Environmental Research 144 (2016) 49–63

Contents lists available at ScienceDirect

Environmental Research
journal homepage: www.elsevier.com/locate/envres

Deforestation scenarios for the Bolivian lowlands

Graciela Tejada a,n, Eloi Dalla-Nora a, Diana Cordoba b, Raffaele Lafortezza c, Alex Ovando a,
Talita Assis a, Ana Paula Aguiar a
a
Earth System Science Center (CCST), National Institute for Space Research (INPE), Av. dos Astronautas 1758, 12227-010 São José dos Campos, SP, Brazil
b
Royal Roads University, 2005 Sooke Road, Victoria, BC, Canada
c
Department of Agriculture and Environmental Science, University of Bari, Via Amendola 165/A, 70126 Bari, Italy

art ic l e i nf o a b s t r a c t

Article history: Tropical forests in South America play a key role in the provision of ecosystem services such as carbon
Received 5 May 2015 sinks, biodiversity conservation, and global climate regulation. In previous decades, Bolivian forests have
Received in revised form mainly been deforested by the expansion of agricultural frontier development, driven by the growing
13 October 2015
demands for beef and other productions. In the mid-2000s the Movimiento al Socialismo (MAS) party
Accepted 13 October 2015
Available online 23 October 2015
rose to power in Bolivia with the promise of promoting an alternative development model that would
respect the environment. The party passed the world’s first laws granting rights to the environment,
Keywords: which they termed Mother Earth (Law No. 300 of 2012), and proposed an innovative framework that was
Deforestation scenarios expected to develop radical new conservation policies. The MAS conservationist discourse, policies, and
Amazon forest
productive practices, however, have since been in permanent tension. The government continues to
Land cover change (LCC) model
guarantee food production through neo-extractivist methods by promoting the notion to expand agri-
LuccME
culture from 3 to 13 million ha, risking the tropical forests and their ecosystem services. These actions
raise major environmental and social concerns, as the potential impacts of such interventions are still
unknown. The objective of this study is to explore an innovative land use modeling approach to simulate
how the growing demand for land could affect future deforestation trends in Bolivia. We use the LuccME
framework to create a spatially-explicit land cover change model and run it under three different de-
forestation scenarios, spanning from the present–2050. In the Sustainability scenario, deforestation
reaches 17,703,786 ha, notably in previously deforested or degraded areas, while leaving forest exten-
sions intact. In the Middle of the road scenario, deforestation and degradation move toward new or paved
roads spreading across 25,698,327 ha in 2050, while intact forests are located in Protected Areas (PAs). In
the Fragmentation scenario, deforestation expands to almost all Bolivian lowlands reaching 37,944,434 ha
and leaves small forest patches in a few PAs. These deforestation scenarios are not meant to predict the
future but to show how current and future decisions carried out by the neo-extractivist practices of MAS
government could affect deforestation and carbon emission trends. In this perspective, recognizing land
use systems as open and dynamic systems is a central challenge in designing efficient land use policies
and managing a transition towards sustainable land use.
& 2015 Elsevier Inc. All rights reserved.

1. Introduction
Tropical forests in South America play a key role in the provi-
sion of ecosystem services (ES) such as carbon sinks, biodiversity
conservation, and climate regulation at local, regional, and global
scales (Nobre, 2014). However, these unique forests and their
services have been threatened by complex, interconnected driving
forces such as agricultural expansion, climate variability, and forest
degradation (Davidson et al., 2012; Malhi et al., 2008). Bolivia, for
example, is listed among the countries with the highest net forest
n
Corresponding author.
E-mail address: graciela.tejada@inpe.br (G. Tejada).
loss during 2000–2010 (FAO, 2010), with 50% of its territory now
covered by lowland forests (Killeen et al., 2007).
Bolivian lowlands experienced intense colonization from the
1950s to the 1970s due to the migration of peasants from the
Andean region (Pacheco, 2006). In the mid-1980s, the agro-in-
dustrial corporations engaged in large-scale deforestations mainly
in the southwestern portion of the Bolivian Amazon, in Santa Cruz,
where current deforestation converted 75% of the land for agri-
cultural purposes (Killeen et al., 2008). More recently, interna-
tional driving forces, such as the growing demand for agricultural
products (mainly soybeans and beef), have been the major cause of
deforestation in Bolivia as well as in other Amazonian countries
(Dalla-Nora et al., 2014; Müller et al., 2012; Pacheco et al., 2010).
In the mid-2000s the Movimiento al Socialismo (MAS) party

http://dx.doi.org/10.1016/j.envres.2015.10.010
0013-9351/& 2015 Elsevier Inc. All rights reserved.
50 G. Tejada et al. / Environmental Research 144 (2016) 49–63
rose to power in Bolivia with the promise of promoting an alter-
native development model that is more respectful of the en-
vironment. The MAS party enacted the Mother Earth Law, (No.
300, of 2012), which recognizes Mother Earth's rights and the
State's obligations to ensure these rights. This law also introduced
a new non-market based mechanism for forest conservation, the
“Joint Mitigation and Adaptation Mechanism for the Integrated
and Sustainable Management of Forests and Mother Earth” (De-
cree 1696 of 2013). This mechanism seeks to ensure sustainable
forest management through the knowledge and rights of in-
digenous people and to become an alternative to dominant mar-
ket-based mechanisms like Reducing emissions from deforestation
and forest degradation and the role of conservation, sustainable
management of forests and enhancement of forest carbon stocks
in developing countries (REDD þ) (Müller et al., 2014b).
MAS conservationist and production policies in Bolivia, how-
ever, have been experiencing continuous tension. The lack of
governance, land tenure conflicts, and conceptual gaps limit the
application of laws and regulations regarding deforestation. Thus,
despite the innovative environmental and legal framework, little
progress has been made in this direction (Müller et al., 2014b,
2013; Pacheco et al., 2010). Moreover, the application of some
environmental laws and regulations contradicts the government’s
aim to guarantee food production and exportation, which is in-
creasing the expansion of the agricultural lands from 3 to
13 million ha in the next 10 years with the “13 Pillars of the Pa-
triotic Agenda 2025” (Bolivia, 2013; Chumacero, 2012; Hoiby and
Zenteno-Hopp, 2014; IBCE, 2013). A further contradiction is the
construction of roads and oil exploration in protected areas (PA)
and indigenous territories (IT) (Chumacero et al., 2010; Jiménez,
2013).
Understanding the deforestation processes occurring in Boli-
vian lowland forests, as in the rest of the world, is a challenging
task. It deserves a multidisciplinary approach, as seen in the work
by Aguiar (2006), Aguiar et al. (2014) and Folhes et al. (2015), that
takes into account the multi-dimensional nature of this topic. Few
studies have addressed the issue of deforestation in Bolivia
through time and space (i.e., Mertens et al., 2004; Müller et al.,
2014a, 2012; Sangermano et al., 2012). The lack of pertinent in-
formation such as multi-temporal land cover change (LCC) data
limits the efforts in this issue because many LCC datasets have only
been available since 2012.
The aim of this study is to explore an innovative modeling
approach for Bolivian lowlands to simulate how the growing de-
mand for agricultural land and different land use policies could
affect future deforestation trends. This study also discusses the
social and environmental implications related to different land use
change scenarios based on deforestation rates and spatial pattern
analyses. Ultimately, we seek to assist in the discussion of broader
land use policies regarding the sustainable development of agri-
culture in the Bolivian forests.
2. Materials and methods
2.1. Study area
In Bolivia, the physiography and altitude determine significant
gradients in temperature, precipitation ( Fig. 2), and consequently,
the rich biodiversity (Ibisch and Merida, 2004). Three main phy-
siographic regions can be distinguished: Andean, Sub-Andean, and
Lowlands. The Andean region is located at 3000 m.a.s.l. between the
Western Range (Cordillera Occidental) and Central Range (Cordil-
lera Central) and is characterized by the Altiplano (high plateau)
and high peaks (Navarro and Maldonado, 2002). The Sub-Andean
region, located in the transition zone between the highlands and
Fig. 1. Study area in the eastern lowlands of Bolivia below the natural montane tree
line (  3000 m). Land cover change (LCC) data from Killeen et al. (2012).
lowlands, is comprised of valleys and piedmont regions. The study
area is the Bolivian lowlands, a territory located approximately
3000 m below the natural montane tree line (Killeen et al., 2008,
2007) (Fig. 1). This area covers almost 70% of the national territory,
including the whole Amazon Basin and portions of La Plata Basin.
Although the lowlands are the most extensive physiographic area in
Bolivia, they account for only 31% of the country's population,
especially in the department of Santa Cruz (26%), followed by Beni
(4%) and Pando (1%) (Fig. 2a) (INE, 2014).
The increasing process of deforestation in the Bolivian lowlands
occurs in several stages and is promoted by national development
policies. First, after the National Revolution of 1952, both the State
and international development agencies channelled capital to en-
courage the development of large-scale cash crop agriculture in the
lowlands. Agriculture for domestic consumption was the main
driver of deforestation during this period. In parallel, the central
government promoted a program to colonize the lowlands called
“March to the East” (Sivila, 1977). This program sought to stimulate
the migration of spontaneous colonists from the Andean highlands
to the lower lands, with the purpose of extending agricultural and
supplementing the need for cheap labor in the rising agro-industry
(Zeballos, 2006). These migrants settled mainly in the northwestern
area of Santa Cruz (Yapacani), north Cochabamba (Chapare) and to
the north of La Paz (Yungas) (Pacheco, 2006). Second, in the middle
of the 1980s, the agro-industrial corporations, Santa Cruz farmers,
and foreign colonies (Menonites and Japanese) engaged in large-
scale deforestation to enable agricultural production (75% of current
Bolivian deforestation) in Santa Cruz (Killeen et al., 2008). Third,
this process continued in the 1990s with the support of the gov-
ernment and the World Bank through the Eastern Lowlands Project
(WB, 1997); the support included investments in silos, processing
facilities, highways, and infrastructure. Finally, during the 2000s,
mechanized agriculture (mainly soybean, which was influenced by
local and international markets), cattle ranching, and small-scale
agriculture were the main drivers of deforestation (Müller et al.,
2011; Pacheco, 2006). Periodic soy booms fuelled by international
markets have played an important role in the expansion of agro-
industrial corporations, which were further triggered by leading
foreign producers and their transnational capital (Medeiros, 2008).
 G. Tejada et al. / Environmental Research 144 (2016) 49–63 51

Fig. 2. (a) Population distribution by department, data from INE (2014); (b) Shuttle Radar Topography Mission (SRTM) digital elevation model, data from Jarvis et al. (2008);
(c) mean annual precipitation, source Ibisch and Merida (2004); (d) mean annual temperature, source Ibisch and Merida (2004).

In the 1990s, a series of policies were enacted to promote
sustainable land and forest management (e.g. Land Law No. 1715
and Forest Law No. 1700 of 1996) and to guarantee better en-
vironmental assessments (e.g. Environmental Law No. 1333 of
1997). These policies have resulted in poor outcomes to prevent
deforestation due to lack of governance and serious conflicts on
land tenure (Müller et al., 2014b, 2013; Pacheco et al., 2010). De-
forestation steadily increased from 2000 to 2010. Cattle ranches
were the main driver of deforestation, affecting mostly the Chi-
quitano region and the Northern Amazon (Cobija, Riberalta), fol-
lowed by mechanized agriculture that decreased from 50% (1990–
2000) to 30% (2000–2010) in Santa Cruz (Müller et al., 2014a,
2012). Small-scale agriculture remained the third driver of defor-
estation in both periods, mainly in the Chapare region, primarily
due to coca cultivation (Chumacero et al., 2010; Müller et al.,
2012).
2.2. Modeling approach
In this study we adopted a top-down modeling approach
(Verburg et al., 2006) where land demand is allocated by space
that is based on cell suitability. The modeling protocol is presented
in five steps: (1) spatial database; (2) model description; (3) model
parameterization; (4) model validation; and (5) scenario as-
sumptions. The impacts of deforestation in terms of carbon
emissions and biodiversity were also considered. All tools and
datasets cited in this section are under open access.
2.2.1. Spatial database
The construction of the spatial database is a relevant part of our
model because it considers land cover and biophysical maps as
potential driving factors of the deforestation process in the Boli-
vian lowlands. The dependent variable is deforestation or LCC,
while the independent variables are those socioeconomic, en-
vironmental, and connectivity factors that influence deforestation
across time and space.
The LCC data were drawn from the Noel Kempff Mercado
Museum of Natural History (NKMMNH) (Killeen et al., 2012). The
NKMMNH has been monitoring the LCC in Bolivia using satellite
images since 2001 using aerial photogrammetry data from pre-
vious decades. A detailed description of the methods to obtain the
LCC data can be found in Killeen et al. (2008, 2007). The LCC data
52 G. Tejada et al. / Environmental Research 144 (2016) 49–63
were organized into three classes of interest (Fig. 1): (1) Natural
vegetation, which includes forests, chaco, cerrado, savannas/wet-
lands, and puna/andean scrublands; (2) Others, including water, no
natural vegetation, and clouds; and (3) Deforestation or LCC (nat-
ural vegetation converted to other land cover classes). The 30 m
resolution LCC data (Killeen et al., 2012) were plugged in our
model cell space of 25 km  25 km, and became a continuous
proportion value (percentage of LCC within the cell). The periods
of LCC data analyzed were from 1976 to 2001 for statistical ana-
lysis, while 2005 was used for calibration and 2008 for validation.
For the independent variables, we constructed a database
with 420 independent variables divided into three main groups:
(1) distances and connections (accessibility and economic attrac-
tiveness); (2) public policies (zoning); and (3) environmental fac-
tors (Supplement 1). These variables were selected based on the
literature regarding deforestation processes in the Amazon
(Aguiar, 2006; Geist and Lambin, 2002; Müller et al., 2012; San-
germano et al., 2012). All the variables were integrated in a spatial
database of 25 km  25 km created in TerraView GIS based on
different spatial operators (Supplement 1).
2.2.2. Model description
We used an open-source modeling framework, LuccME (http://
www.terrame.org/doku.php?id ¼luccme), to build a new spatially-
explicit LCC model for Bolivia called LuccME/Bolivia. LuccME is a
multi-scale model that integrates several inputs, which vary in
resolution, into a cellular space. LuccME allows the modeler to
create different land use and cover change (LUCC) models (e.g.
deforestation, agricultural expansion, desertification, forest de-
gradation, urban sprawl) at different scales and to combine dif-
ferent components, such as demand (calculation of the magnitude
or quantity of change), potential (calculation of the suitability or
propensity of change for each cell), and allocation (spatial dis-
tribution of changes based on land demand and each cell’s po-
tential to change) (Fig. 3) (Aguiar et al., 2012a).
Several well-known LUCC models follow this structure, in-
cluding The Conversion of Land Use and its Effects (CLUE) family
(Veldkamp and Fresco, 1996; Verburg et al., 1999; Verburg and
Veldkamp, 2001), Dinamica- Environment for Geoprocessing Ob-
jects (EGO) (Soares-Filho et al., 2002), and Geomod (Pontius et al.,
2001), which use a range of different approaches and techniques
for their three components. However, these models are im-
plemented in different computing platforms and their code is
generally not open access; therefore, they cannot be easily mod-
ified or combined. In this sense, LuccME allows the construction of
Fig. 3. Generic structure of the main spatially-explicit land use/cover change models (adapted from Verburg et al. (2006)).
new models, combining elements of demand, potential, and allo-
cation components, which are designed according to concepts of
the main models found in the literature.
For the demand component, the amount of deforestation was
defined as an input to the model for 2001–2008 through using the
observed deforestation rates derived from the LCC dataset (Killeen
et al., 2012). For the scenarios, land demand estimates (future
deforestation) were based on different assumptions as described
in Section 2.2.5 (scenario assumptions).
For potential, the LuccME/Bolivia model uses an alternative
component based on Spatial Lag Regression (Aguiar et al., 2007;
Anselin and Smirnov, 1996). This component (SpatialLagRegression)
accounts for spatial auto-correlation and dependence to estimate
the cell’s potential to change. In addition, such a mechanism is
able to dynamically update the potential of change at each step in
time, considering not only the temporal changes in the spatial
drivers (according to the scenario premises) but also the distance
to previously opened areas.
Finally, for allocation we used the LuccME AllocationClueLike
component derived from the CLUE model for continuous land use
variables (Verburg et al., 1999) to generate annual deforestation
maps. In the case of deforestation, cells with a positive change
potential received a percentage of annual change proportionate to
their potential to be allocated to the whole area.
2.2.3. Model parameterization
After compiling the database, we conducted a statistical ana-
lysis for 2001 to select a set of variables to be considered in the
model. Variables that were highly correlated to each other were
excluded, while stepwise selection (Aguiar et al., 2007) was ap-
plied to choose the significant variables. We used the results of
Spatial Lag Regression coefficients to parameterize the model on
GeoDa (Anselin et al., 2006) according to the correlation coeffi-
cient R2 and the significance of each variable (Table 1). The re-
sulting determinant variables were: distance to roads, PAs and ITs,
connectivity index to regional markets, and flat slope (up to 5%);
all of which were highly significant (Table 1). Similar variables
(e.g., PAs, distance to roads, market connection) were also de-
terminants in other studies in the Amazon (Aguiar, 2006; Geist
and Lambin, 2002; Müller et al., 2012).
For distance to roads, we used the Euclidean distance logarithm
while taking into account the closest paved and unpaved roads.
The spatial datasets were obtained from the Bolivian Road Net-
work Administrator (ABC, 2008, 2010). For the connectivity index to
regional markets variable, the Generalized Proximity Matrix (GPM)
Table 1
Description of model components, temporal and spatial resolution, explanatory variables and scenario assumptions regarding deforestation projections.

General parameters Spatial scale Extent Eastern Bolivian lowlands below the natural montane tree line (  3000 m) (based on Killeen et al. (2012))
Resolution Regular cells of 25 km  25 km (625 km2)
Temporal scale Extent 2008–2050
Resolution Yearly
Data for statistical 2001 (Killeen et al., 2012)
analysis
Data for calibration 2005 (Killeen et al., 2012)
Data for validation 2008 (Killeen et al., 2012)
Land use/cover classes Percentage of natural Vegetation, deforestation and other in the cell

Potential: Selected deforestation spatial determinants Regression coefficient Std B Significance Scenario
SpatialLagRegres- dependent
sion parameters
W_log_def Spatial autoregressive coefficient 0.91273280 0.011 0.000
constant Regression constant  0.04853574 0.016 0.002

G. Tejada et al. / Environmental Research 144 (2016) 49–63

GPM to regional markets Connectivity index via the road network to regional markets (cities 720.66800000 114.881 0.000 Y
with 470,000 people)
Distance to roads Euclidean distance to the closest paved or unpaved roads (log10  0.01356610 0.003 0.000 Y
transformed)
Flat slope Percentage of cell area covered by a slope flat (up to 5%) 6.64  10  7 0.004 0.044
Protected Areas and In- Percentage of cell area covered by protected areas and indigenous territories 0.02201879 0.004 0.005 Y
digenous Territories

Allocation: Deforestation allocation parameters – submodel A20 Scenario A Scenario B Scenario C

AllocationClueLike
parameters maxError Maximum allocation error allowed for each land use 500 km2 500 km2 500 km2
minValue Minimum value (percentage) allowed for that land use as a result of new 0% 0% 0%
changes
maxValue Maximum value (percentage) allowed for that land use as a result of new 100% 100% 100%
changes
changeThresholdValue Threshold applied to the level of saturation in each cell. The saturation level 40% 40% 40%
is dynamically computed, according to the available forest in the neigh-
borhood, deconsidering the protected areas. According to the threshold the
speed of change of a given land use in the cell is modified.
maxChange Maximum change in a given land use allowed in a cell in a time step until 6% 6% 6%
(saturation) threshold
maxChangeAbovethreshold Maximum change in a given land use allowed in a cell in a time step after 3% 3% 3%
(saturation) threshold

Demand: Pre- Deforestation Scenario A Scenario B Scenario C

ComputedValues
Trend of 2005–2008 un- Trend of
til 2013, then decrease of 2005–2008
50%
In 2008 starts to increase annually
until it reaches 13 million ha in
2025, then replicates the 2005–
2008 trend

53
54 G. Tejada et al. / Environmental Research 144 (2016) 49–63

Table 2
Description of dependent variables used in the LuccME/Bolivia model.

Category Name Description Actualization Source Operator to compute Dynamic Unit

date variable in each cell

Distances and connec- Distance to roads Euclidean distance to the logarithm of 2001 ABC (2008), Distance to paved and Yes km
tions (accessibility and closest paved or unpaved road. 2005 ABC (2010) unpaved roads (vec-
economic 2008 tor dataset, lines)
attractiveness) GPM to regional Connectivity index via the road net- 2001 INE (2002), Connectivity index Yes n/a
markets work to regional markets (cities ABC (2008) (vector dataset, lines,
with 470,000 people) 2005 INE (2011), points)
ABC (2008)
2008 INE (2011),
ABC (2008),
ABC (2010)

Public Policies (territorial Percentage of Pro- Percentage of cell area covered by 2001 SERNAP Percentage (polygon) Yes %
planning and zoning) tected Areas (PAs) protected areas and indigenous (2005)
and Indigenous Ter- territories 2004 CI (2008) for
ritories (ITs) PA and
2007 FANa, 2012
for IT

Environmental Percentage of flat Percentage of cell area covered by a Not dynamic SRTM (Jarvis Percentage of each No %
slope given slope where X ¼ (1) Steep et al., 2008) preprocessed class X
( 420%); (2) Moderate (between 11% 90 m (raster)
and 20%); (3) Smooth (between 5%
and 10%); (4) Flat (up to 5%)

a
FAN: Friends of Nature Foundation (Fundación Amigos de la Naturaleza) provided the indigenous territories spatial data (not published).

(Aguiar et al., 2007) was applied. The GPM relates neighborhoods
based on Euclidian distance, proximity, and road network relation
(paved or unpaved roads have different weights) and was run for
many assumptions (local, national, and export markets; see Sup-
plements 1 and 4). The GPM to regional markets refers to the road
connections (Table 2, Supplement 4) from cities with more than
70,000 habitants. The population data were taken from the 2001
Census (INE, 2002), and the INE (2011) population projections
were utilized for the years 2005 and 2008. For the GPM, we used
the same road maps of 2001, 2005, and 2008 as in the distance to
roads variable.
Protected areas in Bolivia are divided into two categories: those
with integral protection, where the exploitation of natural re-
sources is prohibited (i.e., National, Departmental and Municipal
Parks; Wildlife Sanctuary and Refuge), and those categories where
the sustainable use of natural resources is allowed (i.e., Integrated
Management Natural Area; Wildlife Reserve; Biosphere Reserve)
(RAISIG, 2013). After the first statistical analysis, we combined
sustainable use and integral protection into one PA category. As PAs
change over time, they are treated as a dynamic variable in the
model. The updated years are 2001 (all PAs created until 2001),
2004 (PAs created from 2001 to 2004), and 2007 (PAs created from
2004 to 2007).
Indigenous Territories in Bolivia are those territories re-
cognized by the State where indigenous people have land and
natural resource management rights according to their customs,
culture, and organization (Bolivia, 2007; Chumacero et al., 2010).
The majority of ITs are in the lowlands and have an important role
in future forest conservation, since most are comprised of primary
forests in areas with low road connectivity (Müller et al., 2013,
2012). They have had recognition (collective land titling) from the
State since 1990 and many are already consolidated (through a
title) or are in the process of consolidation. We considered all ITs
as a single variable, and after the preliminary statistical analysis
we combined the ITs with the PAs into the Protected areas and
indigenous territories variable.
The environmental variable considered in the model is flat slope
(up to 5%). For this variable we used the Shuttle Radar Topography
Mission (SRTM) of 90 m (Jarvis et al., 2008). It was divided into
four slope classes: steep (420%), moderate (11–20%), smooth (5–
10%), and flat (o5%). Only the last class (flat) was included in the
spatial statistical analysis as an indicator of soil fertility and other
environmental variables (see Table 1 and Supplement 1). Flat slope
is the only non-dynamic variable in the deforestation model that
remains stable in the future.
2.2.4. Model validation
For model validation, the multi-resolution analysis (Costanza,
1989; Pontius, 2002) was used to compare model results and ob-
served deforestation during the 2001–2008 period. This method
compares observed data with simulated data at different levels of
coincidence on a scale from 1 to 20.
2.2.5. Scenario assumptions
Scenario assumptions are based on AMAZALERT scenarios for
the Brazilian Amazon (Aguiar et al., 2014) that have been adapted
for Bolivia. This approach combined exploratory (e.g. “Where,
plausibly, are we heading to?”) and normative/anticipatory (e.g.
“What do we want and how do we get there?”) scenario ap-
proaches. The scenarios vary from Low to High Social Development
and High to Low Environmental Development (Fig. 4). We define
High Environmental Development as the responsible management
of natural resources (e.g., environmental stewardship), which in-
cludes high quality and equal access to services, opportunities, and
resources supported by strong institutions. These two axes match
the IPCC AR5 (2014) global SSPs (Socioeconomic pathways) (Arnell
et al., 2011).
This study constructed a storyline for each scenario (Table 3).
The spatial distribution of dynamic variables (roads, connections,
PAs, and ITs) are described in Supplements 2, 3, and 4 and in the
Scenario assumptions in Table 4.
Scenario A, Sustainability, assumes that all the existing en-
vironmental laws are in force, and that policies to reduce defor-
estation and to preserve and create new PAs and ITs are in place. In
this scenario, the country develops an alternative development
model, which respects the environment to a greater degree based
 G. Tejada et al. / Environmental Research 144 (2016) 49–63 55

hydrocarbons (especially gas). Forest governance continues to be

Fig. 4. Representation of AMAZALERT land use scenarios in the context of the
Environmental and Social Development axes (also aligned to the IPCC AR5 (2014)
Socioeconomic pathways). During the project, qualitative participative scenarios
were developed for Brazil and adapted for Bolivia (adapted from Aguiar et al.
(2014)).
on the policy framework initiated in the mid-2000s by the MAS
party. The Mother Earth Law No. 300, enacted by the MAS party in
2012, recognizes Mother Earth's rights and the State's obligations
to ensure these rights. The non-market based mitigation me-
chanism for forest conservation called “Joint Mitigation and
Adaptation Mechanism for the Integrated and Sustainable Man-
agement of Forests and Mother Earth” (Decree 1696 of 2013), in-
troduced by the Mother Earth Law, ensures sustainable forest
management through the knowledge and rights of indigenous
people. This mechanism becomes an alternative to, and/or aligns,
with dominant market-based mechanisms like REDD þ (Müller
et al., 2014b). Moreover, in this scenario, Law No. 3545 on Com-
munity Redirection of the Agrarian Reform Modifies Law 1715
(INRA Law) accelerates collective land titling to indigenous peo-
ples and peasants at the expense of agribusiness and the Supreme
Decree No. 29643 of 2012 empowers community forest organiza-
tions and enables a more effective multi-level forest governance.
Consequently, there are no new roads. Only roads under con-
struction become paved while road maintenance ensures con-
nectivity and development, decreasing deforestation by 50% in
2014.
In Scenario B, Middle of the road, we assume that conserva-
tionist, agricultural, and extractive policies and initiatives continue
to be a motive for tension and contradiction. While the Mother
Earth Law claims harmony between nature and development,
Bolivia’s main source of income stems from the exploitation of
centralized with the national government playing a strong role in
decision-making.
However, in this scenario, land demand remains at the 2005–
2008 rate (Fig. 6), since government economic and exportation
restrictions are imposed on Santa Cruz farmers (Cordoba and
Jansen, 2014). The planned road construction and hydrocarbon
exploitation in PAs and ITs are delayed until 2045 due to the strong
resistance of Amazonian indigenous organizations towards infra-
structure projects and agriculture expansion in their territories
(Chumacero et al., 2010). In addition, PAs and ITs continue to be
protected by Environmental Law No. 1333, the United Nations
Declarations on the Rights of Indigenous Peoples (UNDRIP), and
Agreement 169 of the International Labor Organization (ILO),
among others (Chumacero et al., 2010).
Scenario C, Fragmentation, reflects the alliance between the
government and the agribusiness sector stated in the “13 Pillars of
the Patriotic Agenda 2025” of 2013. This alliance aims to expand
the agricultural frontier from 3 to 13 million ha (Bolivia, 2013;
Chumacero, 2012; Condori, 2013; IBCE, 2013; Hoiby and Zenteno-
Hopp, 2014). Demand for agricultural land and cattle ranching are
increasing considerably. Based on Law No. 337 on the Support of
Food Production and Forest Restoration, financial incentives and
initiatives sustain the commercialization of agricultural products
to benefit and legalize clear-cutting for extensive agriculture. In
addition, extractive industries expand over intact forests as well as
along roads, including PAs and ITs, violating current environ-
mental laws and international conventions (i.e. Agreement 169 of
the ILO). Thus, in line with the international interest to construct
roads under IIRSA (COSIPLAN, 2011), all planned and unpaved
roads will be paved in 2025 (ABC, 2008, 2010). By 2025, PAs and
ITs in oil exploration zones will no longer exist, reflecting the in-
tention to explore oil in these areas (Corz and Lezcano, 2013; Ji-
ménez, 2013). This is the most catastrophic and immediate sce-
nario in terms of deforestation.
2.3. Impacts of deforestation in the key priority areas for biodiversity
conservation
As an indicator of biodiversity, we used a map of the Key
Priority Areas for Biodiversity Conservation of Bolivia (Araujo et al.,
2010), here on referred to as Priority Biodiversity Conservation
Zones. This map combines priority areas for ecological functions,
biological viability, and representation of biodiversity. We esti-
mated the extension of the Priority Biodiversity Conservation
Zones that could be affected by deforestation in 2008 and in each
scenario (Sustainability, Middle of the road, and Fragmentation) in
2050.

Table 3
Brief storylines of each scenario describing the socioeconomic and institutional contexts.

Scenario A: Sustainability Scenario B: Middle of the road Scenario C: Fragmentation

“Relying on strong governance; the Environmental and
Mother Earth laws are enforced. The Amazon rural
landscapes are preserved as protected areas and in-
digenous territories. Existing roads are improved. The
society and diversified economy are well organized and
based on the industrial, forest and agricultural sectors.
The strong agricultural sector uses intensive and en-
vironmentally safe methods”.
“There is the same economic growth trend as in the
period from 2005 to 2008. Santa Cruz farmers are
opponents of the government, slowing down the
growth of the agricultural frontier. The indigenous of
the Amazon defend their territories and delay road
construction and oil exploration in protected areas
and indigenous territories. Deforestation follows the
same trend as in 2005–2008; mechanized agri-
culture and cattle ranching are the main causes of
deforestation”.
“The Government and the Santa Cruz farmers reach
an agreement to expand the agricultural frontier to
13 million ha by 2025 to ensure food sovereignty. All
roads, both planned and unpaved, will be paved in
2025. The economy is based on mechanized agri-
culture, oil exploration and cattle ranching. Road
construction and market commodities for soy are
influenced by international interest”.
 56
Table 4
Assumptions for deforestation scenarios in the LuccME/Bolivia model (Bolivian lowlands).

Variables 2015 2025 2045

Scenario A: Sustainability Roads No new roads Roads under construction are unpaved
GPM to regional No new roads Roads under construction are unpaved
marketsa
PA and IT PAs and ITs maintained New PAs are created

G. Tejada et al. / Environmental Research 144 (2016) 49–63

Deforestation rate Trend of 2005–2008 until 2013, then decrease by 50%

Scenario B: Middle of the road Roads Unpaved roads are paved. Roads under construction are unpaved Planned roads are unpaved Planned roads are paved
Unpaved roads are paved
GPM to regional Unpaved roads are paved. Roads under construction are unpaved Planned roads are unpaved. Unpaved
marketsa roads are paved
PA and IT No new PA or IT PAs and ITs no longer exist in oil
exploration zones
Deforestation rate Trend of 2005–2008 Trend of 2005–2008 Trend of 2005–2008

Scenario C: Fragmentation
Roads Unpaved roads are paved. Roads under construction are unpaved Roads under construction are paved.
Planned roads are paved
GPM to regional Unpaved roads are paved. Roads under construction are unpaved Roads under construction are paved.
marketsa Planned roads are paved
PA and IT No new PA or IT PAs and ITs no longer exist in oil ex-
ploration zones
Deforestation rate Annual increase beginning in 2008 until 13 million ha are removed by 2025 (for in- Reaches 13 million ha, then replicates
tensive agriculture); cattle ranching and small-scale agriculture replicate the 2005– the 2005–2008 trend
2008 trend.

GPM: generalized proximity matrix; PA: protected areas; IT: indigenous territories (Terriotorios Indígenas Originarios Campesinos [TIOC] according to Bolivian Law).
a
Connectivity index via the road network to regional markets (cities with 470,000 people).
 G. Tejada et al. / Environmental Research 144 (2016) 49–63 57

2.4. Carbon emissions from deforestation

For carbon storage, we used the mean above ground biomass

data of Saatchi et al. (2011) (1 km  1 km resolution) for each of
our cells (25 km  25 km) to calculate the biomass in Pg (carbon is
50% of biomass) for 2008 and for each scenario in 2050. For the
deforested areas, we calculated the deforestation gross CO2
(Pg CO2 ¼44/12*Mg C) potential emissions for 2008 and for each
scenario in 2050, assuming that 100% of CO2 emissions take place
when deforestation occurs (not considering regrowth or other
parameters).

3. Results and discussion

Fig. 6. Projection of the demand for each scenario by 2050. Scenario A, Sustain-
3.1. Model performance
ability; Scenario B, Middle of the road; and Scenario C, Fragmentation.

The performance of the LuccME/Bolivia model was satisfactory,

with a spatial adjustment index between the observed and si-
mulated patterns of deforestation resulting in 69% at the first level,
75% at the fifth, and 81% at the tenth level of the validation process
(Fig. 5). The model effectively captures the spatial distribution of
the deforestation process, especially near the previously cleared
areas due to the LuccME allocation component that uses spatial lag
regression. Only in the southern region does the model slightly
amplify the deforestation distribution; this could be due to the
strong influence of the connection to regional markets that gives
more weight to paved roads than unpaved roads (Supplement 4).
3.2. Deforestation scenarios by 2050
The deforestation scenarios for the Bolivian Amazon by 2050
represent future deforestation or forest degradation (Fig. 7). The
different assumptions for each scenario make the deforestation
trends vary in space and time. Deforestation demand has a linear
trend (Fig. 6) compared to other deforestation projections (i.e.
Müller et al., 2014a; Sangermano et al., 2012). Our total deforested
area in each scenario is very large and well represents our attempt
to show the trend (deforestation rate of 2005–2008) under the
MAS party, which has governed since 2006.
Until 2025, the differences among the scenarios have not been
relevant, whereas in 2050 they become considerable. In a sus-
tainable scenario (Scenario A), deforestation reaches 17,703,786 ha
in 2050, which is more than double that of the cleared areas in
2008. By maintaining the same trend as from 2005 to 2008 (Sce-
nario B) deforestation reaches 25,698,327 ha, 7.9 million ha, or 30%
more than for Scenario A in the same period. With weak govern-
ance, new roads, few PAs and ITs, and a large increase in agri-
cultural land, deforestation is more than 50% (20 million ha) than
that in Scenario A, reaching 37,944,434 ha by 2050 (Fig. 6).
The spatial distribution of deforestation is strongly influenced
by road connectivity (consequently from regional markets) and by
previous deforested areas in the three scenarios (Figs. 7 and 8). In
2020, the influence of the recently paved roads appears to not yet
be significant in Scenarios A and B and only slightly in Scenario C.
In 2025 the increment of land demand, reduction of PAs and ITs,
and pavement of planned and unpaved roads in Scenario C make
deforestation and degradation more evident, expanding towards
the north and, more conspicuously, towards the south in 2030.
Scenario B shows the expansion of degradation through new

Fig. 5. Observed (Killeen et al., 2012) versus simulated deforestation in 2008. The red cells indicate completely deforested areas, while the yellow or light green cells (40–
70%) denote progressive degradation (anthropic influence) zones. (For interpretation of the references to color in this figure legend, the reader is referred to the web version
of this article.)
58 G. Tejada et al. / Environmental Research 144 (2016) 49–63

Fig. 7. Deforestation patterns for the Bolivian Amazon in different scenarios. Scenario A, Sustainability; Scenario B, Middle of the road; Scenario C, Fragmentation. Deforested
areas are shown in red (80–100%), while progressive degradation (anthropic influence) zones (40–70%) are shown in yellow or light green. (For interpretation of the
references to color in this figure legend, the reader is referred to the web version of this article.)

roads, as it is intensified in previously deforested areas. By 2040
and 2050, only Scenario A remains with a deforestation patterns
similar as those of 2008, with large areas of natural vegetation. In
Scenario B and more so in Scenario C, the transition from de-
gradation to deforestation is intensified leaving only few areas of
intact forest (Fig. 7). To better understand the impacts of defor-
estation and degradation, we divided Bolivia into regions with
similar causes of deforestation, which are discussed in the next
section.
The Priority Biodiversity Conservation Zones were well pre-
served until 2008, with only 5% being affected by deforestation
(Table 5). In 2050, the situation becomes dramatically different. In
Scenario A, more than 1 million ha will undergo deforestation,
affecting 14% of the total Priority Biodiversity Conservation Zones.
Scenario B shows almost 25% of these zones as being deforested.
Lastly, in Scenario C, deforestation reaches more than 4 million ha
(38%) (Table 5 and Fig. 8).
It is worthy to note the impact of deforestation in terms of
carbon emissions. Before 2008 (considering accumulated defor-
estation), CO2 emissions reached 1.53 Pg (1 Pg C ¼1015 C g); in
2050, Scenario A reached 3.75 Pg, Scenario B reached 5.33 Pg, and
Scenario C 7.65 Pg, more than 5 times total emissions before 2008
(Table 6).
3.3. Discussion: the future of the Bolivian forest
Bolivia’s deforestation dynamics differ according to sub-re-
gional characteristics such as population density, road network,
and biophysical features (slope, precipitation, temperature or soil
fertility). Here, we discuss the scenarios over the forest mask in
five regions: (1) Santa Cruz city, the most populated and deforested
area of Bolivia characterized mainly by mechanized agriculture
(soybean, sugarcane, rice and sunflower production) (Killeen et al.,
2008; Müller et al., 2014a); (2) Chiquitanía, a unique ecosystem in
the eastern Santa Cruz Department (Fig. 8) with large extensions
of chiquitano and cerrado forests, and cattle raising as the main
 G. Tejada et al. / Environmental Research 144 (2016) 49–63 59

Fig. 8. Deforestation scenarios for 2050 over the forest mask, Priority Biodiversity Conservation Zones from Araujo et al. (2010). (a) Observed deforestation in 2008, five
regions with similar deforestation dynamics are shown in light blue; (b) Scenario A, Sustainability; (c) Scenario B, Middle of the road; (d) Scenario C, Fragmentation. The red
cells indicate deforested areas (80–100%), while the yellow or light green cells (40–70%) represent progressive degradation (anthropic influence) zones. (For interpretation of
the references to color in this figure legend, the reader is referred to the web version of this article.)

cause of deforestation (less small-scale agriculture) (Müller et al.,
2014a); (3) the Northern Amazon (near the Brazilian border) forests
with very low population density, where cattle raising and small-
scale agriculture are the main causes of deforestation (Pacheco,
2006); (4) Corridor Amboró-Madidi (CAM) national conservation
corridor, which comprises piedmont and montane forests (Sup-
plement 5), with local and global relevance for biodiversity con-
servation (Araujo et al., 2010; Ibisch et al., 2007; Ibisch and Merida,
2004) and where small-scale agriculture causes most deforesta-
tion (mainly coca cultivation) (Killeen et al., 2008; Müller et al.,
2012); and (5) Chaco, where there are great extensions of boli-
viano-tucumano and chaco dry forests and few paved roads. All
the scenario assumptions for each of the mentioned regions such
as PAs, Priority Biodiversity Conservation Zones, paved and un-
paved roads, and deforestation trend over the forest mask for 2050
are shown in Fig. 8.
60 G. Tejada et al. / Environmental Research 144 (2016) 49–63

Table 5 support unsustainable natural resource exploitation (Cordoba and

Deforestation in the Key Priority Areas for Biodiversity Conservation of Bolivia Jansen, 2015; Gudynas, 2011).
(Araujo et al., 2010) in different deforestation scenarios for 2050.

Deforestation Deforestation in the Key Priority Increased defor-
Scenarios for 2050 Areas for Biodiversity Conservation estation com-
pared to 2008
(ha) (%) Deforestation in (ha)
2008 (ha)
Scenario A: 1,717,898 14 548,160 (5%) 1,169,739
Sustainability
Scenario B: Middle 2,843,401 24 2,295,242
of the road
Scenario C: 4,565,493 38 4,017,333
Fragmentation
Total area of 12,000,100 100
Priority Con-
servation Zones
3.3.1. Scenario A “Sustainability”
In Santa Cruz city, the degraded areas become deforested by
2050; Degradation extends to the east and partly to the west of
the Chiquitanía, while the remaining area is still preserved. In
the Northern Amazon, degradation is observed only in the main
cities where large extensions of forests are preserved. Despite
that 14% of the Priority Biodiversity Conservation Zones are
deforested, the CAM conservation corridor still plays a funda-
mental role in biodiversity conservation at national and inter-
national scales (Ibisch et al., 2007; Ibisch and Merida, 2004)
since only Carrasco and Amboró National Parks are degraded
and there are newly created PAs. Despite some degradation in
Chaco close to the main cities, considerable forest extensions are
still preserved (Fig. 8).
As described in the Sustainability Scenario, Bolivia remains a
country with considerable development and environmental gov-
ernance (Fig. 8). Total deforestation (17.7 million ha) is not as low
as in other sustainable scenarios, but with low connectivity (im-
proving the current road network), ITs, and new PAs, deforestation
is pronounced in previously deforested or degraded areas and
advances slightly towards intact forest areas.
For this scenario, the State has to play an active role in the
environmental and agricultural sectors. First, the State needs to
foster the protection of environmental functions of livelihoods.
This implies securing food without increasing pressures on forests
and establishing a development model that is less sensitive to the
global demands for grains and beef. Second, agricultural policy
recommendations enhanced by the Productive Revolution Law of
2011 need to be implemented to strengthen “sustainable produc-
tion systems”. This includes establishing efficient practices based
on innovation and technological development in current me-
chanized agriculture and cattle raising zones and emphasizing
agro-ecological practices in small-scale agriculture followed by
key actors. This scenario, however, is very unlikely to be im-
plemented due to strong contradictions between the current legal
framework’s focus on the government’s rhetoric of Mother Earth
rights and the dominance of neo-extractivist practices that
3.3.2. Scenario B “Middle of the road”
In the surroundings of Santa Cruz city, deforestation expands in
areas that are already degraded. In Trinidad, Chiquitanía, and in the
Northern Amazon, degradation starts along the roads and main
cities (Fig. 8) due to better connections to regional markets as
perhaps the increase of cattle ranching (the main deforestation
cause from 2000 through 2010 [Müller et al., 2014a]); Only slight
degradation in the CAM is observed, south of TIPNIS, Apolobamba,
and Madidi, which, in this scenario are no longer PAs. The reason
could be that paved roads surrounding the CAM are foreseen only
in 2045. Deforestation affected the Priority Biodiversity Con-
servation Zones by 24%. Chaco has a great deforestation impact on
the southern cities and along the roads (Fig. 8). In the long term,
this scenario could be the same as the Fragmentation scenario,
with the same drivers but in different years.
Although the country retains a law to ensure the rights of
Mother Earth, the government’s efforts seem insufficient at ap-
plying its environmental agenda and instead continue their cur-
rent deforestation trends. First, revenues from hydrocarbons are
crucial for funding the government’s political aims towards re-
distribution and poverty alleviation. Second, global demand on
commodities reverses important initiatives to create more sus-
tainable agricultural production systems. Third, the country's
economy is based on the export of raw materials facilitating the
expansion of extractive industries. Finally, the government tends
to follow the plan of the basic road network integrated with IIRSA
that increases the pressure on forests. However, a sign of the
probable realization of this scenario is the case TIPNIS, where
lowland indigenous communities have contested the construction
of a highway through the territory because of lack of technical and
legal procedures (e.g., an environmental impact study or popular
consultation in breach of current laws). Indigenous organizations
place pressure on the government to revoke plans to build this
road and move forward with the project (Cordoba and Jansen,
2015). Current and future social pressures from indigenous and
peasant populations could oppose increasing deforestation in fa-
vor of a less dramatic scenario.
3.3.3. Scenario C “Fragmentation”
In Santa Cruz city, deforestation expands in all directions,
reaching east Chiquitanía where areas of mechanized agriculture
seem to have expanded. Because of the impact of road construc-
tion in 2025 and the great demand for land, deforestation expands
considerably followed by degradation in all populated areas,
principally in the main cities (Fig. 8). CAM, Carrasco and, Amboró
become deforested extensions of Santa Cruz city. Without legal
protection, current threats such as small-scale agriculture (mainly
coca crops) and cattle ranching activities may dominate, leaving
no intact forests but only degraded areas (except to the east of
what was Madidi PA). The Priority Biodiversity Conservation Zones
are significantly affected by deforestation (38%). Lastly, Chaco de-
forestation intensifies in the degraded areas.

Table 6
Carbon emissions from deforestation for 2008 and 2050 scenarios.

Biomass (Pg) Deforestation (ha) Carbon loss (Pg) CO2 emissions (Pg)

Deforestation until 2008 10.65 7,548,559 0.42 1.53

Deforestation Scenarios for 2050:
Scenario A: Sustainability 8.61 17,703,786 1.02 3.75
Scenario B: Middle of the road 7.75 25,698,327 1.45 5.33
Scenario C: Fragmentation 6.26 37,994,434 2.20 8.06
 G. Tejada et al. / Environmental Research 144 (2016) 49–63
Few forested areas are left intact as the Kaa Iya, Noel Kempff,
and Rios Blanco y Negro PAs (Fig. 8, Supplement 5). This scenario
attempts to show the worst situation in terms of deforestation.
Recent news regarding policies and alliances to extend agricultural
land provide some evidence in support of the likelihood of this
scenario (Condori, 2013; IBCE, 2013). The concomitant rise in food
prices (as the one for 2007–2008) and current fall in oil prices on
the global market strengthens the government’s alliance with the
agribusiness to bring down domestic food prices and to increase
the country's revenues. The strong power of agribusiness elites in
the lowlands, who benefit from large tracts of land and un-
sustainable agricultural practices, is likely to influence and wea-
kened current environmental governance. Also, the road network
of IIRSA will be implemented. Contradictions between these en-
vironmental governances and government practices could speed
the process of deforestation. Revenues from hydrocarbons are
crucial for funding the MAS party’s political aims towards redis-
tribution and poverty reduction, which work against the con-
servation of PAs and ITs.
3.4. Implications for decision-making
Harmonized food, fiber, and beef supply, along with natural
resource conservation in Bolivia, will depend on broader land use
policies and intensification. Also, the recovery of degraded, aban-
doned, or underused lands may lead an important strategy to in-
crease land use efficiency and reduce deforestation (Lambin et al.,
2013). In other words, Bolivian land use policies should treat land
use systems as open systems linked with remote land use drivers
inside and outside the country. Otherwise, region-focused mea-
sures run the risk of simply displacing deforestation pressure
across the country, as previously observed in other tropical
countries (Dalla-Nora et al., 2014; Lambin and Meyfroidt, 2011).
Moreover, institutional challenges still exist in Bolivia, such as the
development of alternative markets or robust incentives for bio-
diversity conservation. Law enforcement by itself cannot ensure
sustainable land use control over the agricultural frontiers. As
suggested in the reviewed literature, farmers are likely to reduce
their managed acreage only if land becomes a scarce resource
(Barretto et al., 2013). In this sense, providing new incentives for
ecosystem service conservation, beyond carbon sequestration,
along with national coverage can become an important mechan-
ism for limiting Bolivian lowland deforestation. However, political
measures following this line of reasoning in Bolivia are strongly
opposed.
4. Conclusions
We assessed deforestation in the Bolivian lowlands by gen-
erating a spatially-explicit land cover change model, which con-
siders deforestation driving factors, different land demands, land
policies and governance arrangements, and ran it under three
scenarios until 2050: Sustainability (optimistic), Middle of the
road (similar to business as usual) and Fragmentation (worst).
In the Sustainability scenario, the conservation corridor Car-
rasco-Amboró and large intact forest areas are preserved, even
though deforestation is pronounced in previously deforested or
degraded areas. Only by applying strict environmental laws
against illegal deforestation and infrastructure projects in PAs, ITs,
and Priority Biodiversity Conservation Zones is this scenario likely
to happen. In the Middle of the road scenario, deforestation and
degradation occur in territories where PAs and ITs no longer exist
along the roads. In the long term, this scenario could be the same
as Scenario C; the difference being that roads will only be paved in
2045, indicating they are a determining factor in the deforestation
61
process. In the Fragmentation scenario, Bolivian lowlands are al-
most completely fragmented by deforestation and degradation.
Intact forest areas in the CAM no longer exist, 38% of Priority
Biodiversity Conservations Zones are deforested, and only small
forests in the Noel Kempff and Kaa Iya PAs remain. Nonetheless,
Scenario C is based on the intentions to establish policies sup-
porting agricultural frontier expansion, oil exploration, and road
construction in PAs (Chumacero, 2012; Chumacero et al., 2010;
Condori, 2013; Corz and Lezcano, 2013; Jiménez, 2013).
Preventing the expansion of the agricultural frontier in the
Bolivian lowlands cannot ensure biodiversity conservation or car-
bon savings in the absence of complementary measures com-
mitted with land use efficiency, controlled land use expansion, and
new economic alternatives. In this perspective, recognizing land
use systems as open and human-driven systems is a first and
central challenge in designing more efficient land use polices.
Otherwise, managing a transition towards more sustainable land
use would become utopian.
Land cover change scenarios are useful in showing how present
and future decisions could affect deforestation trends in the Boli-
vian lowlands. A real-life scenario could be a combination of the
three scenarios presented herein. Observing the potential impacts
of deforestation in a spatially-explicit way, as a valuable discussion
on the existing laws presented in this study, can help to prevent
(or reduce) and influence policy makers’ actions to improve forest
governance. Our data were also used to estimate deforestation
gross carbon emissions of Bolivian lowlands as an initial approach
for 2050 in different scenarios. Using the carbon emissions model
of Aguiar et al. (2012b), we could improve this approximation by
including more parameters to calculate net carbon emissions.
Participatory scenario construction, such as the experience in
Aguiar et al. (2014) that included all the Brazilian deforestation
stakeholders, might enrich our study and help decision makers to
understand the relationship of current policies and future defor-
estation. Moreover, local or regional deforestation analyses could
be accomplished by using a smaller cell size (e.g. 5 km  5 km).
Acknowledgements
This study is part of the Land Use and Cover Change Scenarios
for the Madeira River Basin of the AMAZALERT Project 282664
founded by the European Union's Seventh Framework Programme.
We thank the São Paulo Research Foundation (FAPESP) for Grant
no. 2013/20616-6, Liliana Soria from the Noel Kempff Mercado
Museum of Natural History for providing the Land Use and Cover
Change Data. We also thank Daniel Larrea from the Friends of
Nature Foundation (FAN) and Celso vonRandow from the Earth
System Science Center (CCST-INPE).
Appendix A. Supplementary material
Supplementary data associated with this article can be found in
the online version at 10.1016/j.envres.2015.10.010.
References
ABC, 2010. Mapa de la Red Viaria Fundamental a junio del 2010 – Administración
Boliviana de Carreteras.
ABC, 2008. Vías Principales: Red Fundamental de Bolivia – Administración Bo-
liviana de Carreteras.
Aguiar, A.P., 2006. Modelagem de mudança do uso da terra na Amazônia: ex-
plorando a heterogeneidade intraregional. Sensoriamento Remoto. Instituto
Nacional de Pesquisas Espaciais, São Jose dos Campos.
Aguiar, A.P., Camara, G., Escada, M.I.S., 2007. Spatial statistical analysis of land-use
62 G. Tejada et al. / Environmental Research 144 (2016) 49–63
determinants in the Brazilian Amazonia: exploring intra-regional hetero-
geneity. Ecol. Model. 209, 169–188. http://dx.doi.org/10.1016/j.
ecolmodel.2007.06.019.
Aguiar, A.P., Carneiro, T., Andrade, P.R., Assis, T., 2012a. LuccME-TerraME: an open-
source framework for spatially explicit land use change modelling. GLP News.
Aguiar, A.P., Ometto, J.P., Nobre, C., Lapola, D.M., Almeida, C., Vieira, I.C., Soares, J.V.,
Alvala, R., Saatchi, S., Valeriano, D., Castilla-Rubio, J.C., 2012b. Modeling the
spatial and temporal heterogeneity of deforestation-driven carbon emissions:
the INPE-EM framework applied to the Brazilian Amazon. Glob. Chang. Biol. 18,
3346–3366. http://dx.doi.org/10.1111/j.1365-2486.2012.02782.x.
Aguiar, A.P., Tejada, G., Assis, T., Dalla-Nora, E., 2014. Set of land-use scenarios for Brazil,
linked to implications for policies. AMAZALERT Project, São Jose dos Campos.
Anselin, L., Smirnov, O., 1996. Efficient algorithms for constructing proper higher
order spatial lag operators. J. Reg. Sci. 36, 67–89. http://dx.doi.org/10.1111/
j.1467-9787.1996.tb01101.x.
Anselin, L., Syabri, I., Kho, Y., 2006. GeoDa: an introduction to spatial data analysis.
Geogr. Anal. 38, 5–22. http://dx.doi.org/10.1111/j.0016-7363.2005.00671.x.
Araujo, N., Müller, R., Nowicki, C., Ibisch, P. (Eds.), 2010. Prioridades de conservación
de la biodiversidad de Bolivia., SERNAP, FAN, TROPICO,CEP, NORDECO, GEF II, CI,
TNC, WCS. Universidad de Eberswalde. Editorial FAN, Santa Cruz-Bolivia.
Arnell, N., Kram, T., Carter, T., Ebi, K., Edmonds, J., Hallegatte, S., Kriegler, E., Mathur,
R., O’Neill, B., Riahi, K., Winkler, H., VanVuuren, D., Zwickel, T., 2011. A Frame-
work for a New Generation of Socioeconomic Scenarios for Climate Change
Impact, Adaptation, Vulnerability and Mitigation Research. Working Paper.
Barretto, A.G.O.P., Berndes, G., Sparovek, G., Wirsenius, S., 2013. Agricultural in-
tensification in Brazil and its effects on land-use patterns: an analysis of the
1975–2006 period. Glob. Chang. Biol. 19, 1804–1815. http://dx.doi.org/10.1111/
gcb.12174.
Bolivia, 2013. 13 pilares de la Bolivia digna y soberana: Agenda patriótica del Bi-
centenario 2025.
Bolivia, 2007. Nueva Constitución Política del Estado.
Chumacero, J.P., 2012. ¿Comer de nuestra tierra?; estudios de caso sobre tierra y
producción de alimentos en Bolivia, Informe 2012. FundaciónTierra, La Paz.
Chumacero, J.P., Tinta, E., Salgado, J., Vadillo, A., Colque, G., Ortiz, M. V, Calizaya, O.,
Costas, P., 2010. Informe 2010, Territorios Indígena Originario Campesinos en
Bolivia-Entre la Loma Santa y la Pachamama. FundaciónTierra, La Paz-Bolivia.
CI, 2008. Mapa de Áreas Protegidas Departamentales, Municipales, y Regionales de
Bolivia-Conservación Internacional.
Condori, I., 2013. Agro proyecta ampliar frontera agrícola hasta 15 millones de ha.
Diario de circulación nacional [WWW Document]. La Razón. 〈http://www.la-
razon.com/economia/Agro-proyecta-frontera-agricola-millones_0_
1759624072.html〉.
Cordoba, D., Jansen, K., 2015. Realigning the political and the technical: NGOs and
the politicization of agrarian development in Bolivia. Eur. J. Dev. Res. . http://dx.
doi.org/10.1057/ejdr.2015.12
Cordoba, D., Jansen, K., 2014. The return of the state: neocollectivism, agrarian
politics and images of technological progress in the MAS Era in Bolivia. J. Agrar.
Chang. 14, 480–500. http://dx.doi.org/10.1111/joac.12036.
Corz, C., Lezcano, M., 2013. YPFB explotará petróleo y gas natural en las áreas
protegidas. Diario de circulación nacional [WWW Document]. La Razón. 〈http://
www.la-razon.com/economia/YPFB-explotara-petroleo-natural-protegidas_0_
1838816152.html〉.
COSIPLAN, 2011. Cartera de Proyectos Proyectos para la Integración de la Infra-
estructura Regional Suramericana IIRSA [WWW Document]. 〈http://www.iirsa.
org/Page/Detail?menuItemId ¼32〉.
Costanza, R., 1989. Model goodness of fit: a multiple resolution procedure. Ecol.
Model. 47, 199–215. http://dx.doi.org/10.1016/0304-3800(89)90001-X.
Dalla-Nora, E.L., de Aguiar, A.P.D., Lapola, D.M., Woltjer, G., 2014. Why have land use
change models for the Amazon failed to capture the amount of deforestation
over the last decade? Land Use Policy 39, 403–411. http://dx.doi.org/10.1016/j.
landusepol.2014.02.004.
Davidson, E.A., de Araujo, A.C., Artaxo, P., Balch, J.K., Brown, I.F., Bustamante, C., Coe,
M.M., DeFries, M.T., Keller, R.S., Longo, M., Munger, M., Schroeder, J.W., Soares-
Filho, W., Souza, B.S., Wofsy, S.C., C.M., 2012. The Amazon basin in transition.
Nature 481, 321–328.
FAO, 2010. Global Forest Resources Assessment 2010 Main report – Food and
Agriculture Organization of the United Nations, FAO Forestry Paper 163.
Folhes, R.T., de Aguiar, A.P.D., Stoll, E., Dalla-Nora, E.L., Arauo, R., Coelho, A., do
Canto, O., 2015. Multi-scale participatory scenario methods and territorial
planning in the Brazilian Amazon. Futures 73, 86–99.
Geist, H.J., Lambin, E.F., 2002. Proximate causes and underlying driving forces of
tropical deforestation: tropical forests are disappearing as the result of many
pressures, both local and regional, acting in various combinations in different
geographical locations. Bioscience 52, 143–150. http://dx.doi.org/10.1641/
0006-3568(2002)052[0143:pcaudf]2.0.co;2.
Gudynas, E., 2011. El nuevo extractivismo progresista. In: Acosta, A., Gudynas, E.,
Houtart, F., Ramírez-Sole, H., Martínez-Aller, J., Macas, L. (Eds.), Colonialismos
del Siglo XXI Negocios Extractivos y Defensa del Territorio en América. Icaria
Editorial, pp. 75–92.
Hoiby, M., Zenteno-Hopp, J., 2014. Emerging and traditional elites and the gov-
ernance of the soy sector. In: Bull, B., Aguilar-Stoen, M.C. (Eds.), Environmental
Politics in Latin America: Elite Dynamics, the Left Tide and Sustainable Devel-
opment. Kent, p. 51.
IBCE, 2013. Encuentro Agroindustrial Productivo “Más inversión, más em-
pleos” – Producción Agroalimentaria en Bolivia y el Rol del Sector Privado –
Instituto Boliviano de Comercio Exterior. Comer. Exter. #214.
Ibisch, P., Araujo, N., Nowicki, C. (Eds.), 2007. Visión de Conservación de la Biodi-
versidad del Corredor Amboró-Madidi., FAN/WWF/TN. Editorial FAN, Santa
Cruz.
Biodiversity: The Richness of Bolivia: State of Knowledge and Conservation, FAN. In:
Ibisch, P., Merida, G. (Eds.), Editorial FAN, Santa Cruz-Bolivia.
INE, 2002. CENSO 2001 [WWW Document]. 〈http://www.ine.gob.bo/comunitaria/
comunitaria.aspx〉.
INE, 2011. Proyecciones generales de la población-Instituto Nacional Estadística de
Bolivia [WWW Document]. 〈http://www.ine.gob.bo/indice/indice.aspx?
d1 ¼ 0307&d2 ¼6〉.
INE, 2014. Un pincelazo a las estadísticas con base a datos de censos. Datos re-
levantes del Censo de Población y Vivienda 2012 y de la Boleta Comunal del
Primer Censo Agropecuario del Estado Plurinacional-Instituto Nacional Esta-
dística de Bolivia.
IPCC, 2014. Climate Change 2014: impacts, adaptation, and vulnerability. Part A:
global and sectoral aspects. Contribution of Working Group II to the Fifth As-
sessment Report of the Intergovernmental Panel on Climate Change. In: Field,
C.B., Barros, V.R., Dokken, D.J., Mach, K.J., Mastrandrea, M.D., Bilir, T.E., Chat-
terjee, M., Ebi, K.L., Estrada, Y.O., Genova, R.C., Girma, B., Kissel, E.S., Levy, A.N.,
MacCracken, S., Mastrandrea, P.R., WhiteField, L.L. (Eds.), Climate Change 2014:
Impacts, Adaptation, and Vulnerability. Part A: Global and Sectoral Aspects.
Contribution of Working Group II to the Fifth Assessment Report of the Inter-
governmental Panel on Climate Change. Cambridge University Press, New York,
USA, Cambridge and New York.
Jarvis, A., Reuter, H.I., Nelson, A., Guevara, E., 2008. Hole-filled SRTM for the globe
Version4.
Jiménez, G., 2013. Territorios Indígenas y Áreas Protegidas en la mira: La ampliación
de la frontera de industrias extractivas. PetroPress, p. 5.
Killeen, T.J., Calderon, V., Soria, L., Quezada, B., Steininger, M.K., Harper, G., So-
larzano, L.A., Tucker, C.J., 2007. Thirty Years of Land-cover Change in Bolivia.
AMBIO A J. Hum. Env. 36, 600–606. http://dx.doi.org/10.1579/0044-7447(2007)
36[600:tyolci]2.0.co;2.
Killeen, T., Guerra, A., Calzada, M., Correa, L., Calderon, V., Soria, L., 2008. Total
historical land-use change in eastern Bolivia: who, where, when, and how
much? Ecol. Soc. 13, 36.
Killeen, T., Soria, L., Quezada, B., Guerra, A., Calderón, V., Clazadilla, M., Steininger,
M.K., 2012. Mapa de Cobertura de la Tierra y Deforestación hasta 2008.
Lambin, E.F., Gibbs, H.K., Ferreira, L., Grau, R., Mayaux, P., Meyfroidt, P., Morton, D.C.,
Rudel, T.K., Gasparri, I., Munger, J., 2013. Estimating the world’s potentially
available cropland using a bottom-up approach. Glob. Environ. Change 23,
892–901. http://dx.doi.org/10.1016/j.gloenvcha.2013.05.005.
Lambin, E.F., Meyfroidt, P., 2011. Global land use change, economic globalization,
and the looming land scarcity. Proc. Natl. Acad. Sci. 108, 3465–3472. http://dx.
doi.org/10.1073/pnas.1100480108.
Malhi, Y., Roberts, J.T., Betts, R.A., Killeen, T.J., Li, W., Nobre, C.A., 2008. Climate
change, deforestation, and the fate of the Amazon. Science 319, 169–172. http:
//dx.doi.org/10.1126/science.1146961 (80-. ).
Medeiros, G., 2008. Evolución y caracteristicas del sector soyero en Bolivia. In:
Soruco, X., Plata, W., Medeiros, G. (Eds.), Evolución y caracteristicas del sector
soyero en Bolivia. El Poder en Santa Cruz Ayer y Hoy. Fundación Tierra, Los
Barones del Oriente, pp. 173–240.
Mertens, B., Kaimowitz, D., Puntodewo, A., Vanclay, J., Mendez, P., 2004. Modeling
deforestation at distinct geographic scales and time periods in Santa Cruz,
Bolivia. Int. Reg. Sci. Rev. 27, 271. http://dx.doi.org/10.1177/0160017604266027.
Müller, R., Larrea-Alcazar, D.M., Cuellar, S., Espinoza, S., 2014a. Causas directas de la
deforestación reciente (2000–2010) y modelado de dos escenarios futuros en
las tierras bajas de Bolivia. Ecol. en Boliv 49, 20–34.
Müller, R., Müller, D., Schierhorn, F., Gerold, G., 2011. Spatiotemporal modeling of
the expansion of mechanized agriculture in the Bolivian lowland forests. Appl.
Geogr. 31, 631–640. http://dx.doi.org/10.1016/j.apgeog.2010.11.018.
Müller, R., Müller, D., Schierhorn, F., Gerold, G., Pacheco, P., 2012. Proximate causes
of deforestation in the Bolivian lowlands: an analysis of spatial dynamics. Reg.
Environ. Change 12, 445–459. http://dx.doi.org/10.1007/s10113-011-0259-0.
Müller, R., Pacheco, P., Montero, J.C., 2014b. The context of deforestation and forest
degradation in Bolivia: drivers, agents and institutions. CIFOR . http://dx.doi.
org/10.17528/cifor/004600.
Müller, R., Pistorius, T., Rohde, S., Gerold, G., Pacheco, P., 2013. Policy options to
reduce deforestation based on a systematic analysis of drivers and agents in
lowland Bolivia. Land Use Policy 30, 895–907. http://dx.doi.org/10.1016/j.
landusepol.2012.06.019.
Navarro, G., Maldonado, M., 2002. Geografía ecológica de Bolivia: Vegetación y
ambientes acuaticos. Centro de Ecologia Simon I. Patiño, Santa Cruz-Bolivia.
Nobre, A.D., 2014. O Futuro Climático da Amazônia-Relatório de Avaliação Cientí-
fica, ARA (ARA).
Pacheco, P., 2006. Agricultural expansion and deforestation in lowland Bolivia: the
import substitution versus the structural adjustment model. Land Use Policy
23, 205–225. http://dx.doi.org/10.1016/j.landusepol.2004.09.004.
Pacheco, P., de Jong, W., Johnson, J., 2010. The evolution of the timber sector in
lowland Bolivia: examining the influence of three disparate policy approaches.
Policy Econ. 12, 271–276. http://dx.doi.org/10.1016/j.forpol.2009.12.002.
Pontius, R.G., 2002. Statistical methods to partition effects of quantity and location
during comparison of categorical maps at multiple resolutions. Photogramm.
Eng. Remote Sens. 68, 1041–1050.
Pontius, R.G., Cornell, J.D., Hall, C.A.S., 2001. Modeling the spatial pattern of land-use
change with GEOMOD2: application and validation for Costa Rica. Agric. Ecosyst.
Environ. 85, 191–203. http://dx.doi.org/10.1016/S0167-8809(01)00183-9.
 G. Tejada et al. / Environmental Research 144 (2016) 49–63
RAISIG, 2013. Amazonia under Pressure. Amazonian Network of Georeferenced
Socio-environmental Information.
Saatchi, S.S., Harris, N.L., Brown, S., Lefsky, M., Mitchard, E.T.A., Salas, W., Zutta, B.R.,
Buermann, W., Lewis, S.L., Hagen, S., Petrova, S., White, L., Silman, M., Morel, A.,
2011. Benchmark map of forest carbon stocks in tropical regions across three
continents. Proc. Natl. Acad. Sci. 108, 9899–9904. http://dx.doi.org/10.1073/
pnas.1019576108.
Sangermano, F., Toledano, J., Eastman, J.R., 2012. Land cover change in the Bolivian
Amazon and its implications for REDDþ and endemic biodiversity. Landsc.
Ecol. 27, 571–584. http://dx.doi.org/10.1007/s10980-012-9710-y.
SERNAP, 2005. Áreas Protegidas Nacionales de Bolivia-Servicio Nacional de Áreas
Protegidas.
Sivila, H.O., 1977. La problemática agropecuaria y agroindustrial en la zona del or-
iente boliviano. Universidad Católica Boliviana, Instituto de Investigaciones
Socioecónomicas.
Soares-Filho, B.S., Coutinho Cerqueira, G., Lopes Pennachin, C., 2002. DINAMICA – a
stochastic cellular automata model designed to simulate the landscape dy-
namics in an Amazonian colonization frontier. Ecol. Model. 154, 217–235. http:
//dx.doi.org/10.1016/S0304-3800(02)00059-5.
63
Veldkamp, A., Fresco, L.O., 1996. CLUE-CR: an integrated multi-scale model to si-
mulate land use change scenarios in Costa Rica. Ecol. Model. 91, 231–248. http:
//dx.doi.org/10.1016/0304-3800(95)00158-1.
Verburg, P., Kok, K., Pontius, R., Veldkamp, A., 2006. Modeling land-use and land-
cover change. In: Lambin, E., Geist, H. (Eds.), Land-Use and Land-Cover Change
SE-5, Global Change – The IGBP Series. Springer, Berlin Heidelberg,
pp. 117–135. http://dx.doi.org/10.1007/3-540-32202-7_5.
Verburg, P.H., de Koning, G.H.J., Kok, K., Veldkamp, A., Bouma, J., 1999. A spatial
explicit allocation procedure for modelling the pattern of land use change
based upon actual land use. Ecol. Model. 116, 45–61. http://dx.doi.org/10.1016/
S0304-3800(98)00156-2.
Verburg, P.H., Veldkamp, A., 2001. The role of spatially explicit models in land-use
change research: a case study for cropping patterns in China. Agric. Ecosyst.
Environ. 85, 177–190. http://dx.doi.org/10.1016/S0167-8809(01)00184-0.
WB, 1997. Eastern lowlands natural resource management and agricultural pro-
duction project. World Bank Project database [WWW Document]. 〈http://www.
worldbank.org/projects/P006152/eastern-lowlands-project?lang ¼en〉.
Zeballos, Z.H., 2006. Agricultura y desarrollo sostenible. Plural editores, La Paz,
Bolivia.