Biological Conservation 144 (2011) 1451–1463
Contents lists available at ScienceDirect
Biological Conservation journal homepage: www.elsevier.com/locate/biocon
Local knowledge and species distribution models’ contribution towards mammalian conservation Hugo Fernando López-Arévalo a,b,⇑, Sonia Gallina b,1, Rosario Landgrave c, Enrique Martínez-Meyer d, Lyssette E. Muñoz-Villers e,2 a
Instituto de Ciencias Naturales, Universidad Nacional de Colombia, A.A. 7495, Bogotá D.C., Colombia Instituto de Ecología, INECOL, A.C. Red Biología y Conservación de Vertebrados, km 2.5 Carretera Antigua a Coatepec No. 351, Apartado Postal 91070, Xalapa, Veracruz, Mexico Instituto de Ecología, INECOL, A.C. Red Ecología Funcional, km 2.5 Carretera Antigua a Coatepec No. 351, Apartado Postal 91070, Xalapa, Veracruz, Mexico d Universidad Nacional Autónoma de México, Instituto de Biología, Laboratorio de Análisis Espaciales, Ciudad de México 04510, Mexico e Department of Forest Engineering, Resources and Management, Oregon State University, Corvallis, OR 97331-8615, USA b c
a r t i c l e
i n f o
Article history: Received 12 November 2009 Received in revised form 9 January 2011 Accepted 20 January 2011 Available online 22 February 2011 Keywords: Conservation method Distribution model Local knowledge Medium-sized mammal Mexico
a b s t r a c t Landscape-scale studies facilitate species diversity analysis according to environmental heterogeneity and human activity. This study was aimed at using local knowledge as a tool for testing predictive models’ validity for assessing the spatial distribution of medium-sized mammalian richness, identifying local patterns of species richness and evaluating local protected areas’ role in the conservation of mammals. Distribution maps were generated for historically recorded species using genetic algorithm for rule-set prediction (GARP). The landscape was reclassified as habitat, hospitable matrix and inhospitable matrix in the second scenario and a third scenario was generated limiting species distribution by using the home range. The local richness predicted by all scenarios varied from 1 to 32 species per cell while gamma diversity was 34. The 72 structured interviews led to recording 3–17 species (a total of 27). There have been no reports of nine wild species over the last 2 years. Currently protected areas cannot support viable populations of the species so recorded so shade coffee plantations must adopt conservation strategies. Historical inventories overestimate expected richness; however, combining GARP-generated models with the information obtained from local inhabitants and experts allows rapid regional evaluation of mediumsized mammalian richness and the identification of extinct species, declining populations and abundant species. Ó 2011 Elsevier Ltd. All rights reserved.
1. Introduction Landscape studies involving investigations into spatial patterns and ecological processes converge on broad spatiotemporal scales; they have been identified as a priority for both ecological research and their application to environmental problems (Turner and Gardner, 1991). Species diversity may be analyzed on the landscape scale as a function of (but not just) the physical and biological environment’s heterogeneity and also as the effect of human activities on species’ distribution and abundance (Franklin, 1993; Halffter, 1998).
⇑ Corresponding author. Address: Instituto de Ciencias Naturales, Universidad Nacional de Colombia, Carrera 30 No. 45-03 Bogotá D.C., A.A. 7495, Colombia. Tel.: +57 1 316 5000x11525, fax: +57 1 316 5000x11502. E-mail addresses: hfl
[email protected], hfl
[email protected] (H.F. López-Arévalo),
[email protected] (S. Gallina),
[email protected] (R. Landgrave),
[email protected] (E. Martínez-Meyer), Lyssette.
[email protected] (L.E. Muñoz-Villers). 1 Tel.: +52 228 842 1800x4110. 2 Tel.: +1 541 737 4952. 0006-3207/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.biocon.2011.01.014
Measuring the number of species in the landscape allows the effects of forest fragmentation on species’ permanence or extinction to be detected. Such measures also allow spatial patterns to be identified (Lomolino, 2001), such as those occurring along urban– rural gradients where richness is greater in rural areas (see review by McKinney (2002)). Identifying such patterns leads to forming conservation strategies on different geographic scales (Funk and Richardson, 2002). Simulation models have been useful in selecting biodiversity conservation-promoting action on different scales, thereby predicting species distribution, the effects of climate change and conflicts with human activity (Rodríguez et al., 2007). The copious amount of information available on biodiversity (Bisby, 2000; Edwards et al., 2000) and current digital data processing capacity offer several tools for modeling species distribution. The comparative analysis of different methods using the same data-set highlights how presence-only data are useful for modeling species distributions and demonstrate differences in predictive performance among modeling methods, despite substantial variation at regional and species levels (Elith et al., 2006). The difference between different models’ predictions may be explained
1452
H.F. López-Arévalo et al. / Biological Conservation 144 (2011) 1451–1463
by two modeling techniques’ characteristics: data input requirements (presence/absence vs. presence-only approaches) and the assumptions made by each algorithm when extrapolating beyond the range of data used for building a particular model (Pearson et al., 2006). Genetic algorithm for rule-set prediction (GARP) is a commonly used model for studying species distribution; it is a genetic algorithm that brings together several approaches for modeling artificial intelligence-based distribution. GARP searches for non-random correlations between species presence and environmental parameters using several types of rules (for a detailed explanation see Anderson et al., 2003). Combining GARP results with land use and plant cover data leads to closely approximating species’ real distribution (Sánchez-Cordero et al., 2005). GARP was found to be useful in predicting potential Insectivora, Chiroptera, Rodentia and Artiodactyla species’ distribution in the state of Oaxaca, Mexico; however, as expected, its accuracy depended on the number of available records (Illoldi-Rangel et al., 2004). It has been applied to studying the distribution and ecological relationships of other mammals based on information obtained from scientific collections’ records (Anderson et al., 2003; Anderson and Martinez-Meyer, 2004). GARP also provides better resolution when detecting richness patterns compared to species’ aggregation methods and the use of vegetation type as a descriptor of richness (Stockwell and Peterson, 2003). GARP presents the highest over-prediction (Elith and Graham, 2009) compared to other species distribution models, although comparison results can vary according to the available information, species type, the models being compared and an investigator’s interpretation (Stockwell and Peterson, 2003: Elith et al., 2006; Phillips, 2008). Even though discussion about the parameters for selecting the best model still continue (Segurado and Araújo, 2004; Elith and Graham, 2009), field predictions must be verified. Using local inhabitants’ ecological knowledge regarding their land can be an alternative for verifying a model’s predictions because, in many cases, this source provides more and better information than that obtained by western science (Huntington, 2000). This is more evident in fauna conservation planning and management-related research and projects (Becker and Ghimire 2003; Sheil et al., 2006; Anadón et al., 2009). Interviews with local people have been used to complete local fauna lists and establish regional distribution patterns (Hall and Dalquest, 1963), identify species used for food, medicine and as pets, for detecting variations in their populations and making management proposals (Anadón et al., 2009). Studying Mexican mammals has led to mapping their general distribution patterns (Rodríguez et al., 2003; Escalante et al., 2004) and proposing strategies for their conservation, priority being given to those areas having a greater concentration of endangered and endemic species and those having limited distribution (Ceballos et al., 1998). Records are now thus available for most mammalian species in scientific collections in Mexico (EspinozaMedinilla et al., 2006) and abroad (López-Wilches, 2003). Medium-sized mammals (i.e. those weighing > 200 g) are an ecologically diverse group and most species can be easily identified. Their study requires a variety of methodologies and great efforts are needed for obtaining just a few data which are usually difficult to test statistically. Their natural history is known from studies about their geographical distribution (Ceballos and Oliva, 2005) and these can prove useful for proposing local conservation projects. Veracruz is one of the states in Mexico having the highest mammalian diversity (Gaona et al., 2003; Ceballos and Oliva, 2005). Hall and Dalquest (1963) listed the mammals of Veracruz with information about their natural history; later, more general studies in Mexico updated the number of species in the state (Ramírez-Pulido et al., 1996). Research into the effect of vegetation cover change on this group of mammals has indicated that arboreal species and those
depending on the forest are most affected by fragmentation (Crooks, 2002; Da Silva and Mendes Pontes, 2008; Laurance et al., 2008) while other species’ populations might increase in heterogeneous environments by using forest edges, crop fields and the suburban environment (Crooks and Soulé, 1999; Crooks, 2002; McKinney, 2002; Daily et al., 2003). Some investigations have found that the shade provided by coffee plantations maintains high mammalian diversity and that of other vertebrates in other Mexican regions and the tropics (Gallina et al., 1996, 2008; Greenberg et al., 1997; Moguel and Toledo, 1999; Cruz-Lara et al., 2004). Mexican coffee plantation distribution overlaps the tropical montane cloud forest (TMCF3), an ecological area covering 1% of the country. Although it still occupies most of its original distribution (around 8000 km2), its effective area has been reduced to small fragments incapable of long-term support for this habitat’s typical flora and fauna (Challenger, 1998). TMCF in Veracruz has the highest deforestation rate for this type of tropical forest in the world (Aldrich et al., 2000) and it has been estimated that it contains 10–12% of Mexican plant and animal species (Ramamoorthy et al., 1993; Rzedowski, 1996). TMCF remnants in central Veracruz are immersed in changing matrices which can attenuate the effect of forest fragmentation; matrices include a mixture of shaded coffee crops, disturbed forest and secondary vegetation (Gallina et al., 1996; Williams-Linera et al., 2002) or irreversibly deteriorated vegetation resulting from construction projects and urban sprawl. Accelerated cloud forest destruction is a recent phenomenon in this area and the area west of Xalapa has lost 90% of its natural forests since 1993 (WilliamsLinera et al., 2002). This study’s main goal was to evaluate the effect of current landscape composition and spatial configuration on medium-sized mammals’ species richness. The following three approaches were used. Medium-sized mammalian species distribution in the upper Antigua river basin in central Veracruz was modeled to obtain species richness models which were tested for their predictive value by using local knowledge. Possible species richness patterns related to altitude and human presence were identified. Local protected areas’ current role regarding medium-sized mammals in the region was evaluated. 2. Materials and methods 2.1. Study area The study was carried out between February 2007 and July 2008 in the upper Antigua river basin (UARB)4 in Veracruz, Mexico (1325 km2); it is located between 19°100 -19°340 N and 96°500 97°160 W (Fig. 1), lying at over 3600 m (600–4200 masl). The upper part of the basin (2500–4200 masl) is highly dissected, having 20– 45° slopes while slopes range from 3° to 10° in the lower part. The climate is humid temperate in almost all the basin (Muñoz-Villers and López Blanco, 2007). According to Rzedowski (1978, 1990), the basin’s wooded areas are pine, oak-pine, cloud and deciduous forest. Cloud forest was the dominant cover from 1990 to 2003 for 26.5% of the area; 21,100 ha of cloud forest were transformed into pastures and crop fields during a 13-year period (Muñoz-Villers and López Blanco, 2007). 2.2. Species list The following literature was consulted for obtaining a list of medium-sized mammals (those weighing > 200 g) for the study area: Hall and Dalquest (1963), Gallina et al. (1996), González3 4
Tropical montane cloud forest. Upper Antigua river basin.
H.F. López-Arévalo et al. / Biological Conservation 144 (2011) 1451–1463
1453
Fig. 1. Location of the upper Antigua river basin in the state of Veracruz, Mexico and the model test sites evaluated. 1. Yerbabuena, 2. Xico, 3. Cosautlan, 4. Ixhuacan de los Reyes. Protected areas: (A) Mount Orizaba (Pico de Orizaba), (B) the Cofre de Perote Mountain, (C) San Juan del Monte, (D) Texolo Waterfalls, (E) Cerro de las culebras, (F) Fco. Javier Clavijero (including the Botanical Garden) (G) Barragán, (H) El Tejar Garnica, (I) Macuiltepetl, (J) Molino de San Roque, (K) Cerro de la Galaxia, (L) La Martinica, (M) El Castillo.
Romero and López-González (1993), Gaona et al. (2003) and Gallina et al. (2008). Mexican mammal collection databases from the USA and Canada (López-Wilches, 2003) were also used. Unpublished literature (theses, technical reports) containing mammalian inventories within or near the study area were also checked. The geographical information for each species recorded in Mexican specimen collections was obtained from CONABIO (2007).
2.3. Biophysical information Cartographic information regarding Mexican climate and topography was used for generating the distribution models. In total, 19 climate data layers were used from the WorldClim database 1,3 (http://www.worldclim.org; 1 km2/resolution). A description of this information can be found in Hijmans et al. (2005). The topographical layers were obtained from a digital elevation model of
1454
H.F. López-Arévalo et al. / Biological Conservation 144 (2011) 1451–1463
North America, obtained from the United States Geological Survey (30 s resolution) (USGS, 2007). The digital cartographic information (20 m grid) regarding the study area’s land cover types was obtained from Muñoz-Villers and López Blanco (2007).
2.4. Designing the model 2.4.1. Modeling the niche with GARP Distribution maps for each species were obtained by modeling the niche using a desktop genetic algorithm for rule-set prediction (GARP) (http://www.lifemapper.org/desktopgarp/). One hundred distribution models were generated for each species; 60% of the records were used as training data with extrinsic omission for species having more than 30 unique localities. The models were generated using 100% of the data with intrinsic omission for species having less than 30 localities. The ten best models having the least omission errors were selected (Anderson et al., 2003) and their geographic predictions (criteria – at least 90% of the records for each species) were added to obtain a final potential distribution map for each species in Mexico. The study area was delimited using Arcview 3.2 (ESRI, 1996) on the country-scale maps so generated. The maps were then combined to obtain expected species richness and their composition values for different areas of the basin; the results will hereinafter be referred to as the GARP scenario (GARPS).
2.4.2. Modeling using vegetation cover and land use type Vegetation cover and land use maps were reclassified, considering published information about each species’ natural history (Table 1), vegetation type where it had been reported (Ceballos and Oliva, 2005) and local experts’ advice for determining whether there were any relationship between theoretical species richness distribution and current land cover in the study area. Such reclassification of cover type was based on Tischendorf et al.’s proposal (2003) in which habitat is defined as being the cover type where species could establish viable populations, hospitable matrix is the cover type which could facilitate movement between areas having desirable habitat and inhospitable matrix is where species’ presence is low or null. A map was obtained for each species by using this new classification (Table 1). The habitat distribution maps for each species and the distribution map obtained for each species with GARP were spatially added with Arcview 3.2 (ESRI, 1996) to adjust the final fundamental niche distribution model (generated by GARP) to the species’ real niche (Sánchez-Cordero et al., 2005; Soberón and Peterson, 2005; Peterson et al., 2006). The niche modeled with GARP and the area overlapping between the niche and the existence of habitat was taken for finally defining species’ presence (i.e. species absence was assumed in areas where species’ presence was not predicted by niche variables but there was suitable habitat). A new species richness map called HABITATS was obtained for the basin by adding each of these maps.
2.5. Testing the model Four circular model test sites were selected (Yerbabuena, Xico, Ixhuacán and Cosautlán) for evaluating the scenarios generated with the models; they covered around 50 km2 corresponding to 15% of the entire area (Fig. 1). The selected sites did not overlap, were accessible most of the time and contained most of the land use types common to the region. Field surveys supported by a photographic guide of the mammals were used with local adult inhabitants, with more than 5 years living in the area. We focus on the presence or absence of species, their newest record and sites were observed. A 1:50,000 map and a GPS (Garmin Etrex) were used for locating the sites indicated by interviewees (<30 m error). Researchers who worked in model test sites were also interviewed. Field interviews were complemented by looking for pelts, other hunting trophies and pets. Tracks, scats and observations in the wild were recorded during trekking while direct observations of run-over animals were recorded on the highways. Ten camera traps were set up at each test site in different parts of the forest and their adjacent vegetation cover during three periods of five consecutive nights. Species accumulation curves (using the Chao 1 non-parametric estimator) were generated with EstimateS ver. 8.0 software (Colwell, 2006) based on the information obtained during the interviews. Similarity between model test sites and field observations was compared for each scenario using Jaccard’s index (Magurran, 1988). The difference between the expected richness for each scenario and that observed in the field via interviews was evaluated using Dunnett’s a posteriori comparison of means (Zar, 1996) where a zero value indicates equality. Different landscape characteristics were calculated for each site using the Patch Analyst extension for Arcview 3.2 (MacGarigal and Marks, 1994). The impact of road density was calculated on a radius of 50 km per cell (Table 2). 2.6. Identifying species richness patterns 100 samples were randomly re-sampled with 10,000 repetitions from all cells in each scenario to determine the existence of an altitude-associated species richness pattern. A correlation test (Zar, 1996) was also made for species richness obtained in the field, human population density and road density to identify a possible urban–rural gradient. 2.7. The role of the protected areas The federal, state-owned and municipal-protected areas in the basin were identified; these were superimposed on the HRANGES scenario and the richness verification points obtained in the field. The potential for conserving local mammalian biodiversity and possible management action according to the present species’ biological characteristics, location and size was discussed. 3. Results
2.4.3. Modeling based on the home range A third scenario was generated by selecting contiguous habitat areas which were equal to or larger than twice the smallest home range recorded for each species. A value of 1 was assumed at these sites for the probability of species presence (HRANGES). Species diversity and composition were calculated inside the studied area based on the resulting distribution model. The size of the distribution areas obtained with the GARPS and HRANGES scenarios were compared for each species.
3.1. Species richness Thirty-four medium-sized mammal species had previously been recorded for the basin: three marsupials, two edentates, one primate, one lagomorph, seven rodents, three artiodactyls and 17 carnivores (Table 1), giving a total of 159 records obtained from unpublished and published information for the UARB for 16 different localities (Ramírez-Pulido et al., 1996; González-Romero and López-González, 1993; Gallina et al., 1996; García, 2007;
1455
H.F. López-Arévalo et al. / Biological Conservation 144 (2011) 1451–1463
Table 1 Medium-sized mammal that are probably present in the studied area and the number of records used in the predictive distribution model. Degree of dependence on the forest: High, H, Moderate, M, Low, L. Locomotion: Arboreal, Ar, Scansorial, S, Terrestrial, T, Aquatic, Aq, Fossorial, F. Rain-fed agriculture, 1, Coffee plantations, 2, Agroforestry systems, 3, Sugar cane, 4, Coniferous forest, 5, Pine-oak forest, 6, Tropical montane cloud forest, 7, Water bodies, 8, Alpine grassland, 9, Cultivated grassland, 10, Tropical deciduous forest, 11, Bare soil, 12, Urban areas, 13. Species
Common name
Records Dependence Locomotion Habitat
Hospitable matrix
Inhospitable matrix
Didelphis marsupialis Opossum Tlacuache
77
L
S
3,11
1,2,4,7,8,10,13
5,6,9,12
Didelphis virginiana
Opossum Tlacuache
170
L
S
3,5,6,7,11
1,2,8,13
4,9,10,12
Philander opossum
Four-eyed opossum Tlacuache cuatro ojos, chipe
84
M
Ar
3,7,11
8
1,2,4,5,6,9,10,12,13
Dasypus novemcinctus
Armadillo
105
M
T
3,5,6,7,11
1,2,4,10
8,9,12,13
Armadillo, Tochi Tamandua mexicana
Mexican collared anteater Oso hormiguero, Chupa miel, brazo fuerte
54
H
Ar
3,7,11
1,2,4,5,6,8,9,10,12,13
Ateles geoffroyi
Spider monkey Mono araña
68
H
Ar
7,11
1,2,3,4,5,6,8,9,10,12,13
Canis latrans
Coyote Coyote
153
M
T
3,5,6,7,10,11,12
1,2,9
4,8,13
Urocyon cinereoargenteus
Grey fox
210
M
T
3,5,6,7,9,10,11,
1,2,4,12,13
8
Zorra gris Puma yagouaroundi
Jaguarundi Leoncillo, jaguarundi, Onza real
45
M
T
7,11
2,3,
1,4,5,6,8,9,10,12,13
Leopardus pardalis
Ocelot Ocelote
55
H
T
7,11
2,3
1,4,5,6,8,9,10,12,13
Leopardus wiedii
Margay Tigrillo, gato montes
36
M
Ar
6,7,11
3
1,2,4,5,8,9,10,12,13
a
Mountain lion Gato montes
127
L
T
5,6,9
2,3,7
1,4,8,10,11,12
Puma concolor
Puma Puma
65
M
T
5,6,7,11
9
1,2,3,4,8,10,12,13
Lontra longicaudis
River otter, water dog Nutria de río, perro de agua
187
H
Aq
7,8,11
3
1,2,4,5,6,9,10,12,13
Eira barbara
Tayra Viejo del bosque
25
H
S
7,11
Mustela frenata
Long-Tailed Weasel Comadreja
100
L
T
3,5,6,7,11
1,2,4,8,9,13
12
Galictis vittata
Greater grison Grisón
19
H
T - Ar
7,8,11
1,2,3,4,
5,6,9,10,12,13,
Conepatus leuconatus
Hog-nosed skunk
166
M
T
5,6,7,11
1,2,3,9,10
4,8,12
Lynx rufus
1,2,3,4,5,6,8,9,10,12,13
Zorrillo Mephitis macroura
Hooded skunk Zorrillo listado
135
H
T
2,5,6,7,11
1,2,3,9,10
4,8,12,13
Potos flavus
Kinkajou, honey bear Martucha, mico de noche
88
H
Ar
7,11
3
1,2,4,5,6,8,9,10,12,13
Bassariscus astutus
Ringtail, miner’s cat Cacomixtle, Sietillo
138
L
S
3,5,6,7,11
1,2,9,10,13
4,8,12
Bassariscus sumichrasti
Cacomistle
16
H
Ar
3,5,6,7,11
1,2,4,8,9,10,12,13
Cacomistle, Cacomixtle, Sietillo Nasua narica
White-nosed coatimundi Tejón, Coatí
144
H
S
3,5,6,7,11
1,2,10
4,8,9,12,13
Procyon lotor
Racoon Mapache
170
L
S
5,6,7,8,11
1,2,3,
4,9,10,12,13
Odocoileus virginianus
White-tailed deer
248
M
T
5,6,9,10
1,2,3,7,11
4,8,12,13
Venado cola blanca Mazama americana
Red brocket Temazate
47
H
T
7,11
1
2,3,4,5,6,8,9,10,12,13
Pecari tajacu
Peccary Pecari de collar
89
H
T
6,7,11
1,2,3,4,5
8,9,10,12,13 (continued on next page)
1456
H.F. López-Arévalo et al. / Biological Conservation 144 (2011) 1451–1463
Table 1 (continued) Species
Common name
Records Dependence Locomotion Habitat
Hospitable matrix
Inhospitable matrix
Sciurus aureogaster
Grey squirrel Ardilla gris
306
L
Ar
3,5,6,7,11
2,13
1,4,8,9,10,12
Sciurus deppei
Deppe’s squirrel Ardilla
113
L
Ar
3,5,6,7,11
Spermophilus variegatus
Rock squirrel
213
L
T
4,5,6,7,9,11
1,2,10
3,8,12
99
L
F
2,3,4,7.9.11
1,10
5,6,8,9,12,13
1,2,4,8,9,10,12,13
Ardillón Orthogeomys hispidus
Hispid pocket gopher Tuza
Sphiggurus mexicanus
Mexican hairy dwarf porcupine, Mexican tree porcupine Puerco espín, Viztlacuache
58
M
Ar
7,11
3
1,2,4,5,6,8,9,10,12,13
Cuniculus paca
Agouti, paca Tepezcuintle
45
M
T
7,8,11
2,3
1,4,5,6,9,10,12,13
Dasyprocta mexicana Mexican agouti Guaqueque negro
18
H
T
7,11
1,2,3,4,
5,6,8,9,1012,13
a
32
M
T
7,11
3,10
1,2,4,5,6,8,9,12,13
382
L
T
3,4,5,6,7,9,10,11 1,2,
Sylvilagus brasiliensis
Brazilian forest rabbit Conejo
Sylvilagus floridanus a
Eastern cottontail Conejo
8,12,13
Indicates species with no record in the area but present in nearby areas and that were included in the interview.
Table 2 Landscape variables for each of the four model test sites evaluated. Variable
Description
YERBABUENA
XICO
IXHUACAN
COSAUTLAN
Total area (ha) Number of forest fragments patches Mean forest fragments size Shape index Distance Y km (Crooks, 2002)
Total area of woody canopy Total number of forest fragments
851.44 1311
598.2 674
2297.56 663
590 1031
The size of each forest fragments in the model test sites Area: perimeter ratio Distance to the closest habitat that is equal to or larger in size (measured from the edge of each patch) The sum of the linear length of roads in the sector multiplied by a weighting factor depending on the type of road. Total area occupied by shaded coffee crops in the model test sites
0.65 ± 3.14 (Coef var 482.92) 3.69 39.6
0.89 ± 7.73 (Coef var 870.75) 5.96 54.86
3.47 ± 45.71 (Coef var 1318.91) 7.99 34.04
0.89 ± 4.19 (Coef var 731.59) 3.38 51.94
0.39 a 1.38; mean 1.07 + 0.199
0.34 a 1.14; mean 0.66 + 0.203
0.368 a 0.745; mean 0.5161 + 0.066
0.71 a 1.34; mean 1.03 + 0.190
1573.64
673.84
186.44
2091.56
824.28 521.84 319
3326.6 65.88 131.84
1551.76 1117.36 3
500.92 437.16 33.28
Road density
Area of shaded coffee crops (ha) Area of pastures (ha) Agricultural area (ha) Residential area (ha)
Total area occupied in the model test sites
Tlapaya, 2008; Gallina et al., 2008). Sylvilagus brasiliensis and Lynx rufus were also included in the surveys and analysis due to their presence in areas very near to the river basin (Table 1). The zoological nomenclature used followed Wilson and Reeder (2005). Eira barbara, Galictis vittata, Bassariscus sumichrasti and Dasyprocta mexicana presented less than 30 unique records for the country based on CONABIO information (2007). Eighteen species had been recorded more than 100 times and the remaining 14 species had intermediate record values (Table 1). The predictability values for the ten best country-scale models for the 32 species having trained and validated data varied from 0.85 to 1, having 20.3 to 455.8 Chi square values (p < 0.001). Canis latrans and Puma concolor distribution was not predicted on the UARB scale whilst the presence of S. brasiliensis and L. rufus was predicted. The basin had three main species distribution patterns: species widely distributed throughout the basin (for exam-
ple, Didelphis marsupialis, Mustela frenata), species having continuous distribution and a well-defined limit, possibly related to altitude (e.g. Sphiggurus mexicanus, E. barbara) and species whose distribution was fragmented (e.g. Odocoileus virginianus, B. sumichrasti). Local species richness values using GARP (GARPS) ranged from 6 to 32 species (34 gamma diversity). Expected local richness values for HABITATS and HRANGES fell between 1 and 32 species having the same value (34 spp.) for gamma diversity (Fig. 2). The highest number of species recorded by an interviewee was 20 from the 72 interviews conducted between 2007 and 2008 whereas the lowest was three (median 13 species). The total number of species for the basin recorded via interviews was 27 (Table 3) while the median obtained with the Chao 1 non-parametric estimator was 28 species (Fig. 3). Twenty-two species were recorded using the other methods (Table 4).
1457
H.F. López-Arévalo et al. / Biological Conservation 144 (2011) 1451–1463
Fig. 2. Spatial patterns for the potential distribution of medium-sized mammal species diversity in the upper Antigua river basin. On the left, the results of the model generated using the environmental variables used in GARPS and, on the right, the results of the model generated by combining available habitat and the minimum home range size (HRANGES).
Table 3 Jaccard similarity index values between the different model test sites and between predicted richness by scenarios and field information. GARP scenario GARPS
Model test sites
Expected richness
YERB
XICO
1 0.97 0.87 0.94
1 0.84 0.97
1 0.82
1
Home range scenario HRANGES
YERB
XICO
IXHU
COSA
YERB XICO IXHU
1 0.96 0.90
1 0.87
1
COSA Total
0.90
0.93
0.76
1
Field information
YERB
XICO
IXHU
COSA
Observed richness
YERB XICO IXHU COSA
1 0.86 0.69 0.79
1
21 20 23 22
YERB XICO IXHU COSA
IXHU
COSA 30 31 28 32
Total
34
1 0.79 0.83
1 0.73
Expected richness 28 29 27 31 34
Total
27
Scenarios and field data
YERB
XICO
IXHU
COSA
GARPS HABITATS HRANGES
0.63 0.70 0.75
0.58 0.63 0.71
0.66 0.85 0.88
0.61 0.69 0.71
The interviewees had not seen nine species in the wild for at least the last 2 years: Ateles geoffroyi, Mazama americana, P. concolor, Lontra longicaudis, E. barbara, Cuniculus paca and D. mexicana. Captive species like O. virginianus or Tayassu pecari which had been accidentally liberated during the past year were also mentioned by the interviewees. The most frequently recorded species included Sciurus aureogaster, Urocyon cinereoargenteus, Orthogeomys hispidus, M. frenata, D. virginiana, D. marsupialis, D. novemcinctus, S. floridanus and S. brasiliensis (Fig. 4). Although it had not been recorded in previous studies, S. brasiliensis presence was mentioned at all the test sites by 63% of the interviewees. Examination of a pelt confirmed its existence in the basin. L. rufus was mentioned in 14% of the interviews and scats which might have belonged to this species were observed in the upper part of the basin. C. latrans distribution was not predicted by GARP, but its presence in the basin was mentioned by 25% of
those interviewed. The reintroduction of a pair of coyotes 7 years ago was documented for the Xico model test site; a female was captured 5 years ago in Yerbabuena and a hunted animal was reported in January 2009. The size of the potential distribution area calculated in GARPS for 13 species was reduced by more than 40%. D. novemcinctus, S. floridanus, M. frenata and M. macroura were the only species presenting a reduction of less than 10% of their potential distribution area (Fig. 5). 3.2. Comparing scenarios by model test site Although most current vegetation cover in the basin occurs at the selected test sites, these represent a gradient extending from urban to suburban areas in Yerbabuena, agricultural areas in Cosautlan (mainly coffee plantations and sugarcane fields),
1458
H.F. López-Arévalo et al. / Biological Conservation 144 (2011) 1451–1463
Upper Basin of the La Antigua River
100
Sobs (Mao Tau) Chao 1 Mean Chao 1 95% CI Lower Bound Chao 1 95% CI Upper Bound Singletons Mean Doubletons Mean
Number of species
80
60
40
20
Expected richness values for each model test site in the HABITATS and HRANGES scenarios were the same as those predicted by GARPS, so they were omitted from Table 3. Similarity values were high in all modeled scenarios (0.76 being the lowest value). According to information obtained in the field, the lowest similarity value was 0.69 and the highest 0.86 (Table 3). Comparing species richness values between sites revealed a spatial change between theoretical GARPS species richness and that obtained from field information. The Ixhuacán sector (IXHU) had the lowest richness value (28 species) in the GARPS model compared to the other GARPS model test sites. This site had the highest richness value obtained with field information (23 species). Jaccard values for predicted richness by scenario and data collected in the field were higher than 0.50 at all test sites. HRANGES and field data had the highest similarity values (0.71–0.88) (Table 3). 3.3. Identifying species richness patterns
0 0
15
30
45
60
75
Number of interviews Fig. 3. Species accumulation curves obtained using the Sobs function (Mao Tau) and the Chao 1 non-parametric estimator, based on information gathered from 72 interviews.
pastured areas for livestock in Xico and forested areas in Ixhuacán (Table 2). The species richness predicted by the different scenarios was different from the richness found in the field (F = 30.643; p < 0.001). The GARP scenario was different from the others, having more similarity with HABITATS and HRANGES. When the nine species which had not been recorded as living in the wild were removed from each scenario, the differences between the models and the information from the interviews decreased, but the difference was still significant (F = 7.6044; p < 0.001). The species accumulation curves and Chao 1 non-parametric estimator indicated that asymptote was reached for all model test sites. There were small differences between observed and estimated richness and loss in the number of species having a single individual (Fig. 6).
All scenarios were negatively correlated with altitude (p < 0.001). This was seen most clearly in GARPS (0.974–0.951 95% correlation interval) and decreased in HABITATS (0.7294– 0.472) and HRANGES (0.721–0.468). Species richness tended to decrease with increased population density (correlation coefficient = 0.402) and road density (correlation coefficient = 0.22). 3.4. The role of protected areas There are six protected areas within the UARB, having different conservation categories and objectives. Five of these lie completely within the basin, covering 635 ha, extending from 1164 to 1200 masl (Table 5). Part of the Cofre de Perote Mountain National Park (about 4500 ha) lies in the basin, its lower limit being 3000 masl (Fig. 1), representing the largest protected area in the basin. The smallest protected area is the 1-ha Barragán Ecological Park. According to the HRANGES scenario, the RAMSAR site and Clavijero Park include the areas having the greatest potential number of species (27 and 28, respectively). When expected richness was compared to that observed in the field, all the areas had fewer species than expected (Table 5). Cerro de las Culebras (40 ha), the San Roque Mill (18 ha) and Barragán Park (1 ha) were
Table 4 Species recorded by other methods in the Upper Basin of the La Antigua river, obtained during the same period by García (2007). Species
Camera
Capture
Didelphis marsupialis Didelphis virginiana Philander opossum Dasypus novemcinctus Tamandua mexicana Urocyon cinereoargenteus Puma yagouaroundi Leopardus pardalis Leopardus wiedii Mustela frenata Conepatus leuconatus Mephitis macroura Potos flavus Bassariscus astutus Bassariscus sumichrasti Nasua narica Procyon lotor Sciurus aureogaster Orthogeomys hispidus Sphiggurus mexicanus Sylvilagus brasiliensis Sylvilagus floridanus
X X
X X X
Species total (22) a b
Pet
Furb
Scat and tracks
X X X X
X
X X
X X
Observationa
X X
X
X
X X X X X X X
X
X X
X X X
X X X
X
X X X
X 7
X X X
X 6
Includes observations in the wild and animals that had been run over by vehicles. Observed when the field interviews were being done.
3
11
13
4
1459
H.F. López-Arévalo et al. / Biological Conservation 144 (2011) 1451–1463
70
Number of interviews
60 50 40 30 20 10
At C p D m Eb Lw Ll M a O v Pc G v Pt Ph Lp H y C l Sv Lr Tm Pf Bs M m C l Ba N n C m Sd Pr Sb Sf D v D n D m M f O h U c Sa
0
Species Fig. 4. Distribution of species records in the interviews (n = 72) in the upper Antigua river basin. The capital letter represents the genus and the lower case letter the species, see Table 1.
100
80
Lw
Percentage change
Sd Lp Ll Pf
60
DmCp Sa Dv Ag
Hy Po Nn Uc
40
Bs Tm Ba
Dm
Sb
Pl Lr
20
Cm
Sb Oh Cn
Ma Pt Gv Mm
Mf
Ov Sf
Eb Cl Pc
0 Lw
Lp
Pf
Cp
Dv
Hy
Nn
Bs
Ba
Sb
Lr
Sb
Cn
Pt
Mm Ov
Eb Pc Dn
-20
Species Fig. 5. Percentage change in potential distribution area between the minimum area for the home range scenario (HRANGES) and the GARP scenario (GARPS). The capital letter represents the genus and the lower case letter the species, see Table 1.
immersed in an urban matrix, inferring that mammal populations in these places are isolated. The Francisco Javier Clavijero Park and the Texolo Waterfalls could support populations if some of the areas surrounding them were preserved for allowing animals to move between them, as suggested by Halffter (2007) in his Archipelago Reserves proposal as a complement to protected areas. The Cofre de Perote Mountain National Park, covering more than 11,000 ha, can support local medium-sized mammal populations; however, its geographical location in the upper part of the basin means that sites having lower expected richness are protected. There are no protected areas at sites lying below 1000 masl where the highest number of species were predicted for the basin (Figs. 1 and 2).
4. Discussion 4.1. Species richness and composition Medium-sized mammal richness in the area arises from a combination of Nearctic and Neotropical species, a characteristic that
has been acknowledged as being one of the factors promoting terrestrial Mexican mammals’ high gamma and beta diversity (Rodríguez et al., 2003). This combination exemplifies the complex biogeographical history of the cloud forest in this area (Rzedowski, 1991, 1996). GARP allowed us to model the distribution of most species previously recorded (GARPS) in the UARB. This led to identifying differences throughout the basin regarding medium-sized mammals’ theoretical species richness. GARP’s ability to predict mammals’ macro-distribution has been demonstrated in several studies using the type of environmental variables presented in this study (for example Anderson and Martinez-Meyer, 2004). The results obtained by combining GARP with the actual vegetation maps have demonstrated their usefulness in evaluating the effects of habitat transformation on a detailed scale given by fragmentation for endemic Mexican mammals (Sánchez-Cordero et al., 2005). Two species (P. concolor and C. latrans) were not found in the basin according to the models generated with GARP, even though being characterized by broad geographic distribution, wide home ranges and potential distribution covering the entire country. Although puma density is lower in
1460
H.F. López-Arévalo et al. / Biological Conservation 144 (2011) 1451–1463
Yerbabuena
Sobs (Mao Tau) Chao 1 Mean Chao 1 95% CI Lower Bound Chao 1 95% CI Upper Bound Singletons Mean
0
2
4
6
Number of species
Number of species
Xico
Sobs (Mao Tau)
80 70 60 50 40 30 20 10 0
Chao 1 Mean Chao 1 95% CI Lower Bound Chao 1 95% CI Upper Bound Singletons Mean
61 41 21 1
0 1 2 3 4 5 6 7 8 9 10 11 12 1314 15 16 17 18
8 10 12 14 16 18 20 22 24 26 28 30
Number of interviews
Number of interviews
Ixhuacán
Cosautlán Chao 1 95% CI Lower Bound Chao 1 95% CI Upper Bound Singletons Mean
60 40 20 0
Sobs (Mao Tau)
80
Chao 1 Mean
80
Number of species
Number of species
Sobs (Mao Tau)
Chao 1 Mean Chao 1 95% CI Lower Bound Chao 1 95% CI Upper Bound Singletons Mean
60 40 20 0
0
1
2
3
4
5
6
7
0
8
2
4
6
8
10
12
14
16
18
20
Number of interview
Number of interviews
Fig. 6. Species accumulation curves obtained with the Sobs funciton (Mao Tau) and the Chao 1 non-parametric estimator, based on the interviews held at each of the four model test sites.
Table 5 Currently protected areas in the Upper Basin of the La Antigua river, Veracruz, Mexico. Protected area
Area (ha)
Altitudinal range m a.s.l.
Expected species HRANGES
Observed species
Cofre de Perote Mountain National Park Texolo Waterfalls RAMSAR site Francisco Javier Clavijero Ecological Park Snake Hill (Cerro de las culebras) Ecological Park San Roque Mill Ecological Park Barragán Ecological Park
11,700 500 76 40 18 1
3000–4282 1093–1164
2–9 11–27 7–28 7–15 7–15 7
7a 22b 5a 6a 3a 2a
1200–1325 1350 1200
References: a Subsecretaria del Medio Ambiente, 2000. b Gordillo and Cruz, Unpublished results.
the southern part of the country and the coyote is naturally absent from the tropical rainforest and the cloud forest in southern Veracruz as well as the tropical evergreen forest (Leopold, 2000), the presence of both species has been noted in the region (Gallina et al., 1996; Gómez, unpublished result). The exclusion of both species from the area generated by GARP can be explained by the fact that collection record distribution was mainly from northern Mexico (CONABIO, 2007). This could have resulted from a lack of digitizing scientific collections, other kind of records and probably from inadequate sampling which can create artificial absences in species distribution models (Ponder et al., 2001). Godown and Peterson (2000), Loiselle et al. (2003) and Elith et al. (2006) give examples of GARP use and limitations in the conservation and study of other biological groups. All the scenarios overestimated the richness found in the test sites due to historic records allowing the modeling of available habitat for nine species. However, those species were not recorded during the field work, suggesting the existence of sites where the species have been locally extirpated; however, such places could possibly be used for reintroducing them, following a posteriori analysis (Anderson et al., 2003). Six of these species are character-
ized by having broad distribution and low density, two of them have limited distribution and high density and one is widely distributed with high density (Arita et al., 1990). The historical records and the existence of potential distribution areas for these species indicated that other causes such as hunting or population isolation could explain their absence from the model test sites. The effect of hunting on the disappearance of species from areas that still offer suitable habitat has been described for different tropical areas (IUCN, 2002). Overestimated species richness based on historical data has been analyzed in different scenarios, including the richness of species in protected areas in Canada where it was found that the historical maps produced an overestimation in the area of species’ occupancy, this being more evident on a fine scale than large spatial scales (Habib et al., 2003). Using different types of distribution data and identifying novel tools for application to existing distribution data-sets can minimize uncertainty about target attainment (Underwood et al., 2010). Compared to other species distribution models, GARP presented the highest over-prediction (Elith and Graham, 2009). Nevertheless, both the quality and availability of environmental data and
H.F. López-Arévalo et al. / Biological Conservation 144 (2011) 1451–1463
modeling techniques used can result in uncertainty and can overor under-estimate a species’ distribution (e.g. Loiselle et al., 2003; Rodríguez et al., 2007). Testing models developed from presence– absence data has been a recurrent focus in ecological discussion (e.g. Vaughan and Ormerod, 2005) including distribution and relative abundance models’ application (Royle et al., 2007; Wilson et al., 2010). The inverse relationship found between mammalian richness and elevation in UARB has been empirically recognized by several biological research groups (Graham, 1990; Stevens, 1992; Hunter and Yonzon, 1993). The relationship for small nonvolant mammals is curvilinear and richness is greatest at intermediate elevations between 2000 and 2500 masl (Sánchez-Cordero, 2001; McCain, 2004). The tendency towards a higher number of species in the lowlands than in the mountains has been identified for mammals in Asia (Steinmetz et al., 2007). The information gathered in the field allowed us to detect variations along an urban–rural gradient, greater species richness occurring in rural areas. This type of relationship has been documented in several studies (for a review, see McKinney, 2002) and coincides with the results of this study which found that the site least altered by fragmentation and urbanization (IXHU) had more species and its species richness was more similar to expected richness (Tables 2 and 3). Changes in species composition for terrestrial mammals regarding different land use (such as that reported for the UARB) have also been documented in Costa Rica where 60 species have been historically recorded, 37 recorded in the field and at least six species being locally extinct. Species richness and composition were related to habitat type, with pasture providing the least diverse sites while forest remnants and coffee plantations had similar richness to that of extensive forests (Daily et al., 2003). Mammals such as U. cinereoargenteus, D. marsupialis and D. virginiana and M. frenata were most frequently recorded and had the largest distribution throughout the area. An increased abundance of these species in fragmented forests has been reported in California where fragment area and the degree of isolation are the main factors explaining variation (Crooks, 2002). On the other hand, a high rate of land transformation and human activity in more than 60% of the area (Muñoz-Villers and López-Blanco, 2007) could be affecting populations on a regional scale, thereby promoting the disappearance of big carnivores and increasing the presence of generalist mesopredators, such as the grey fox U. cinereoargenteus or domestic mammals such as dogs and cats (Crooks and Soulé, 1999). The most frequent species (i.e. those that mentioned in at least half of the interviews, Fig. 4) had wide distribution and high density in tropical forest (Arita et al., 1990). However, the most abundant ones, U. cinereoargenteus, S. aureogaster, O. hispidus and M. frenata, are not limited to the forest and they do well in heterogeneous environments, such as those in the study area including coffee plantations (Gallina et al., 1996; Cruz-Lara et al., 2004), fragment edges and suburban environments (Crooks, 2002; Daily et al., 2003). These species fall into the categories of being exploiters and adaptable, as defined by McKinney (2002), sometimes becoming crop pests (González-Romero, 1980). The nine species that were not recorded have broad distribution but low density (Arita et al., 1990), making them more susceptible to anthropogenic effects in the sites evaluated here. Less common species, such as felids and other carnivores depending on the forest, fall into the categories that avoid urban and semi-urban environments, thereby being more susceptible to habitat changes (McKinney, 2002). Local inhabitants’ ability to quickly recognize medium-sized mammal species provided valuable information about the current richness at each model test site. The sampling effort put into the interviews was sufficient, given that the asymptote was reached in the species accumulation curve and in non-parametric estima-
1461
tors’ curves. The differences between estimated richness in the different scenarios and observed richness indicated local inhabitants’ ability to identify rare and extinct species as well as abundant and pest species. The inclusion of so-called traditional ecological knowledge often provides more abundant, reliable information than formal research (Huntington, 2000). Interviews are a traditional method in the social sciences and have also been used in ecological research. This is most evident in projects focused on conservation management and planning (Becker and Ghimire, 2003; Anadón et al., 2009). Interviews have been used for studying endangered species, evaluating their traditional uses and monitoring fauna by local communities (Lizcano et al., 2002; Anadón et al., 2009). Camera traps did not provide sufficient data to allow statistical analysis even if there have been good experiences of sampling elusive or rare species with this technique (Silver et al., 2004). This happened because there were many domestic animals and people in the area who activated the cameras, thereby giving useless pictures. The collection of medium and big mammals has little justification at the moment which is why systematizing direct or indirect observation would allow us to model the present distribution of species. These models would offer proposals regarding specific areas and concrete local action, complementing existing national proposals (Ceballos et al., 1998; Ceballos, 2007; Vázquez et al., 2009). According to some authors (Williams-Linera et al., 2002; Muñoz-Villers and López-Blanco, 2007), almost all the basin is being used for production activities negatively affecting the cloud forests. Nevertheless, the surrounding natural remnants matrix has been able to maintain its current medium-size mammal richness. The presence of different vegetation types is recognized as being important for maintaining landscape-scale assemblage of mammals (Velázquez et al., 2001). Given the low viability of the protected areas in the basin, the species recorded survive due to the coffee plantations (especially those using shade) and the inaccessibility of certain forested areas. A conservation alternative is to protect different sized areas permitting species establishment or dispersal to more suitable habitats, as if they were islands for a regional conservation scheme. This proposal is considered to be a good strategy in environments which have been highly modified by humans, especially in areas having high species turnover (Halffter, 2007; Williams-Linera et al., 2007). Even though species turnover is low in the basin, this would be an appropriate strategy given the species’ vagility, increasing its viability if it sought to increase the area’s structural connectivity by designing corridors. Municipal and private initiatives thus acquire more relevance since they have been referenced in different scenarios (Meisel and Woodward, 2005; Ochoa-Ochoa et al., 2009).
5. Conclusions The historical inventories used for analyzing species distribution with GARP overestimate the expected richness. Combining GARP with the information obtained from local inhabitants and experts allows rapid evaluation of medium-sized mammal richness on a regional scale, permits extirpated species to be easily recognized and those populations which have become considerably decreased (as well as abundant species) to be identified. Combining predictive distribution models of species (GARP), delimiting species distribution according to habitat type and area and local knowledge allow quick evaluation of medium-sized mammalian species richness. These tools can be combined for conservation goals and identifying research priorities.
1462
H.F. López-Arévalo et al. / Biological Conservation 144 (2011) 1451–1463
Reduced habitat in the area being evaluated seemed to be the main cause for the local disappearance of medium-sized mammals having a broad home range and which depended on forested areas, although hunting, isolation of remnant populations and the introduction of non-native fauna may be causing less perceptible damage to all UARB mammalian species. It is unlikely that wild medium-sized mammal populations will persist in currently protected areas, except for generalists or species having intermediate home range, traits increasing their probability of surviving in the existing landscape mosaic. Their long-term existence will depend on the permanence of a set of habitats arranged as archipelago reserves. Species depending on arboreal structures or forests, having small home ranges, will also present a higher probability of survival. Nevertheless, the currently protected areas are too small to support viable populations in the long-term. Acknowledgements The first author would like to thank the Universidad Nacional de Colombia for permission to pursue PhD studies and is grateful to the Russell E. Train Education for Nature Program, run by the World Wildlife Fund (Grant RL 27) and the Consejo Nacional de Ciencia y Tecnología, México (CONACYT) for partial scholarships awarded for carrying out his PhD studies at the Instituto de Ecología, A. C. in Xalapa, Veracruz, Mexico. Grateful acknowledgement is extended to Dr. Gonzalo Halffter and Dr. Octavio Pérez Maqueo for their valuable contributions. We thank three anonymous reviewers for providing helpful comments on previous drafts of this manuscript. Also to Bianca Delfosse and Bibiana López Cano who translated the text from the original in Spanish and Jason Garry who extensively revised it. Some of the data was obtained from CONABIO projects (Comisión Nacional de Biodiversidad): Q068, T9, J123, P130, J121 and A26. References Aldrich, M.P., Bubb, P., Hostettler, S., Van De Wiel, H., 2000. Bosques nublados tropicales montanos. Tiempo para la acción. WWF International/UICN, The World Conservation Union. Cambridge, England. Anadón, J.D., Gimenez, A., Ballestar, R., Pérez, I., 2009. Evaluation of local ecological knowledge as a method for collecting extensive data on animal abundance. Conserv. Biol. 3 (3), 617–625. Anderson, R.P., Martinez-Meyer, E., 2004. Modeling species’ geographic distributions for preliminary conservation assessments: an implementation with the spiny pocket mice (Heteromys) of Ecuador. Biol. Conserv. 116, 167– 179. Anderson, R.P., Lew, D., Peterson, A.T., 2003. Evaluating predictive models of species’ distributions: criteria for selecting optimal models. Ecol. Model. 162, 211–232. Arita, H.T., Robinson, J.G., Redford, K.H., 1990. Rarity in Neotropical forest mammals and its ecological correlates. Conserv. Biol. 4, 181–192. Becker, C.D., Ghimire, K., 2003. Synergy between traditional ecological knowledge and conservation science supports forest preservation in Ecuador. Conserv. Ecol. 8, 1. Bisby, F.A., 2000. The quiet revolution: biodiversity informatics and the Internet. Science 298, 2309–2312. Ceballos, G., 2007. Conservation priorities for mammals in megadiverse Mexico: the efficiency of reserve networks. Ecol. Appl. 17, 569–578. Ceballos, G., Oliva, G. (Eds.), 2005. Los mamíferos silvestres de México. CONABIO and Fondo de Cultura Económica, México. Ceballos, G., Rodríguez, P., Medellín, R.A., 1998. Assessing conservation priorities in megadiverse Mexico: mammalian diversity, endemicity, and endangerment. Ecol. Appl. 8, 8–17. Challenger, A., 1998. Utilización y conservación de los ecosistemas terrestres de México. Pasado presente y futuro. CONABIO, Instituto de Ecología y Agrupación Sierra Madre, México. Colwell, R.K., 2006. Estimate S: Statistical Estimation of Species Richness and Shared Species from Samples.
. CONABIO, Comisión Nacional de Biodiversidad, 2007. Base de datos de mamíferos de México, consulted on November 1, 2007. Crooks, K.R., 2002. Relative sensitivities of mammalian carnivores to habitat fragmentation. Conserv. Biol. 16, 488–502. Crooks, K.R., Soulé, M.E., 1999. Mesopredator release and avifaunal extinctions in a fragmented system. Nature 400, 563–565.
Cruz-Lara, L.E., Lorenzo, C., Soto, L., Naranjo, E., Ramírez-Marcial, N., 2004. Diversidad de mamíferos en cafetales y selva mediana de las cañadas de la selva Lacandona, Chiapas, México. Acta Zool. Mex. 20, 63–81. Da Silva, A.P., Mendes Pontes, A.R., 2008. The effect of a mega-fragmentation process on large mammal assemblages in the highly-threatened Pernambuco Endemism Centre, north-eastern Brazil. Biodiversity Conserv. 17, 1455–1464. Daily, G.C., Ceballos, G., Pacheco, J., Suzan, G., Sánchez-Azofeifa, A., 2003. Countryside biogeography of neotropical mammals: conservation opportunities in agricultural landscape of Costa Rica. Conserv. Biol. 17, 1814– 1826. Edwards, J.L., Lane, M.A., Nielsen, E.S., 2000. Interoperability of biodiversity database: biodiversity information on every desktop. Science 298, 2312–2314. Elith, J., Graham, C.H., 2009. Do they? How do they? WHY do they differ? On finding reasons for differing performances of species distribution models. Ecography 32, 66–77. 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.M., Peterson, A.T., Phillips, S.J., Richardson, K., Scachetti-Pereira, R., Schapire, R.E., Soberón, J., Williams, S., Wisz, M.S., Zimmermann, N.E., 2006. Novel methods improve prediction of species’ distributions from occurrence data. Ecography 29, 129–151. Escalante, T., Rodríguez, G., Morrone, J., 2004. The diversification of Nearctic mammals in the Mexican Transition Zone. Biol. J. Linn. Soc. 83, 327–339. Espinoza-Medinilla, E., Lorenzo, C., Briones-Salas, M., 2006. Integración del conocimiento de las colecciones mastozoológicas de México. In: Lorenzo, E., Espinoza-Medinilla, M., Briones-Salas, F.A., Cervantes (Eds.), Colecciones mastozoológicas de México. Asociación Mexicana de Mastozoología, A.C. México, D.F. pp. 537–548. ESRI (Environmental Systems Research Institute), 1996. Arcview GIS 3.2. Franklin, J.F., 1993. Preserving biodiversity: species, ecosystem or landscapes? Ecol. Appl. 3, 202–205. Funk, V., Richardson, K., 2002. Systematic data in biodiversity studies: use it or lose it. Syst. Biol. 51, 303–316. Gallina, S., Mandujano, S., González-Romero, A., 1996. Conservation of mammalian biodiversity in coffee plantations of central Veracruz, Mexico. Agrof. Syst. 33, 13–27. Gallina, S., González-Romero, A., Manson, R.H., 2008. Mamíferos pequeños y medianos. In: Manson, R., Hernández-Ortíz, V., Gallina, S., Melhtreter, K. (Eds.), Agroecosistemas cafetaleros de Veracruz: biodiversidad, manejo y conservación. INECOL, INE-SEMARNAT, Mexico. pp. 161-180. Gaona, S., González-Christen, A.L., López-Wilchis, R., 2003. Síntesis del conocimiento de los mamíferos silvestres del Estado de Veracruz, México. Rev. Soc. Mex. Hist. Nat. 3 época 1, 91–123. García, B. J., 2007. Comparación de la riqueza de mamíferos medianos en un gradiente de manejo de cafetales del centro de Veracruz. Master thesis. Instituto de Ecología A.C. Xalapa, Mexico (Unpublished results). Godown, M.E., Peterson, A.T., 2000. Preliminary distributional analysis of US endangered bird species. Biodivers. Conserv. 9, 1313–1322. González-Romero, A., 1980. Roedores plaga de las zonas agrícolas del Distrito Federal. Instituto de Ecología, Museo de Historia Natural de la Ciudad de México, Mexico. González-Romero, A., López-González, C., 1993. Reconocimiento preliminar de la mastofauna asociada a las zonas suburbanas de Xalapa y Coatepec. In: LópezMoreno, I. (Ed.), Ecología urbana aplicada a la ciudad de Xalapa Instituto de Ecología A.C., Xalapa, Mexico. pp. 221–238. Graham, G.L., 1990. Bats vs. birds: comparisons among Peruvian vertebrate faunas along an elevational gradient. J. Biogeogr. 17, 657–668. Greenberg, R., Bichier, P., Sterling, J., 1997. Bird populations in rustic and planted shade coffe plantations of Eastern Chiapas, México. Biotropica 29 (4), 501–514. Habib, L.D., Wiersma, Y.F., Nudds, T.D., 2003. Effects of errors in range maps on estimates of historical species richness of mammals in Canadian national parks. J. Biogeogr. 30, 375–380. Halffter, G., 1998. Una estrategia para medir la biodiversidad a nivel del paisaje. In: Halffter, G. (Ed.), La diversidad Biológica de Iberoamérica, vol. II, Acta Zool. Mex., Vol. Esp., Mexico, pp. 3–18. Halffter, G., 2007. Reservas archipiélago: un nuevo tipo de área protegida. In: Halffter, G., Guevara, S., Melo, A. (Eds.), Hacia una cultura de conservación de la diversidad biológica. Monografías Tercer Milenio, Zaragoza, España. pp. 281– 286. Hall, E.R., Dalquest, W.W., 1963. The mammals of Veracruz, vol. 14. University of Kansas Publications, Museum of Natural History. pp. 165–362. 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. Int. J. Climatol. 25, 1965–1978. Hunter Jr., M.L., Yonzon, P., 1993. Altitudinal distributions of birds, mammals, people, forests, and parks in Nepal. Conserv. Biol. 7, 420–423. Huntington, H.P., 2000. Using traditional ecological knowledge in science: methods and applications. Ecol. Appl. 10, 1270–1274. Illoldi-Rangel, P., Sánchez-Cordero, V., Peterson, A.T., 2004. Predicting distributions of Mexican mammals using ecological niche modeling. J. Mammal. 85, 658–662. IUCN Species Survival Commission, 2002. Links between Biodiversity Conservation, Livelihoods and Food Security: The Sustainable Use of Wild Species for Meat. International Union for Conservation of Nature and Natural Resources, Gland, Switzerland and Cambridge, UK. Laurance, W., Laurance, S.G., Hilbert, D.H., 2008. Long-term dynamics of a fragmented rainforest mammal assemblage. Conserv. Biol. 22, 1154–1164.
H.F. López-Arévalo et al. / Biological Conservation 144 (2011) 1451–1463 Leopold, A.S., 2000. Fauna Silvestre de México, second ed., Editorial Pax Mex. Lizcano, D.J., Pizarro, V., Cavelier, J., Carmona, J., 2002. Geographic distribution and population size of the mountain tapir (Tapirus pinchaque) in Colombia. J. Biogeogr. 29, 7–15. Loiselle, B.A., Howell, C.A., Graham, C., Goerck, J.M., Brooks, T., Smith, K.G., Williams, P.H., 2003. Avoiding pitfalls of using species distribution models in conservation planning. Conserv. Biol. 17, 1591–1600. Lomolino, M.V., 2001. Elevation gradients of species diversity: historical and prospective views. Global Ecol. Biogeogr. 10, 3–13. López-Wilches, R., 2003. Base de datos de los mamíferos de México depositados en colecciones de Estados Unidos y Canadá. Universidad Autónoma MetropolitanaIztapalapa. . MacGarigal, K., Marks, B.J., 1994. Fragstats: Spatial Pattern Analysis Program for Quantifying Landscape Structure. Reference Manual. Forest Science Department, Oregon State University. Corvallis, Oregon. 62 p + Append. Magurran, A.E., 1988. Ecological Diversity and its Measurement. Princeton University Press, New Jersey. McCain, C.M., 2004. The mid-domain effect applied to elevational gradients: species richness of small mammals in Costa Rica. J. Biogeogr. 31, 19–31. McKinney, M.L., 2002. Urbanization, biodiversity, and conservation. BioScience 52, 883–890. Meisel, J.E., Woodward, C.L., 2005. Andean orchid conservation and the role of private lands: a case study from Ecuador. Selbyana 26, 49–57. Moguel, P., Toledo, V.M., 1999. Biodiversity conservation in trafitional coffe systems of México. Conserv. Biol. 13, 1–21. Muñoz-Villers, L.E., López-Blanco, J., 2007. Land use/cover changes using Landsat TM/ETM images in a tropical and biodiverse mountainous area of centraleastern Mexico. Int. J. Remote Sens. 29, 71–93. Ochoa-Ochoa, L., Urbina-Cardona, J.N., Vázquez, L.B., Flores-Villela, O., BezauryCreel, J., 2009. The effects of governmental protected areas and social initiatives for land protection on the conservation of Mexican amphibians. PLoS One 4 (9), e6878. doi:10.1371/journal.pone.0006878. Pearson, R.G., Thuiller, W., Araujo, M.B., Martinez-Meyer, E., Brotons, L., McClean, C., Miles, L., Segurado, P., Dawson, T.P., Lees, D.C., 2006. Model-based uncertainty in species range prediction. J. Biogeogr. 33, 1704–1711. Peterson, A.T., Sánchez-Cordero, V., Martínez-Meyer, E., Navarro-Siguenza, A.G., 2006. Tracking population extirpations via melding ecological niche modeling with land-cover information. Ecol. Modell. 195, 229–236. Phillips, S.J., 2008. Transferability, sample selection bias and background data in presence-only modelling: a response to Peterson et al. (2007). Ecography 31, 272–278. Ponder, W.F., Carter, G.A., Flemons, P., Chapman, R.R., 2001. Evaluation of museum collection data for use in biodiversity assessment. Conserv. Biol. 15, 648–657. Ramamoorthy, T.P., Bye, R., Lot, A., Fa, J. (Eds.), 1993. Biological Diversity of Mexico: Origins and Distribution. Oxford University Press, New York. Ramírez-Pulido, J.A., Castro-Campillo, Arroyo-Cabrales, J., Cervantes, F.A., 1996. Lista taxonómica de los mamíferos terrestres de México. Occas. papers, Mus. Texas Univ. 158, 1–62. Rodríguez, P., Soberón, J., Arita, H.T., 2003. El componente beta de la diversidad de mamíferos de México. Acta Zool. Mex. 89, 241–259. Rodríguez, J.P., Brotons, L., Bustamante, J., Seoane, J., 2007. The application of predictive modelling of species distribution to biodiversity. Divers. Distrib. 13, 243–251. Royle, J.A., Kéry, M., Gautier, R., Schmid, H., 2007. Hierarchical spatial models of abundance and occurrence from imperfect survey data. Ecol. Monogr. 77, 465– 481. Rzedowski, J., 1978. Vegetación de México. Ed. Limusa, México. Rzedowski, J., 1990. Vegetación Potencial, IV.8.2. Atlas Nacional de México. Vol II. Escala 1:4 000 000. Instituto de Geografía, UNAM. México. Rzedowski, J., 1991. Diversidad y orígenes de la flora fanerogámica de México. Acta Bot. Mex. 14, 3–21.
1463
Rzedowski, J., 1996. Análisis preliminar de la flora vascular de los bosques mesófilos de montaña de México. Acta Bot. Mex. 35, 25–44. Sánchez-Cordero, V., 2001. Elevational gradients of diversity for rodents and bats in Oaxaca, México. Global Ecol. Biogeogr. 9, 63–76. Sánchez-Cordero, V., Illoldi-Rangel, P., Linaje, M., Sarkar, S., Peterson, A.T., 2005. Deforestation and extant distributions of endemic Mexican mammals. Biol. Conserv. 126, 464–473. Segurado, P., Araújo, M.B., 2004. An evaluation of methods for modeling species distributions. J. Biogeogr. 31, 1555–1568. Sheil, D., Puri, R., Wan, M., Basuki, I., van Heist, M., Liswanti, N., Rukmiyati, Rachmatika, I., Samsoedin, I., 2006. Local people’s priorities for biodiversity: examples from the forests of Indonesian Borneo. Ambio 15, 17–24. Silver, S.C., Ostro, L.E.T., Marsh, L.K., Maffei, L., Noss, A.J., Kelly, M.J., Wallace, R., Gómez, H., Ayala, G., 2004. The use of camera traps for estimating jaguar Panthera onca abundance and density using capture/recapture analysis. Oryx 38, 148–154. Soberón, J., Peterson, A.T., 2005. Interpretation of models of fundamental ecological niches and species’ distributional areas. Biodivers. Inform. 2, 1–10. Steinmetz, R., Chutipong, W., Seuaturien, N., 2007. Community structure of large mammals in tropical montane and lowland forest in the Tenasserim–Dawna Mountains, Thailand. Biotropica 40, 344–353. Stevens, G.C., 1992. The elevational gradient in altitudinal range: an extension of Rapoport’s latitudinal rule to altitude. Am. Nat. 140, 893–911. Stockwell, D., Peterson, A.T., 2003. Comparison of resolution of methods used in mapping biodiversity patterns from point-occurrence data. Ecol. Indic. 3, 213–221. Subsecretaria del Medio Ambiente, 2000. Áreas protegidas de Veracruz. Gobierno del Estado de Veracruz. p. 171. Tischendorf, L., Bender, D.J., Fahrig, L., 2003. Evaluation of patch isolation metrics in mosaic landscapes for specialist vs. generalist dispersers. Landscape Ecol. 18, 41–50. Tlapaya, R. L., 2008. Efecto de la cacería sobre la diversidad de mamíferos medianos en cafetales del centro de Veracruz. undergraduate thesis. Escuela de Biología. Benemérita Universidad Autónoma de Puebla, Mexico (Unpublished results). Turner, M.G., Gardner, R.H., 1991. Quantitative Methods in Landscape Ecology. Springer-Verlag, New York. Underwood, J.D., D’agrosa, C., Gerber, L.R., 2010. Identifying conservation areas on the basis of alternative distribution data sets. Conserv. Biol. 24, 162–170. USGS, United States Geological Service, 2007. . Vaughan, I.P., Ormerod, S.J., 2005. The continuing challenges of testing species distribution models. J. Appl. Ecol. 42, 720–730. Vázquez, L.B., Bustamante-Rodríguez, C.G., Bahena Arce, D.G., 2009. Area selection for conservation of Mexican mammals. Anim. Biodiv. Conserv. 32, 29–39. Velázquez, A.F., Romero, J., Rangel-Cordero, H., Heil, G.W., 2001. Effects of landscape changes on mammalian assemblages at Izta-Popo Volcanoes, Mexico. Biodiver. Conserv. 10, 1059–1075. Williams-Linera, G., Manson, R.H., Isunza, E., 2002. La fragmentación del bosque mesófilo de montaña y patrones de uso del suelo en la región oeste de Xalapa, Veracruz, México. Madera y Bosques 8, 73–89. Williams-Linera, G., Guillén Servent, G.A., Gómez García, O., Lorea Hernández, F., 2007. Conservación en el centro de Veracruz, México. El bosque de niebla: Reserva archipiélago o corredor biológico?. In: Halffter, G., Guevara, S., Melo, A. (Eds.), Hacia una cultura de conservación de la diversidad biológica, Monografías Tercer Milenio, Zaragoza, pp. 303–310. Wilson, G.J., Reeder, D.M., 2005. Mammal Species of the World: A Taxonomic and Geographic Reference, third ed. The John Hopkins University Press, Baltimore. Wilson, T.L., Odei, J.B., Hooten, M.B., Edwards Jr., T.C., 2010. Hierarchical spatial models for predicting pygmy rabbit distribution and relative abundance. J. Appl. Ecol. 47, 401–409. Zar, J.H., 1996. Biostatistical Analysis, second ed.. Prentice-Hall Inc., Englewood Cliffs, New Jersey.
