Modeling deforestation at distinct geographic scales and time periods in Santa Cruz, Bolivia

http://irx.sagepub.com
International Regional Science Review
DOI: 10.1177/0160017604266027
2004; 27; 271 International Regional Science Review
BenoÎt Mertens, David Kaimowitz, Atie Puntodewo, Jerry Vanclay and Patricia Mendez
Modeling Deforestation at Distinct Geographic Scales and Time Periods in Santa Cruz, Bolivia
http://irx.sagepub.com/cgi/content/abstract/27/3/271
The online version of this article can be found at:
Published by:
http://www.sagepublications.com
On behalf of:
American Agricultural Editors' Association
can be found at:International Regional Science Review Additional services and information for
http://irx.sagepub.com/cgi/alerts Email Alerts:
http://irx.sagepub.com/subscriptions Subscriptions:
http://www.sagepub.com/journalsReprints.navReprints:
http://www.sagepub.com/journalsPermissions.navPermissions:
http://irx.sagepub.com/cgi/content/refs/27/3/271
SAGE Journals Online and HighWire Press platforms):
(this article cites 8 articles hosted on the Citations
© 2004 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
by Jerome Vanclay on November 17, 2007 http://irx.sagepub.comDownloaded from

10.1177/0160017604266027 ARTICLEINTERNATIONAL REGIONAL SCIENCE REVIEW (Vol. 27, No. 3, 2004)Mertens et al. / MODELING DEFO ES ATION IN BOLIVIA
MODELING DEFORESTATION AT DISTINCT
GEOGRAPHIC SCALES AND TIME PERIODS
IN SANTA CRUZ, BOLIVIA
BENOÎT MERTENS
Center for International Forestry Research (CIFOR), France, benoit.mertens@cgiar.org
DAVID KAIMOWITZ
Center for International Forestry Research (CIFOR), Indonesia, d.kaimowitz@cgiar.org
ATIE PUNTODEWO
Center for International Forestry Research (CIFOR), Indonesia, a.puntodewo@cgiar.org
JERRY VANCLAY
Southern Cross University, Australia, jvanclay@scu.edu.au
PATRICIA MENDEZ
UTD-PLUS Prefectura del Departamento de Santa Cruz, Bolivia, utdplus@hotmail.com
This article analyzes geo-referenced data to elucidate the relations between deforestation and
access to roads and markets, attributes of the physical environment, land tenure, and zoning poli-
cies in Santa Cruz, Bolivia. It presents separate models for Santa Cruz as a whole and for seven
different zones within Santa Cruz, as well as for two different time periods (pre-1989 and 1989 to
1994). The relation between deforestation and the explanatory variables varies depending on
geographic scale and the zone and time period analyzed. At the department scale, locations
closer to roads and the city and places that have more fertile soils and wetter climates have a
greater probability of being deforested. The same applies to colonization areas. Protected areas
and forest concessions are less likely to be deforested. Nevertheless, in many specific zones, these
variables had no significant impact or actually had the opposite impact than in the entire depart-
ment. Most of these relations were weaker between 1989 and 1994 than in the previous period.
Keywords: Amazon; Bolivia; deforestation; frontier; agriculture development; spatial analy-
sis; tenure systems
1. INTRODUCTION
Research on tropical deforestation has increased rapidly during the past twenty
years, but we still do not understand many aspects of where, why, and how fast for-
est clearing occurs. This reflects the issue’s inherent complexity and limited data
availability, as well as the weaknesses of available research methodologies. Recent
INTERNATIONAL REGIONAL SCIENCE REVIEW 27, 3: 271–296 (July 2004)
DOI: 10.1177/0160017604266027
© 2004 Sage Publications
© 2004 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
by Jerome Vanclay on November 17, 2007 http://irx.sagepub.comDownloaded from

literature reviews that have surveyed the approaches to modeling deforestation
have emphasized the great opportunity presented by the rapidly growing availabil-
ity of spatially referenced data (Lambin 1994; Kaimowitz and Angelsen 1998).
Regression models that analyze such data can be especially good for looking at the
relations between deforestation and explanatory variables such as access to mar-
kets, land tenure, climate, soils, topography, and zoning restrictions. These models
are well suited for predicting where deforestation will occur and generally involve
large samples and reasonably reliable data. Researchers can often test the models’
robustness by measuring what percentage of time they correctly predict which
areas will be deforested.
Previous models that look at the effects of spatially referenced explanatory vari-
ables on deforestation have shown that landholders convert more forest to agricul-
tural use in locations with better access to markets, favorable environmental condi-
tions, and no government restrictions on forest clearing. People are more likely to
clear forests when the forests are physically closer to roads and railroads and it
takes less time to reach them (Chomitz and Gray 1996; Deininger and Minten 1996;
Liu, Iverson, and Brown 1993; Ludeke, Maggio, and Reid 1990; Mamingi et al.
1996; Mertens and Lambin 1997; Nelson and Hellerstein 1995; Sader and Joyce
1988; Rosero-Bixby and Palloni 1996). Similarly, forests near urban markets and
villages have a higher chance of being cleared (Chomitz and Gray 1996; Mertens
and Lambin 1997; Nelson and Hellerstein 1997). So do forests in areas with better
soils (flat, fertile, and adequately drained) and drier climates (Chomitz and Gray
1996; Gastellu-Etchegorry and Sinulingga 1988; Sader and Joyce 1988; Rosero-
Bixby and Palloni 1996). Forest fragments are more at risk than forests in large con-
tinuous areas, and those close to the forest edge are especially vulnerable (Brown,
Iverson, and Lugo 1993; Liu, Iverson, and Brown 1993; Ludeke, Maggio, and Reid
1990; Mertens and Lambin 1997; Rosero-Bixby and Palloni 1996). Protected areas
have a smaller probability of being deforested than nonprotected areas (Deininger
and Minten 1996; Chomitz and Gray 1996). Since deforestation processes have a
lot of inertia, places close to previously deforested locations are more likely to be
deforested (Mertens and Lambin 2000).
This article presents the results from logistical multiple regression models that
analyze the relation between deforestation and spatially referenced explanatory
variables for the Department of Santa Cruz, Bolivia. It has two main objectives: (1)
to provide an empirical analysis of the deforestation processes in that important
region and (2) to demonstrate the potential benefits of examining deforestation at
multiple geographic scales and time periods.
From a practical point of view, it is important to understand deforestation pro-
cesses in Santa Cruz because forest clearing there has accelerated rapidly since the
early 1990s, converting the region into one of the world’s thirty-five main defores-
tation “hot spots” (Achard et al. 1998). Moreover, deforestation in Santa Cruz may
follow quite a distinct pattern from other Latin American regions studied previ-
ously because it has more large-scale mechanized agriculture than in most of those
272 INTERNATIONAL REGIONAL SCIENCE REVIEW (Vol. 27, No. 3, 2004)
© 2004 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
by Jerome Vanclay on November 17, 2007 http://irx.sagepub.comDownloaded from

other regions (Kaimowitz, Mendez, et al. 2002). Farmers have also cleared large
areas of Brazil and Paraguay for mechanized agriculture, and the results from Santa
Cruz may provide insights into deforestation in those regions.
From a methodological perspective, the article highlights the fact that explana-
tory variables influence deforestation in different ways depending on the particular
zone, time period, and geographic scale one looks at. It compares results for pre-
1989 deforestation with those for deforestation that occurred between 1989 and
1994 and compares results from models for Santa Cruz as a whole with those from
seven distinct zones within Santa Cruz. This allows it to show that the same explan-
atory variables can appear to have quite different impacts on deforestation depend-
ing on the scale one looks at or the specific zone or time period. Previous modeling
exercises of this type have largely failed to take that into account.
The article begins with a descriptive analysis of deforestation in Santa Cruz.
Then it discusses the economic theory underlying deforestation models of this type
and the role that scale plays in that theory. Then it present the deforestation models,
including the data and methodology used and the results.
2. DEFORESTATION IN SANTA CRUZ, BOLIVIA
2.1. BASIC BACKGROUND INFORMATION
The Republic of Bolivia is divided into nine departments, of which Santa Cruz is
the largest (see Figure 1). It includes fourteen provinces and covers 364,000 square
kilometers, about one-third of the entire country (Montes de Oca 1989). Although
certain portions of western Santa Cruz reach elevations of more than two thousand
meters above sea level (masl), most of the department lies below five hundred masl,
and geographers have traditionally classified the department as part of Bolivia’s
lowland tropics. The lowland tropics house the great majority of Bolivia’s forest.
In 1994, Santa Cruz had 30.7 million hectares of forest. This accounted for 84.3
percent of the total territory. An additional 3.2 million hectares (8.8 percent) was in
pasture or savanna, most of which is natural and has not had forest cover for a long
time, if ever. Farmers used 2.1 million hectares (5.8 percent) for agriculture. Water
covered most of the remaining 0.4 million hectares (Morales 1996).
As these figures suggest, historically forest clearing in Santa Cruz was limited.
As recently as 1950, the entire department had less than sixty thousand hectares of
cultivated land. Forest clearing slowly accelerated between 1950 and the early
1980s. The government constructed a road between the cities of Cochabamba and
Santa Cruz in the early 1950s and implemented policies encouraging sugar and rice
cultivation in the 1960s. In the 1970s, it provided subsidized agricultural credit. For
much of the period, it also promoted settlement by large and small farmers as part of
formal colonization schemes. Japanese and Mennonites made up a large portion of
the large farmers involved in these schemes (Pacheco 1998). All of these policies
Mertens et al. / MODELING DEFORESTATION IN BOLIVIA 273
© 2004 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
by Jerome Vanclay on November 17, 2007 http://irx.sagepub.comDownloaded from

encouraged forest clearing. Nevertheless, by international standards, the region’s
deforestation rates remained low.
In the past fifteen years, deforestation rates have risen rapidly. Between 1986
and 1990, Capacidad de Uso Mayor de la Tierra (CUMAT; 1992) found that the
Amazonian portion of Santa Cruz (the area north of the 18° parallel) lost 38,000
hectares of forest annually. That region covers 61 percent of Santa Cruz but proba-
bly accounted for a higher percentage of forest clearing during that period.
Between 1989 and 1992, deforestation in the entire department of Santa Cruz had
risen to around 78,000 hectares annually. Between 1992 and 1994, annual defores-
tation reached 117,000 hectares (Morales 1993, 1996).
Since 1986, deforestation in Santa Cruz has occurred under distinct economic
conditions and government policies. The structural adjustment policies initiated in
that year have favored export production more than production for domestic con-
sumption. There has been less government support for agricultural colonization
schemes and less subsidized credit. In addition, favorable exchange rates, preferen-
tial access to the Andean Common Market, improved transportation infrastructure,
and the development of the necessary technology and processing facilities have
greatly stimulated soybean production (Pacheco 1998). This has led to an increase
in deforestation by large-scale farmers who grow soybeans for export while dis-
couraging forest clearing by small-scale agricultural colonists (Kaimowitz, Thiele,
and Pacheco 1999).
274 INTERNATIONAL REGIONAL SCIENCE REVIEW (Vol. 27, No. 3, 2004)
FIGURE 1. Study Area
© 2004 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
by Jerome Vanclay on November 17, 2007 http://irx.sagepub.comDownloaded from

2.2. DATA USED IN THE DESCRIPTIVE ANALYSIS AND THE MODELS
The remainder of this article relies heavily on the geographical information sys-
tem (GIS) data sets of the Natural Resource Protection Project of the Departmental
Government of Santa Cruz, the Bolivia Sustainable Forest Management Project
(BOLFOR), and the Center for Research on the Management of Renewable
Resources (CIMAR). This section describes those data sets.
The data sets include information on (1) land cover (1989 and 1994), (2) soil
aptitude for agriculture, (3) rainfall, (4) transport infrastructure, (5) forest conces-
sions, (6) colonization zones, and (7) protected areas. Most of the data was digitized
from 1:250,000 scale maps and converted into ARC Info format.
The land cover data comes from interpretations of Landsat satellite images. It
was provided in raster form, at a spatial resolution of one by one square kilometer
(Morales 1993). With such low spatial resolution, one can only analyze relatively
large changes in forest cover. Hence, the analysis may not fully reflect the clearing
of small areas of forest for shifting cultivation. Cloud cover was minimal in the
1994 images. However, no data was available for certain parts of eastern Santa Cruz
(see Figure 2).
Mertens et al. / MODELING DEFORESTATION IN BOLIVIA 275
FIGURE 2. Deforestation in Santa Cruz, Bolivia (1989-94)
© 2004 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
by Jerome Vanclay on November 17, 2007 http://irx.sagepub.comDownloaded from

The 1989 land cover data has separate categories for forest, agricultural land,
savanna and pasture, areas with little or no vegetation, water, and urban areas. The
1994 land cover data further divides the agricultural area into traditional agricul-
ture, commercial agriculture, anthropogenic pastures, mixed agriculture, and agri-
culture with forests (Morales 1996). The land cover data provides no information
about forest degradation resulting from logging and other activities. It only pro-
vides information about deforestation, which involves the complete removal of for-
est cover. In the discussion that follows, all references to deforestation prior to
1989 refer to the agricultural area category of the 1989 land cover data set. The
assumption is that all land in agriculture in 1989 had previously been in forest. The
data for the variable deforestation between 1989 and 1994 was derived by subtract-
ing the 1989 agricultural area from the 1994 agricultural area. The variable dis-
tance from previous deforested areas was computed as one-kilometer-size buffer
zones from the agricultural areas in the 1989 data set.
The soil aptitude data follow the U.S. Department of Agriculture (USDA) clas-
sification scheme, which classifies land on a scale from I to VIII. Type I areas have
the highest agricultural potential. Type VIII areas have the lowest potential. The
classification takes into account soil fertility, depth, texture, slope, salinity, and
chemical toxicity. The Natural Resource Protection Project of the Departmental
Government of Santa Cruz assembled the data using secondary sources, satellite
interpretation, aerial photography and observation, ground truthing, and soil sam-
pling (Prefectura del Departamento—Consorcio IP/CES/KWC 1996).
Rainfall refers to average annual precipitation and has been divided into discrete
classes by rounding off to the nearest one hundred millimeters. No information was
available regarding what sources the government of Santa Cruz used to prepare its
rainfall map.
The Santa Cruz Natural Resource Protection Project assembled its information
on primary and secondary roads, trails, and railroads using secondary sources and
ground truthing. The road data include all classified roads for the year 1993. It
would have been preferable to use road data from prior to 1989 to avoid the possi-
bility that certain roads that existed in 1993 might have been constructed in
response to deforestation that occurred between 1989 and 1994 rather than the
other way around. Unfortunately, no such data was available. The trails data
include temporary roads constructed for logging and petroleum exploration.
The GIS data set includes the three protected areas with forest that existed in
Santa Cruz in the early 1990s: the Amboro National Park, the Noel Kempff
Mercado National Park and Biological Reserve, and the Rios Blanco y Negro Wild-
life Reserve. The Bolivian government established both Amboro and Noel Kempff
Mercado prior to 1989, although it expanded Amboro in 1991. The Rio Blanco and
Negro Wildlife Reserve was created in 1990.
The forest concession boundaries used in the analysis were obtained from the
BOLFOR GIS database and come from Bolivia’s Forestry Development Center
(CDF), a government agency. During the period covered, forest concessions fre-
276 INTERNATIONAL REGIONAL SCIENCE REVIEW (Vol. 27, No. 3, 2004)
© 2004 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
by Jerome Vanclay on November 17, 2007 http://irx.sagepub.comDownloaded from

quently overlapped with other types of property, including private land holdings,
mining concessions, and even protected areas. Most of the forest concessions were
granted before 1989.
Santa Cruz has government-sponsored small-farm colonization zones and Men-
nonite and Japanese agricultural colonies, populated by larger farmers. The coloni-
zation zone data include both and comes from a map CIMAR produced. The map
has a scale of 1:1,000,000 and should be considered a first approximation (de Vries
1994).
2.3. DEFORESTATION TRENDS IN SEVEN SELECTED ZONES OF SANTA CRUZ
For the purposes of this study, the authors divided a portion of the department of
Santa Cruz into seven separate zones based on the type of farmer (large, small), pro-
duction system (agriculture, ranching), and settlement patterns (spontaneous or
directed settlement, new or old). The seven zones are the (1) commercial integrated
zone, (2) southern expansion zone, (3) northern expansion zone, (4) northwestern
colonization zone, (5) northern colonization zone, (6) western colonization zone,
and (7) eastern ranching zone (see Table 1 and Figure 3).
The first six zones, which fall within a radius of approximately two hundred
kilometers from the city of Santa Cruz, cover only 20 percent of the department’s
total area, but together they accounted for 78 percent of total deforestation between
1989 and 1994. The average annual rate of deforestation in these zones during that
period was 1.5 percent, almost five times higher than the departmental average,
which was 0.33 percent for the same period (Morales 1993, 1996).
The rest of the department, which includes zone 7 and areas not included in any
of the seven zones, has a very low deforestation rate of less than 0.1 percent of forest
loss per year. The authors modeled zone 7 separately because it seemed important
to compare the factors influencing deforestation in a low deforestation/ranching
area with those that affect in the higher deforestation, more agricultural, zones. No
Mertens et al. / MODELING DEFORESTATION IN BOLIVIA 277
TABLE 1. Characteristics of the Seven Defined Zones of Santa Cruz
Type of
Zone Farmer Settlement Colonization Soil Accessibility
1 Large, commercial Old Spontaneous and directed Good Good
2 Large, commercial New Spontaneous and directed Good Good but
recent
3 Small Old Directed Medium Good
4 Small Old Directed Good Mixed
5 Small New Spontaneous Poor Mixed
6 Small Mixed Directed Poor Mixed
7 Ranchers Mixed Spontaneous Medium Poor
© 2004 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
by Jerome Vanclay on November 17, 2007 http://irx.sagepub.comDownloaded from

separate model was developed for the 51 percent of the department outside the
seven zones. Not only does it have extremely low deforestation rates, but the great
diversity in the social processes and production systems in these areas would have
made it difficult to interpret the results. The areas left out of the seven zones include
the Chaco region to the south and the area southwest of the city of Santa Cruz.
Annual precipitation in the Chaco often falls below eight hundred millimeters. Dry
forest and scrub makes up a large portion of the natural vegetation in that region,
and the arid climate discourages crop production. Steep slopes limit agricultural
development in the area southwest of the city of Santa Cruz. That area has some of
the department’s oldest settlements but lacks economic dynamism and has a poor
road network (Davies 1994).
Zone 1 (commercial integrated zone). The commercial integrated zone consti-
tutes Santa Cruz’s traditional center of large-scale commercial agriculture and was
one of the earliest areas to undergo large-scale deforestation. Farmers had already
cleared a large portion of its forest by 1980. It has good soils and a favorable cli-
mate, as well as the department’s most developed transportation infrastructure and
its largest market. More than a quarter of the area in this zone lies less than five kilo-
meters from a primary road. Large farms owned by both native-born Bolivians and
Japanese and Mennonite farmers that settled in the government-sponsored coloni-
zation schemes dominate the area (Pacheco 1998). In the 1970s, the farmers grew
278 INTERNATIONAL REGIONAL SCIENCE REVIEW (Vol. 27, No. 3, 2004)
FIGURE 3. Subregions (Zones) Considered
© 2004 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
by Jerome Vanclay on November 17, 2007 http://irx.sagepub.comDownloaded from

mostly sugar cane, maize, cotton, and sorghum and grazed cattle. More recently,
cotton has declined, while wheat and soybeans have emerged as major crops
(Camara Agropecuaria del Oriente [CAO] 1996). Between 1989 and 1994, this
zone had the highest annual deforestation rate, measured as a percentage of the
remaining area in forest. However, the absolute amount of forest clearing was only
a fraction of that found in the following zone (see Table 2). As a result of the wide-
spread deforestation over the years, the great majority of the remaining forest is
close to previously deforested areas, the so-called forest edge.
Zone 2 (southern expansion zone). Following the mid-1980s, the major focus
for forest clearing in Santa Cruz shifted eastward from the commercial integrated
zone to the southern expansion zone. This zone had by far the highest absolute level
of deforestation between 1989 and 1994 (see table 2). Large-scale farmers cleared
big blocks of land, mostly for mechanized soybean and wheat production. The zone
has an excellent climate and the soils are good for growing soybeans, although they
are fragile and easily susceptible to compaction and wind erosion (Kaimowitz,
Thiele, and Pacheco 1999). Poor access to markets and the fact that there were still
cheap and fertile lands available closer to Santa Cruz limited expansion into this
zone prior to the mid-1980s. Since then, the Bolivian government has greatly
improved the road network in the area and given out large land grants to wealthy
farmers, and it has become increasingly expensive to purchase land in the inte-
grated commercial zone. In addition, the exchange rate devaluation in the mid-
1980s and technological progress in soybeans encouraged the expansion of mecha-
nized soybean production in this area, which was particularly suited for the crop
(Kaimowitz, Thiele, and Pacheco 1999). As a result, forest clearing advanced
rapidly eastward.
Zone 3 (northern expansion zone). This area lies northwest of the southern
expansion zone. Government road building and agricultural colonization schemes
promoted small-farmer settlement there in the 1970s. More than 60 percent of all
the land in this zone forms part of colonization zones, and these zones account for
an even higher percentage of total deforestation (see Table 3). As a result, forest
clearing follows the classic fish-bone pattern along the secondary roads. Since the
1970s, spontaneous migration to the area has continued, and has extended beyond
the original colonization zones. As late as 1989, 75 percent of the zone was still for-
ested, but between 1989 and 1994, forest clearing proceeded extremely rapidly. By
1994, only 64 percent of the land still had forest cover. Most of the farmers practice
shifting cultivation and grow rice as their main crop.
Zone 4 (northwestern colonization zone). Colonization schemes represent only
13.2 percent of the total area in this zone. Nevertheless, the zone includes some of
the oldest and most successful small-farm agricultural colonization schemes, dat-
ing back to the early 1960s (Thiele 1995; Pacheco 1998). Spontaneous migration
Mertens et al. / MODELING DEFORESTATION IN BOLIVIA 279
© 2004 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
by Jerome Vanclay on November 17, 2007 http://irx.sagepub.comDownloaded from

TA
BL
E
2.
Fo
re
st
a
nd
F
o
re
st
C
ov
er
C
ha
ng
es
in
S
an
ta
C
ru
z i
n
th
e S
ev
en
D
ef
in
ed
Z
on
es
a
nd
O
ut
sid
e t
he
Z
on
es
To
ta
l
Ra
te
o
f
To
ta
l A
re
a
Fo
re
st
1
98
9
Pe
rc
en
ta
ge
o
f
Pe
rc
en
ta
ge
o
f
D
efo
res
ta
tio
n
D
efo
re
st
at
io
n
(sq
ua
re
(sq
ua
re
La
nd
in
F
o
re
st
Fo
re
st
1
99
4
La
nd
in
F
o
re
st
19
89
-9
4
(an
n
u
a
l p
er
ce
n
ta
ge
Zo
ne
ki
lo
m
et
er
s)
ki
lo
m
et
er
s)
(19
89
)
(sq
ua
re
ki
lo
m
et
er
s)
(19
94
)
(he
cta
re
s)
o
f fo
res
t c
ov
er
)
1
9,
36
7
3,
26
7
34
.
9
2,
59
6
27
.
7
67
,1
00
4.
1
2
12
,0
41
10
,5
78
87
.
8
8,
99
9
74
.
7
15
7,
90
0
3.
0
3
4,
22
1
3,
17
0
75
.
1
2,
69
2
63
.
8
47
,8
00
3.
0
4
15
,6
99
12
,5
19
79
.
7
12
,0
61
76
.
8
45
,8
00
0.
7
5
21
,3
86
16
,7
82
78
.
5
16
,3
87
76
.
6
39
,5
00
0.
5
6
10
,4
54
8,
58
4
82
.
1
8,
19
3
78
.
4
39
,1
00
0.
9
7
10
5,
87
1
86
,0
36
99
.
3
85
,7
17
98
.
8
31
,9
00
0.
1
O
ut
sid
e
zo
ne
s
18
5,
57
6
17
3,
36
6
99
.
1
17
2,
58
7
98
.
7
77
,9
00
0.
1
To
ta
l
36
4,
61
5
31
4,
45
9
86
.
2
30
9,
34
2
84
.
4
51
1,
70
0
0.
3
280
© 2004 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
by Jerome Vanclay on November 17, 2007 http://irx.sagepub.comDownloaded from

followed these schemes in the 1970s. It is a humid region with fertile soils, where
rice does well. Other crops and livestock have become more important as the small
farmers have become more capitalized over time. Access to primary and secondary
roads in this zone is poorer than in the other zones near Santa Cruz. In recent years,
the annual deforestation, measured as a percentage of total forest area, has been has
been much lower in this zone than in the previously mentioned zones, although a
significant amount of forest has been lost nonetheless. Much of the forest loss has
been within forest concessions, particularly the El Chore Forest Reserve, where
there have been many conflicts between timber companies and spontaneous
migrants. Between 1989 and 1994, fully 80 percent of all deforestation occurred in
forest concessions (see Table 4).
Mertens et al. / MODELING DEFORESTATION IN BOLIVIA 281
TABLE 3. Colonization Zones and Deforestation
Percentage of Percentage of
Percentage of Deforested Area Deforested Area
Total Area in in Colonization in Colonization
Zone Colonization Zones Zones, pre-1989 Zones, 1989-94
1 33.2 34.3 37.3
2 20.0 48.5 28.5
3 62.5 76.1 70.7
4 13.2 34.5 9.6
5 0.5 1.3 0.8
6 12.1 17.7 7.7
7 0.0 0.0 0.0
Santa Cruz 3.3 29.3 22.2
TABLE 4. Forest Concessions and Deforestation
Percentage of Percentage of
Percentage of Deforested Area in Deforested Area in
Total Area in Forest Concessions, Forest Concessions,
Zone Forest Concessions pre-1989 1989-94
1 0.0 0.0 0.0
2 1.0 0.1 0.0
3 2.3 0.4 3.3
4 75.0 37.7 79.0
5 75.1 68.9 72.4
6 2.6 5.2 0.3
7 45.0 13.6 14.7
Santa Cruz 43.0 11.0 17.6
© 2004 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
by Jerome Vanclay on November 17, 2007 http://irx.sagepub.comDownloaded from

Zone 5 (northern colonization zone). This area has been settled more recently
than the previously mentioned zones and with less government support. It tends to
have poorer soils. In part as a result, it has the lowest deforestation rate as a percent-
age of total forest of any of the four colonization zones. Forest still covered 76.6
percent of the area in 1994. Forest concessions account for fully three-quarters of
the zone’s area (see Table 4).
Zone 6 (western colonization zone). This zone has some small-farm coloniza-
tion zones, but they only cover 12 percent of the total area. The soils are poor. The
Amboro National Park covers almost 40 percent of the entire zone (see Table 5).
Amboro has been the subject of conflict between local farmers and government
authorities. Even so, less than 2 percent of all the deforestation in the zone has
occurred within the protected area.
Zone 7 (eastern ranching zone). Moving east from the expansion zone, one hits
the “Brazilian” shield, characterized by infertile and acidic soils. Cattle ranching
and logging are the main activities there, although some small-farm agriculture
exists. During the period under study, a large portion of the land was in timber con-
cessions. Ranchers have cleared a moderate amount of forest for pasture, but it is
difficult to distinguish these areas from natural savanna in the satellite images.
2.4. ADDITIONAL DESCRIPTIVE STATISTICS ABOUT
DEFORESTATION IN SANTA CRUZ
Government-sponsored colonization zones cover only 3 percent of Santa Cruz
but account for 20 percent of all deforestation (see Table 3). Zones 1, 2, and 3
include the majority of colonization zones. A large proportion of the colonization
area had already been deforested prior to 1989. Nevertheless, large areas of forest
continued to be lost in these areas between 1989 and 1994.
282 INTERNATIONAL REGIONAL SCIENCE REVIEW (Vol. 27, No. 3, 2004)
TABLE 5. Protected Areas and Deforestation
Percentage of Percentage of Percentage of
Total Area in Deforested Area in Deforested Area in
Zone Protected Areas Protected Areas, pre-1989 Protected Areas, 1989-94
1 1.0 0.3 2.2
2 0.0 0.0 0.0
3 0.0 0.0 0.0
4 0.0 0.0 0.0
5 0.0 0.0 0.0
6 39.3 2.0 1.5
7 7.6 0.0 0.3
Santa Cruz 7.9 0.3 0.6
© 2004 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
by Jerome Vanclay on November 17, 2007 http://irx.sagepub.comDownloaded from

As of 1989, 43 percent of the land in the department was in forest concessions
(see Table 4). These were mainly located far from the city of Santa Cruz in the
north, northwest, and eastern parts of the department (zones 4, 5, and 7). About 17
percent of all deforestation has occurred in forest concession areas, most of which
is concentrated in zones 4 and 5, to the north and northwest of the city of Santa
Cruz. In fact, in these two zones, more than 70 percent of all deforested areas are
located within forest concessions. Protected areas cover approximately 8 percent of
the department, but only 1 percent of the total deforestation has taken place within
these areas (see Table 5).
Almost 56 percent of total deforestation in the period from 1989 to 1994
occurred less than one hundred kilometers from the city of Santa Cruz (see Table 6).
Although less than 20 percent of all land in Santa Cruz was within five kilometers of
a previously forested area that had been cleared for agriculture, almost 80 percent
of the deforestation during that period was in those areas (see Table 7).
Only 10.5 percent of the land in Santa Cruz is less than ten kilometers from a pri-
mary road. Still, 30.5 percent of the deforestation between 1989 and 1994 was in
such areas. This relation does not apply, however, to zones 1, 2, and 3, which have
already lost a fair portion of their forest and experienced the highest rates of defor-
estation during that period. Most of the land in these zones close to primary roads
that has favorable characteristics for agriculture had already been deforested prior
to 1989, and that may explain this seemingly contradictory result (see Table 8).
Mertens et al. / MODELING DEFORESTATION IN BOLIVIA 283
TABLE 6. Deforestation and Distance from Santa Cruz, 1989-94 (in percentages)
Zone 0-50 km 50-100 km 100-150 km 150-200 km >200 km
1 63.0 31.6 5.4 0.0 0.0
(71.1) (22.5) (6.4) (0.0) (0.0)
2 2.7 65.9 31.4 0.0 0.0
(2.5) (46.5) (49.0) (2.0) (0.0)
3 0.0 86.0 14.0 0.0 0.0
(0.0) (87.6) (12.4) (0.0) (0.0)
4 4.1 48.9 45.4 1.5 0.0
(4.8) (38.7) (48.8) (7.6) (0.0)
5 0.0 6.8 41.0 50.1 2.0
(0.0) (1.0) (33.3) (57.7) (7.9)
6 12.0 78.5 9.5 0.0 0.0
(16.6) (64.0) (19.4) (0.0) (0.0)
7 0.0 0.0 0.0 0.7 99.3
(0.0) (0.0) (0.0) (4.4) (95.6)
Santa Cruz 10.4 45.5 21.1 6.0 16.9
(2.6) (7.3) (10.8) (12.4) (66.7)
Note: Values shown represent total deforested area within the zone located less than a certain distance
from Santa Cruz. Values in parentheses correspond to total area of each zone located within the same dis-
tance threshold.
© 2004 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
by Jerome Vanclay on November 17, 2007 http://irx.sagepub.comDownloaded from

284 INTERNATIONAL REGIONAL SCIENCE REVIEW (Vol. 27, No. 3, 2004)
TABLE 7. Deforestation in Santa Cruz and Distances from the Forest/Nonforest Edge, 1989-
94 (in percentages)
Zone 0-2.5 km 2.5-5 km 5-7.5 km 7.5-10 km >10 km
1 90.3 9.7 0.0 0.0 0.0
(94.5) (4.8) (0.6) (0.0) (0.0)
2 44.9 25.8 13.0 7.7 8.7
(32.9) (17.5) (10.8) (8.3) (30.4)
3 87.2 12.5 0.2 0.0 0.0
(79.0) (16.5) (3.2) (1.0) (0.2)
4 45.4 26.2 17.2 6.8 4.4
(31.4) (15.4) (10.9) (8.4) (33.9)
5 63.8 19.5 7.3 6.3 3.0
(21.2) (16.9) (13.0) (11.0) (37.8)
6 76.2 19.7 2.8 0.3 1.0
(32.1) (22.2) (13.0) (9.0) (23.6)
7 48.0 15.1 9.2 8.2 19.4
(4.8) (6.7) (7.4) (7.9) (73.2)
Santa Cruz 59.8 19.8 8.3 4.6 7.4
(11.6) (7.5) (6.3) (6.1) (68.5)
Note: Values shown represent total deforested area within the zone located less than a certain distance
from the forest/nonforest. Values in parentheses correspond to total area of each zone located within the
same distance threshold.
TABLE 8. Deforestation in Santa Cruz and Distances from Primary Roads, 1989-94 (in per-
centages)
Zone 0-5 km 5-10 km 10-15 km 15-20 km >20 km
1 21.9 21.5 20.1 15.6 20.9
(26.7) (20.9) (16.6) (13.7) (22.1)
2 9.7 6.3 8.9 8.3 66.7
(8.1) (6.3) (5.4) (4.7) (75.5)
3 13.6 9.0 8.6 21.1 47.7
(10.3) (10.6) (13.5) (14.6) (51.0)
4 3.5 9.6 6.1 7.6 73.1
(4.5) (3.0) (3.2) (3.5) (85.9)
5 23.3 21.0 27.8 11.4 16.5
(10.4) (10.1) (10.4) (9.6) (59.5)
6 48.3 25.8 13.8 7.2 4.9
(18.9) (18.4) (16.7) (14.8) (31.2)
7 25.0 7.9 16.8 11.2 39.1
(4.3) (3.9) (3.8) (3.9) (84.0)
Santa Cruz 19.1 11.6 11.9 10.1 47.3
(5.6) (4.9) (4.8) (4.5) (80.1)
Note: Values shown represent total deforested area within the zone located less than a certain distance
from a primary road. Values in parentheses correspond to total area of each zone located within the same
distance threshold.
© 2004 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
by Jerome Vanclay on November 17, 2007 http://irx.sagepub.comDownloaded from