.
b. Number of cattle percentage change and population percentage change from 2005 to
2012.
Jose Guzman_Possible Causes and Reducers of Deforestation in the Brazilian Legal Amazon:
The Role of Environmental Policies from 2005 to 2012_ PA 388K-59935_Fall_2015
Page 9
- Added Legal_Amazon_Muni.shp with joined tables.
- Added Amazon_River_Network.shp.
- Added shapefile for urban centers for with populations above 100 thousand.
- Added field PopChange. Used field calculator to calculate population percent change for
each municipality from 2005 to 2012.
- Added field CowChange. Used field calculator to calculate number of cattle percent
change for each municipality from 2005 to 2012.
- Classified PopChange into three classes. First, classified by Jenks method and then
manually classified percent change ranges based on Jenks for easier interpretation.
- Classified CowChange into three classes. First, classified by Jenks method and then
manually classified percent change ranges based on Jenks for easier interpretation.
c. Deforestation Monitoring Mechanisms in the Legal Amazon Map
- Added Legal_Amazon_Muni.shp with joined tables.
- Added Amazon_River_Network.shp.
- Selected municipalities that MMA established as priority municipalities. Added
PRIORITY field the attribute table. Started a new editing session, classifying priority
municipalities as 1 and non-priority municipalities as 0.
- Added shapefile for embargoes and fines related to illegal deforestation
embargo_IBAMA.shp
- Selected features by year 2005, 2007, 2010, and 2012. Omitted features without
description of fine or embargo. Created separate shapefiles by years. Merged shapefiles.
Used the feature to point operation to turn the polygon shapefiles into points. Points
were used instead of polygons because not all features were embargoes that confiscated
a certain land area.
d. Indigenous Lands and Protected Areas Map
- Added Legal_Amazon_Muni.shp with joined tables.
- Added Brazil_indigenous_lands.shp
- Added Brazil_Protected_areas.shp
- Clipped Brazil_indigenous_lands.shp into Legal_Amazon_Muni.shp. Created
Legal_Amazon_Ind_Lands.shp.
- Clipped Brazil_Protected_areas.shp into Legal_Amazon_Muni.shp. Created
Amazon_Protected_Areas.shp.
Jose Guzman_Possible Causes and Reducers of Deforestation in the Brazilian Legal Amazon:
The Role of Environmental Policies from 2005 to 2012_ PA 388K-59935_Fall_2015
Page 10
- Selected clipped protected area shapefile by attributes, choosing categories la, II, and V,
and those designated as World Heritage Sites. Created Amazon_Protect_select.shp.
3. Spatial Analysis Maps
a. Ordinary Least Squares (OLS) – First ran OLS regression using variables from the
descriptive map section to see which values were statistically significant, and which
could be omitted in the second step of the analysis. The unit of analysis selected was Km
Sq per municipality in the Legal Amazon.
- Added Legal_Amazon_Muni.shp with joined tables.
- Added Amazon_River_Network.shp.
- Added 2005_Deforest_PRODES.shp, 2007_Deforest_PRODES.shp,
2010_Deforest_PRODES.shp, and 2012_Deforest_PRODES.shp
- Added 2005_embargoes.shp, 2007_embargoes.shp, 2010_embargoes.shp, and
2012_embargoes.shp.
- Spatially joined 2005_Deforest_PRODES.shp, 2007_Deforest_PRODES.shp,
2010_Deforest_PRODES.shp, and 2012_Deforest_PRODES.shp, summarizing by sum
attributes. This process allowed having a sum of all the deforestation that falls within
each municipality for the joined year. To normalize the sum results, a field was added
for each year, and field calculator was used to divide the sum of deforestation for each
year by the municipality shape_area attribute table. Added field named AGG_Def and
used filed calculator to find the aggregate percentage area deforested for the four years
analyzed. Named final joined shapefile RegressionMap_1.shp
- Spatially joined 2005_embargoes.shp, 2007_embargoes.shp, 2010_embargoes.shp, and
2012_embargoes.shp, and the merged shapefile of all the embargoes/fines, summarizing
by sum attributes. This process allowed having a sum of all the counts of
fines/embargoes that falls within each municipality for the joined year. Named final
Joined shapefile RegressionMap_2.shp.
- Added Amazon_Protect_Select.shp. Performed intersect operation with
Legal_Amazon_Muni.shp, to separate/split protected areas that overlapped multiple
municipalities. The created intersected polygon shapefile was turned into a point
shapfile using the feature to point operation. This was done to avoid extra addition of
protected areas that fell outside municipalities, in order to have only the area summed
that falls within each municipality once it was joined into the regression shapefile.
Spatially Joined Amazon_Protected_Area_point.shp with RegressionMap_2.shp,
summarizing by sum attributes. This process allowed having a sum of the area in each
municipality that falls under a protected area. Added attribute field PROPERAR, and
used field calculator to divide the sum of protected areas in each municipality by the
shape_area attribute to normalize the percentage area of municipality that falls under a
designated protected area. Named final joined shapefile RegressionMap_3.shp.
Jose Guzman_Possible Causes and Reducers of Deforestation in the Brazilian Legal Amazon:
The Role of Environmental Policies from 2005 to 2012_ PA 388K-59935_Fall_2015
Page 11
- Added Legal_Amazon_Ind_Lands.shp. Performed intersect operation with
Legal_Amazon_Muni.shp to separate/split indigenous lands that overlapped multiple
municipalities. The created polygon intersected polygon shapefile was turned into a
point shapefile using the feature to point operation. This was done to avoid extra
addition of indigenous lands that fell outside municipalities, in order to have only the
summed area that falls within once each municipality it was joined into the regression
shapefile. Spatially Joined Legal_Amazon_Ind_Lands_point.shp with
RegressionMap_2.shp, summarizing by sum attributes. This process allowed having a
sum land area in each municipality that falls under an indigenous land. Added attribute
field INDPERAR, and used field calculator to divide the sum of indigenous lands in
each municipality by the shape_area attribute to normalize the percentage area of
municipality that falls under an indigenous land. Named final joined shapefile
RegressionMap_4.shp.
- Added Brazil_Road_network.shp, clipped it to Legal_Amazon_Muni.shp. Intersected
resulting shapefile to Legal_Amazon_Muni to separate road lines into each
municipality. Turned line feature into points using the feature to point operation. Joined
road_point shapefile, summarizing by sum attributes. Added attribute field Road_Den
and used field calculator to measure road density (sum road length divided by
shape_area).
- Ran OLS regression.
Dependent Variable
AGG_DeF – Aggregate percentage area deforested per munitipality
Independent Variables
PopChange – Population percent change from 2005 to 2012
Count_1 – Embargo/Fine instances
CowChange – Number of cattle percentage change from 2005 to 2012
COWDEN12 – Cattle Density 2012
PRIORITY – Whether deforestation falls within priority municipalities
IND_PER_AR – Percentage area of municipality that falls under indigenous lands
PROPERAR – Percentage area of municipality that falls under a designated protected area
b. Geographically Weighted Regression (GWR) – After running the OLS regression,
performed a GWR regression with only the statistically significant result variables from
the OLS Regression. While OLS regression is based on one single equation that assumes
that the relationship between variables is exactly the same everywhere in the legal
Amazon, the GWR regression constructs a separate equation for every feature
(municipalities) in the dataset.
Jose Guzman_Possible Causes and Reducers of Deforestation in the Brazilian Legal Amazon:
The Role of Environmental Policies from 2005 to 2012_ PA 388K-59935_Fall_2015
Page 12
Dependent Variable
AGG_deF – Aggregate percentage area deforested per municipality
Independent Variables
PRIORITY – Whether deforestation falls within priority municipalities
COWDEN12 – Cattle Density 2012
IND_PER_AR – Percentage area of municipality that falls under indigenous lands.
PROPERAR – Percentage area of the municipality that falls under a designated protected area
V. Map Findings
1. Context Maps –
Maps 1.1 through 1.4 demonstrate the deforestation decrease levels for the years selected. The
progression shows an overall decrease in deforestation from the years 2005 to 20012.
Deforestation kilometers for the year 2005 totaled 23,815 square kilometers. Deforestation for
the year 2007 totaled 11,451 square kilometers. Deforestation for the year 2010 totaled 6,735
square kilometers. Finally, deforestation for the year 2012 totaled 4,435 square kilometers.
An important note regarding annual deforestation figures acquired and used for subsequent
analysis. These figures are based on the statistics gathered from the attribute tables the the
polygon shapefiles. These statistics numbers resulted to be higher than official figures and tables
in Brazilian government estimates.
Jose Guzman_Possible Causes and Reducers of Deforestation in the Brazilian Legal Amazon:
The Role of Environmental Policies from 2005 to 2012_ PA 388K-59935_Fall_2015
Page 13
Map 1.1
Jose Guzman_Possible Causes and Reducers of Deforestation in the Brazilian Legal Amazon:
The Role of Environmental Policies from 2005 to 2012_ PA 388K-59935_Fall_2015
Page 14
Map 1.2
Jose Guzman_Possible Causes and Reducers of Deforestation in the Brazilian Legal Amazon:
The Role of Environmental Policies from 2005 to 2012_ PA 388K-59935_Fall_2015
Page 15
Map 1.3
Jose Guzman_Possible Causes and Reducers of Deforestation in the Brazilian Legal Amazon:
The Role of Environmental Policies from 2005 to 2012_ PA 388K-59935_Fall_2015
Page 16
Map 1.4
Jose Guzman_Possible Causes and Reducers of Deforestation in the Brazilian Legal Amazon:
The Role of Environmental Policies from 2005 to 2012_ PA 388K-59935_Fall_2015
Page 17
2. Descriptive Maps
a. 2012 number of cattle and population density per municipality (Sq Kms)
- 670 of the 772 municipalities had population densities ranging from 1 to 20 people per
square kilometer. 80 municipalities had population densities ranging from more than 20
to a 100 people per square kilometer. 13 municipalities had population densities ranging
from more than 100 to 400 people per square kilometer. 7 municipalities had
municipality densities ranging from 400 to 3000 people per square kilometers (Map 2.1)
- 381 of the 772 municipalities had cattle densities ranging from 1 to 20 cattle per square
kilometer. 231 municipalities had cattle densities ranging from more than 20 to 60 cattle
per square kilometers. 111 municipalities had cattle densities ranging from more than 60
to 100 cattle per square kilometers. 49 of the municipalities had cattle densities ranging
from more than 100 to 200 cattle per square kilometers. (Map 2.2)
b. Cattle and Population Percent Change from 2005 to 2012
- 293 of the 772 municipalities in the Legal Amazon experienced negative percent change
in number of cattle, but no more than 1 percent change. 478 municipalities experienced
up to 10 percent number of cattle change. Only 1 municipality experienced a 47 percent
change in the number of cattle from 2005 to 2012 (Map 2.3)
- 237 of the 772 municipalities in the Legal Amazon experienced negative percent change
in population, but no more than 1 percent change. 533 of the municipalities experienced
up to 10 percent number of cattle change. Only 1 municipality experienced a 46 percent
change in population from 2005 to 2012 (Map 2.4)
c. Deforestation Monitoring Mechanisms in the Legal Amazon
- Fines/embargoes totaled 2225 when adding the four years under study (41 in 2005, 531
in 2007, 519 in 2010, and 1134 in 2012). The highest number of fines/embargoes in a
municipality was 305 (Map 2.5)
d. Protected Areas and Indigenous Lands
- 141 out of the designated Amazon protected areas are located within the Legal Amazon,
including 45 natural protected areas (IUNC category V), 54 national parks, 33 Strict
Nature Reserves, and 2 World Heritage Sites.
- The indigenous lands are composed of 377 different tribes for each area. The summed
area represents around 22 percent of the Legal Amazon area (Map 2.6)
Jose Guzman_Possible Causes and Reducers of Deforestation in the Brazilian Legal Amazon:
The Role of Environmental Policies from 2005 to 2012_ PA 388K-59935_Fall_2015
Page 18
Map 2.1
Jose Guzman_Possible Causes and Reducers of Deforestation in the Brazilian Legal Amazon:
The Role of Environmental Policies from 2005 to 2012_ PA 388K-59935_Fall_2015
Page 19
Map 2.2
Jose Guzman_Possible Causes and Reducers of Deforestation in the Brazilian Legal Amazon:
The Role of Environmental Policies from 2005 to 2012_ PA 388K-59935_Fall_2015
Page 20
Map 2.3
Jose Guzman_Possible Causes and Reducers of Deforestation in the Brazilian Legal Amazon:
The Role of Environmental Policies from 2005 to 2012_ PA 388K-59935_Fall_2015
Page 21
Map 2.4
Jose Guzman_Possible Causes and Reducers of Deforestation in the Brazilian Legal Amazon:
The Role of Environmental Policies from 2005 to 2012_ PA 388K-59935_Fall_2015
Page 22
Map 2.5
Jose Guzman_Possible Causes and Reducers of Deforestation in the Brazilian Legal Amazon:
The Role of Environmental Policies from 2005 to 2012_ PA 388K-59935_Fall_2015
Page 23
Map 2.6
Jose Guzman_Possible Causes and Reducers of Deforestation in the Brazilian Legal Amazon:
The Role of Environmental Policies from 2005 to 2012_ PA 388K-59935_Fall_2015
Page 24
3. Spatial Analysis (Map 3.1)
a. Ordinary Least Squares (OLS) - Table 1 shows the results of the regression based on
the variables explained above.
Table 1
- The variables for priority municipality, population change, and cattle density for 2012,
showed a significant positive correlation to aggregate percentage area deforested.
- The variables for indigenous lands, protected areas, and road density show a negative
significant correlation to aggregate percentage area deforested.
- The variables above are statistically significant because the t-statistics are large and the
p-values are below .05. This means that we are 95% confident that these coefficients are
statistically different from zero, or that relationship between aggregate deforestation and
the variables is caused by something other than mere random chance.
- The OLS regression resulted in an R² value of 0.167077 and an adjusted R² value of
0.158344. The variables used in the regression thus underperform in explaining
aggregate percent deforestation.
- The OLS regression appears show clustering along the deforestation corridor noticed in
the context maps as well as the priority municipalities. The spatial cluster of standard
residuals means that the model is missing one or more key explanatory variables.
- Results with Standard Deviation > 1.5 indicate that 51 municipalities experience higher
deforestation higher than the model predicts. Results with Standard Deviation -1.5
indicate that 2 municipalities experience less deforestation than the model predicts.
Jose Guzman_Possible Causes and Reducers of Deforestation in the Brazilian Legal Amazon:
The Role of Environmental Policies from 2005 to 2012_ PA 388K-59935_Fall_2015
Page 25
b. Geographically Weighted Regression (GWR)
- The GWR regression resulted in an R² value of 0.371 and an adjusted R² value of 0.317.
Using GWR increases the goodness of fit compared to OLS regression, showing that
more than 30 percent of the variation is explained by the significant variables used in the
regression.
- Results with Standard Deviation > 1.5 indicate that 47 municipalities experience higher
deforestation higher than the model predicts. Results with Standard Deviation -1.5
indicate that 4 municipalities experience less deforestation than the model predicts.
Map 3.1
Jose Guzman_Possible Causes and Reducers of Deforestation in the Brazilian Legal Amazon:
The Role of Environmental Policies from 2005 to 2012_ PA 388K-59935_Fall_2015
Page 26
VI. Discussion
1. Context Maps
The context maps show a clear decrease in annual deforestation for the years observed. The maps
exhibit a pattern of deforestation along mixed forested lands. This is likely due, among other
factors, to the border of such forest types with savannas and pastured land used for cattle grazing
and soy production for cattle feed.
2. Descriptive Maps
a. Population density for 2012 was not a significant factor when compared to the degree of
deforestation in the Brazilian Legal Amazon. Although some urban centers and the most
densely populated areas are present in small municipalities along the western coastal region,
they do not appear to be correlated to increased deforestation. Furthermore, densely
populated municipalities are spatially spread across the region.
b. Visually, cattle density appears to be correlated to increased deforestation levels, since the
municipalities with the highest cattle density (70 to 200 cattle per square kilometer) are
located along the same municipalities that experience high percentages of annual
deforestation.
c. The majority of municipalities in the Brazilian Legal Amazon have experienced a positive
percentage change in both in the number of cattle and populations. This could be due to the
profitability in cattle ranching, increased development infrastructure, and population
migration towards the Amazon.
d. The priority municipalities appear to be located along the areas most affected by
deforestation. Most counts of embargoes and fines fall within the 41 priorities designated by
decree, indicating increased monitoring efforts in the areas selected.
e. Protected areas and indigenous lands seem to be spread across the Legal Amazon, perhaps
showing some cluster in the northern section bordering Colombia, Venezuela, Guyana,
Suriname, and French Guiana, as well as the Peruvian border. Other clusters appear to be
located centrally. Spatial analysis methods, such as hot spot analysis could help clarify where
the clusters of indigenous lands and protected areas are located.
3. Spatial Analysis
With the OLS regression results, there is 95% confidence that the significant coefficients are
statistically different from zero, or that relationship between aggregate deforestation and the
variables is caused by something other than mere random chance. The presence of priority
municipalities is positively correlated to deforestation levels, indicating that the Brazilian federal
government has effectively designated municipalities that are most likely to experience the
highest levels of deforestation. However, the presence of priority municipalities does not seem to
disperse or reduce deforestation.
Jose Guzman_Possible Causes and Reducers of Deforestation in the Brazilian Legal Amazon:
The Role of Environmental Policies from 2005 to 2012_ PA 388K-59935_Fall_2015
Page 27
There is a negative correlation between fines/embargo instances and deforestation levels,
meaning that these monitoring measures can decrease deforestation levels. However, the
association is not statistically significant. Separating fines by count, and embargoes by
percentage of area embargoed might lead to more significant results than just counting all the
instances of embargoes and fines together. Furthermore, adding continuous data on
embargoes/fines over the 8-year period instead of 4 years might have yielded more significant
results
As expected, protected areas and indigenous lands show a significant negative correlation with
deforestation levels. This means that the higher percentage area of a municipality that falls under
a protected area or indigenous land, the lower the level of percentage area deforested per
municipality.
An interesting finding is the negative correlation between road density and deforestation. This is
likely the case because the OLS correlation measures all the road networks in the Legal Amazon.
Road networks in the Legal Amazon, such as those in the south of Mato Grosso state, are located
far away from forested areas. Hence, the results are likely to show a negative correlation. Better
classification of roads close to deforested areas, or close to the forested Legal Amazon region
would likely result in a significant positive correlation between roads and deforestation.
The GWR improved the R² value because it ran regressions with the variables specified for each
municipality, instead of a global regression, which assumes no variation of the variables in each
municipality. Vegetation, road networks, population densities, and cattle densities differ
depending on the region of the Legal Amazon. It therefore made sense to run a GWR, which
takes these differences into account.
An issue associated with highly deforested municipalities is that they are highly clustered,
resulting in issues of auto correlation. The highest (> 1.5) and lowest (< -1.5) standard residual
results appear to cluster, meaning that the variables included in the regressions do not result in a
proper model, which might be better adjusted with other key explanatory variables.
The OLS and GWR regressions only use four years as samples from the eight-year period.
Continuous aggregate percentage deforestation levels from the years 2005 to 2012, as well as
more robust datasets for the independent variables might result in more robust and significant
models.
VII. Conclusion
Although the variables used for the OLS and GWR regressions do not create a proper model for
analysis (the model is missing other key variables), the purpose of the report was not to include
the best explanatory variables of deforestation, which has been extensively analyzed, but rather
to analyze specific environmental policies, such as the establishment protected areas, the role
indigenous land management and administration of their own lands, and monitoring mechanisms
in priority municipalities. Number of cattle, which was included as a proxy of cattle ranching,
demonstrated to be the leading factor in the increase of deforestation in the Legal Amazon.
Population was also included as a possible variable. However, as the literature points out,
Jose Guzman_Possible Causes and Reducers of Deforestation in the Brazilian Legal Amazon:
The Role of Environmental Policies from 2005 to 2012_ PA 388K-59935_Fall_2015
Page 28
demographic trends should also include other sociodemographic variables, such as wage/income
earnings, and education levels. Further analysis on such variables could better explain the effect
of population changes in the Amazon on deforestation rates. Nonetheless, most of the variables
used for the regression turned out as significant values in the regression, meaning that they do
explain increases or decreases in deforestation levels.
Other questions of interest for further analysis are the following: (1) How much of the land is
going being reforested rather than continuing being used for cattle ranching or soy production?
(2) How would the OLS or GWR differ if deforested area was not aggregated and analyzed by
year? Do previous years of deforestation future deforestation?
Although the Brazilian Legal Amazon experienced a decrease in annual deforestation from 2005
to 2012, the rates have picked up since then, even though monitoring efforts have increased.xvi
As the report hints, the creation of priority municipalities might not be enough to curb
deforestation rates. Policy recommendations based on the findings and current trends in
deforestation include further analysis on the effectiveness of illegal deforestation monitoring
mechanisms, and enhancement or decentralization of management and administration efforts for
protected areas in indigenous lands.
Footnotes
i Brasileiros, “Desmatamento Na Amazônia Legal Diminuiu 82% Na última Década.” ii Kirby et al., “The Future of Deforestation in the Brazilian Amazon,” 434. iii Fearnside, “Deforestation in Brazilian Amazonia,” June 1, 2005. iv “Calculating Deforestation in the Amazon.” v Fearnside, “Deforestation in Brazilian Amazonia,” December 1, 1993. vi Bouchardet and Porsse, “An Exploratory Spatial Data Analysis for Deforestation in Brazilian Amazon.” vii Pfaff, “What Drives Deforestation in the Brazilian Amazon?” viii Kaimowitz et al., “Spatial Regression Analysis of Deforestation in Santa Cruz, Bolivia.” ix Fearnside, “Deforestation in Brazilian Amazonia,” June 1, 2005, 683. x Ibid., 683–684. xi Kaimowitz et al., “Spatial Regression Analysis of Deforestation in Santa Cruz, Bolivia.” xii Nepstad et al., “Slowing Amazon Deforestation through Public Policy and Interventions in Beef and Soy Supply Chains.” xiii Macedo et al., “Decoupling of Deforestation and Soy Production in the Southern Amazon during the Late 2000s.”
xiv Assunção and Gandour, “Deforestation Slowdown in the Brazilian Amazon,” 35. xv Kintisch, “Improved Monitoring of Rainforests Helps Pierce Haze of Deforestation.” xvi Watts, “Amazon Deforestation Report Is Major Setback for Brazil ahead of Climate Talks.”
Jose Guzman_Possible Causes and Reducers of Deforestation in the Brazilian Legal Amazon:
The Role of Environmental Policies from 2005 to 2012_ PA 388K-59935_Fall_2015
Page 29
Works Cited
Assunção, Juliano, and Clarissa Gandour. “Deforestation Slowdown in the Brazilian Amazon:
Prices or Policies?” CPI. Accessed December 8, 2015.
http://climatepolicyinitiative.org/publication/deforestation-slowdown-in-the-legal-
amazon-prices-or-policie/.
Bouchardet, Daniel, and Alexandre Porsse. “An Exploratory Spatial Data Analysis for
Deforestation in Brazilian Amazon.” ERSA conference paper. European Regional
Science Association, 2015. https://ideas.repec.org/p/wiw/wiwrsa/ersa15p845.html#biblio.
Brasileiros. “Desmatamento Na Amazônia Legal Diminuiu 82% Na última Década.” Brasileiros.
Accessed December 8, 2015. http://brasileiros.com.br/2015/08/desmatamento-na-
amazonia-legal-diminuiu-82-na-ultima-decada/.
“Calculating Deforestation in the Amazon.” Mongabay.com. Accessed November 9, 2015.
http://rainforests.mongabay.com/amazon/deforestation_calculations.html.
Fearnside, Philip M. “Deforestation in Brazilian Amazonia: History, Rates, and Consequences.”
Conservation Biology 19, no. 3 (June 1, 2005): 680–88. doi:10.1111/j.1523-
1739.2005.00697.x.
———. “Deforestation in Brazilian Amazonia: The Effect of Population and Land Tenure.”
Ambio 22, no. 8 (December 1, 1993): 537–45.
Kaimowitz, D., P. Mendez, A. Puntodewo, and J. K. Vanclay. “Spatial Regression Analysis of
Deforestation in Santa Cruz, Bolivia.” Center for International Forestry Research, 2002.
http://www.cifor.org/library/603/spatial-regression-analysis-of-deforestation-in-santa-
cruz-bolivia/.
Kintisch, Eli. “Improved Monitoring of Rainforests Helps Pierce Haze of Deforestation.” Science
316, no. 5824 (2007): 536–37.
Kirby, Kathryn R., William F. Laurance, Ana K. Albernaz, Götz Schroth, Philip M. Fearnside,
Scott Bergen, Eduardo M. Venticinque, and Carlos da Costa. “The Future of
Deforestation in the Brazilian Amazon.” Futures, Futures of Bioregions, 38, no. 4 (May
2006): 432–53. doi:10.1016/j.futures.2005.07.011.
Macedo, Marcia N., Ruth S. DeFries, Douglas C. Morton, Claudia M. Stickler, Gillian L.
Galford, and Yosio E. Shimabukuro. “Decoupling of Deforestation and Soy Production in
the Southern Amazon during the Late 2000s.” Proceedings of the National Academy of
Sciences of the United States of America 109, no. 4 (January 24, 2012): 1341–46.
doi:10.1073/pnas.1111374109.
Nepstad, Daniel, David McGrath, Claudia Stickler, Ane Alencar, Andrea Azevedo, Briana
Swette, Tathiana Bezerra, et al. “Slowing Amazon Deforestation through Public Policy
and Interventions in Beef and Soy Supply Chains.” Science 344, no. 6188 (June 6, 2014):
1118–23. doi:10.1126/science.1248525.
Pfaff, Alexander S. P. “What Drives Deforestation in the Brazilian Amazon?: Evidence from
Satellite and Socioeconomic Data.” Journal of Environmental Economics and
Management 37, no. 1 (January 1999): 26–43. doi:10.1006/jeem.1998.1056.
Watts, Jonathan. “Amazon Deforestation Report Is Major Setback for Brazil ahead of Climate
Talks.” The Guardian, November 27, 2015, sec. World news.
http://www.theguardian.com/world/2015/nov/27/amazon-deforestation-report-brazil-
paris-climate-talks.
Jose Guzman_Possible Causes and Reducers of Deforestation in the Brazilian Legal Amazon:
The Role of Environmental Policies from 2005 to 2012_ PA 388K-59935_Fall_2015
Page 30
Sources
South America (Local Copy) [Computer File] Redlands, CA: Esri ArcGIS, 2013.
http://www.arcgis.com/home/item.html?id=d3d2bae5413845b193d038e4912d3da9
Brazil [Computer File] Redlands, CA: Esri ArcGIS, 2014.
https://www.arcgis.com/home/item.html?id=1e45bf47c2f542cda9d507ce38ba354a
Population Data [Computer File] Brasilia, DF: IBGE, 2014.
http://www.ibge.gov.br/home/estatistica/populacao/estimativa2015/default.shtm
Agricultural Data [Computer File] Brasilia, DF: IBGE - Pesquisa Pecuária Municipal, 2015.
http://www.sidra.ibge.gov.br/bda/tabela/listabl.asp?c=73&z=p&o=34
Brazil Municipalities [Computer File] Brasilia, DF: IBGE, 2011.
http://www.ibge.gov.br/english/geociencias/default_prod.shtm
Brazil Vegetation [Computer File] Brasilia, DF: IBGE, 2011.
http://www.ibge.gov.br/english/geociencias/default_prod.shtm
Brazil Hydrology [Computer File] Brasilia, DF: IBGE, 2011.
http://www.ibge.gov.br/english/geociencias/default_prod.shtm
Streets [Computer File] Open Street Map - © OpenStreetMap contributors.
http://www.openstreetmap.org/copyright,
Annual Deforestation [Computer File] Brasilia, DF: INPE – PRODES, 2015.
http://www.obt.inpe.br/prodes/index.php
Embargoes [Computer File] Brasilia, DF: IBAMA, 2015.
http://siscom.ibama.gov.br/novo/mapas/
IUCN and UNEP-WCMC (2014-2015), The World Database on Protected Areas (WDPA) [On-
line], [December, 2015], Cambridge, UK: UNEP-WCMC. Available at:
www.protectedplanet.net.
Indigenous Lands [Computer File] Washington, DC: WRI - Global Forest Watch, 2015.
http://data.globalforestwatch.org/datasets/322d13636595466883421d553f2af65a_2