This is not the document you are looking for? Use the search form below to find more!

Report home > Science

Ecological-Niche Modeling and Prioritization of Conservation-Area Networks for Mexican Herpetofauna

0.00 (0 votes)
Document Description
One of the most important tools in conservation biology is information on the geographic distribution of species and the variables determining those patterns. We used maximum-entropy niche modeling to run distribution models for 222 amphibian and 371 reptile species (49% endemics and 27% threatened) for which we had 34,619 single geographic records. The planning region is in southeastern Mexico, is 20% of the country's area, includes 80% of the country's herpetofauna, and lacks an adequate protected-area system. We used probabilistic data to build distribution models of herpetofauna for use in prioritizing conservation areas for three target groups (all species and threatened and endemic species). The accuracy of species-distribution models was better for endemic and threatened species than it was for all species. Forty-seven percent of the region has been deforested and additional conservation areas with 13.7% to 88.6% more native vegetation (76% to 96% of the areas are outside the current protected-area system) are needed. There was overlap in 26 of the main selected areas in the conservation-area network prioritized to preserve the target groups, and for all three target groups the proportion of vegetation types needed for their conservation was constant: 30% pine and oak forests, 22% tropical evergreen forest, 17% low deciduous forest, and 8% montane cloud forests. The fact that different groups of species require the same proportion of habitat types suggests that the pine and oak forests support the highest proportion of endemic and threatened species and should therefore be given priority over other types of vegetation for inclusion in the protected areas of southeastern Mexico.
File Details
  • Added: April, 28th 2010
  • Reads: 1122
  • Downloads: 59
  • File size: 308.35kb
  • Pages: 11
  • Tags: amphibians, area prioritization, conservation planning, maxent, niche based distribution models, protected areas, reptiles, site selection
  • content preview
Submitter
Embed Code:

Add New Comment




Related Documents

Modeling and Simulation of Dynamic Systems, (March 21, 1997), by Robert L. Woods, Kent L. Lawrence, SOLUTION MANUAL

by: bestsmtb, 34 pages

Modeling and Simulation of Dynamic Systems, (March 21, 1997), by Robert L. Woods, Kent L. Lawrence, SOLUTION MANUAL --------------------------------------------------------- My email is: ...

Component Based Modeling and Simulation of Value Stream Mapping for Lean Production Systems

by: shinta, 8 pages

Value Stream Mapping is an important tool in the implementation of lean manufacturing. It identifies the waste in the system, paving the way for a successful lean implementation. VSM is a paper ...

ASSESSING THE DIET, EXERCISE, BODY IMAGE, AND WEIGHT OF ADOLESCENTS: A GUIDE FOR OUT-OF-SCHOOL TIME PROGRAM PRACTITIONERS

by: shinta, 4 pages

This brief discusses diet, exercise, body image, and weight and also provides information for practitioners on how to measure these factors among youth in their program. Most adolescents do not eat ...

Modeling and Simulation of a Four Bar Mechanism and a Crane using ProM - PDF

by: tetsuo, 16 pages

Modeling and Simulation of a Four Bar Mechanism and a Crane using ProM - PDF

Body Concerns In and Out of the Bedroom: Implications for Sexual ...

by: joline, 46 pages

Objectification theory (Fredrickson & Roberts, 1997) proposes that body image concerns impair sexual function and satisfaction. The present study was designed to test whether body shame was ...

IMPROVEMENT AND POPULARIZATION OF DIVERSIFIED CASSAVA PRODUCTS FOR INCOME GENERATION AND FOOD SECURITY : A CASE STUDY OF KIBABU

by: shinta, 15 pages

Kibabu is a traditional, plain, deep fried cassava product prepared from cassava paste. It was noticed in Muheza district, Tanzania, to be nutrient-deficient as it was prepared from plain ...

Sustainable Aboriginal Communities: Balancing Economic Vitality, Ecological Integrity, Social and Cultural Well-being

by: samanta, 9 pages

Sustainable resource use and development of societal well-being have been two interwoven threads since the beginning of human activity in the North. Non-renewable resource development activities ...

Foundations of Finance The Logic and Practice of Financial Management Keown 6th Edition Solutions Manual

by: gordonbarbier, 48 pages

Foundations of Finance The Logic and Practice of Financial Management Keown 6th Edition Solutions Manual

Theory and Practice of Counseling and Psychotherapy Corey 8th Edition Test Bank

by: gordonbarbier, 48 pages

Theory and Practice of Counseling and Psychotherapy Corey 8th Edition Test Bank

Theory and Practice of Counseling and Psychotherapy Corey 8th Edition Test Bank

by: georgesheslers, 48 pages

Theory and Practice of Counseling and Psychotherapy Corey 8th Edition Test Bank

Content Preview
Contributed Paper
Ecological-Niche Modeling and Prioritization
of Conservation-Area Networks for Mexican
Herpetofauna

J. NICOL´
AS URBINA-CARDONA,∗†§ AND OSCAR FLORES-VILLELA∗‡
∗Museo de Zoolog´ıa “Alfonso L. Herrera,” Departamento de Biolog´ıa Evolutiva, Facultad de Ciencias, UNAM. A. P. 70-399,

exico DF 04510
†Conservation International, Colombia, Carrera 13 # 71-41, Bogot´
a, Colombia
‡Department of Biology, University of Texas at Arlington, Box 19498, 337 Life Science Building, Arlington, TX 76019-0498,U.S.A.
Abstract:
One of the most important tools in conservation biology is information on the geographic distri-
bution of species and the variables determining those patterns. We used maximum-entropy niche modeling to
run distribution models for 222 amphibian and 371 reptile species (49% endemics and 27% threatened) for
which we had 34,619 single geographic records. The planning region is in southeastern Mexico, is 20% of the
country’s area, includes 80% of the country’s herpetofauna, and lacks an adequate protected-area system. We
used probabilistic data to build distribution models of herpetofauna for use in prioritizing conservation areas
for three target groups (all species and threatened and endemic species). The accuracy of species-distribution
models was better for endemic and threatened species than it was for all species. Forty-seven percent of the
region has been deforested and additional conservation areas with 13.7% to 88.6% more native vegetation
(76% to 96% of the areas are outside the current protected-area system) are needed. There was overlap in 26 of
the main selected areas in the conservation-area network prioritized to preserve the target groups, and for all
three target groups the proportion of vegetation types needed for their conservation was constant: 30% pine
and oak forests, 22% tropical evergreen forest, 17% low deciduous forest, and 8% montane cloud forests. The
fact that different groups of species require the same proportion of habitat types suggests that the pine and
oak forests support the highest proportion of endemic and threatened species and should therefore be given
priority over other types of vegetation for inclusion in the protected areas of southeastern Mexico
.
Keywords: amphibians, area prioritization, conservation planning, MaxEnt, niche-based distribution models,
protected areas, reptiles, site selection
Modelado del Nicho Ecol´
ogico y Priorizaci´
on de Redes de ´
Areas de Conservaci´
on para la Herpetofauna Mexicana
Resumen: La informaci´on sobre la distribuci´on geogr´
afica de las especies y de las variables que determinan
esos patrones es una de las herramientas m´
as importantes de la biolog´ıa de la conservaci´
on. Utilizamos el
modelado de la m´
axima entrop´ıa del nicho para correr modelos de la distribuci´
on de 222 especies de
anfibios y 371 de reptiles (49% end´
emicas y 27% amenazadas) de las que contamos con 34,619 registros
geogr´
aficos individuales. La regi´
on de planificaci´
on est´
a en el sureste de M´
exico, comprende 20% de la
superficie del pa´ıs, incluye 80% de la herpetofauna del pa´ıs, y carece de un sistema de ´
areas protegidas
adecuado. Utilizamos datos probabil´ısticos para construir modelos de distribuci´
on de la herpetofauna para
utilizarlos en la priorizaci´
on de ´
areas de conservaci´
on para tres grupos focales (todas las especies, especies
end´
emicas y especies amenazadas). La precisi´
on de los modelos de distribuci´
on de especies fue mejor para
especies end´
emicas y amenazadas que para todas las especies. Cuarenta y siete porciento de la regi´
on ha sido
deforestada y se requieren ´
areas de conservaci´
on adicionales con 13.7% a 88.6% de m´
as vegetaci´
on nativa
(76% a 96% de las ´
areas est´
an afuera del actual sistema de ´
areas protegidas). Hubo traslape en 26 de las
principales ´
areas seleccionadas en la red de ´
areas de conservaci´
on priorizada para preservar a los grupos
§email nurbina@yahoo.com
Paper submitted May 18, 2009; revised manuscript accepted September 8, 2009.
1
Conservation Biology, Volume **, No. **, ***–***
C 2010 Society for Conservation Biology
DOI: 10.1111/j.1523-1739.2009.01432.x

2
Niche Models and Conservation-Area Prioritization
focales, y para los tres grupos focales la proporci´
on de tipos de vegetaci´
on requeridos para su conservaci´
on
fue constante: 30% bosques de pino-encino, 22% bosque tropical perennifolio, 17% bosque bajo deciduo y 8%
bosques montanos de niebla. El hecho de que grupos diferentes de especies requieren la misma proporci´

on de
tipos de h´
abitat sugiere que los bosques de pino y encino soportan la mayor proporci´
on de especies end´
emicas
y amenazadas y, por lo tanto, deben ser priorizados por encima de otros tipos de vegetaci´
on para su inclusi´
on
en las ´
areas protegidas del sureste de M´
exico.
Palabras Clave: anfibios, ´areas protegidas, MaxEnt, modelos de distribuci´on basados en nichos, planificaci´on
de la conservaci´
on, priorizaci´
on de ´
areas, reptiles, selecci´
on de sitios
Introduction
existing PAs, and only the periphery of their distribution
is protected (Urbina-Cardona & Loyola 2008).
Changes in land use resulting from anthropogenic activ-
We identified conservation areas for herpetofauna on
ities are the most critical cause of the biodiversity crisis.
the basis of geographic-niche distribution models for
The current wave of species extinction is being caused
southeastern Mexico, a region that contains 80% of Mex-
by irreversible transformation of the landscape and this is
ico’s herpetofauna. Mexico had one of the highest her-
making habitats unsuitable for many life forms (Margules
petofauna diversity in the world, and species are not
& Sarkar 2007). Fragmentation and loss of habitats threat-
distributed homogeneously in natural areas. Rather, to-
ens both amphibians and reptiles with extinction, and
pographic isolation determines their diversity patterns,
habitat specialists—such as endemic or rare species—are
which means it is necessary to include climatic factors in
more vulnerable to changes in habitat gradients (Urbina-
studies of species distribution across the country (Flores-
Cardona et al. 2006; Urbina-Cardona 2008).
Villela 1993). Niche-based distribution modeling is an in-
Even though their critical population decline has been
novative analytical approach that relates species locality
documented, the herpetofauna is not often taken into
records and environmental data (Elith et al. 2006). Be-
account as a conservation objective (Pawar et al. 2007;
cause climate and topography influence the geographical
Urbina-Cardona 2008). Unfortunately, many natural pro-
distribution of amphibians and reptiles (Duellman 1966),
tected areas (PAs) were established for reasons other than
we expected to obtain robust models for ecological niche
the protection of biodiversity and without clear conser-
modeling with environmental and topographic variables.
vation objectives or systematic, scientific prioritization
We sought to determine the relative contribution of
of the areas to include in a network of PAs. Generally,
certain environmental and topographic variables to the
these ad hoc reserves are situated on lands with poor eco-
geographic distribution of the herpetofauna and priori-
nomic and productive value (Margules & Sarkar 2007). In
tize CAN scenarios on the basis of 10% and 30% target
Mexico there are 521 established PAs that cover 9.5%
of representation of the total geographic range of all her-
of Mexico’s land mass (Bezaury-Creel et al. 2007). Over
petofauna, endemic species, and threatened species in
54% of Mexican PAs effectively prevent human-driven
southeastern Mexico.
change in land cover (Figueroa & S´
anchez-Cordero 2008),
The new method used identifies CAN for biodiversity
but most PAs are located on mountains tops with poor
through niche-based distribution models and fine-scale
soils, whereas other, more vulnerable habitats are un-
“self-learning tabu search” for conservation areas. This
represented (Cantu et al. 2004). Currently Mexican PAs
novel analytical approach restricts species niche models
cannot protect the country’s herpetofauna (Garc´ıa 2006)
to natural remnant vegetation through the use of national
because they protect only 31% of the amphibians (29%
land-use patterns to determine small-scale conservation
of endemics) and 76% of the reptiles (46% of endemics)
scenarios and includes a large number of species as con-
(Santos-Barrera et al. 2004).
servation objectives. This approach could be generalized
Mexican PAs are isolated from one another, and in
to other geographic regions or taxa for realistic conser-
many cases natural gradients for animal movement are
vation management plans.
blocked by anthropogenic barriers disrupting species’
natural cycles and interactions. The identification of con-
servation units that include and connect several ecosys-
tems is crucial to keeping biological processes and
Methods
ecosystem services operating on broad spatial scales
Study Area
(Margules & Sarkar 2007). One of the most important
priorities for amphibian conservation is reinforcement of
The study region is composed of the Mexican states
PA management and expansion of conservation-area net-
of Oaxaca, Chiapas, Veracruz, Guerrero, Puebla, and
works (CAN) to include the entire distribution range of
Michoac´
an and covers 396,311 km2 (between lat 14◦
threatened species (Urbina-Cardona 2008). Currently, the
33 N and lat 32◦ 43 N and from long 86◦46 W to long
core distribution of endangered hylid species lies outside
117◦19 W). This region has 188 PAs (31% of the region is
Conservation Biology
Volume **, No. **, 2010

Urbina-Cardona & Flores-Villela
3
Table 1. Number of herpetofauna species and protected areas (PAs) in the study region in southeastern Mexico.
Number
PA area
Percentage of state
Number of
State
Area (ha)
of PAs
(ha)
protected
species
Oaxaca
9,314,700
10
356,696
3.83
425
Chiapas
7,362,800
103
1,129,987
15.35
369
Veracruz
7,200,500
24
253,546
3.52
357
Guerrero
6,479,100
5
5,852
0.09
270
Puebla
3,415,500
6
234,904
6.88
246
Michoac´
an
5,858,500
40
79,898
1.36
224
Total
39,631,100
188
2,060,883
31

Values obtained from Bezaury-Creel et al. (2007).
protected; Table 1) covering 20,608.83 km2. We selected
projects that contained unpublished geographic records
this area for study because it has the greatest number of
pertinent to our study (see Acknowledgments).
species of herpetofauna (Table 1), has the highest annual
deforestation rates in the country (Cantu et al. 2004), is
Ecological Niche Modeling
geographically interconnected, thus facilitating the pri-
oritization of CAN for herpetofauna on the basis of areas
MaxEnt ecological niche modeling (Phillips et al. 2006,
with high habitat quality (primary natural vegetation) and
2009) uses known occurrences and pseudo-absence data,
complementary areas that could be restored (secondary
resampled from the set of pixels where the species is
vegetation), and has the highest number and best quality
not known to occur, to make inferences about proba-
of historical, geographic species records (Ochoa-Ochoa
bility of distribution of a species on the basis of prob-
& Flores-Villela 2006).
ability of distribution of maximum entropy, association
between species, and environmental variables in a geo-
Herpetofauna
graphic space. The resulting model represents the rela-
tive probability of the species’ distribution over all grid
For herpetofauna, Mexico is one of the most biodiverse
cells in the defined geographic space where a high prob-
countries in world with 361 species of amphibians and
ability indicates that the space is predicted to have suit-
804 species of reptiles (Flores-Villela & Canseco-M´
arquez
able environmental conditions for this species (Elith et
2004); however, it is in second place for the number of
al. 2006).
extinct and threatened amphibians and reptiles (IUCN
Nineteen environmental variables were obtained from
2009). One of the most important components of Mexi-
the WorldClim database (http://www.worldclim.org) in-
can biodiversity is the high species turnover (beta diver-
terpolated from data sets of global climate (Hijmans et
sity) of endemic species: 65% of amphibians and 57% of
al. 2005). We also used a spatial layer of elevation from
reptiles in Mexico are endemics. About 70% of endemic
the U.S. Geological Survey’s Hydro-1K (http://edc.usgs.
amphibians and 53% of endemic reptiles in Mexico have
gov/products/elevation/gtopo30/gtopo30.html).
Slope
restricted distributions (Flores-Villela 1993). The study
and topoindex were calculated on the basis of elevation
region has the highest endemic herpetofauna diversity
in the Spatial Analyst extension in ArcView 3.3 (ESRI
in the country (Ochoa-Ochoa & Flores-Villela 2006) and
2002). We resampled 22 layers of environmental and
represents 50% of the physiographic provinces where
topographic variables at a resolution of 0.01◦, or ap-
Mexican herpetofauna occurs (Flores-Villela 1993).
proximately 1 km2 and clipped the layers to the study
region bounded by 14.5341036 to 22.4714165 N and
Species’ Distribution Database
−103.737823 to −90.3728172 W (Supporting Informa-
We initiated the study with a secondary data set of species
tion).
localities for 830 species of Mexican herpetofauna. This
We ran MaxEnt under the “auto-features” mode as sug-
data set was compiled from 26 research projects and
gested by Phillips and Dudik (2008). Our use of the de-
106 biological collections and was corrected by Ochoa-
fault settings was reasonable given that it was validated
Ochoa & Flores-Villela (2006), who standardized scien-
in studies with a wide range of species, environmen-
tific names, eliminated dubious locality records, and dis-
tal conditions, individual species records, and in cases
carded records for species with dubious synonymy.
with sample-selection bias (Phillips & Dudik 2008). By
To complement the data set, we conducted a search
default, MaxEnt chose uniformly and at random 10,000
in three academic search engines (BIOONE, ISI Web of
background samples of pseudo absences from the study
Knowledge, and REDALYC). We systematically searched
area and used them in place of absences during mod-
paper titles and abstracts to find additional geographic
eling to represent the environmental conditions in the
records in the literature. Finally, we compiled data from
region (Phillips et al. 2006, 2009). We configured the
theses, manuals, book chapters, and ongoing research
machine-learning algorithm to use 75% of the species
Conservation Biology
Volume **, No. **, 2010

4
Niche Models and Conservation-Area Prioritization
records for training the data set and 25% for testing the
CAN because they are presently unsuitable for survival of
model. We determined the heuristic estimate of the rela-
most species.
tive contribution of each variable to species’ distribution.
The conservation scenarios were evaluated across
We selected the logistic output format because it is robust
153,229 cells of the planning region (i.e., 287,987 cells
when prevalence is unknown and easier to interpret as
of the study area minus the 134,758 anthropogenic cells
the estimated probability of a species’ presence given the
excluded from the analysis) for three target groups: all
constraints imposed by environmental variables (Phillips
herpetofauna (537 species) and endemic (276 species)
& Dudik 2008). In this case, grid cells with a small logistic
and threatened species (162 species). For each of the
value were predicted to be unsuitable or only marginally
three target groups, we evaluated species’ representa-
suitable for the species under study given their assumed
tion values of 10% and 30%, expressed as percentage of
ecological niche.
the total representation obtained from the probability of
We used MaxEnt software instead of other presence-
occurrence from the MaxEnt models and resulting in a
only methods because its algorithm constrains predicted
total of six conservation scenarios.
species ranges and thus reduces and avoids commission
ConsNet software uses a metaheuristic algorithm
errors that could lead to erroneous conservation deci-
called self-learning tabu search that uses memory to avoid
sions (Urbina-Cardona & Loyola 2008). We produced a
revisiting solutions that were discovered in previous iter-
geographic distribution of the ecological-niche model for
ations of the algorithm (Ciarleglio et al. 2008). It supports
each species with MaxEnt software (Phillips et al. 2006)
objectives based on rules and a dynamic neighborhood
in order to include each species as conservation objec-
selection that controls possible movements during the
tives in the CAN scenarios.
search for solutions, and intelligently arranges the struc-
We chose the models to be included in the CAN prior-
ture of the spatial problems (Ciarleglio et al. 2009). Using
itization on the basis of three criteria: higher area-under-
the species probability distribution for each cell in a geo-
the-curve (AUC) test values and lower standard deviation;
graphic grid, ConsNet makes a binary decision (to select
higher test-data curve (in the receiver–operating charac-
or not a cell to be put under a conservation plan) and or-
teristic [ROC] sensitivity-specificity plot [Phillips et al.
ders each cell hierarchically on the basis of its biodiversity
2006]) than the random-prediction curve; lower p val-
value. We built an objective that reduced the number of
ues (<0.05) in the threshold test; and expert opinions
selected cells, maximized CAN compactness (called min
of herpetologists to validate the distribution model for
cells and shape intransitive shape objective [ITS]), and
each species across the planning region and correct for
met one of the two required targets of representation
possible overprediction by the models. The models that
(i.e., 10% or 30%). For the six conservation-network sce-
did not meet these criteria were eliminated from the pro-
narios, the MDS2 adjacency algorithm gave the best initial
cess of area prioritization. We used test AUC values to
solutions on the basis of the fewest cells and lowest shape
estimate of the standard deviation of the AUC so that
value. The MDS2 adjacency algorithm first selected cells
standard deviation values could be used as a measure
that contained the species more distant from the repre-
of the uncertainty of the AUC parameter (DeLong et al.
sentation target, and it reduced the representation deficit
1988). Thus, if standard deviation is low, then one can
in the targets (“most deficient surrogates” [Ciarleglio et al.
have greater confidence in the estimate of the AUC. We
2008]) and attempted to select cells that were contigu-
compared model performance between target groups by
ous with other previously selected cells. In this sense,
comparing the AUC values with Student’s t tests.
the MDS2 algorithm uses rules similar to the concepts of
rarity (i.e., the presence of taxa that are geographically
restricted, less abundant, or have niche specificity) and
Prioritization of CAN
complementarity (i.e., determining whether a new site
We used ConsNet software (Ciarleglio et al. 2008, 2009)
maximizes the representation of taxa by measuring what
to design the CAN scenarios for herpetofauna. In Con-
that new site adds to the biodiversity represented by a
sNet, the probability of a species’ presence in each cell
previously existing set of sites [Margules & Sarkar 2007]).
is obtained from the MaxEnt models and total represen-
On the basis of the initial solution generated by the
tation for each species is the sum of all the probabilities.
MDS2 adjacency algorithm and using the same objec-
The polygons of the PAs (1:250 000) were obtained
tive of min cells and ITS over the course of 306,500
from Bezaury-Creel et al. (2007) and were included in
iterations, we refined the initial solution with the “tabu
the ConsNet prioritization analysis as cells (n = 5217)
search” for the six conservation scenarios on the basis
from which site selection was initiated. Anthropogenic
of a strategic selection of “escape with spatial neigh-
areas (human settlements, forest plantations, urban, agri-
borhood.” This allowed a compact CAN to be created
cultural, pastures) were identified with the third series
when adding or deleting cells in an effort to attain a
of the land-use and vegetation layer (1:250 000; INEGI
lower area to perimeter relationship (i.e., shape) with a
2005) and were set in ConsNet as cells (n = 134,758)
strategy called adaptative tabu reactor. We ran 306,500
to be excluded from the prioritization and design of the
iterations to perform a thorough search with twice as
Conservation Biology
Volume **, No. **, 2010

Urbina-Cardona & Flores-Villela
5
many iterations as the number of cells in the network
cipitation of driest month (34.6%), temperature season-
(n = 153,229).
ality (34.2%), mean diurnal range (31.1%), and precipita-
In the final step of the CAN optimization, we began
tion seasonality (30%; Supporting Information).
from the best solution that was currently available for
Endemic amphibians and reptiles had a higher AUC
the previous specific objective function and used the
value (mean [SD] 0.97 [0.083]) than all species taken to-
same objective (min cells and shape ITS) to refine prior-
gether (mean [SD] 0.95 [0.095]) (t = 3.6; p ≤ 0.001).
itization over 153,229 iterations and used a basic strate-
There was no difference in AUC values of endemic
gic selection of “large neighborhood only” (i.e., a larger
species (t = −1.91; p = 0.055) compared with the values
neighborhood is used to thoroughly explore the space
for threatened species (mean [SD] 0.98 [0.071]; Support-
around the current solution). This search is useful for re-
ing Information). The AUC value for threatened species
fining best solutions because it assesses a large number of
was higher than that of all species (t = 4.7; p ≤ 0.001).
possible moves at each step and can make improvements
that might otherwise have been missed (Ciarleglio et al.
Prioritization of Conservation Area Networks
2008, 2009).
The results of the six conservation scenarios solved by
To protect herpetofauna in the planning region, it is
ConsNet were transformed into polygons, and we calcu-
necessary to preserve between 13.8% (for 10% target
lated the area and perimeter of the polygons with the
of threatened species) and 88.6% (for 30% target of all
Projector extension in ArcView (version 3.3; ESRI 2002)
herpetofauna) of the total area under native vegetation
with an equal-area cylindrical projection.
(Table 2). The CAN generated for the six conservation
scenarios required 76% to 96% more area than is currently
covered by the national system of natural PAs (Table 2).
Results
For the target of 10% species representation, CAN sce-
narios showed 74.7% overlap of endemic species CAN
with that of all herpetofauna, 62.6% overlap of endemic
Ecological niche was modeled for 222 amphibians
species CAN with that of threatened species, and 58.8%
(11,170 records, mean [SD] 50.3 [88.8]) and 371 reptiles
overlap of herpetofauna CAN with that of threatened
(23,449 records, mean [SD] 63.2 [97.06]) and was rep-
species (Fig. 1a–f). For the target of 30% representation,
resented by 34,619 independent records (range 4–848,
CAN scenarios showed 89.6% overlap of all herpetofauna
median 27). We identified 136 species that were both
CAN with that of endemic species, 82% overlap of en-
endemic and threatened with extinction to some degree
demic species CAN with that of threatened species, and
(Supporting Information).
71.2% overlap of all herpetofauna CAN with that of threat-
ened species (Fig. 1a–f).
Main Variables in the Species’ Distribution Models
The priority areas for the preservation of conservation
The most important variables contributing to more than
scenarios with the target of 10% of the species’ distri-
30% of the 223 amphibian species distribution models
butions included 32 native vegetation types (including
were temperature seasonality (40.3%), precipitation of
preserved primary natural vegetation and secondary veg-
driest month (39%), precipitation seasonality (34.5%),
etation). For conservation of the three target groups at
and elevation (34.1%, Supporting Information). In the
the 10% level, it was necessary to preserve almost the
case of reptiles, the most important variables contributing
same proportion of habitat types: pine-oak forest (in-
to more than 30% of the 370 distribution models were el-
cludes pine, oak, pine-oak, and oak-pine forests; range
evation (36.2%), annual temperature range (35.1%), pre-
of values for three scenarios: 31.2–29.7%), tropical rain
Table 2. Solutions for six scenarios of conservation-area networks for the herpetofauna (all, endemic, and threatened species) of southeastern
Mexico.

Conservation scenarios / representation targets (10% or 30%)
all
endemic
threatened
Planning
region
10
30
10
30
10
30
No. selected cellsa
153,229
27,737
148,961
27,362
134,617
24,613
107,381
No. clustersa
5,594
3,488
5,815
2178
6,899
2,609
6,071
Perimeter (km)a
127,732
43,752
113,651
34,871
140,712
35,597
113,651
Area (km2)b
180,605
27,533
160,120
28,251
143,514
24,901
114,140
Area out of PAs (%)b
88.6
78.91
96.00
79.50
95.60
76.76
94.55
Area of native vegetation types (%)b
100
15.24
88.65
15.64
79.46
13.78
63.20
aValues obtained from the ConsNet solution.
bValues calculated from the refined and projected solutions (see Methods) (PA, protected area).
Conservation Biology
Volume **, No. **, 2010

6
Niche Models and Conservation-Area Prioritization
Figure 1. Prioritization of the conservation-area network along southeastern Mexico for protection of (a) all herp-
etofauna species, 10% target (i.e., species representation; 10% coverage of the total geographic range), (b) all
herpetofauna species, 30% target, (c) endemic species of herpetofauna, 10% target, (d) endemic species of herp-
etofauna, 30% target, (e) threatened species of herpetofauna, 10% target, and (f) threatened species of
herpetofauna, 30% target. Site selection was based on ConsNet analyses, including protected areas (PAs) and
excluding anthropogenic vegetation types. Current Mexican PA networks are in gray, and the needed additional
conservation-area networks are in black. The political limits of the planning region are shown in the inset of (a).
See Table 2 and Supporting Information for a numerical summary of these results. Shape files can be obtained on
request from J.N.U.-C.

forest (21.5–23.3%), tropical dry forest (15.8%–18.8%),
and endemic species (Table 4). Almost two-thirds (19)
and cloud forest (7.4–9.6%). The remaining area
of the 26 main areas were located in the highlands or
(19.8–23.1%) was covered by the other 28 vegetation
mountain regions. The other one-third (seven) were in
types (Table 3).
the lowlands. Twelve of the selected areas fell within the
We identified 26 regions that were constant and sys-
limits of at least one of the existing PAs (Table 4), but
tematically prioritized for the conservation scenarios
only six target areas included a significant proportion of
with 10% representation of all species and threatened
a PA (areas 3, 18, 19, 21, 25, and 26 in Table 4). The other
Conservation Biology
Volume **, No. **, 2010

Urbina-Cardona & Flores-Villela
7
Table 3. Area (km2) of the vegetation types prioritized for the conservation scenarios for the herpetofauna (all, endemic, and threatened species)
of southeastern Mexico.

Representation target
10%
30%
Planning
Vegetation type
region
all
endemic
threatened
all
endemic
threatened
Natural
pine forest
16, 952.8
2, 623.0
2, 864.5
2, 343.7
14, 900.5
13, 315.0
9, 225.1
oak forest
20, 723.4
2, 327.5
2, 376.3
1, 673.6
16, 532.8
14, 326.2
11, 015.9
oak-pine forest
8, 793.5
1, 030.9
1, 009.8
683.9
7, 386.3
5, 992.9
3, 945.8
pine-oak forest
36, 092.6
3, 244.8
3, 437.8
3, 057.7
30, 833.2
27, 301.8
18, 367.7
fir forest
373.6
80.6
58.3
76.3
329.3
313.1
327.4
drooping juniper (tascate) forest
802.5
123.9
154.3
123.3
666.9
656.7
474.6
mountain cloud forest
15, 578.7
2, 206.1
2, 084.8
2, 397.9
13, 692.8
11, 661.3
11, 155.7
chaparral
803.2
225.0
201.8
216.1
693.3
691.8
693.3
mangrove
1, 415.6
365.1
318.1
378.8
454.8
454.7
454.7
crasicaule scrub
1, 096.9
311.2
321.5
307.8
606.0
606.0
606.0
desert scrub
1, 423.3
335.0
311.2
424.2
1, 019.6
1, 002.5
1, 019.6
submontane scrub
1.4
0
0.7
0.7
0.9
0.9
0.9
mesquite
38.4
1.3
0.8
0.9
17.2
17.2
17.2
tropical cedar forest
13.4
0
0
0
9.4
9.3
9.4
gallery forest
32.3
9.7
7.1
5.8
19.3
15.8
11.5
halophilic vegetation
454.1
57.2
31.6
76.9
175.1
175.1
175.1
natural grassland
40.3
1.4
0
0.5
33.9
33.9
33.9
swamp
399.8
238.94
124.4
98.3
315.3
315.3
315.3
zacatonal grassland
75.6
25.9
38.1
29.5
70.0
69.9
70.0
savannah
702.4
220.8
58.9
154.3
601.4
600.9
568.5
tropical evergreen rainforest
31, 815.8
5, 931.0
6, 591.6
5, 552.1
29, 054.7
27, 424.2
28, 440.5
lower montane rainforest
549.2
318.7
319.3
347.6
375.3
375.3
375.3
(selva alta subperennifolia)
tropical deciduous forest
54, 782.2
4, 867.7
5, 320.6
3, 933.5
32, 710.4
29, 734.0
20, 405.7
arid tropical scrub
404.2
41.9
28.0
154.4
221.4
221.1
216.0
(selva baja espinosa caducifolia)
arid tropical scrub
36.7
25.7
24.3
25.2
31.0
31.0
31.0
(selva baja espinosa
subperennifolia)
tropical deciduous forest
78.6
17.7
2.7
56.1
70.7
70.7
70.7
(selva baja perennifolia)
tropical deciduous forest
28.4
0
0
0
25.3
25.3
2.5
(selva baja subcaducifolia)
gallery rainforest
25.6
0
0
0
3.4
3.4
3.4
tropical semideciduous forest
1921.3
511.6
442.9
702.6
1, 035.4
980.9
852.2
(selva mediana caducifolia)
tropical semideciduous forest
3.5
1.1
0
0
2.2
2.2
2.2
(selva mediana perennifolia)
tropical semideciduous forest
7502.6
609.2
631.6
401.8
3, 424.2
2, 908.8
1, 565.1
(selva mediana subcaducifolia)
tropical semideciduous forest
5193.7
1, 128.4
981.8
1, 132.2
3, 914.2
3, 282.8
2, 793.6
(selva mediana subperennifolia)
cattalis fields
1350.0
492.0
386.5
457.7
697.7
697.7
697.7
dune grassland
246.9
18.2
17.9
19.2
38.9
38.9
38.9
gallery vegetation
25.1
3.0
2.8
5.1
8.1
8.1
8.1
halophilic vegetation
215.7
138.5
100.0
63.0
149.0
149.0
149.0
Anthropogenic
not applicable
158, 233.5
0
0
0
0
0
0
palm plantations
924.8
0
0
0
0
0
0
induced pasture
22, 700.2
0
0
0
0
0
0
introduced grass
1, 131.8
0
0
0
0
0
0
no vegetation
540.6
0
0
0
0
0
0
Conservation Biology
Volume **, No. **, 2010

8
Niche Models and Conservation-Area Prioritization
Table 4. Areas that deserve attention in solutions with targets of 10% representation for the conservation scenarios for the herpetofauna (all,
endemic, and threatened species) of southeastern Mexico (see Fig. 1a–f)
.∗
Area
Representation target
Michoac´
an
Guerrero
Oaxaca
Puebla
Veracruz
Chiapas
10% all spp.
1, 2, 3
4, 5, 6
7, 8, 9, 10, 11
12, 13, 14, 18
15, 16, 17, 18, 19, 20
21, 22, 23, 24, 25, 26
10% endemic spp.
1, 2, 3
4, 5, 6
7, 8, 9, 10, 11
12, 13, 14, 18
15, 16, 17, 18, 19, 20
21, 22, 23, 26
10% threatened spp.
1, 3
5, 6
8, 9, 10, 11
12, 18
16, 17, 18, 19, 20
21, 23, 24, 25, 26
Key: , natural protected areas whose limits fall within those of any of the selected areas; ∗∗, cover a significant portion of the target; , marginal
or cover a small area of the selected targets. Number key: 1, region between Penjamillo (NE)
, Tingambato (SE) , Paracho , Pajuacar´
an
(NW), and Periban (SW) in Michoac´
an (Pico de Tanc´ıtaro , Barranca del Cupatitzio ); 2, western Sierra de Coalcom´
an between Villa
Victoria and Aquila, Michoac´
an; 3, central Balsas Basin on the Temascaltepec River , near the Infiernillo Dam (south of G´
ambara , Zicuri´
an
, and La Huacana ), Michoac´
an (Zicuar´
an-Infiernillo ,∗∗); 4, Arcelia , Guerrero; 5, Teloloapan , Guerrero; 6, central Guerrero on Sierra
Madre del Sur between Tlacotepec, Yextla, Chichihualco, and Chilapa Guerrero (General Juan N Alvarez ); 7, Sierra de Tlaxiaco , South of
Huajuapan de Le´

on , Oaxaca (Boquer´
on de Tonal´
a ); 8, extension of Sierra de Tlaxiaco between Santa Mar´ıa Asunci´
on and Santa Mar´ıa
Zaniza , Oaxaca; 9, surroundings of the Presidente Miguel Alem´
an Dam , Oaxaca; 10, triangle between Pluma Hidalgo , San Miguel del
Puerto , and Sta. Mar´ıa Huatulco , Oaxaca (Huatulco ); 11, Los Chimalapas , Oaxaca; 12, Sierra Norte de Puebla (between Huauchinango ,
Tlatlauquitepec
, Zacapuaxtla , and Cuetzalan ), Puebla (Cuenca H. del R´ıo Necaxa ); 13, between Izucar de Matamoros and Huahuapan de
Le´

on , Oaxaca, between the drainage basins of the Atoyac and Acatl´
an rivers , Puebla; 14, Sierra del Monumento , Central Puebla; 15, Delta
of the Panuco River , Veracruz; 16, Tuxpan , Veracruz; 17, coastal central Veracruz along Highway 180 and towns of Emilio Carranza-Cardel
, Calipa , Juchique de Ferrer , and Las Hayas , Veracruz; 18, eastern Trans-Mexican Volcanic Belt between Cofre de Perote and Ca˜

on
del R´ıo Blanco , Puebla-Veracruz (Cofre de Perote mountain ,∗∗, Mount Orizaba ,∗∗ and R´ıo Blanco Canyon ,∗∗); 19, Sierra de Los Tuxtlas
, Veracruz; 20, Nanchital , Veracruz; 21, Agua Azul-Ocosingo region , Chiapas (Chan-Kin ,∗∗, Lacan-Tum ,∗∗); 22, Marqu´ez de Comillas ,
Chiapas; 23, Salto de Agua
, Tumbal´
a , Amatl´
an , Chiapas; 24, El Ocote, Chiapas (El Ocote ∗∗); 25, Sierra del Soconusco , Chiapas (El Triunfo
,∗∗); 26, Motozintla, Tacan´
a , Chiapas (Tacan´
a Volcano) ,∗∗.
six targets fell marginally within the limits of an existing
Ours is one of the few studies to use probability data on
PA or the target areas were significantly larger than the
species’ distribution to systematically plan the conserva-
existing PAs (areas 1, 6, 7, 10, 12, 24 in Table 4).
tion of herpetofauna on a fine scale, and this is important
because herpetofauna is the most threatened group of all
terrestrial vertebrates. The areas selected to preserve 30%
Discussion
of the geographic range of each species, represent 63%
to 88% of the areas with native vegetation and would re-
Ecological Niche Modeling and Representation Targets
quire implementing a CAN that covers 94% to 96% of the
areas outside of current PAs, an area of land that would
It is likely that the accuracy of niche modeling varies sys-
not be cost-effective to acquire. It would be impossible
tematically across biological groups (McPherson & Jetz
to protect these areas because of logistical, political, eco-
2007; Pawar et al. 2007). It has been suggested that more-
nomic, and financial restrictions. This reinforces the need
accurate niche models can be generated for species with
to implement conservation strategies not only within the
narrow ecological niches, such as microendemics, be-
PAs, but also by collaborating in private local initiatives
cause they have narrow distribution ranges (Tsoar et al.
(Ochoa-Ochoa et al. 2009b). Although it is possible that a
2007). We found that the AUC prediction value for dis-
target of 10% would not represent a sufficient portion of
tribution models for the endemic and threatened species
the geographic range for maintaining viable populations
of herpetofauna had higher values than the AUC values
of species, working with conservation scenarios with tar-
for all herpetofauna.
gets of 10% species representation allow a more feasible
We also found differences in the variables that in-
alternative for implementing CANs.
fluence species’ geographic distribution in this study
when we compared the relative contributions of the
Prioritization of CAN and Spatial Scale
most important environmental variables influencing 10
anuran species’ distribution in southeastern Mexico (Ap-
In the study region, there are 32 native primary and sec-
pendix S2) with those variables influencing the distri-
ondary vegetation types, and these cover 53% of the study
bution of the same species at the broader spatial scale
area. Nevertheless, only 11.4% of this native vegetation is
of Mesoamerica (Urbina-Cardona & Loyola 2008). Al-
currently under protection within a natural PA (Table 2).
though elevation, temperature seasonality, precipitation
Given that 47% of the original vegetation of the study re-
of driest month, and precipitation seasonality broadly
gion has been modified by anthropogenic activities, rep-
influenced amphibian and reptile distribution, for rep-
resenting the herpetofauna properly would require up
tiles temperature annual range and mean diurnal range
to 80% of their range to be covered by native vegetation
were important because these variables affect the thermal
(Appendix S2 and S3).
environment requirements for reptiles to efficiently ther-
With the increase in annual deforestation rate, it
moregulate.
has become increasingly difficult to design areas for
Conservation Biology
Volume **, No. **, 2010

Urbina-Cardona & Flores-Villela
9
conservation in environments with native vegetation. Be-
of these areas are outside current PA limits. A previous
cause of the need to preserve a significant portion of
analysis showed that centers of endemic species are lo-
biodiversity in natural areas, more human and financial
cated in our study area (Ochoa-Ochoa & Flores-Villela
resources are needed to accomplish this goal.
2006), but a comparison with our results showed that
When comparing our results with other analyses of
only 38.4% of our target areas are within the limits of
conservation-area prioritization done with other biodiver-
those centers of endemism. This discrepancy could be
sity surrogates and at different spatial scales, we found
due to the influence and importance of range size for
that CANs to preserve Mexican biodiversity and ecore-
species that are not endemic, and for threatened species
gions on a semicontinental scale (Sarkar et al. 2009) were
the discrepancy could be because they do not necessarily
coincidental with our solution in almost 30% of the 26
occupy areas of high endemism.
main selected areas in our study; CANs selected to con-
The majority of the areas selected in our study were
serve Mexican biodiversity from a study at a country level
located in mountainous regions in southeastern Mexico.
(Koleff et al. 2009) overlap with over 50% of our main
Despite the fact that many of the Mexican PAs are located
selected areas (Table 4); results of a regional study on
above 3000 m (Cantu et al. 2004), the mountainous re-
mammals in Mexico (Fuller et al. 2007) show very little
gions in southeastern Mexico are underrepresented in
coincidence with our results. It seems that the prioritized
the current PA system and represent areas of high diver-
scenarios in this study are not similar to those of Mexican
sity for many species of endemic salamanders and frogs.
mammals at finer scales, but the scenarios do agree more
Between 1993 and 2003, 55 species from Chiapas, Guer-
with those that include other species groups at broader
rero, and Oaxaca were described as new and another 23
spatial scales.
new species were discovered in Michoac´
an, Puebla, and
Biological processes differ from broad to fine scales,
Veracruz (Flores-Villela & Canseco-M´
arquez 2004). The
and conservation priorities and decision-making scenar-
survival of many of these species depends on the pro-
ios may also vary with scale for the same region and same
tection of montane areas with pine-oak forest and cloud
species group (Margules & Sarkar 2007). A recent GAP
forest and lowland areas with tropical rain and dry forest.
analysis of Mexico revealed that 74% of amphibians and
Establishing representative CANs where biodiversity
81.7% of reptiles are present in at least one natural PA, al-
can persist should be a policy goal for the government
though different algorithms (e.g., ResNet, CPlex, Marxan)
agencies responsible for conservation and natural re-
led to a different sites being selected (Ochoa-Ochoa et al.
source management, intergovernmental, and nongovern-
2009a). Comparing Ochoa-Ochoa et al.’s (2009a) results
mental organizations (Margules & Sarkar 2007). Conser-
in a broad-scale study with 10% representation with our
vation at the local level needs to be recognized as an
results at a finer scale, the area prioritized for all herpeto-
essential component in addressing the biodiversity cri-
fauna species overlaps only 25%. The area for endemic
sis (Ochoa-Ochoa et al. 2009b). Our minimum-area ap-
species overlaps 32%, and the area for threatened species
proach to prioritizing areas for conservation is based on
overlaps only 28%. These differences can be partially ac-
the use of rarity and complemetarity to obtain a quan-
counted for by the difference in scale, but occur mainly
titative solution to herpetofauna species representation,
because Ochoa-Ochoa et al. (2009a) prioritized conserva-
sets objectives at a fine spatial scale, and takes into ac-
tion areas throughout the country regardless of the type
count land-use patterns. This protocol could be included
of land use rather than forcing the algorithm to prioritize
in the selection of CANs to strengthen the practice of
exclusively natural vegetation, which masks the acute
conservation planning elsewhere in the world.
problem of fine-scale deforestation that is eliminating the
Our results show that the CANs needed to preserve
last relicts of native vegetation that support populations
Mexican herpetofauna should not be used to identify
of species with narrow distributional ranges.
priority areas for other groups of Mexican biota. One
We detected some agreement between the selected
group of species may be a good surrogate for species
sites with a 10% target in our study and those of other
in a region, but there is no guarantee that it will be a
fine-scale studies of herpetofauna hotspots along the Pa-
good predictor elsewhere, and the distribution of species
cific coast. For all herpetofauna species, we found some
among higher taxa can change from place to place (Gas-
priority areas in common with Garc´ıa (2006) in south-
ton 1996). To set real targets of representation for each
western Michoac´
an, central Oaxaca, and the interior val-
species and to ensure the persistence of biodiversity in
leys of Guerrero. For endemic species, there was some
selected areas, more work is needed to quantify viable
similarity between our areas and those of Garc´ıa (2006) in
population sizes, phylogenetic diversity, species home
southwestern Michoac´
an and the interior valleys of Guer-
ranges, and source–sink population structure to define
rero. For threatened species, only some of our selected
design criteria: size, shape, compactness, connectivity,
areas in central Oaxaca were similar to those in Garc´ıa
dispersion, width of buffers, spacing, replication, and
(2006).
alignment (Margules & Sarkar 2007). To fill knowledge
For the consistently prioritized regions in our study
gaps, better distributional data and models of biodiver-
(Table 4), the data are not very promising because most
sity for predicting biological and socioeconomic features
Conservation Biology
Volume **, No. **, 2010

10
Niche Models and Conservation-Area Prioritization
(e.g., poverty, natural disasters, and distance to roads) of
ful comments on an earlier version of this manuscript.
the region being analyzed are needed. Although current
L. Canseco-M´
arquez and C. Hern´
andez-Jim´
enez helped
data sets are far from ideal, they must be used to mitigate
with the taxonomic data cleaning of the herpetofauna
the loss of natural habitat in the face of inadequate pol-
geographic distribution data base. J. M. Garza-Castro,
icy and planning decisions that are being made every day
U. O. Garc´ıa-V´
azquez, L. Ochoa-Ochoa, E. Pineda, A. J.
(Margules & Sarkar 2007).
Gonz´
alez-Hern´
andez, O. Hern´
andez, F. H. Carmona, J. J.
Alvarado-D´ıaz, I. Suazo, J. Reyes-Velasco, C. Hern´
andez,
F. Mendoza-Quijano, M. R. Harfush-Melendez, M. R¨
oss,
Conclusions
N. R. V´
azquez-Cisneros, L. Oliver-L´
opez, M. L´
opez-Luna,
E. Beltr´
an-S´
anchez, I. L´
opez-L´
opez, G. Santos-Barrera pro-
Use of conservation planning tools can improve existing
vided valuable information from the literature and species
PAs by showing decision makers how PAs can be trans-
geographic records from their projects. This study repre-
formed into better and more efficient networks (Margules
sents part of the results from the postdoctoral research of
& Sarkar 2007). We identified conservation areas that are
J.N.U.-C. J.N.U.-C. thanks Direcci´
on General de Asuntos
not included in the current national system of PAs that en-
del Personal Acad´
emico (DGAPA) of the Universidad Na-
sure representation of 80% of the Mexican herpetofauna
cional Aut´
onoma de M´
exico (UNAM) for the award of a
and provide tools to guide delineation and refinement of
full postdoctoral fellowship. O.F.V. thanks the authorities
policy alternatives.
of the UNAM and the University of Texas, Arlington, for
There are more than 126 institutions working on
the facilities to complete this research while on sabbat-
biodiversity conservation in the planning region (http://
ical. Consejo Nacional de Ciencia y Tecnolog´ıa (CONA-
www.directorio-delaconservacion.org.mx/directorio/
CyT) and DGAPA-UNAM provided support for O.F.V. Part
instituciones.php [accessed April 2009]). If conservation
of the information used in this paper was financed by U.S.
NGOs purchase land, they need to know what the true
National Science Foundation grant DEB-0613802.
biodiversity priorities are because just purchasing the
cheapest land or whatever land becomes available will
not necessarily contribute to biodiversity conservation
Supporting Information
and may only increase the number of inefficient ad
hoc reserves (Margules & Sarkar 2007). Our planning
“Codes for 22 environmental variables layers used to
protocols are support tools that can help local experts
model species distribution” is available as part of the on-
make good policy decisions. Nevertheless, before
line article (Appendix S1). “Herpetofauna species distri-
implementation of a CAN, it is important to identify
bution in southeast Mexican states, number of localities,
previous conservation-area selection exercises that have
threat and endemic categories and relative contribution
been done in the region to redefine the ideal spatial
of the environmental variables to the distribution mod-
configuration of the network. Multicriteria analysis is
els” is available as part of the online article (Appendix
needed to satisfy divergent socioeconomic criteria and to
S2). “Values for the representation targets for herpeto-
identify natural hazards, vulnerability, and conservation
fauna species in the conservation scenarios in southeast-
goals of stakeholders during implementation at the
ern Mexico” is available as part of the online article (Ap-
local scale. Budgetary, ethical, and other sociopolitical
pendix S3).
constraints will determine whether prioritized sites can
represent and ensure persistence of biodiversity with
Literature Cited
minimum overlap with human activities (Sarkar et al.
2006). It is only by managing whole regions and natural
Bezaury-Creel, J. E., J. F. Torres, N. Moreno, and L. M. Ochoa-Ochoa.
gradients on different spatial scales and by taking into
2007. Base de datos geogr´
afica de ´
areas naturales protegidas es-
tatales, del Distrito Federal y Municipales de M´
exico para an´
alisis
account socioeconomic, ecosystem, and evolutionary
espaciales. The Nature Conservancy, Programa M´
exico, Pronatura,
processes and the mechanisms that support biodiversity
A.C., Comisi´
on Nacional para el Conocimiento y Uso de la Biodi-
that species will be protected from the threats they face
versidad (CONABIO), and Comisi´
on Nacional de ´
Areas Naturales
and be assured of survival for generations to come.
Protegidas (CONANP), Mexico.
Cantu, C., R. G. Wright, J. M. Scott, and E. Strand. 2004. Assessment
of current and proposed nature reserves of Mexico based on their
capacity to protect geophysical features and biodiversity. Biological
Acknowledgments
Conservation 115:411–417.
Ciarleglio, M., S. Sarkar, and J. W. Barnes. 2008. ConsNet manual. Ver-
M. Ciarleglio, T. Fuller, and S. Sarkar provided vital
sion 1.0. University of Texas, Austin. Available from http://uts.cc.
information and advice for the use of ConsNet soft-
utexas.edu/∼consbio/Cons/consnet_home.html (accessed April
2009).
ware. B. Delfosse assisted in editing the English ver-
Ciarleglio, J., W. Barnes, and M. S. Sarkar. 2009. ConsNet: new software
sion of this manuscript and offered valuable suggestions.
for the selection of conservation area networks with spatial and
M.C. Londo˜
no-Murcia and L. Ochoa-Ochoa made use-
multi-criteria analyses. Ecography 32:205–209.
Conservation Biology
Volume **, No. **, 2010

Download
Ecological-Niche Modeling and Prioritization of Conservation-Area Networks for Mexican Herpetofauna

 

 

Your download will begin in a moment.
If it doesn't, click here to try again.

Share Ecological-Niche Modeling and Prioritization of Conservation-Area Networks for Mexican Herpetofauna to:

Insert your wordpress URL:

example:

http://myblog.wordpress.com/
or
http://myblog.com/

Share Ecological-Niche Modeling and Prioritization of Conservation-Area Networks for Mexican Herpetofauna as:

From:

To:

Share Ecological-Niche Modeling and Prioritization of Conservation-Area Networks for Mexican Herpetofauna.

Enter two words as shown below. If you cannot read the words, click the refresh icon.

loading

Share Ecological-Niche Modeling and Prioritization of Conservation-Area Networks for Mexican Herpetofauna as:

Copy html code above and paste to your web page.

loading