Patterns and causes of deforestation in the Colombian Amazon
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Ecological Indicators 6 (2006) 353–368
This article is also available online at:
www.elsevier.com/locate/ecolind
Patterns and causes of deforestation in the Colombian Amazon
Dolors Armenteras a,b,*, Guillermo Rudas c, Nelly Rodriguez a,
Sonia Sua a, Milton Romero a
a
Biological Resources Research Institute Alexander von Humboldt, GIS Unit, Carrera 7#35-20, Bogotá, Colombia (South America)
b
Department of Geography, King’s College London Strand, London WC2R 2LS, UK
c
Department of Economics, Javeriana University, Calle 40 N 6-23, Bogotá, Colombia (South America)
Accepted 29 March 2005
Abstract
Ecosystem information on the Colombian Amazonia is poor in comparison with that on the Brazilian Amazon. We examined
patterns of ecosystem diversity, deforestation and fragmentation and provided an estimate on their possible causes through a
temporal and spatial analysis of biotic and abiotic data using remote sensing and geographical information systems in six pilot
areas covering a total of 4,200,000 ha. Ecological, demographic and socio-economic data were analysed to establish the local
conditions. We used a landscape ecology approach to calculate indicators of ecosystem diversity, cover and forest fragmentation
such as number of patches, mean patch size, mean shape index and mean nearest neighbour distance. Patterns of deforestation
did not run parallel to access roads; instead the typical pattern of unplanned colonization follows the only transportation network
existing in many areas in the Colombian Amazonia: rivers. In addition, we have used indicators of human influence such as
demographic pressure, quality of life and economic activity indicators. Results show that the extent and rate of change varies
between areas depending on population density. Annual deforestation rates were 3.73 and 0.97% in the high population density
growth areas of Alto Putumayo and Macarena respectively, and 0.31, 0.23, and 0.01% in the relatively unpopulated areas of
indigenous population. These changes are related to land use history as well as to environmental and historical socio-economic
factors such as oil extraction, deforestation, cattle ranching or illegal cropping. The current situation in the region suggests that
tropical deforestation rates in the Colombian Amazon are substantially higher than those found in previous studies in the rest of
the Amazon.
# 2005 Elsevier Ltd. All rights reserved.
Keywords: Fragmentation; Satellite imagery; Tropical deforestation; Land use change; Biodiversity; Indicators; Amazonia; Colombia
1. Introduction
* Corresponding author. Tel.: +57 1 6086900x238; The destruction of tropical forests has received
fax: +57 1 6086900.
E-mail addresses: darmenteras@humboldt.org.co,
worldwide attention due to the significance of forest
dolors.armenteras@kcl.ac.uk (D. Armenteras), on global climate, carbon sequestration, water cycles,
grudas@javeriana.edu.co (G. Rudas). biodiversity and the potential global effects on climate
1470-160X/$ – see front matter # 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ecolind.2005.03.014354 D. Armenteras et al. / Ecological Indicators 6 (2006) 353–368
change (Fearnside, 1995; Fearnside et al., 2001). transformation in Colombia remains largely unknown,
Globally, some estimates suggest that 9 million km2 of such that entire ecosystems could be under threat of
tropical humid forests have been lost in less than 50 disappearance. Global conservation prioritisation
years and that current rates of extinctions are not only (Myers et al., 2000) proposed the northern Andes
high but accelerating (Pimm et al., 2001). Achard et al. and the Choco regions as two hotspots in Colombia,
(2002) estimate an annual deforestation rate of 0.38% while the Amazonia was classified merely as a ‘‘major
of humid tropical forest in Latin America. wilderness area’’. This ‘‘hotspot’’ approach is directed
Deforestation has led to the fragmentation of towards decision-makers, but devalues and detracts
natural ecosystems throughout the world causing attention away from non-designated areas (Bates and
further loss of original forests, reduction of the size of Demos, 2001).
forest fragments and increasing isolation. Most studies There is evidence that ecosystems in half of the
on ecosystem cover and fragmentation are centred on Colombian Amazon are experiencing high rates of
the quantification of those changes (Vogelmann, 1995; deforestation. Ruiz (1989) estimated that 2.5 mil-
Ranta et al., 1998; Sierra, 2000; Steininger et al., lion ha of forest were lost in the late 1980s in
2001a, 2001b) and on the effects that these can have on Colombia. Sierra (2000) analysed the extent and rate
ecological processes (Klein, 1989; Carvalho and of deforestation and the level of forest fragmentation
Vasconcelos, 1999; Gascon et al., 1999; Davies and in the Napo region of western Amazonia, which
Margules, 1998; Laurance et al., 1998, 2000; Nepstad included a small portion of the Colombian Amazon,
et al., 1999). The Amazon hosts over half of the and concluded that deforestation was advancing faster
world’s remaining tropical forests and it is currently on the Colombian side of his study area (0.9%/year)
subject to accelerating deforestation and changing due to population growth from the foot of the Andes
patterns of ecosystem loss (Laurance, 1998; Whit- towards the Amazon.
more, 1997; Lima and Gascon, 1999). Laurance et al. Detailed and updated studies are still lacking in
(2002) suggest, among others, that Brazilian Amazo- Colombia. There are no regional geographic databases
nia has the world’s highest absolute rates of forest of current information on the dynamics and patterns of
deforestation and fragmentation. land cover change and the levels and patterns of
However, while Brazilian Amazonia deforestation fragmentation and ecosystem integrity for this part of
has been widely analysed (Fearnside, 1990; Fearnside the world (Sierra, 2000). Hence the aim of this study is
et al., 1990; Fearnside, 1995; Reis and Margulis, 1991; to provide an estimate of both deforestation and
Laurance, 1998; Parayil and Tong, 1998; Laurance ecosystem transformation and probable causes of
et al., 2002) and much emphasis has been placed on change using an analysis which incorporates both
this part of the world, information on other parts of the biotic and anthropogenic data for a portion of the
Amazon, in particular the Colombian Amazonia, is Colombian Amazon. An approach that incorporates
scarce or non-existent. While there are a number of both factors is critical for understanding of the
studies on social, demographic and economic deter- consequences of human activities on the natural
minants of the Amazonian deforestation which environment in the Colombian Amazon. As a first
suggest that deforestation is primarily determined attempt towards monitoring and analysing the state of
by human population, accessibility, land use and land natural ecosystems in the Amazonia, we have chosen
tenure issues (Reis and Margulis, 1991; Fearnside, to use remote sensing (RS) and a geographic
1993; Wood and Skole, 1998; Laurance, 1998; information system (GIS). In addition, we have
Fearnside, 2001; Nepstad et al., 2001; Portela and evaluated possible economic and socio-demographic
Rademacher, 2001; Laurance et al., 2002), these determinants of deforestation.
studies are largely confined to Brazil. The patterns of ecosystem diversity, the spatial
Colombia, having one of the most diverse regions patterns of deforestation and the resulting forest
in flora and fauna in the world, has been identified as a fragmentation patterns that occur in Colombian
‘‘mega-diverse’’ country (IAvH, 1998). While tropical rainforests are totally different to those documented
ecosystem transformation is occurring all over the in Brazil (Batistella et al., 2000) or even in Ecuador
tropics, the loss of biodiversity and landscape (Sierra, 2000). In Colombia, pasture-led deforestationD. Armenteras et al. / Ecological Indicators 6 (2006) 353–368 355 for ranching activities occurs, but in the absence of a Due to variation in geology and geomorphology, the clear state policy, spontaneous colonization has also region yields environments with varying drainage occurred since the early 1970s and both follow rivers systems and soil qualities. This has led to very as the only existing communication network in some significant differences in ecosystem composition and areas. structure that supports a high degree of biological The more recent – and more significant – threats to diversity. The region can be divided into five broad the eastern slopes of the Andes and thus the adjacent vegetation categories (Kalliola et al., 1993; Domı́n- Amazon lowland are the cropping of illicit cultivars. guez, 1987; Prance, 1985; Huber, 1981; Sierra, 1999): Coca cultivation in the Andes, in particular in Peru, Bolivia and Colombia, has been expanding over the (1) Lowland forests
356 D. Armenteras et al. / Ecological Indicators 6 (2006) 353–368
Fig. 1. Location of the study pilot areas (1–6) and 10 protected areas (national park, PNN and national nature reserve, RNN) in the Colombian
Amazonia.
of the study areas were delimited without considering secondary information on climate and geomorphol-
any kind of political boundary. We also analysed the ogy, vectorisation and finally ground-truthing. Areas
information available in the ten national protected transformed by human activities were defined using
areas in the Amazon. These areas represent over 65% the spectral characteristics of deforested sites. This
of all current protected areas in Colombia (Fig. 1). classification included agricultural areas. Standard
methods of accuracy assessment, based on contin-
2.2. Data collection gency tables, were used. Ground truth data were taken
at 250 points at each of the five different transects, one
2.2.1. Ecosystem mapping for each pilot area. In one of the sites (Macarena) field
Remote sensing data from satellite imagery for the work was cancelled due to increased social unrest, so
period 1985–2001 (Landsat MSS, TM, ETM) were accuracy value could not be calculated. Extensive field
used to generate ecosystem maps for all the pilot areas. work was carried out between June and October 2001
In the case of La Chorrera, cloud free satellite in order to verify classes. Misclassified polygons were
information was only available for the year 1985. identified and corrected manually in a GIS. Overall
Major ecosystem types were determined by a accuracy after correction with ground truth data was
combination of supervised classification and manual 93% in the 1980s and 95% in the 2000s. We identify
interpretation of satellite images supplemented with accuracy of 1980s map based on natural ecosystemsD. Armenteras et al. / Ecological Indicators 6 (2006) 353–368 357
that remain untouched in the 2002. We used both an index of quality of life, with values between 0
ERDAS Imagine (ERDAS Inc., 2000) remote sensing and 100 that represent the minimum and maximum
processing software and the GIS software Arcview possible level of population quality of life
(ESRI, 2000) to integrate the data using standard GIS respectively.
features. demographic indices expressed as absolute popula-
As a result of this interpretation, maps of ecosystem tion (number of inhabitants), population density
cover for two different time periods for five of the six (inhabitants/km2) as well as annual population
pilot areas were produced at the 1:250,000 scale. growth rate (%/year).
Ecosystems were identified and classified into biomes an economic activity index, or the percentage of
adapting Walter’s classification (1985; Walter and land area devoted to ranching and farming.
Breckle, 1986) as follows (a biome is defined here as a violence index, the annual percentage of deaths
an assembly of ecosystems with similar structural and that were violent deaths.
functional characteristics): Table 1 presents the results of the demographic and
socio-economic indicators generated for the pilot a-
Amazonia and Orinoquia Tropical Forests; reas and protected areas analysed in this study.
Amazonian Orobiomes;
Andean Orobiomes; 2.3. Data analysis
Amazonian Helobiomes;
Amazonian Litobiomes; Our goal was to offer region specific information to
Peinobiomes; support decision-making in the Colombian Amazon
Transformed ecosystems. with maximum cost effectiveness under budget
restrictions. Furthermore, the data had to be as up
Maps of deforestation were also produced for each to date as possible and presented in an easily
of the pilot areas. In order to facilitate reporting, interpretable way. This paper focuses on quantifying
ecosystem classes were aggregated into (a) biomes and both ecosystem changes and the changes in the spatial
(b) three major ecosystem types (natural, transformed patterns of ecosystems that have taken place over time
and water). Landscapes dominated by land uses in the Amazon. It also points out how they might be
associated with agriculture, pasture or urban sites were related to the changes in the demographic and
assigned the category of transformed ecosystems. economic structure of this area. We reported
quantitative data of land cover change over the last
2.2.2. Census estimates and other socio-economic 20 years in this part of the world. The measures were
indicators generated as part of the Indicators Project (Armenteras
Demographic and economic structure indicators et al., 2002; Rudas et al., 2002) at the Biological
were derived from the population and agricultural Resources Research Institute Alexander von Hum-
census at the municipal level, the only source of this boldt of Colombia. Biotic indicators such as ecosys-
information for the Amazon in Colombia (CGR, 1951; tem extent (ha), change rates (%) and fragmentation
DANE, 1964, 1973, 1985 and1993). The population indices were based on major biomes types map that we
data is split into ‘rural’ and ‘total’ for each of the derived from the remote sensing and ground-truthing
municipal areas in the pilot areas. The information on studies described earlier. In addition, we also analysed
population quality of life was obtained directly from ecosystem information for the ten national protected
the Colombian Departamento Nacional de Planea- areas obtained from a general ecosystem map of
ción. This provided us with a synthetic index that Colombia (Etter, 1998).
includes information on education, family size, With the exception of fragmentation indicators,
household building quality material, water availabil- which were calculated using the software Fragstats
ity, garbage collection, household density and income. (McGarigal and Marks, 1995), the other measures
We used four types of indices to relate levels of these were calculated using standard GIS functions in
indicators to levels of deforestation with the municipal Arcview and ERDAS Imagine and statistical analy-
area as the spatial analysis unit: tical tools such as SPSS v.10.358
Table 1
Socio-economic, demographic and biological indicators: pilot and protected areas in the Amazon
Pilot (PL) and Quality of Population (1993) Population density Annual population growth (%) Pasture area Violent Natural ecosystems
protected areas (PT)a life index (number of (inhabitants/km2)b (% total area) deaths (%) remaining
D. Armenteras et al. / Ecological Indicators 6 (2006) 353–368
inhabitants) (% total area) 2000s
Rural Total Rural Total Rural Total 1973–1985 1985–1993 1973–1993
Alto Putumayo (PL) 42.9 62.4 213549 388062 9.98 18.11 3.7 4.0 3.8 56.6 27.9 27.9
Macarena (PL) 33.4 50.5 96690 127306 2.45 3.16 8.2 6.0 7.4 26.1 38.0 68.5
Inirida-Mataven (PL) n.a. n.a. 13942 18367 0.76 1.00 2.6 6.8 4.3 0.0 0.0 94.3
La Chorrera (PL) n.a. n.a. 20146 22963 0.30 0.32 3.6 7.9 5.3 0.0 0.0 98.3c
Mitú (PL) n.a. n.a. 9768 13896 0.47 0.67 7.7 4.5 2.8 0.0 0.0 91.2
Puré (PL) n.a. n.a. 3028 5227 0.13 0.22 8.3 4.7 6.9 0.0 0.0 99.2
Amacayacu (PT) n.a. 71.7 14539 35083 1.90 5.10 8.8 5.3 7.4 0.4 0.0 99.7
La Paya (PT) 78.6 64.7 32778 54740 2.42 4.03 12.2 6.6 4.7 1.5 15.1 86.6
Nukak (PT) n.a. 61.8 28628 33424 1.18 1.36 0.0 0.0 0.0 13.4 65.3 96.8
Sumapaz (PT)d 44.3 61.0 78597 132308 2.76 4.23 3.8 2.0 1.5 16.3 24.9 99.1
Chiribiquete (PT) n.a. 57.5 8280 10078 0.15 0.18 6.6 8.4 7.3 1.4 31.2 100.0
La Macarena (PT) 36.0 42.4 91162 116369 2.85 3.63 6.6 2.8 5.1 24.8 26.8 79.9
Los Picachos (PT) 35.2 42.1 89963 115084 2.19 2.77 8.3 1.8 5.7 14.1 62.2 97.0
Tinigua (PT) 34.5 40.1 61275 76874 2.45 3.08 7.6 1.0 5.0 9.2 42.4 82.4
Puinawal (PT) n.a. n.a. 22422 26847 0.33 0.39 1.7 12.7 4.1 0.0 0.0 100.0
Cahuinari (PT) n.a. n.a. 3322 4489 0.06 0.08 19.7 10.0 15.8 0.0 0.0 100.0
n.a.: not available.
a
Total indicators for municipal areas with territory in selected pilot and protected areas.
b
Weighted mean by municipal territory participation in total pilot or protected area.
c
1980s data (no information available for the 2000s).
d
Without Bogotá’s indicators.D. Armenteras et al. / Ecological Indicators 6 (2006) 353–368 359
The rate and extent of natural ecosystem loss and independent variables. The percentage of ecosystem
fragmentation was calculated in five of the pilot areas, loss (NED, natural ecosystem degradation, includes
because for one (La Chorrera) cloud free satellite only changes from natural to anthropogenic) was used
information was not available. Matrix information was as the dependent variable. We analysed the correlation
generated to obtain total area of each ecosystem type between variables and undertook complementary
and estimates of patterns and average rates of regression analysis to clarify the levels of statistical
deforestation throughout the study period (1985– confidence in the relationships between some of the
2001) for each pilot area. In order to calculate the analysed variables. Further, we analysed the most
average annual deforestation rate, we assumed that significant determinant of deforestation and projected
this rate is not constant and the following formula was the ecosystem changes over 50 years from the present,
employed: assuming that the same tendencies will prevail.
½LnðAt1 Þ LnðAt0 Þ 100
Loss rate ¼ ;
t1 t0 3. Results and discussion
where A equals the ecosystem area (ha), t1 final year
and t0 initial. The most representative biomes of the six pilot areas,
Forest fragmentation was analysed for the period are tropical humid forests (61.81%) and helobiomes of
1985–2001 using the following landscape metrics: the Amazonia (11.57%). The pilot areas with the
fragment number, size (mean patch size, patch size highest percentage of natural ecosystems over the 15-
standard deviation), shape (mean shape index, mean year period of analysis (1985–2000) are Puré (99.23%),
perimeter/area ratio and mean patch fractal dimension) Inı́rida-Matavén (94.37%) and Mitú (91.23%)
and edge (total edge, edge density and mean patch edge). (Table 2). The area of greatest transformation is the
The results for each index were grouped using 0.5 zone of Alto-Putumayo near the Andes, with only 28%
standard deviation limits into three classes (high, of natural ecosystems left in 2001, followed by
average, and low). We constructed maps showing the Macarena with 68.57% (Table 2). These areas also
degree of transformation and the fragmentation of had the highest average annual rate of natural
natural ecosystems for each pilot area using these three ecosystem loss: the highest rate corresponds to the
categories and combining the results for each index into a Putumayo (3.73%), followed by Macarena (0.97%) and
single fragmentation level that could be illustrated in a followed by Mitú (0.31%), Inı́rida-Matavén (0.23%)
map. Fragstats softwarewas also used to studylandscape and finally Puré with 0.01% annual loss rate (Table 2,
diversity within the five sites, and three different indices Fig. 2).
of landscape diversity were used in order to compare the In general, the relative degree of fragmentation of
pilot areas: the number of ecosystem types, the Shannon each site follows the same order as the above-mentioned
diversity and the evenness index. deforestation rates, with the highly fragmented natural
The smallest spatial unit for which economic and ecosystems in Putumayo and Macarena pilot areas and
demographic data is available is the municipio, which less fragmentation in the other three. The pattern of
has a purely administrative boundary. The census data fragmentation follows the colonization and develop-
were aggregated into a single record for each pilot area ment associated with the rivers, the only transportation
and protected area by weighting the indices by the network in Amazonia (Fig. 3). This pattern of
percentage of the total area of interest that is covered fragmentation and deforestation is clearly very different
by the municipio. This aggregated record data to the ‘‘fishbone’’ patterns in areas of the Brazilian
therefore refers to the characteristics of all the Amazonia and some parts of Ecuador (Sierra, 2000),
municipios within these pilot and protected areas. where the construction of roads is one of the main
In order to analyse the impact of human pressures drivers of deforestation (e.g. in Rondonia, Batistella
on natural ecosystems and find the possible determi- et al., 2000), a determinant that is not apparent yet in the
nants of forest ecosystem loss we undertook simple Colombian Amazonia. These differences may reflect
ordinary least squares (OLS) regression analysis. different periods in the evolution of fragmentation that
Demographic and socio-economic data were treated as in Colombia may be in an earlier stage than some areas360 D. Armenteras et al. / Ecological Indicators 6 (2006) 353–368
Table 2
Extent, natural ecosystem (NE) cover and landscape level metrics for natural ecosystems in six pilot areas of the Colombian Amazonia
Pilot areas Study Natural ecosystems NEa annual Number of Shannon’s Shannon’s
area (ha) remaining (ha) loss rate (%) NE a diversity index evenness index
(% of study area) 1980s 2000s 1980s 2000s 1980s 2000s 1980s 2000s
La Macarena 713,386 560,658 (78.6%) 489179 (68.5%) 0.97% 41 41 3.32 3.28 0.89 0.88
Mataven-Inirida 640,887 628,026 (97.9%) 604826 (94.3%) 0.23% 41 41 3.01 3 0.81 0.81
Mitu 626,786 595,480 (95.0%) 571384 (91.2%) 0.31% 39 39 2.96 2.97 0.81 0.81
Pure 705,056 700,770 (99.4%) 699638 (99.2%) 0.01% 40 40 2.80 2.80 0.78 0.76
Putumayo 802,477 336, 912 (42.2%) 224799 (27.9%) 3.74% 45 45 3.34 3.28 0.88 0.77
Chorrerab 712,116 700,057 (98.3%) – – 30 – 2.66 – 0.77 –
a
NE, natural ecosystems.
b
No information available for the 2000s.
Fig. 2. Deforestation patterns and natural ecosystem loss in five pilot areas of the Colombian Amazonia for a period of 15–20 years (between the
1980s and 2000s).D. Armenteras et al. / Ecological Indicators 6 (2006) 353–368 361
Fig. 3. Degree of natural ecosystem fragmentation in the Macarena Pilot Area for the year 2000.
in Brazil. Further, there are several small, scattered and Landscape diversity results are especially surpris-
isolated deforestation patches usually related to coca ing considering that although ecosystem richness is
growing in the Amazon, present in some of the areas but very similar between areas, ecosystem diversity and
to a major degree present in Alto Putumayo and evenness are higher in the more threatened areas such
Macarena areas. as the Putumayo and Macarena regions. In fact, this
Table 2 also summarizes the indices of landscape result is logical because these two areas are the closest
diversity for all pilot areas: number of ecosystem types, to the Andes and, being topographically much more
Shannon’s diversity (SD) and Shannońs evenness (SE) heterogeneous than the lowlands of the Amazon areas,
index. The areas all have similar ecosystem richness, they contain a higher number of different environ-
but ecosystem diversity and evenness are high in the ments and ecosystems.
more threatened Putumayo and Macarena areas with an In the Putumayo region there has been a major
SD ranging between 3.28 and 3.34 for the two areas, and decrease in the percentage of natural ecosystems
an SE of between 0.88 and 0.89 for the Macarena (the (from 42 to 28%), coincident with the annual change
highest), and between 0.77 and 0.88 for the Putumayo rate of 3.73% discussed above. Fragmentation of
area. On the contrary, the less degraded areas have natural forest has increased dramatically. Current rural
lower SD and SE: (a) Mataven-Inirida has an SD of 3 population density is 9.98 inhabitants/km2 and has
and an SE of 0.81, (b) Mitu has an SD of 2.97 and SE of increased from 1.98 (1951) to 2.97 (1964) to 4.63
0.81 and (c) Pure, the most preserved pilot area has an (1973) to 7.24 (1985). There has also been a
SD of 2.80 and SE of 0.76. substantial decrease in the percentage of natural362 D. Armenteras et al. / Ecological Indicators 6 (2006) 353–368
areas with current high population density (e.g. Alto
Putumayo and Macarena), and in areas with low
population and naturally high growth rates (Vaupés,
La Chorrera and Inı́rida-Matavén). In the former case,
the results indicate that not only do the areas have high
populations but also that the growth processes are still
highly dynamic. In the latter case this result reflects
the fact that although important changes in population
are taking place, population density remains low.
The significant effect of economic activity on
natural ecosystems is also reflected in these results.
Eq. (5) (Table 4) shows that the pasture area (PA, main
legal land use activity) has a significant positive
relationship with degraded ecosystems. However, it is
Fig. 4. Evolution of rural population density around pilot areas over not appropriate, to undertake a multivariate analysis
the last five population censuses in Colombia. Source: Rudas et al.
(2002) and Armenteras et al. (2002).
using this variable and population variables since
these variables are highly correlated (Eqs. (7)–(10),
Table 4). Another significant result is that neither
ecosystems (from 79 to 68%) similar to the Macarena quality of life nor violence levels are statistically
region, although the annual change rate is 0.86% which related with ecosystem degradation (Eqs. (11) and
is lower than Macarena. Natural ecosystems are (12), Table 4).
becoming increasingly fragmented. Rural population Overall, population processes and the main
density in the Macarena region is lower than in the economic activity have a very significant impact on
Putumayo pilot area at 2.45 inhabitants/km2, a popula- natural ecosystems degradation in these areas. For
tions increased from 0.56 in 1973 to 1.51 in 1985. instance, an increment of one inhabitant per square
The results also show that areas that are more kilometer would generate a loss of natural ecosystems
degraded coincide with areas of high population of more than 7% (Eq. (4b), Table 3). As a result, a
density pressure (Fig. 4) and low quality of life. deforestation simulation model was developed in
Information analysis results demonstrate a statistically order to project future tendencies of natural ecosystem
significant relationship between demographic pres- degradation in the area. Since rural population density
sures and the percentage of natural cover lost (Fig. 2). (RPD) was the most significant determinant of natural
In fact, as shown in Eqs. (1)–(4) of Table 3, there is a ecosystem degradation (NED) with a r2 = 0.86
statistically significant correlation of both absolute (Eq. (4a), Table 3), we used this factor to simulate
population and population density with natural ecosystem loss. The values were area weighted.
ecosystem degradation (NED). Furthermore the result Roughly speaking, NED equals 6.61 RPD. RPD was
is significant for total population (TP), rural popula- linearly projected over the next 10, 20, 30, 40 and 50
tion (RP) as well as for total and rural population years using the minimum (0.33%), maximum
density (TPD and RPD). It is important to note that for (17.25%) and average (6.8%) RPD of the last five
models with demographic variables the intercept was demographic censuses in the study area. We then
not significant, and so these models were re-estimated constructed three different future scenarios assuming
ignoring the intercept. This is based on the assumption that in the 1990s the ecosystem degradation was zero.
that with zero population, ecosystem degradation The results of applying this model can be seen in
would also be zero. Fig. 5. According to the model, the 16 sites of the
Contrary to the significant results of the above- Colombia Amazonia studied will suffer significant
mentioned demographic variables, annual population natural ecosystem cover loss (Fig. 5). In fact, only
growth rates (APG) are not significantly related to under a scenario of low rural population density will
natural ecosystem degradation NED (Eq. (6), Table 4). Amazonian forests be relatively protected. Both
This is because high growth rates are present in both average and high rural population density projectionsD. Armenteras et al. / Ecological Indicators 6 (2006) 353–368 363
Table 3
Statistical results of the regression analysis of population and economic indicators as predictors of natural ecosystem degradation
Equation Est Var Predictor Coefficient t P > jtj n F P>F R2 R2adjusted
(1a) NED 16 59.35 0.000 0.81 0.80
ŷ ¼ 1:75TP
Intercept 1.76 0.65 0.526
TP 1.75 7.70 0.000
(1b) NED 16 90.62 0.000 0.86 0.85
ŷ ¼ 1:66TP
TP 1.66 9.52 0.000
(2a) NED 16 48.15 0.000 0.77 0.76
ŷ ¼ 2:97RP
Intercept 3.42 1.10 0.290
RP 2.97 6.94 0.000
(2b) NED 16 69.87 0.000 0.82 0.81
ŷ ¼ 2:65RP
RP 2.65 8.36 0.000
(3a) NED 16 48.49 0.000 0.78 0.76
ŷ ¼ 3:76TPD
Intercept 0.17 0.06 0.953
TPD 3.76 6.96 0.000
(3b) NED 16 77.69 0.000 0.84 0.83
ŷ ¼ 3:74TPD
TPD 3.74 8.81 0.000
(4a) NED 16 6.30 0.000 0.83 0.82
ŷ ¼ 7:08RPD
Intercept 2.25 0.88 0.393
RPD 7.08 8.32 0.000
(4b) NED 16 102.08 0.000 0.87 0.86
ŷ ¼ 6:61RPD
RPD 6.61 10.10 0.000
(5) NED 16 55.75 0.000 0.80 0.79
ŷ ¼ 1:08PA
Intercept 0.11 0.04 0.968
PA 1.08 7.47 0.000
NED = natural ecosystem degradation (%) in 1990s. TPD = total population density, inhabitants/km2 (1993). TP = total population, 10.000
inhabitants (1993). RPD = rural population density, inhabitants/km2 (1993). RP = rural population, 10.000 inhabitants (1993). PA = pasture
area (%). NED = natural ecosystem degradation which is the percentage of natural land cover converted to anthropogenic land cover; TP = total
population; TPD = total population density; RP = rural population and PA = percentage of the area under pasture. Sample size = 16 (6 study
plots and 10 natural protected areas).
Table 4
Control analysis statistics of regressions between population and economic indicators and natural ecosystems degradation
Equation Est Var Predictor Coefficient t P > jtj n F P>F R2 R2adjusted
(6) PA 1.09 9.41 0.000
ŷ ¼ 0:72APG NED 16 0.25 0.623 0.02 0.05
Intercept 15.11 1.65 0.120
APG 0.72 0.50 0.623
(7a) PA 16 151.84 0.000 0.92 0.91
ŷ ¼ 1:54TP
Intercept 1.13 0.77 0.968
TP 1.54 12.32 0.000
(7b) PA 16 236.05 0.000 0.94 0.94
ŷ ¼ 1:48TP
TP 1.48 15.36 0.000364 D. Armenteras et al. / Ecological Indicators 6 (2006) 353–368
Table 4 (Continued )
Equation Est Var Predictor Coefficient t P > jtj n F P>F R2 R2adjusted
(8a) PA 16 210.97 0.000 0.94 0.93
ŷ ¼ 2:70RP
Intercept 3.04 2.25 0.041
RP 2.70 14.52 0.000
(8b) PA 16 245.58 0.000 0.94 0.94
ŷ ¼ 2:41RP
RP 2.41 15.60 0.000
(9a) PA 16 39.15 0.000 0.74 0.72
ŷ ¼ 3:02TPD
Intercept 1.10 0.44 0.668
TPD 3.02 6.26 0.000
(9b) PA 16 67.70 0.000 0.82 0.81
ŷ ¼ 3:15TPD
TPD 3.15 8.23 0.000
(10a) PA 16 72.35 0.000 0.84 0.83
ŷ ¼ 5:87RPD
Intercept 0.90 0.44 0.670
RPD 5.87 8.51 0.000
(10b) PA 16 119.38 0.000 0.89 0.88
ŷ ¼ 5:68RPD
RPD 5.68 10.93 0.000
(11) NED 16 0.69 0.420 0.05 0.02
ŷ ¼ 0:18VD
Intercept 7.48 1.15 0.269
VD 0.18 0.83 0.420
(12) NED 10 0.01 0.933 0.001 0.12
ŷ ¼ 0:66QLI
Intercept 19.65 0.48 0.641
QLI 0.06 0.09 0.933
(13) VD 10 3.32 0.106 0.29 0.21
ŷ ¼ 0:98QLI
Intercept 87.79 2.89 0.641
QLI 0.98 1.82 0.106
(14a) VD 16 0.51 0.489 0.03 0.03
ŷ ¼ 0:97TPD
Intercept 17.93 2.53 0.024
TPD 0.97 0.71 0.489
(14b) VD 16 5.29 0.036 0.26 0.21
ŷ ¼ 1:09TPD
TPD 1.09 9.41 0.000
(15a) VD 16 1.33 0.268 0.09 0.02
ŷ ¼ 2:79RPD
Intercept 15.57 2.15 0.050
RPD 2.79 1.15 0.268
(15b) VD 16 8.49 0.011 0.36 0.32
ŷ ¼ 6:07RPD
RPD 6.07 2.91 0.011
NED = natural ecosystem degradation in 1990s (%). APG = anual population growth, 1973–993 (%). TP = total population, 10,000 inhabitants
(1993). PA = pasture area (%). RP = rural population, 10,000 inhabitants (1993). VD = violent death (%). TPD = total population density,
inhabitants/km2 (1993). QLI = quality of life index (0 < QLI < 100; best = 100). RPD = rural population density, inhabitants/km2 (1993).
NED = natural ecosystem degradation which is the percentage of natural land cover converted to anthropogenic land cover; TP = total
population; TPD = total population density; RP = rural population, PA = percentage of the area under pasture; APG = annual population growth;
QLI = quality of life. Sample size = 16 (6 study plots and 10 natural protected areas).D. Armenteras et al. / Ecological Indicators 6 (2006) 353–368 365
Fig. 5. Three possible scenarios of natural ecosystem degradation in the study area under three different projections of rural population density
(area weighted).
are very pessimistic and suggest that in 50 years uses are located near the Andes where more than half
between 85 and 100% of natural ecosystems will be of the population lives. In the 1950s oil extraction
lost. brought population immigration to the Alto Putumayo
Furthermore, the model shown in Eq. (5) (Table 3) region. Since the 1970s, partly due to a lack of
confirms the hypothesis that pasture and cattle government policies, illegal activities such as coca
ranching are the main cause of ecosystems degrada- growing have been taking place and violent para-
tion in the Colombian Amazon. Both variables were military and guerrilla groups have spread. The current
measured by independent procedures, and the model patterns on the landscape are clearly due to forest
estimates that an increase of one percent of pasture extraction associated with cattle farming combined
areas would produce a decrease of one percent of with illegal cropping. In the 1970s and 1980s there
natural ecosystem cover. However, statistical informa- was an unsuccessful attempt at establishing slash and
tion on the area of illicit crops per municipality is not burnt agriculture in the Macarena region. This land
available and this must be taken into account. It was slowly transformed into grazing encouraged partly by
impossible to incorporate these data into the analysis a land reform. Recently, these areas have also been
to further explain deforestation rates in the study area. transformed by illegal cropping due to, in part, by
establishment of a peace talk territory with no
government presence that has lasted a few years.
4. Conclusions Analysts have applied different approaches to study
change and the causes of change in natural ecosys-
Ecosystem loss rates reported in this project are tems. Over the 15-year period covered by this study,
much higher than recent estimates which suggest an there has been an increase in the magnitude of natural
annual rate of change of forest cover of between ecosystem loss. This loss is a function of the change in
0.38% (Achard et al., 2002) and 0.4% in tropical pressure on the natural resources in the region. The
South America (FAO, 2001). This suggests that population processes and the main economic activity
greater attention at a national and international level related to population have very significant impacts on
should be directed towards the Colombian Amazon. natural ecosystem degradation in these areas.
Furthermore, not only the extent and rate of We expect that the GIS methods developed during
deforestation is worrying but also the degree of this project will prove to be a very valuable tool for
fragmentation. policy and decision makers in the Amazon. We expect
Clearly, the Macarena and Alto Putumayo areas that they will assist in both identifying further data and
have undergone dramatic changes since the mid-1980s solving interpretation problems, as well as developing
mainly due to deforestation associated with, cattle new indicators according to the area local conditions
farming and illegal cropping. Both predominant land and data availability. This study covered less than 10%366 D. Armenteras et al. / Ecological Indicators 6 (2006) 353–368
of Amazonia. Clearly, many important sites could and Pedro Botero and Jaime Forero for their work in this
need to be studied further. project. We also want to thank the following participant
Ecosystem loss rates reported in this project are institutions: CDA, Corpoamazonia, Cormacarena,
much higher than recent estimates which suggest an Instituto Sinchi, Unidad de Parques and the Ministerio
annual rate of change of forest cover of between del Medio Ambiente. Finally we want to thank Sophia
0.38% (Achard et al., 2002) and 0.4% in tropical Burke and Scott Newey for their corrections and
South America (FAO, 2001). This suggests that comments on the manuscript.
greater attention at a national and international level
should be directed towards the Colombian Amazon.
Furthermore, not only the extent and rate of References
deforestation is worrying but also the degree of
fragmentation. This is critically important due to its Achard, F., Eva, H.D., Stibig, H., Mayaux, P., Gallego, J., Richards,
effect on forest degradation and ecosystem function- T., Malingreau, J.P., 2002. Determination of deforestation rates
of the world’s humid tropical forests. Science 297, 999–1002.
ality. Armenteras, D., Rodriguez, N., Rudas, G., Sua, S., 2002. An
Hopefully, the results of this work will provide indicator system with GIS to explain and prevent biodiversity
some much-needed answers in our efforts to under- losses in the Colombian Amazonia. In: Proceedings of the 22nd
stand the spatial pattern and probable causes of ESRI Users Conference, San Diego, CA, July.
Bates, J.M., Demos, T.C., 2001. Do we need to devalue Amazonia
deforestation. At the same time we also hope that
and other large tropical forests? Divers. Distrib. 7, 249–255.
natural resource managers and planners use the Batistella, M., Brondizio, E.S., Moran, E.F., 2000. Comparative
information presented in this paper as to undertake analysis of landscape fragmentation in Rondonia, Brazilian
biodiversity policies in the Colombian Amazonia. Amazon. Int. Arch. Photogr. Rem. Sens. Spatial Inform. Sci.
Remote sensing offers rapid and accurate sampling 33, 148–155.
and is perhaps the only affordable means of looking at Carvalho, K.S., Vasconcelos, H.L., 1999. Forest fragmentation in
central Amazonia and its effects on litter-dwelling ants –
processes over large spatial areas. GIS analysis and analysis by ant communities. Biol. Conserv. 91 (2), 151–157.
mapping are a good way of informing and warning Contralorı́a General de la República (CGR), 1951. XII Censo de
managers. It is imperative to keep in mind that Población y I de Vivienda.
indicators have to be quantitative so that results can be Davies, K.F., Margules, C., 1998. Effect of habitat fragmentation on
carabid beetles: experimental evidence. J. Anim. Ecol. 67, 460–
compared over time and space. It is important to
471.
continue further interpretation, modeling and predic- Departamento Administrativo Nacional de Estadı́stica (1964, 1973,
tions of the dynamics of our natural resources through 1985 & 1993). Censos de población de 1964, 1973. 1985 y 1993.
analysis involving both remotely sensed and social Colombia.
data. Although this research is just a small contribu- Domı́nguez, C., 1987. Colombia y la Panamazonia. In: Colombia
tion to the still scarce knowledge regarding the Amazonica. Universidad Nacional de Colombia y Fondo FEN de
Colombia, Bogota, Colombia.
Colombian Amazonia, we hope that this article will Etter, A., 1998. Mapa general de ecosistemas de Colombia
capture the attention of both public and researchers (1:1.500.000). In: IAvH (1998) Informe nacional sobre el estado
towards this part of the world. de la biodiversidad 1997. Instituto de Investigación de Recursos
Biológicos Alexander von Humboldt, PNUMA and Ministerio
de Medio Ambiente, 3 vol. Bogotá, Colombia.
FAO, 2001. Global Forest Resource Assessment. FAO, Rome.
Acknowledgements Fearnside, P.M., 1990. The rate and extent of deforestation in
Brazilian Amazonia. Environ. Conserv. 17 (3), 213–226.
This work was the result of the project Diseño e Fearnside, P.M., 1993. Deforestation in Brazilian Amazonia: the
Implementación del Sistema de Indicadores de Segui- effect of population and land tenure. Ambio 22 (8), 537–545.
Fearnside, P.M., 1995. Potential impacts of climatic change on
miento de la Polı́tica de Biodiversidad en la Amazonia
natural forests and forestry in Brazilian Amazonia. Forest Ecol.
Colombiana (Instituto Humboldt – Ministerio del Manage. 78, 51–70.
Medio Ambiente, Crédito BID 774 OC/CO). Our Fearnside, P.M., Lima de Alencastro Graca, P.M., Alves Rodrı́guez,
thanks to Fernando Gast, General Director of Instituto F.J., 2001. Burning of Amazonian rainforests: burning efficiency
Alexander von Humboldt. We want to specially thank and charcoal formation in forest cleared for cattle pasture near
Juan Carlos Betancourth and Carol Franco and also Manaus, Brazil. Forest Ecol. Manage. 146, 115–128.D. Armenteras et al. / Ecological Indicators 6 (2006) 353–368 367
Fearnside, P.M., 2001. Land-tenure issues as factors in environ- Parayil, G., Tong, F., 1998. Pasture-led to logging-led deforestation
mental destruction in Brazilian Amazonia: the case of Southern in the Brazilian Amazon: the dynamics of socio-environmental
Pará. World Dev. 29 (8), 1361–1372. change. Glob. Environ. Change 8 (1), 63–79.
Fearnside, P.M., Tardin, A.T., Meira Filho, L.G., 1990. Deforesta- Pimm, S.L., Ayres, M., Balmford, A., Branch, G., Brandon, K.,
tion Rate in Brazilian Amazonia. Instituto de Pesquisas Espa- Brooks, T., Bustamante, R., Costanza, R., Cowling, R., Curran,
ciais (INPE), Sao José dos Campos, Sao Paulo, 8 pp. L.M., Dobson, A., Farber, S., da Fonseca, G.A.B., Gascon, C.,
Gascon, C., Lovejoy, T.E., Bierregaard Jr., R.O., Malcolm, J.R., Kitching, R., McNeely, J., Lovejoy, T., Mittermeier, R.A.,
Stouffer, P.C., Vasconcelos, H.L., Laurance, W.F., Zimmerman, Myers, N., Patz, J.A., Raffle, B., Rapport, D., Raven, P., Roberts,
B., Tocher, M., Borges, S., 1999. Matrix habitat and species C., Rodrı́guez, J.P., Rylands, A.B., Tucker, C., Safina, C.,
richness in tropical forest remnants. Biol. Conserv. 91 (2), 223– Samper, C., Stiassny, M.L.J., Supriatna, J., Wall, D.H., Wilcove,
229. D., 2001. ENVIRONMENT: Can We Defy Nature’s End?
Huber, O., 1981. Significance of savanna vegetation in the Amazon Science 293, 2207–2208.
territory of Venezuela. In: Prance, G.T. (Ed.), Biological Diver- Portela, R., Rademacher, I., 2001. A dynamic model of patterns of
sification in the Tropics. Columbia University Press, New York. deforestation and their effect on the ability of the Brazilian
Instituto, de Investigación de Recursos Biológicos Alexander von Amazonia to provide ecosystem services. Ecol. Model. 143,
Humboldt (IavH), 1998. Informe nacional sobre el estado de la 115–146.
biodiversidad 1997. Instituto de Investigación de Recursos Prance, G.T., 1985. A review of the phytogeographic evidences for
Biológicos Alexander von Humboldt, PNUMA and Ministerio pleistocene climate changes in the neotropics. Ann. Missouri
de Medio Ambiente, 3 vol. Bogotá, Colombia. Botanical Garden 69 .
Kalliola, R., Maarit, P., Danjoy, W., 1993. Amazonia Peruana. Ranta, P., Blom, T., Niemela, J., Joensuu, E., Siitonen, M., 1998. The
Vegetación húmeda tropical en el llano subandino. Proyecto fragmented Atlantic rain forest of Brazil: size, shape and dis-
Amazonia Universidad de Turku, Findland. tribution of forest fragments. Biodiv. Conserv. 7, 385–403.
Klein, B.C., 1989. Effect of forest fragmentation on dung and Reis, E.J., Margulis, S., 1991. Perspectivas Economicas de desflor-
carrion beetle communities in central Amazonia. Ecology 70 estamiento de Amazonia. Textos para Discussao No. 215.
(6), 1715–1725. Insituto de Pesquisa Economica Aplicada (IPEA), Brasilia, 47
Laurance, W.F., Delamonica, P., Laurance, S.G., Vasconcelos, H.L., pp.
Lovejoy, T.E., 2000. Conservation: rainforest fragmentation Rudas, G.D., Armenteras, S.M., Sua, Rodrı́guez, N., 2002. Indi-
kills big trees. Nature (404), 836. cadores de Seguimiento de la Polı́tica de Biodiversidad en la
Laurance, W.F., Laurance, S.G., Delamonica, P., 1998. Tropical Amazonia Colombiana 2001. Informe Final de Resultados.
forest fragmentation and greenhouse gas emissions. Forest Ecol. Proyecto Diseño e Implementación del Sistema de Indicadores
Manage. 110 (1), 173–180. de Seguimiento de la Polı́tica de Biodiversidad en la Amazonia
Laurance, W.F., 1998. A crisis in the making: responses of Ama- Colombiana. Instituto Humboldt, CDA, Corpoamazonia,
zonian forests to land use and climate change. TREE 13 (10), Cormacarena, Instituto Sinchi, Unidad de Parques, Ministerio
412–451. del Medio Ambiente (Crédito BID 774 OC/CO), Bogotá,
Laurance, W.F., Albernaz, A.K.M., Schroth, G., Fearnside, P.M., Colombia.
Bergen, S., Venticinque, E.M., Da Costa, C., 2002. Predictors of Ruiz, J., 1989. Desarrollo regional en Colombia: ordenamiento
deforestation in the Brazilian Amazon. J. Biogeogr. 29 (5), 737– ambiental y desarrollo regional. Cuadernos de Industria y Econ-
748. omia Rural 22.
Lima, M.G., Gascon, C., 1999. The conservation value of linear Sierra, R. (Ed.), 1999. Propuesta preliminar de un sistema de
forest remnants in central Amazonia. Biol. Conserv. 91, 241– clasificación de vegetación para el Ecuador continental. Project
247. INEFAN/GEF-BIRF and EcoCiencia, Ecuador.
Mcgarigal, K., Marks, B.J., 1995. Fragstats: spatial pattern analysis Sierra, R., 2000. Dynamics and patterns of deforestation in the
program for quantifying landscape structure. Gen. Tech. Re western Amazon: the Napo deforestation front, 1986–1996.
PNW-GTR-351. U.S. Departament of Agriculture, Forest Ser- Appl. Geogr. 20, 1–16.
vice, Pacific Northwest Research Station, Portland, OR, USA. Steininger, M.K., Tucker, C.J., Ersts, P., Killeen, T.J., Villegas, Z.,
Myers, N., Mittermeier, R.A., Mittermeier, C.G., da Fonseca, Hecht, S.B., 2001a. Clearance and fragmentation of tropical
G.A.B., Kent, J., 2000. Biodiversity hotspots for conservation deciduous forest in the Tierras Bajas, Santa Cruz, Bolivia.
priorities. Nature 403, 853–858. Conserv. Biol. 15 (4), 856–866.
Nepstad, D.C., Veroassimo, V., Alencar, A., Nobre, C., Lima, E., Steininger, M.K., Tucker, C.J., Townshend, J.R.G., Killeen, T.J.,
Lefebvre, P., Schlesinger, P., Potterk, C., Moutinho, P., Mendoza, Desch, A., Bell, V., Ersts, P., 2001b. Tropical deforestation in the
E., Cochrane, M., Brooks, V., 1999. Large-scale impoverishment Bolivian Amazon. Environ. Conserv. 28 (2), 127–134.
of Amazonian forests by logging and fire. Nature 398, 505– U.S. Department of State, 1999. 1999 International Narcotics Con-
508. trol Strategy Report. USA.
Nepstad, D.C., Carvalho, G., Barros, A.C., Alencar, A., Capobianco, Vogelmann, J.E., 1995. Assessment of forest fragmentation in
J.P., Bishop, J., Moutinho, P., Lefebvre, P., Silva Jr., U.L., Prins, Southern New England using remote sensing and geographic
E., 2001. Road paving, fire regime feedbacks, and the future of information systems technology. Conserv. Biol. 9 (2), 439–
Amazon forests. Forest Ecol. Manage. 154, 395–407. 449.368 D. Armenteras et al. / Ecological Indicators 6 (2006) 353–368 Walter, H., 1985. Vegetation of the Earth and Ecological Systems of (Eds.), Tropical Forest Remnans: Ecology. Management and the Geo-biosphere, 3rd ed. Springer-Verlag, New York, 318 pp. Conservation of Fragmented Communitites, pp. 3–12. Walter, H., Breckle, S.W., 1986. Ecological Systems of the Geobio- Wood, C.H., Skole, D., 1998. Linking satellite, census and sphere. Tropical and Subtropical Zonobiomes, vol. 2. Springer- survey data to study deforestation in the brazilian amazon. In: Verlag, Berlin/Heidelburg. Liverman, D. (Ed.), People and Pixels: Linking Remotes Whitmore, T.C., 1997. Tropical forest disturbance, disappearance, Sensing and Social Sciences. National Research Council, and species loss. In: Laurance, W.F., Bierregaard, Jr., R.O. USA, pp. 70–93.
You can also read
