Role of a campesine reserve zone in the Magdalena Valley (Colombia) in the conservation of endangered tropical rainforests

Abstract

Tropical forests of Colombia have one of the highest deforestation rates in the world. The humid forest of the Magdalena valley region is one of the ecosystems with the highest risk of landscape transformation, despite being home to many endemic and threatened species. The aim of this study was to evaluate the role of a Peasant Reserve Zone in the conservation of tropical humid forests and endangered species in the Magdalena valley region. To reach this aim, we performed a multi-temporal analysis of the forest dynamics in the Peasant Reserve Zone-Cimitarra River Valley (PRZ-CRV) and assessed the extinction risk of eight species endemic to Colombia. Our outcomes indicated that the most extended land cover in the PRZ-CRV is the forest (56.30%), followed by open areas (38.75%). The forest dynamics analysis indicated that the forest cover has decreased by 3.82% between 2017 and 2019, being the area with redoubts from the Serranía de San Lucas Forest the most conserved. Finally, our results indicated that less than 50% of the climatically suitable areas for each species are covered by forests and that less than 10% of those areas are covered by Protected Areas, while for such species as Agalychnis terranova and Ateles hybridus the PRZ-CRV covered a higher percentage of their distribution than all Protected Areas together in this ecosystem. In conclusion, our results have indicated that the PRZ-CRV could be an important area for the maintenance and conservation of humid forests and their associated fauna, playing an important role as an ally to the Protected Area system in the Magdalena valley region.

Full text

ROLE OF A CAMPESINE RESERVE ZONE IN THE MAGDALENA
VALLEY (COLOMBIA) IN THE CONSERVATION
OF ENDANGERED TROPICAL RAINFORESTS
Natalia Trujillo-Arias1,2,* , Víctor H. Serrano-Cardozo1, Martha P. Ramírez-Pinilla1
1Universidad Industrial de Santander, Colombia
2Instituto de Investigación en Recursos Biológicos Alexander von Humboldt, Colombia
*e-mail: natitrujillo@gmail.com
Received: 15.02.2022. Revised: 14.09.2022. Accepted: 17.09.2022.
Tropical forests of Colombia have one of the highest deforestation rates in the world. The humid forest of
the Magdalena valley region is one of the ecosystems with the highest risk of landscape transformation,
despite being home to many endemic and threatened species. The aim of this study was to evaluate the
role of a Peasant Reserve Zone in the conservation of tropical humid forests and endangered species in the
Magdalena valley region. To reach this aim, we performed a multi-temporal analysis of the forest dynamics
in the Peasant Reserve Zone-Cimitarra River Valley (PRZ-CRV) and assessed the extinction risk of eight
species endemic to Colombia. Our outcomes indicated that the most extended land cover in the PRZ-
CRV is the forest (56.30%), followed by open areas (38.75%). The forest dynamics analysis indicated that
the forest cover has decreased by 3.82% between 2017 and 2019, being the area with redoubts from the
Serranía de San Lucas Forest the most conserved. Finally, our results indicated that less than 50% of the
climatically suitable areas for each species are covered by forests and that less than 10% of those areas are
covered by Protected Areas, while for such species as Agalychnis terranova and Ateles hybridus the PRZ-
CRV covered a higher percentage of their distribution than all Protected Areas together in this ecosystem.
In conclusion, our results have indicated that the PRZ-CRV could be an important area for the maintenance
and conservation of humid forests and their associated fauna, playing an important role as an ally to the
Protected Area system in the Magdalena valley region.
Key words: Andes, forest loss, land cover, Peasant Reserve Zone, threatened species
Introduction
Deforestation is increasing in the tropics
with severe implications for biodiversity conser-
vation, climate regulation, and maintenance of
ecosystem services (Etter et al., 2006; Da Ponte
et al., 2017; Negret et al., 2019). Deforestation
of tropical forests is responsible for the massive
extinction of species (Myers, 1988; Pimm et al.,
1995) and the decrease in biological diversity
(Tucker & Townshend, 2000; Marsh et al., 2016),
more than half of the species of plants and ani-
mals being housed in this biome type (Gallery,
2014), being of great interest for knowing the
current extent of tropical forests and their defor-
estation rates (Tucker & Townshend, 2000).
The highland and lowland tropical forests
of Colombia have one of the highest defores-
tation rates, with clearing accelerating to 1.4%
annually during the second half of XX cen-
tury (Etter et al., 2006, 2008; Armenteras et
al., 2011). Specically, the tropical humid for-
est of the Magdalena Valley Region (MVR) is
one of the ecosystems with the highest risk of
landscape transformation (Etter et al., 2006) and
is one of the areas where clearing of highland
and lowland forests has been documented (Etter
& van Wyngaarden, 2000; Negret et al., 2019).
These forests are home to a large number of en-
demic species, many of them being threatened
(e.g. Sachatamia punctulata (Ruiz-Carranza &
Lynch, 1995), Crax alberti Fraser, 1852, Ateles
hybridus I. Georoy, 1829) (Cuervo Maya et
al., 1999; Rojas-Morales et al., 2014; Marsh et
al., 2016) and also are important as a potential
bridge between populations of fauna and ora
from biodiverse hotspots such as the Choco,
Central America and Guyana Amazon regions
(Chapman, 1917; Echeverry & Morrone, 2013).
The MVR is a critical priority area to con-
serve according to various national and interna-
tional organisations, such as the Instituto Alex-
ander von Humboldt and World Wildlife Fund
(Sanchez-Cuervo & Aide, 2013; Quintero-Vallejo
et al., 2017). Despite this, the region is under-
represented in the national system of Protected
Areas (PAs) (Forero-Medina & Joppa, 2010;
Sanchez-Cuervo & Aide, 2013) and its threats
(such as illicit crops, glyphosate fumigation, un-
Nature Conservation Research. Заповедная наука 2023. 8(1) https://dx.doi.org/10.24189/ncr.2023.003

controlled deforestation for timber and hunting,
mining activities and armed conict) have in-
tensied in the last decades (Álvarez, 2002; Ar-
menteras et al., 2011; Dávalos et al., 2011; San-
chez-Cuervo & Aide, 2013; Chadid et al., 2015;
Murillo-Sandoval et al., 2020). In order to dimin-
ish these threats, the Colombian government has
declared some important natural areas as Peasant
Reserve Zones (PRZs) where, according to law
the expansion of the agricultural frontier must be
controlled to improve the human well-being and
assure ecosystem and biodiversity conservation
(Quintero-Vallejo et al., 2017; Tocancipá Falla &
Ramírez Castrillón, 2018).
The PRZs are found throughout Colombia,
and depending on their size and location, many
constitute areas of great importance for the con-
servation of tropical forests and endangered spe-
cies. The PRZ of the Cimitarra River Valley (PRZ-
CRV) (Asociación campesina del valle del rio
Cimitarra, 2008) located in the Middle Magdalena
valley (Fig. 1a), comprises a total area of ~ 5800
km2 that contains a peasant area (~ 2670 km2) and
a forest reserve (~ 3130 km2) (Quijano-Mejía &
Linares-García, 2017) (Fig. 1b). This PRZ is one
of the largest in the country, and because it con-
tains humid forests, among them forests of the San
Lucas Mountain Range (Serranía de San Lucas),
which is one of the most diverse and unprotected
areas of the country, the PRZ-CRV could be a key
component for the maintenance and conservation
of biodiversity in the region. The aim of this study
was to evaluate the role of the PRZ-CRV in the
conservation of tropical humid forests and endan-
gered species in the MVR. For reaching this goal,
we have performed a multi-temporal analysis of
the forest dynamics in the PRZ-CRV using Senti-
nel images. Additionally, we have assessed the po-
tential extinction risk of eight species endemic to
Colombia and distributed in the MVR employing
species distribution models.
Material and Methods
Study area
The Middle Magdalena Valley (MMV) is lo-
cated between the central and western cordilleras
(Central and Eastern Cordillera) of Colombia,
and it is part of the Magdalena-Urabá ecoregion
(Olson et al., 2001). The MMV includes various
types of habitats. Among them are premontane
wet forests and tropical moist forests that span an
altitudinal gradient of 0–2500 m a.s.l. These for-
ests link the northern ecoregions of Mesoamerica
and the Chocó with the Andean and Amazonian
ecoregions (Echeverry & Morrone, 2013). There
are no national parks in the region (Forero-Me-
dina & Joppa, 2010), but several areas of intact
habitat remain, e.g. San Lucas Mountain Range.
These areas are under pressure from high popula-
tions of humans and are threatened by timber op-
erations, cattle ranching, and illicit drug cultiva-
tion (Etter et al., 2006). The PRZ-CRV is located
in the Middle Magdalena Valley by covering a
part of the Antioquia and Bolivar departments
(provinces) (Fig. 1a). This PRZ comprises a total
area of ~ 5800 km2. It is distributed in i) peas-
ant area (~ 2670 km2), where the peasants mainly
carry out their productive activities (e.g. agricul-
tural and cattle raising activities) and ii) forest
reserve (~ 3130 km2), where the peasants them-
selves demarcated an area around the redoubts of
forests of the San Lucas Mountain Range, known
as «yellow line», to restrict the colonisation ad-
vance; i.e. it is prohibited any type of interven-
tion on the fauna and ora (Fig. 1b).
Multi-temporal analysis of forest dynamics
in the PRZ-CRV
Land cover classication for the 2019
In this study, Sentinel 2 image data were ac-
quired between the reference years of 2017 to
2019. Four tiles were needed to cover the entire
extent of the PRZ-CRV. A total of 116 images
with less than 40% cloud cover were obtained
from the Copernicus Open Access Hub (https://
scihub.copernicus.eu/) (Table A1). Atmospheric
correction was performed to Level-1C images
using Sen2Cor (v. 2.5.5), which converts the
top-of-atmosphere reectance Level-1C data to a
bottom-of-atmosphere (BOA) reectance Level-
2A product (Müller-Wilm, 2016). Individual sat-
ellite images were merged using Sen2Mosaic (v.
0.2) to create a continuous and cloud-free cover-
age across the study area. A mosaicking process
consisted of systematically selecting the most
similar and cloud-free pixels across each time
period. Only for the year 2019, it was possible
to obtain a cloud-free image. For other years, the
images had about 2% cloud cover. After atmo-
spheric correction, the bands 5, 6, 7, 11 and 12
were re-sampled at a 20-m spatial resolution and
georeferenced to the Transverse Mercator projec-
tion, Bogota Observatory Datum. Finally, the im-
ages were cut to the study area (PRZ-CRV). All
processing of multispectral images was carried
out using ESRI ArcGIS and Python scripts.
Nature Conservation Research. Заповедная наука 2023. 8(1) https://dx.doi.org/10.24189/ncr.2023.003

Fig. 1. The location of the Magdalena region in the northern Andes, showing the Magdalena River, the San Lucas Mountain
Range and the Peasant Reserve Zone of the Cimitarra river Valley (PRZ-CRV) (a.). Map of the Peasant Reserve Zone of the
Cimitarra river Valley, showing the forest reserve area, the peasant area and the area delimited as «Yellow line» (b.).
We classied the study area into ve land cover
categories, namely forest (a), open areas (b), rivers
(c), wetlands (d), other covers (e). «Forest» areas
correspond to primary and secondary vegetation
grouped as a single natural land use. The «open
areas» designation was used to identify pastures,
crops and grasslands for cattle ranching and farm-
ing. «Other cover» was represented by urban areas,
sands, and small dense clouds. For each land cover, a
minimum of 15 training polygons were targeted be-
tween 45-pixels and 100-pixels using ground truth
data collected during a eld trip in 2019, as well as
on the basis of very high-resolution (VHR) imagery
in GoogleEarth®. Finally, to determine and quantify
the land cover of each category, we rst tested the
performance of three non-parametric classier algo-
rithms Support Vector Machines (Cortes & Vapnik,
1995), Neural Network (Yoshida & Omatu, 1994)
and Random Forest (Breiman, 2001) using «svm»
(Becker et al., 2009), «nnet» (Venables & Ripley,
2002), «randomForest» (Liaw & Wiener, 2002) and
«caret» (Kuhn, 2008) packages in R version 4.0.2 (R
Core Team, 2020). For this purpose, a dataset from
the previously targeted polygons was employed to
train the model and another independent dataset
was used to test the model. After this, we assessed
the three algorithms based on the confusion matri-
ces, accuracy statistics, such as accuracy, Kappa,
Sensibility and Specicity (Olofsson et al., 2013;
Foody, 2020). In addition, based on the Bonferroni
test, we assessed whether there were signicant dif-
ferences between the performance of the algorithms
(p-value < 0.05). After those analyses, we selected
«random forest» as the best method to perform the
nal land cover classication due to its bigger accu-
racy with smaller intervals (Fig. A1), greater values
of Sensibility and Specicity for almost all categories
(values > 0.98) (only the wetland category was 0.83,
but much higher than for the other two algorithms
(0.55 or 0.76)), and because the dierences between
this algorithm with the other two were signicant.
Forest dynamic analysis (2017–2019)
To analyse the forest dynamics between 2017
and 2019, we compared the estimated cover area
of each category between years. For this purpose,
we used «random forest» and the same training
polygons for all land cover types employed for the
Nature Conservation Research. Заповедная наука 2023. 8(1) https://dx.doi.org/10.24189/ncr.2023.003

2019 classication (those polygons were designed
on constant areas through the three years). The ex-
ception was for «wetlands» and «other covers»,
because these land covers were highly dynamic
through space and time. Specically, to quantify
the area of the forest cover that has been lost from
2017 to 2019, we followed Chadid et al. (2015).
We did not consider for the land cover quantica-
tion areas where there was presence of clouds in
any of the analysed years to avoid under- or over-
estimations of the forest loss. The total area cov-
ered by clouds was 2%. Finally, we quantied in
square kilometres (km2) the lost forest area and
generated a forest dynamic map by identifying ar-
eas, where the forest had been conserved through
years, as well as the areas where have occurred de-
forestation or forest recover events using the clas-
sication images and ESRI ArcGIS.
Extinction risk: Species distribution models
and gap analysis
For the extinction risk analysis, we selected
eight species endemics (or near endemics) to Colom-
bia distributed in the MVR. They are under threat
(or near threat) according to the IUCN: three am-
phibians (Sachatamia punctulata (Ruíz-Carranza &
Lynch, 1995) (Vulnerable, VU), Diasporus anthrax
(Lynch, 2001) (VU), and Agalychnis terranova Ri-
vera-Correa, Duarte-Cubides, Rueda-Almonacid &
Daza-R., 2013 (Near Threatened, NT)), three birds
(Crax alberti (Critically Endangered, CR), Capito
hypoleucus Salvin, 1897 (VU) and Habia guttura-
lis (Sclater, 1854) (Least Concern, LC, but this spe-
cies was assessed as NT in 2017)), and two primate
species (Saguinus leucopus (Günther, 1877) (VU),
and Ateles hybridus (CR)). These species were se-
lected based on their ecological characteristics, such
as low vagility, dependence on forest, and sensitivity
to changes in vegetation cover, which makes them
excellent models to evaluate the dynamics of the for-
ests and its consequences on the fauna of the region.
We modelled the climatic distribution of the
eight species using MaxEnt (Phillips et al., 2006).
We used MaxEnt because it performs well, compared
to other modelling approaches (Elith et al., 2006).
We compiled occurrence records from: i) national
biological collections (e.g. Universidad Industrial
de Santander (UIS), SiB Colombia), ii) Global Bio-
diversity Information Facility (GBIF) (www.gbif.
org), and iii) scientic articles (e.g. Cuervo Maya
et al., 1999; Salaman et al., 2002; Laverde-R et al.,
2005; Ochoa-Quintero et al., 2005; Rojas-Morales
et al., 2014). To reduce the spatial autocorrelation,
for each species we randomly removed duplicate
occurrence records that were less than 1 km apart
from each other. The database has been resulted in a
nal dataset of 13–98 presence records (see number
of occurrences by species in Table A2).
For MaxEnt analyses, we used 19 climate vari-
ables available at WordClim 1.4 (Hijmans et al.,
2005) with a resolution of 30 arc sec. For the nal
analysis, we selected the climatic variables by rst
rejecting highly correlated variables (r > 0.85) (Pe-
terson et al., 2011), and then by selecting relevant
variables by a rationale of permutation importance
> 5% (Trujillo-Arias et al., 2017). The general con-
ditions for analyses were random test points = 20;
replicates = 10; replicate type: subsample; maxi-
mum iterations = 5000; background points = de-
fault (max: 10 000). For species with less than 50
occurrences, we used the bootstrap replicate type.
We did not use bias le because the background
manipulation usually results in higher commission
errors (Kramer-Schadt et al., 2013). We validated
models by evaluating AUC values (Pearson et al.,
2007) and the True Skill Statistic (TSS) (Allouche et
al., 2006) (see examples of receiver operating char-
acteristic curves in Fig. A2). Then, we projected the
species distribution models to the future (year 2050)
using three emission models of Representative Con-
centration Paths 4.5 (RCP 4.5). They represent a
conservative scenario of greenhouse emissions (van
Vuuren et al., 2011), namely 1) HadGEM2-ES pre-
dicting a high temperature increase and a reduction
in precipitation (i.e. hotter / drier scenario) (Jones
et al., 2011); 2) GISS-E2-R predicting a moderate
increase in temperature and rates of relatively con-
stant precipitation (i.e. moderate scenario) (Schmidt
et al., 2014); 3) IPSL-CM5a-LR predicting warm
temperatures and reduced precipitation (i.e. warmer
/ drier scenario) (Dufresne et al., 2013). Models were
generated in ASCII format and exported directly to
the GIS to obtain binary maps using the threshold of
equally training sensitivity and specicity. For this
purpose, all pixels with a value under that threshold
were assigned a value of zero (0), which would rep-
resent the absence of the species.
Finally, we estimated habitats for the present
and the habitat loss between the future and present
for each species employing two measures. First is
the currently available habitat based on climate and
forest cover. This estimation was based on deter-
mining how much (in km2) of the climatically suit-
able area according to MaxEnt models is covered
with forest. For this purpose, we employed the bi-
nary maps for the present period and the layer of
Nature Conservation Research. Заповедная наука 2023. 8(1) https://dx.doi.org/10.24189/ncr.2023.003

forest cover for 2019 produced by the Instituto de
Hidrología, Meteorología y Estudios Ambientales
de Colombia (IDEAM). Basically, using GIS tools,
we overlapped the forest layer over the climatically
suitable area for each species, and we quantied
how much of that climatic area is actually covered
by forests. This measure allows us to have a quanti-
cation of the available area in the present, not only
based on climate, but also on forest cover in the
region. The second measure is the potential future
habitat change (in km2). This estimate was based on
comparing the climatically suitable areas indicated
by the MaxEnt models for the present and the fu-
ture. Specically, from the binary maps, we quanti-
ed the climatically suitable area for each species
in each period and made a description of whether
its area, based on climate, increases or decreases.
This measure allows us to know how much and
what type of change (expansion or contraction) the
species might face according to various future cli-
mate scenarios. In addition, we also evaluated how
much of the range of each species is covered by PAs
employing gap analysis, as a methodological tool
frequently used in conservation to identify «gaps»
in the network of PAs. For this purpose, we con-
sidered the National Protected Areas (PNN), Forest
Reserves (comprised of public and private areas),
and the Natural Reserves of Civil Society. We also
calculated how much of the current distribution of
each species is found within the current distribution
covered by the PRZ-CRV.
Results
Land cover and forest dynamics
In general, classication accuracies obtained
from Sentinel images in 2017–2019 uctuated
from 96% to 99%, with kappa coecients ranging
from 0.96 to 0.98. The 2019 classication indicated
that the land covers with the highest percentage of
area were forest with 56.30% (3307.18 km2), fol-
lowed by open areas with 38.75% (2276.25 km2),
while the other land covers received percentages
lower than 3% (rivers: 2.83% (166 km2), wetlands:
1.77% (103.96 km2), and other covers: 0.31%
(18.21 km2) (Fig. 2a).
The forest dynamics map for 2017–2019 in-
dicated that the forest cover in the PRZ-CRV de-
creased by 3.82%, while the open areas increased by
3.5% (Table 1). This estimation was compared with
values of the forest cover loss obtained on the global
forest watch platform (https://www.globalforest-
watch.org/), by nding a slightly higher loss (4.1%).
This dierence may be related to areas, which were
not quantied in our analysis due to the presence of
clouds. Finally, the forest dynamics map (Fig. 2b)
has identied that the conserved forest areas (with-
out change) were located west of the PRZ-CRV,
mainly in the area dened as «yellow line» charac-
terised by presenting redoubts of forests of the San
Lucas Mountain Range. Areas with the greatest loss
of the forest were distributed around the «yellow
line», both to the north and to the south, with some
spotlights of deforestation within this area or very
close to its limits. This analysis also allowed us to
observe some areas of forest restoration, which are
mainly concentrated east of the study area, a region
characterised by wetlands.
Species distribution models and gap analysis
The species distribution models for the eight
species have been performed better than a random
model with AUC and TSS values greater than 0.83
(Fig. A3). Specically, the species distribution mod-
els for the present period were consistent with the
distribution range of each species. However, for
some species such as Diasporus anthrax and Capito
hypoleucus, the models also indicated other areas
outside the distribution range as climatically viable
areas for their distribution (e.g. the Pacic or the
southeast of the Colombian Amazon) (see Fig. A3).
However, although for some species the climatic
niche for the present period (i.e. fundamental niche)
was a little larger than the known (i.e. realised)
niche, the estimations of the currently available
habitat indicated for most of the species, that less
than 50% of their climatic suitable areas are covered
by forests suggesting that in fact its optimal distri-
bution area is smaller (Table 2a). For instance, for
D. anthrax and Crax alberti only 33% and 36% of
their distributions were covered by forests, respec-
tively. Other species, such as Agalychnis terranova,
showed a higher forest cover along their distribu-
tions (59%). For the potential future habitat change
measure, the ecological niche models suggested
variable responses among the species and the cli-
matic scenarios (Table 2b). However, in most spe-
cies (except for C. hypoleucus and Habia gutturalis)
a contraction of their distribution range is observed
for the hotter / drier scenario (HadGEM2-ES). For
other scenarios (GISS-E2-R and IPSL-CM5a-LR),
the response was highly variable, with some species
showing the range expansion, while others showed
contraction. It should be noted that these results
should be taken with caution considering the small
number of records used to model the distribution of
some species (e.g. amphibians).
Nature Conservation Research. Заповедная наука 2023. 8(1) https://dx.doi.org/10.24189/ncr.2023.003

Fig. 2. Land cover classication for the year 2019 using the Random Forest algorithm (a.); designations: «Forest» corresponds
to primary and secondary vegetation grouped as single natural land use. «Open areas» represents pastures, crops and grasslands
for cattle ranching and farming. «Other covers» represents urban areas, sand, and small clouds. Forest dynamics in the PRZ-
CRV in 2017–2019 (b.); designations: the blue and black colours represent forest areas and open areas respectively, which
have not changed during the study time; the yellow and light blue colours identify areas where the land cover has changed
because of the loss of forest (deforestation areas) or because of restoration of the vegetation cover (forest recovery areas).
Table 1. Changes in land cover at the Peasant Reserve Zone of the River Cimitarra Valley in 2017–2019
Years Percent change
Land covers 2017 (%) 2018 (%) 2019 (%) Lost Gain
Forest 59.64 57.92 55.82 3.82% –
Open areas 35.58 36.88 38.81 – 3.50%
Rivers 2.97 2.71 2.97 – –
Wetlands 1.56 2.06 1.71 – –
Other covers 0.25 0.34 0.43 – 0.18%
Table 2. Available habitats for each species during the present (2019) (based on the potential distribution polygons obtained
from MaxEnt and the forest layer from the IDEAM) and the future (2050) (based on the future potential distribution polygons
obtained from MaxEnt) scenarios. The area is indicated in km2 and in percent (in brackets). See potential distribution of
species in Fig. A3
Scenarios Amphibians Birds Primates
Sachatamia
punctulata
Diasporus
anthrax
Agalychnis
terranova Crax alberti Capito
hypoleucus Habia gutturalis Saguinus
leucopus Ateles hybridus
a. Present
Total area 127 118 (100.0) 64 458 (100.0) 60 962 (100.0) 87 902 (100.0) 131 256 (100.0) 121 495 (100.0) 97 893 (100.0) 113 667 (100.0)
Forest 53 238 (41.8) 21 783 (33.80) 36 256 (59.47) 31 968 (36.4) 56 913 (43.3) 49 190 (40.5) 38 807 (30.5) 28 893 (22.7)
No forest 73 182 (57.6) 42 219 (65.50) 24 646 (40.43) 55 853 (63.5) 73 896 (56.3) 71 686 (59) 58 750 (46.2) 84 699 (6.6)
No data 697 (0.55) 455 (0.70) 59 (0.10) 79 (0.09) 447 (0.34) 618 (0.51) 335 (0.26) 74 (0.06)
PAs (%) 6.95 8.31 1.67 6.20 9.43 8.89 8.26 3.74
PRZ-CRV (%) 4 4 4 6 3 4 4.85 5.12
b. Future
HadGEM2-ES 51 267 (↓) 29 454 (↓) 33 233 (↓) 84 940 (↓) 149 778 (↑) 129 379 (↑) 96 947 (↓) –
GISS-E2-R 60 267 (↓) 41 913 (↓) 104 409 (↑) 69 019 (↓) 141 572 (↑) 145 043 (↑) 97 370 (↓) 200 153 (↑)
IPSL-CM5a-LR 171 872 (↑) 38 279 (↓) 120 287 (↑) 148 628 (↑) 148 628 (↑) 124 276 (↑) 95 747 (↓) 298 501 (↑)
Note: PAs – Protected Areas, PRZ-CRV – Peasant Reserve Zone of the Cimitarra River Valley. Arrows indicate expansion (↑) or contraction (↓) of the
distribution range in the future in relation to the range distribution in the present (i.e. total area).
Nature Conservation Research. Заповедная наука 2023. 8(1) https://dx.doi.org/10.24189/ncr.2023.003

Finally, the gap analyses indicated that for
all species less than 10% of their current distri-
butions are covered by PAs (Table 2a). The spe-
cies with the lowest percentages were Agalych-
nis terranova with only 1.67% of its distribution
covered by PAs, followed by Ateles hybridus
(3.74%) and Crax alberti (6.20%). For A. ter-
ranova and A. hybridus, the PRZ-CRV covered a
higher percentage than the cover of PAs.
Discussion
We evaluated the role of the PRZ-CRV in the
conservation of tropical humid forests and en-
dangered species in the MVR. Our ndings sug-
gest that the PRZ-CRV is an important area for
the maintenance and conservation of humid for-
ests and its associated fauna, playing an impor-
tant role as a complement to the system of PAs to
preserve biodiversity.
Role of the PRZ-CRV in conservation of
tropical humid forests
The main land cover in the PRZ-CRV
through the years has been represented by forests
followed by open areas. This dynamics between
these land covers reveals a deforestation rate
of approximately 1.3% annually, which is very
similar to deforestation rates estimated by other
authors for the MVR, suggesting that this region
is one of the ecosystems with the highest risk of
landscape transformation in Colombia (Etter &
van Wyngaarden, 2000; Etter et al., 2006, 2008;
Negret et al., 2019). The areas with cleared lands
are mainly found in the peasant area and to a
lesser extent in the forest reserve zone (Fig. 1,
Fig. 2), which is expected since the main tasks
of PRZs is not only to ensure the conservation of
ecosystems and biodiversity but also to improve
human well-being through the strengthening of
the local economy with agricultural and cattle
raising activities, among others (Quijano-Mejía
& Linares-García, 2017; Ortiz, 2018).
Despite the rate of deforestation in the area,
the PRZ-CRV is an important area for the main-
tenance and conservation of humid forests, espe-
cially those that belong to the San Lucas Moun-
tain Range. The conserved forest areas (without
change) are mainly located within the «yellow
line» (Fig. 2b), an area located in the forest reserve
zone (Fig. 1b). The «yellow line» is a demarca-
tion made by the peasants themselves around the
redoubts of forests of the San Lucas Mountain
Range, which restricts the advance of colonisa-
tion. The peasants have reached an agreement
that any type of intervention on the fauna and
ora is prohibited in this area (Quijano-Mejía &
Linares-García, 2017). Even though our results
support this initiative as successful and as a case
of community management of the territory with
specic eects on the conservation of ecosys-
tems, some spotlights of deforestation have ap-
peared within this area or very close to its limits
(Fig. 2b). This could be related to high levels of
armed conict in the region, which contributed
to a lack of governance and to an increase in the
level of deforestation (Dávalos, 2001; Dávalos
et al., 2011; Castro-Nunez et al., 2017; Liévano-
Latorre et al., 2021). For instance, Chadid et
al. (2015) evaluated the dynamics in deforesta-
tion in the San Lucas Mountain Range, report-
ing a high increase in the transition from forests
to coca crops or pastures in 2002–2010. These
results were also corroborated by Negret et al.
(2019), where they generated spatial predictions
of deforestation in Colombia from 2000–2015,
by nding a high deforestation pressure induced
by armed conict and coca cultivation in the San
Lucas Mountain Range. Even though we did not
evaluate illicit crops as a land cover, we identi-
ed cleared areas in remote sites (e.g. into the
«yellow line» area) that could be associated with
the increase of illegal practices, which has in-
tensied since the signing of the peace accords,
where persistent illegal groups ght to gain con-
trol (Yagoub, 2018; Murillo-Sandoval et al.,
2020; Liévano-Latorre et al., 2021).
The PRZ-CRV hosts a great diversity of plant
species, and it is important for conservation of
humid forests in the region. Both forest remnants
in the peasant area and in the forest reserve area
harbour endemic species of ecological impor- endemic species of ecological impor-
tance. For instance, a recent study carried out in
forest remnants has been located in the south of
the Bolívar department (i.e. in the peasant area)
indicated that despite being in an intermediate
successional stage, these forests harbour a high
richness of species and are important for the con-
servation of threatened and endemic species, such
as Wettinia hirsuta Burret, Astrocaryum malybo
H.Karst., Chamaedorea ricardoi R.Bernal, Gale-
ano & Hodel, and Unonopsis aviceps Maas. Ad-
ditionally, this study also demonstrated that for-
est remnants still retain their similarity in species
composition with the Amazon and Choco forests
(Ortiz-Lozada, 2020), which suggests that de-
spite their alteration, these forest remnants might
Nature Conservation Research. Заповедная наука 2023. 8(1) https://dx.doi.org/10.24189/ncr.2023.003

continue to play an important role as a potential
bridge between populations of fauna and ora
from other regions (Chapman, 1917; Echeverry
& Morrone, 2013).
Role of the PRZ-CRV in conservation of
threatened and endemic species
Our results, taken together with other stud-
ies (e.g. Marsh et al., 2016; Negret et al., 2019),
suggest that threatened and endemic Colombia’s
species with distributions in the MVR are fac-
ing a dramatic loss of their habitats. Our results
indicated that less than 50% of the climatically
suitable areas for each species are covered by
forests, indicating that in fact their optimal dis-
tribution area is smaller. This pattern could get
worse in the future under climatic scenarios with
a higher temperature and lower precipitation (Ta-
ble 2b). Additionally, our results showed that less
than 10% of the species distribution in the pres-
ent period is covered by PAs, which aggravates
the situation of maintenance and conservation of
these species.
The forest loss is one of the main threats
to the species conservation. During the last de-
cades, more than 2 000 000 km2 of tropical forest
have been lost, threatening the survival of for-
est specialist species (Hansen et al., 2010; Symes
et al., 2018; Agudelo-Hz et al., 2019; Donald et
al., 2019; Negret et al., 2019). In Colombia, hu-
mid forests are one of the most diverse and frag-
mented habitats. A recent study indicated that
more than half of Colombia’s regionally endemic
bird species are projected to lose at least half of
their habitats by 2040 (Negret et al., 2021), in-
cluding Crax alberti, a Cracidae species threat-
ened by habitat loss and also assessed in our
study. According to our analysis, this species has
only 36% of its climatically suitable area covered
by forests (i.e. less than 60% of its distribution)
and it is expected that in future scenarios with
higher temperature and lower precipitation, its
climatically suitable area decreases (Table 2b).
For forest-dependent amphibian species, a dra-
matic impact is also observed in their distribution
range, with a reduction of 50% (e.g. Sachatamia
punctulata) and 70% (e.g. Diasporus anthrax) by
2050. Additionally, ecological characteristics of
species could worsen the situation, since some of
them have a reduced vagility (e.g. amphibians)
or little tolerance to open habitats (e.g. primates),
which limit their ability to respond to changes in
their habitat and climatic conditions.
An additional problem is the lack of PAs in the
MVR to ensure the conservation of habitats suit-
able for species. Our results indicated that for the
eight analysed species less than 10% of their distri-
bution is covered by PAs. This evidences the need
to create PAs in this region of the country, where
a high number of endangered and endemic species
live. It should be noted that this problem has not
been ignored by researchers and national institu-
tions (e.g. Fundación Proyecto Primates, Panthera,
and Wildlife Conservation Society of Colombia),
which since 2016 have requested the Colombian
government to incorporate the San Lucas Moun-
tain Range into the National Park system (Parques
Nacionales Naturales de Colombia, 2021). Unfor-
tunately, the San Lucas Mountain Range has still
remained unprotected and its forests and fauna are
subject to armed conict and illegal activities in
the region (Negret et al., 2019).
Finally, our study suggests that the PRZ-
CRV is a signicant territory for conservation
and maintenance of endemic and threatened spe-
cies in the MVR. It plays an important role as an
ally to the PAs system. Our results indicated that
the PRZ-CRV covers a part of distribution range
of studied species. Even for some of them, this
PRZ covers a higher percentage than PAs (Table
2a). Additionally, if we consider the change in
the species distribution under future climatic sce-
narios, this region might present suitable climatic
areas for the maintenance of these species (Fig.
A3). Although in the PRZ-CRV there are pro-
cesses of changes in the natural cover, forests of
the region conserve a structure and oristic com-
position typical for tropical humid forests (Ortiz-
Lozada, 2020). This is also supported by some
recent studies that report the presence of bioin-
dicator species of high-quality habitats in relict
forests in southern Bolivar (Arbeláez-Cortés et
al., 2021). Therefore, the higher forest coverage
in the PRZ-CRV, the greater species diversity,
and the peasant commitment to the conservation
of their natural resources make this area a poten-
tial ally for the biodiversity conservation of the
region. Even more, if we consider that until now
there are no formal PAs in the San Lucas Moun-
tain Range (one of the most diverse regions of
Colombia) and that a few PAs around the MVR
have a low-moderate impact in avoiding defor-
estation by their small size (Liévano-Latorre et
al., 2021), this PRZ is a valuable community tool
for biodiversity management that can support the
mission of PAs.
Nature Conservation Research. Заповедная наука 2023. 8(1) https://dx.doi.org/10.24189/ncr.2023.003

Conclusions
Our ndings suggest that the PRZ-CRV is an
important area aiming to maintain and protect hu-
mid forests (especially those belonging to the San
Lucas Mountain Range) and its fauna, acting as
an ally to the PA system in the MVR. Addition-
ally, we showed that the capacity of this peasant
organisation to recognise, dene, and generate
conservation agreements (e.g. the establishment
of the «yellow line») is a successful case of com-
munity management in the area with concrete ef-
fects on the conservation of ecosystems. It should
be noted that among the aims of PRZs are con-
trolling the inadequate expansion of the country’s
agricultural frontier, creating and building a com-
prehensive proposal for sustainable human devel-
opment, and assuring ecosystem and biodiversity
conservation. Thus, these peasant agreements
could serve as allies in the biodiversity conserva-
tion of such areas as MVR where there is a poor
representation of PAs and buer zones around
PAs with controlled activities.
Acknowledgements
The authors thank funding from the Ministerio de
Ciencia, Tecnología e Innovación, Ministerio de Educación
Nacional, Ministerio de Industria, Comercio y Turismo,
and ICETEX, Programme Ecosistema Cientíco-Colombia
Cientíca from Fondo Francisco José de Caldas (Grant RC-
FP44842-212-2018). We are also grateful for the postdoctoral
stay programme of the Universidad Industrial de Santanter
(Colombia) for the support received. Finally, we thank Sergio
Cordoba and Björn Reu (Industrial University of Santander)
for their advice on the pre-processing of satellite images.
References
Agudelo-Hz W.J., Urbina-Cardona N., Armenteras-Pascual
D. 2019. Critical shifts on spatial traits and the risk
of extinction of Andean anurans: an assessment of the
combined eects of climate and land-use change in
Colombia. Perspectives in Ecology and Conservation
17(4): 206–219. DOI: 10.1016/j.pecon.2019.11.002
Allouche O., Tsoar A., Kadmon R. 2006. Assessing the ac-
curacy of species distribution models: prevalence,
kappa and the true skill statistic (TSS). Journal of Ap-
plied Ecology 43(6): 1223–1232. DOI: 10.1111/j.1365-
2664.2006.01214.x
Álvarez M.D. 2002. Illicit Crops and Bird Conservation Pri-
orities in Colombia. Conservation Biology 16(4): 1086–
1096. DOI: 10.1046/j.1523-1739.2002.00537.x
Arbeláez-Cortés E., Villamizar-Escalante D., Trujillo-Arias
N. 2021. New voucher specimens and tissue samples
from an avifaunal survey of the Middle Magdalena
Valley of Bolívar, Colombia, bridge geographical and
temporal gaps. Wilson Journal of Ornithology 132(3):
773–779. DOI: 10.1676/20-56
Armenteras D., Rodríguez N., Retana J., Morales M. 2011. Un-
derstanding deforestation in montane and lowland forests
of the Colombian Andes. Regional Environmental Change
11(3): 693–705. DOI: 10.1007/s10113-010-0200-y
Asociación campesina del valle del rio Cimitarra. 2008. La
zona de reserva campesina del valle del río Cimitarra:
una iniciativa agraria de paz. Available from https://
www.prensarural.org/spip/spip.php?article1098
Becker N., Werft W., Toedt G., Lichter P., Benner A. 2009.
penalizedSVM: a R-package for feature selection SVM
classication. Bioinformatics 25(13): 1711–1712. DOI:
10.1093/bioinformatics/btp286
Breiman L. 2001. Random Forests. Machine Learning
451(45): 5–32. DOI: 10.1023/A:1010933404324
Castro-Nunez A., Mertz O., Buritica A., Sosa C.C., Lee S.T.
2017. Land related grievances shape tropical forest-
cover in areas aected by armed-conict. Applied Geog-
raphy 85: 39–50. DOI: 10.1016/j.apgeog.2017.05.007
Chadid M.A., Dávalos L.M., Molina J., Armenteras D. 2015.
A Bayesian Spatial Model highlights distinct dynamics
in deforestation from coca and pastures in an Andean
biodiversity hotspot. Forests 6(11): 3828–3846. DOI:
10.3390/f6113828
Chapman F. 1917. The distribution of bird-life in Colombia: a
contribution to a biological survey of South America. Bul-
letin of the American Museum of Natural History 36: 1–728.
Cortes C., Vapnik V. 1995. Support-vector networks. Machine
Learning 20: 273–297. DOI: 10.1007/bf00994018
Cuervo Maya A.M., Manuel Ochoa J., Salaman P. 1999. Últi-
mas evidencias del Paujil de Pico Azul (Crax alberti)
con anotaciones sobre su historia natural, distribución
actual y amenazas especícas. Boletín de la Sociedad
Antioqueña de Ornitología 10: 18–19.
Da Ponte E., Mack B., Wohlfart C., Rodas O., Fleckenstein M.,
Oppelt N., Dech S., Kuenzer C. 2017. Assessing forest
cover dynamics and forest perception in the Atlantic For-
est of Paraguay, combining remote sensing and household
level data. Forests 8(10): 389. DOI: 10.3390/f8100389
Dávalos L.M. 2001. The San Lucas mountain range in Co-
lombia: How much conservation is owed to the vio-
lence?. Biodiversity and Conservation 10(1): 69–78.
DOI: 10.1023/A:1016651011294
Dávalos L.M., Bejarano A.C., Hall M.A., Correa H.L., Cor-
thals A., Espejo O.J. 2011. Forests and drugs: Coca-
driven deforestation in tropical biodiversity hotspots.
Environmental Science and Technology 45(4): 1219–
1227. DOI: 10.1021/es102373d
Donald P.F., Arendarczyk B., Spooner F., Buchanan G.M.
2019. Loss of forest intactness elevates global extinc-
tion risk in birds. Animal Conservation 22(4): 341–347.
DOI: 10.1111/acv.12469
Dufresne J.L., Foujols M.A., Denvil S., Caubel A., Marti O.,
Aumont O., Balkanski Y., Bekki S., Bellenger H., Ben-
shila R., Bony S., Bopp L., Braconnot P., Brockmann P.,
Nature Conservation Research. Заповедная наука 2023. 8(1) https://dx.doi.org/10.24189/ncr.2023.003

Cadule P., Cheruy F., Codron F., Cozic A., Cugnet D.,
de Noblet N., Duvel J.P., Ethé C., Fairhead L., Fichefet
T., Flavoni S., Friedlingstein P., Grandpeix J.Y., Guez L.,
Guilyardi E., Hauglustaine D. et al. 2013. Climate change
projections using the IPSL-CM5 Earth System Model:
From CMIP3 to CMIP5. Climate Dynamics 40(9–10):
2123–2165. DOI: 10.1007/s00382-012-1636-1
Echeverry A., Morrone J.J. 2013. Generalized tracks, area
cladograms and tectonics in the Caribbean. Journal of
Biogeography 40(9): 1619–1637. DOI: 10.1111/jbi.12117
Elith J., Graham C.H., Anderson R.P., Dudík M., Ferrier
S., Guisan A., Hijmans R.J., Huettmann F., Leathwick
J.R., Lehmann A., Li J., Lohmann L.G., Loiselle B.A.,
Manion G., Moritz C., Nakamura M., Nakazawa Y.,
Overton J., Peterson A.T., Phillips S.J., Richardson
K.S., Scachetti-Pereira R., Schapire R.E., Soberón J.,
Williams S., Wisz M.S., Zimmermann N. 2006. Novel
methods improve prediction of species’ distributions
from occurrence data. Ecography 29(2): 129–151. DOI:
10.1111/j.2006.0906-7590.04596.x
Etter A., van Wyngaarden W. 2000. Patterns of land-
scape transformation in Colombia, with emphasis
in the Andean region. Ambio 29(7): 432–439. DOI:
10.1579/0044-7447-29.7.432
Etter A., McAlpine C., Wilson K., Phinn S., Possingham
H. 2006. Regional patterns of agricultural land use
and deforestation in Colombia. Agriculture, Eco-
systems and Environment 114(2–4): 369–386. DOI:
10.1016/j.agee.2005.11.013
Etter A., McAlpine C., Possingham H. 2008. Historical pat-
terns and drivers of landscape change in Colombia
since 1500: A regionalized spatial approach. Annals of
the Association of American Geographers 98(1): 2–23.
DOI: 10.1080/00045600701733911
Foody G.M. 2020. Explaining the unsuitability of the kappa
coecient in the assessment and comparison of the ac-
curacy of thematic maps obtained by image classica-
tion. Remote Sensing of Environment 239: 111–630.
DOI: 10.1016/J.RSE.2019.111630
Forero-Medina G., Joppa L. 2010. Representation of global
and national conservation priorities by Colombia’s pro-
tected area network. PLoS ONE 5(10): e13210. DOI:
10.1371/journal.pone.0013210
Gallery R.E. 2014. Ecology of Tropical Rain Forests. In:
R. Monson (Eds.): Ecology and the Environment. The
Plant Sciences. Vol. 8. New York: Springer. P. 247–272.
DOI: 10.1007/978-1-4614-7501-9_4
Hansen M.C., Stehman S.V., Potapov P.V. 2010. Quanti-
cation of global gross forest cover loss. Proceedings
of the National Academy of Sciences of the United
States of America 107(19): 8650–8655. DOI: 10.1073/
pnas.0912668107
Hijmans R.J., Cameron S.E., Parra J.L., Jones P.G., Jarvis A.
2005. Very high resolution interpolated climate surfaces
for global land areas. International Journal of Climatol-
ogy 25(15): 1965–1978. DOI: 10.1002/joc.1276
Jones C.D., Hughes J.K., Bellouin N., Hardiman S.C., Jones
G.S., Knight J., Liddicoat S., O’Connor F.M., Andres
R.J., Bell C., Boo K.O., Bozzo A., Butchart N., Cadule
P., Corbin K.D., Doutriaux-Boucher M., Friedlingstein P.,
Gornall J., Gray L., Halloran P.R., Hurtt G., Ingram W.J.,
Lamarque J.F., Law R.M., Meinshausen M., Osprey S.,
Palin E.J., Parsons Chini L., Raddatz T., Sanderson M.G.
et al. 2011. The HadGEM2-ES implementation of CMIP5
centennial simulations. Geoscientic Model Development
4(3): 543–570. DOI: 10.5194/gmd-4-543-2011
Kramer-Schadt S., Niedballa J., Pilgrim J.D., Schröder B.,
Lindenborn J., Reinfelder V., Stillfried M., Heckmann
I., Scharf A.K., Augeri D.M., Cheyne S.M., Hearn A.J.,
Ross J., Macdonald D.W., Mathai J., Eaton J., Marshall
A.J., Semiadi G., Rustam R., Bernard H., Alfred R.,
Samejima H., Duckworth J.W., Breitenmoser-Wuersten
C., Belant J.L., Hofer H., Wilting A. 2013. The impor-
tance of correcting for sampling bias in MaxEnt spe-
cies distribution models. Diversity and Distributions
19(11):1366–1379. DOI: 10.1111/ddi.12096
Kuhn M. 2008. Building predictive models in R using the
caret package. Journal of Statistical Software 28(5):
1–26. DOI: 10.18637/jss.v028.i05
Laverde-R O., Múnera-R C., Renjifo L.M. 2005. Prefer-
encia de hábitat por Capito hypoleucus, ave colom-
biana endémica y amenazada. Ornitología Colom-
biana 3: 62–73.
Liaw A., Wiener M. 2002. Classication and Regression by
randomForest. R News 2(3): 18–22.
Liévano-Latorre L.F., Brum F.T., Loyola R. 2021. How ef-
fective have been guerrilla occupation and protected
areas in avoiding deforestation in Colombia?. Biologi-
cal Conservation 253: 108916. DOI: 10.1016/j.bio-
con.2020.108916
Marsh C., Link A., King-Bailey G., Donati G. 2016. Eects
of fragment and vegetation structure on the population
abundance of Ateles hybridus, Alouatta seniculus and
Cebus albifrons in Magdalena Valley, Colombia. Folia
Primatolgica 87(1): 17–30. DOI: 10.1159/000443929
Müller-Wilm U. 2016. Sen2Cor conguration and user
manual, Ref. S2-PDGS-MPC-L2A- SUM-V2.3. Euro-
pean Space Agency. Available from http://step.esa.int/
thirdparties/sen2cor/2.3.0/[L2A-SUM]%20S2-PDGS-
MPC-L2A-SUM%20[2.3.0].pdf
Murillo-Sandoval P.J., Van Dexter K., Van Den Hoek J.,
Wrathall D., Kennedy R. 2020. The end of gunpoint
conservation: forest disturbance after the Colombian
peace agreement. Environmental Research Letters
15(3): 034033. DOI: 10.1088/1748-9326/AB6AE3
Myers N. 1988. Threatened biotas: «Hotspots» in tropical
forests. Environmentalist 8: 187–208.
Negret P.J., Sonter L., Watson J., Possingham H.P., Jones
K.R., Suarez C., Ochoa-Quintero J.M., Maron M. 2019.
Emerging evidence that armed conict and coca cultiva-
tion inuence deforestation patterns. Biological Conser-
vation 239: 108176. DOI: 10.1016/j.biocon.2019.07.021
Nature Conservation Research. Заповедная наука 2023. 8(1) https://dx.doi.org/10.24189/ncr.2023.003

Negret P.J., Maron M., Fuller R.A., Possingham H.P., Watson
J.E.M., Simmonds J.S. 2021. Deforestation and bird
habitat loss in Colombia. Biological Conservation 257:
109044. DOI: 10.1016/j.biocon.2021.109044
Ochoa-Quintero J.M., Melo-Vásquez I., Palacio-Vieira J.A.,
Patiño Á.M. 2005. Nuevos registros y notas sobre la
historia natural del paujil colombiano Crax alberti al
nororiente de la cordillera central colombiana. Orni-
tología Colombiana 3: 42–50.
Olofsson P., Foody G.M., Stehman S.V., Woodcock C.E.
2013. Making better use of accuracy data in land
change studies: Estimating accuracy and area and quan-
tifying uncertainty using stratied estimation. Remote
Sensing of Environment 129: 122–131. DOI: 10.1016/j.
rse.2012.10.031
Olson D.M., Dinerstein E., Wikramanayake E.D., Burgess
N.D., Powell G.V., Underwood E.C., D’Amico J.A., Itoua
I., Strand H.E., Morrison J.C., Loucks C.J., Allnutt T.F.,
Ricketts T.H., Kura Y., Lamoreux J.F., Wettengel W.W.,
Hedao P., Kassem K.R. 2001. Terrestrial Ecoregions of the
World: A New Map of Life on Earth: A new global map of
terrestrial ecoregions provides an innovative tool for con-
serving biodiversity. BioScience 51(11): 933–938. DOI:
10.1641/0006-3568(2001)051[0933:TEOTWA]2.0.CO;2
Ortiz S. 2018. Las Zonas de Reserva Campesina: retos y ex-
periencias signicativas en su implementación. Bogotá:
FAO/ANT. 545 p.
Ortiz-Lozada J.D. 2020. Estructura y composición orística
de un bosque húmedo tropical, en el sur del departa-
mento de Bolívar, Colombia. B.Sc. Thesis. Bucara-
manga, Santander, Colombia: Universidad Industrial de
Santander. 48 p.
Parques Nacionales Naturales de Colombia. 2021. Serranía
de San Lucas: busca una estrategia de conservación.
Parques Nacionales. Available from https://www.par-
quesnacionales.gov.co/portal/es/serrania-de-san-lucas-
busca-una-estrategia-de-conservacion
Pearson R., Raxworthy C., Nakamura M., Peterson A.T.
2007. Predicting species distributions from small num-
bers of occurrence records: a test case using cryptic
geckos in Madagascar. Journal of Biogeography 34(1):
102–117. DOI: 10.1111/j.1365-2699.2006.01594.x
Peterson A.T., Soberón J., Pearson R.G., Anderson R.P.,
Martínez-Meyer E., Nakamura M., Araújo M. 2011.
Ecological Niches and Geographic Distributions
(MPB-49). Princeton: Princeton University Press. 328
p. DOI: 10.23943/princeton/9780691136868.001.0001
Phillips S.J., Anderson R.P., Schapire R.E. 2006. Maxi-
mum entropy modeling of species geographic distribu-
tions. Ecological Modelling 190(3–4): 231–259. DOI:
10.1016/j.ecolmodel.2005.03.026
Pimm S.L., Russell G.J., Gittleman J.L., Brooks T.M. 1995.
The future of biodiversity. Science 269(5222): 347–350.
DOI: 10.1126/science.269.5222.347
Quijano-Mejía C., Linares-García J. 2017. Zonas de Reserva
Campesina: territorialidades en disputa. El caso del
Valle del río Cimitarra, Colombia. Revista De Tra-
bajo Social e Intervención Social 24: 225–251. DOI:
10.25100/prts.v0i24.4478
Quintero-Vallejo E., Benavides A., Moreno N., Gonzalez-
Caro S. 2017. Bosques Andinos, estado actual y retos
para su conservación en Antioquia. Medellín, Colom-
bia: Fundación Jardín Botánico de Medellín Joaquín
Antonio Uribe-Programa Bosques Andinos. 542 p.
R Core Team. 2020. R: A language and environment for statisti-
cal computing. Vienna, Austria: R Foundation for Statisti-
cal Computing. Available from https://www.R-project.org/
Rojas-Morales J., Castro-Gómez A., Ramírez-Castaño V. 2014.
Sachatamia punctulata (Ruiz-Carranza y Lynch 1995).
Catálogo de Anbios y Reptiles de Colombia 2: 30–36.
Salaman P., Donegan T., Cuervo A. 2002. New distributional
bird records from Serranía de San Lucas and adjacent
Central Cordillera of Colombia. Bulletin of the British
Ornithologists’ Club 122: 285–303.
Sanchez-Cuervo A.M., Aide T.M. 2013. Identifying hotspots
of deforestation and reforestation in Colombia (2001–
2010): implications for protected areas. Ecosphere
4(11): 1–21. DOI: 10.1890/ES13-00207.1
Schmidt G.A., Kelley M., Nazarenko L., Ruedy R., Russell
G.L., Aleinov I., Bauer M., Bauer S.E., Bhat M.K., Bleck
R., Canuto V., Chen Y.H., Cheng Y., Clune T.L., Del Genio
A., De Fainchtein R., Faluvegi G., Hansen J.E., Healy
R.J., Kiang N.Y., Koch D., Lacis A.A., Legrande A.N.,
Lerner J., Lo K.K., Matthews E.E., Menon S., Miller
R.L., Oinas V., Oloso A.O. et al. 2014. Conguration and
assessment of the GISS ModelE2 contributions to the
CMIP5 archive. Journal of Advances in Modeling Earth
Systems 6(1): 141–184. DOI: 10.1002/2013MS000265
Symes W.S., Edwards D.P., Miettinen J., Rheindt F.E.,
Carrasco L.R. 2018. Combined impacts of deforestation
and wildlife trade on tropical biodiversity are severely
underestimated. Nature Communications 9(1): 4052.
DOI: 10.1038/s41467-018-06579-2
Tocancipá Falla J., Ramírez Castrillón C.A. 2018. The
new rural dynamics in peasants reserve zones in Co-
lombia. Perspectiva Geográca 23(1): 31–52. DOI:
10.19053/01233769.5796
Trujillo-Arias N., Dantas G.P.M., Arbeláez-Cortés E., Naoki
K., Gómez M.I., Santos F.R., Miyaki C.Y., Aleixo A.,
Tubaro P.L., Cabanne G.S. 2017. The niche and phylo-
geography of a passerine reveal the history of biologi-
cal diversication between the Andean and the Atlantic
forests. Molecular Phylogenetics and Evolution 112:
107–121. DOI: 10.1016/j.ympev.2017.03.025
Tucker C., Townshend J. 2000. Strategies for monitoring
tropical deforestation using satellite data. International
Journal of Remote Sensing 21(6–7): 1461–1471. DOI:
10.1080/014311600210263
van Vuuren D.P., Edmonds J., Kainuma M., Riahi K., Thom-
son A., Hibbard K., Hurtt G.C., Kram T., Krey V., La-
marque J.F., Masui T., Meinshausen M., Nakicenovic
N., Smith S.J., Rose S.K. 2011. The representative con-
Nature Conservation Research. Заповедная наука 2023. 8(1) https://dx.doi.org/10.24189/ncr.2023.003

centration pathways: An overview. Climatic Change
109(1): 5. DOI: 10.1007/s10584-011-0148-z
Venables W.N., Ripley B.D. 2002. Modern Applied Statistics
with S. 4th ed. New York: Springer. 498 p.
Yagoub M. 2018. With Armed Strike in Colombia, ELN's War-
mongers Seek Upper Hand. In: InSight Crime. Available
from https://www.insightcrime.org/news/ analysis/armed-
strike-colombia-eln-warmongers-seek-upper-hand/
Yoshida T., Omatu S. 1994. Neural Network Approach to
Land Cover Mapping. IEEE Transactions on Geosci-
ence and Remote Sensing 32(5): 1103–1109. DOI:
10.1109/36.312899
Appendix. Sentinel images employed for forest classication analyses and performance of three non-
parametric classier algorithms in the classication of coverage, and number of species occurrences,
performance and validation of ecological niche models obtained from MaxEnt.
Fig. A1. Performance of the random forest (rf), support
vector machines (svm) and neural network (nnet) algorithms
in the classication of coverage based on the accuracy and
kappa statistics.
Fig. A2. Some examples of receiver operating characteristic (ROC) curves obtained from MaxEnt models for the present
period. The AUC (area under ROC curve) values vary from 0 to 1. Values < 0.5 indicate that the model performance is
worse than random; the value 0.5 indicates the performance that is not better than random; values 0.5–0.7 indicate the poor
performance; values 0.7–0.9 indicate the reasonable or moderate performance; values > 0.9 indicates the high performance.
Nature Conservation Research. Заповедная наука 2023. 8(1) https://dx.doi.org/10.24189/ncr.2023.003

Fig. A3. Ecological niche models for the present and the future (2050) of eight endemic species in Colombia. Designations:
i) Sachatamia punctulata, AUC: 0.96 / TSS: 0.90; ii) Diasporus anthrax, AUC: 0.97 / TSS: 0.95; iii) Agalychnis terranova,
AUC: 0.98 / TSS: 0.94; iv) Crax alberti, AUC: 0.96 / TSS: 0.92; v) Capito hypolucus, AUC: 0.94 / TSS; 0.84; vi) Habia
gutturalis, AUC: 0.94 / TSS; 0.87; vii) Saguinus leucopus, AUC: 0.96 / TSS: 0.88; viii) Ateles hybridus, AUC: 0.95 / TSS:
0.80). Points on the maps in the present models represent the species occurrence.
Nature Conservation Research. Заповедная наука 2023. 8(1) https://dx.doi.org/10.24189/ncr.2023.003

Table A1. Sentinel images with less than 40% cloud cover employed for the forest classication analyses of the Peasant
Reserve Zone-Cimitarra River Valley, Colombia
Satellite Image level Date Tile 18NWN 18NWP 18NXN 18NXP
S2A MSIL1C 04.01.2017 18NWN X X X X
S2A MSIL1C 03.02.2017 18NWN X X X X
S2A MSIL1C 13.02.2017 18NWN X X X X
S2A MSIL1C 04.05.2017 18NWN X X X X
S2A MSIL1C 24.05.2017 18NWN X X X –
S2A MSIL1C 01.09.2017 18NWN X X X X
S2A MSIL1C 31.10.2017 18NWN X X X X
S2B MSIL1C 15.11.2017 18NWN X X X X
S2B MSIL1C 05.12.2017 18NWN X X X X
S2A MSIL1C 10.12.2017 18NWP – X – –
S2B MSIL1C 15.12.2017 18NWN X X X X
S2B MSIL1C 25.12.2017 18NWN X X X X
S2A MSIL1C 08.02.2018 18NWP – X – X
S2B MSIL1C 13.02.2018 18NWP – X – X
S2A MSIL1C 18.02.2018 18NWP – X – X
S2B MSIL1C 23.06.2018 18NWN X X – X
S2A MSIL1C 18.07.2018 18NXN – – X X
S2B MSIL1C 02.08.2018 18NXN – – X –
S2A MSIL1C 07.08.2018 18NWN X – – X
S2A MSIL1C 27.08.2018 18NXN – – X –
S2A MSIL1C 05.12.2018 18NXP – – – X
S2B MSIL1C 10.12.2018 18NWN X X X –
S2A MSIL2A 15.12.2018 18NXP – – – X
S2A MSIL2A 25.12.2018 18NWN X X X X
S2B MSIL2A 30.12.2018 18NWP – X – –
S2A MSIL2A 04.01.2019 18NWP – X – –
S2A MSIL2A 24.01.2019 18NWP – X – –
S2A MSIL2A 04.04.2019 18NWP – X – –
S2A MSIL2A 24.05.2019 18NWP – X – X
S2A MSIL2A 03.06.2019 18NXP – – – X
S2B MSIL2A 08.06.2019 18NWN X X X X
S2B MSIL2A 18.06.2019 18NXP – – – X
S2A MSIL2A 03.07.2019 18NWN X – – –
S2B MSIL2A 18.07.2019 18NWN X X X X
S2B MSIL2A 28.07.2019 18NWN X X X X
S2A MSIL2A 02.08.2019 18NXN – – X X
S2B MSIL2A 07.08.2019 18NWN X X X X
S2A MSIL2A 13.08.2019 18NXP – – – X
S2A MSIL2A 22.08.2019 18NWN X – X X
S2B MSIL2A 27.08.2019 18NWN X X X X
S2A MSIL2A 09.09.2019 18NWN X – – –
Table A2. The number of occurrences used to create MaxEnt
models of eight species endemic (or near endemic) to Colombia
distributed in the Magdalena Valley Region, Colombia
Class Species Occurrences
Birds
Habia gutturalis 98
Capito hypoleucus 77
Crax alberti 38
Amphibians
Sachatamia
punctulata 13
Diasporus anthrax 21
Agalychnis
terranova 15
Primates Saguinus leucopus 60
Ateles hybridus 72
Nature Conservation Research. Заповедная наука 2023. 8(1) https://dx.doi.org/10.24189/ncr.2023.003

РОЛЬ ЗАПОВЕДНОЙ ЗОНЫ КАМПЕСИНО В ДОЛИНЕ РЕКИ
МАГДАЛЕНА (КОЛУМБИЯ) В СОХРАНЕНИИ НАХОДЯЩИХСЯ
ПОД УГРОЗОЙ ИСЧЕЗНОВЕНИЯ ВЛАЖНЫХ ТРОПИЧЕСКИХ ЛЕСОВ
Н. Трухильо-Ариас1,2,* , В. Х. Серрано-Кардозо1, М. П. Рамирез-Пинилла1
1Индустриальный университет Сантандера, Колумбия
2Институт исследования биологических ресурсов Александра фон Гумбольдта, Колумбия
*e-mail: natitrujillo@gmail.com
Тропические леса Колумбии имеют один из самых высоких показателей сокращения площади лесов в
мире. Влажный тропический лес в долине реки Магдалена является одной из экосистем с самым вы-
соким риском трансформации ландшафта, несмотря на то, что он является местом обитания для мно-
гих эндемичных и находящихся под угрозой исчезновения видов. Целью данного исследования было
оценить роль крестьянской заповедной зоны в сохранении влажных тропических лесов и видов, на-
ходящихся под угрозой исчезновения, в районе долины реки Магдалена. Для достижения этой цели мы
провели многовременной анализ динамики лесов в крестьянской заповедной зоне – долине реки Чими-
тарра (КЗЗ-ДРЧ) и оценили риск исчезновения восьми видов, эндемичных для Колумбии. Полученные
результаты показали, что наиболее обширным типом ландшафта в КЗЗ-ДРЧ является лес (56.30%), за
которым следуют открытые территории (38.75%). Анализ динамики лесов показал, что лесной покров
сократился на 3.82% с 2017 по 2019 гг. При этом наиболее сохранившейся является территория с реду-
тами леса Серрания-де-Сан-Лукас. Наконец, результаты показали, что менее 50% климатически под-
ходящих территорий для каждого вида покрыты лесами и менее 10% этих площадей занимают особо
охраняемые природные территории (ООПТ). В то же время для таких видов, как Agalychnis terranova и
Ateles hybridus, КЗЗ-ДРЧ охватил более высокую долю их ареала, чем все вместе взятые ООПТ в этой
экосистеме. Таким образом, полученные результаты показали, что КЗЗ-ДРЧ может быть ключевой тер-
риторией для поддержания и сохранения влажных тропических лесов и связанной с ними фауны, играя
важную роль «союзника» системы ООПТ в регионе долины реки Магдалена.
Ключевые слова: Анды, крестьянская заповедная зона, ландшафт, сокращение лесов, угрожаемый вид
Nature Conservation Research. Заповедная наука 2023. 8(1) https://dx.doi.org/10.24189/ncr.2023.003