Ethics Statement

Field samplings were done on private lands, with the corresponding permission of owners. Field sampling was authorized by the Secretaría del Medio Ambiente y Recursos Naturales (Mexican Council for the Environment and Natural Resources), which legislates scientific field samplings in Mexico, through permission 06108/09, according to the document SEMARNAT-08-049-A. We did not perform any other activities that required specific permissions, and the field sampling did not involve endangered or protected species. We did not sacrifice any organism for this study, because all of the individuals were released in the same location of capture. For field sampling in Mexico, the approval by an Institutional Animal Care and Use Committee (IACUC) or equivalent animal ethics committee is not required.

Field Sampling

Data were gathered from direct field sampling and from databases and literature review (see next section). Field work was carried out from July to September 2009 in 14 sampling sites along a transect of ca. 130 km, covering five vegetation types. For sampling anurans and reptiles, we carried out a direct search during day and night periods (from 9:00 to 13:00, 16:00 to 18:00, and 20:00 to 01:00). We captured individuals by hand or using herpetological hooks while animals were on the floor, below rocks, in caves, or ponds.

For birds, we did bird counts and captured individuals with mist nets. Bird counts were carried out from 17:00 to 19:00 h. At each sampling site a ca. 300 m transect was walked slowly for 20 min recording all birds seen. We captured birds using seven mist nets (12 x 3 m) at each sampling site from 6:30 to 12:30 h, in two consecutive days.

We used Sherman livetraps for capturing terrestrial small mammals, camera traps and sign surveys (tracks and scats) for medium sized mammals, and mist nets for bats. At each sampling site, 50 Sherman traps baited with oats and vanilla extract were set on a 500 m transect; the distance between traps was 10 m. Traps were active for two consecutive nights. We located 22 camera traps along trails that remained active for five months,; the cameras were checked every 20 days. Tracks and scats were recorded over non-restricted day and night walks at each sampling site. Finally, for bats we used five 12 x 3 m mist nets at each sampling site, that remained open over two consecutive nights from 19:00 to 00:00.

Biological Data and Study Area Delimitation

We compiled a database of vertebrate species for the Isthmus of Tehuantepec using georeferenced data for the states of Oaxaca and Veracruz (Fig. 1) from different sources, including literature and digital databases with information from scientific collections in Mexico and abroad: GBIF (Global Biodiversity Information Facility, http://www.gbif.org), UNIBIO (Unidad de Informática para la Biodiversidad of the UNAM’s Institute of Biology, http://unibio.unam.mx), and records provided by CONABIO (Comisión Nacional para el Uso y Conocimiento de la Biodiversidad: projects A14, A26, A27, AA3, B2, CC2, CE6, DC5, DC6, E18, G15, H245, J121, J123, L47, P130, P132, P60, UAZ, R246, S137, T9, U14, V9, W36). We then reviewed the current taxonomic nomenclature and the distribution of each species in specialized literature [23,32-37], the webpage of the American Ornithologist’s Union (http://www.aou.org) and that of the Amphibian Species of the World (http://research.amnh.org). Finally, we discarded all duplicate or triplicate records gathered from our different information sources, and kept only one record. Thus, our analyses do not include duplicate records.

In order to delimit the study area, we set up a grid with cells measuring 0.083 x 0.083 degrees. This grid size was used to clearly observe the cells in which most of the records are concentrated, and because this grain size has also been used in diversity analyses done at the national level [4,38]. For each cell we calculated the density of records, using the neighborhood method, excluding those cells with a low concentration of records from the analysis. The results showed that the areas with the highest density of records are located along the Panamericana Federal Transítsmica #185 highway. After delimiting the study area (i.e. fixed extent), we proceeded to divide the surface using grids of varying grain size, following one of the conceptual approaches recently revised by Barton et al. [2]. We used grids with three grain sizes: 10 x 10 km (122 cells), 20 x 20 km (28 cells) and 40 x 40 km (seven cells), which we refer to as fine, intermediate, and coarse grain hereafter (Figure 1).

For each cell we counted the number of species and the total number of records of each biological group. Cell species richness varied from 0 to 108 (the maximum number was for reptiles in the Los Tuxtlas region). Cells with no records were not included in the analysis. With this information we built matrices with species presence-absence information, and with the number of records for each one of the five vertebrate groups.

Finally, we identified the cells for which species inventories were adequately complete, using the total number of records of all the species as a measure of sampling effort [38]. For each biological group and grain size, inventory completeness was measured as the sample coverage percentage, which in our context is the proportion of the total number of records in a cell that belong to the species represented in our database [39]. To avoid under-sampling biases, only those cells with inventory completeness equal to or greater than 75% were used in the dissimilarity analysis.

Inventory completeness at the fine grain level was greater than 75% in 23 cells for amphibians, in 21 cells for reptiles, 12 for birds, 24 for nonflying mammals, and 30 for bats. For the intermediate grain grid the number of cells with complete inventories was: 12 for amphibians, 13 for reptiles, eight for birds, 15 for nonflying mammals, and 14 for bats. For the coarse grain all seven cells were analyzed for reptiles, nonflying mammals and bats, but only six cells for amphibians and four for birds reached the requested sample coverage (Figure S1). The final database includes 67 species of amphibians, 203 reptiles, 459 birds, 89 nonflying mammals, and 87 bats, for a total of 905 terrestrial vertebrate species (Checklist S1).

Data Analysis

Total dissimilarity was calculated with the Jaccard index (β_cc sensu [26,28], β_jac sensu [25]), and it was broken into two additive components: the dissimilarity derived from species turnover or replacement (β_-3) and the dissimilarity derived from richness differences (β_rich), following the framework proposed by J. C. Carvalho [26,28]. The sum of the values derived from turnover and richness differences must be equal to total dissimilarity, which ranges from 0 to 1. These indices were calculated using R [40] with a specific script [28].

We assessed differences in total dissimilarity (β_cc), turnover dissimilarity (β_-3) and richness differences dissimilarity (β_rich) among the five biological groups for each grain size by means of Kruskal-Wallis tests. When statistical differences were found, we performed Bonferroni corrected pairwise comparisons.

To test whether dissimilarity changes with spatial grain size, we compared the mean values of total dissimilarity (β_cc), turnover dissimilarity (β_-3) and richness differences dissimilarity (β_rich), among the three grain sizes using one way ANOVA tests, with Tukey tests for pairwise multiple comparisons in case of significant differences. As these values are means, they passed the normality test, so there was no need to transform the data. We calculated the geographic distance between cells using the centroid method, locating the central point of each cell and measuring its distance to the other central points, for the three grain sizes. Finally, for each biological group and each spatial grain size, we assessed the relationships between geographic distance and dissimilarity (β_cc, β_-3 and β_rich) using Mantel correlations. The ANOVA and Mantel test were run in the Past program [41].

Dissimilarity and Dispersal Ability

Contrary to our expectations, dispersal ability does not exert a clear influence on species dissimilarity. In fact, as shown in Figure 2, birds have the highest dissimilarities at fine and intermediate grains, even when this group is assumed to be the most vagile. In contrast, amphibians have very low values of dissimilarity for all three grain sizes, in spite of having the most limited dispersal ability.

Dissimilarity and Spatial Grain Size

According to our prediction, grain size is directly related to species dissimilarity, as shown in Figures 2 and 3. Using a fine grain we detect higher values of dissimilarity, while a coarse grain returns lower values of dissimilarity. Total dissimilarity (β_cc) and dissimilarity derived from richness differences (β_rich) changed significantly, decreasing as spatial grain size increased (Figure 3). In contrast, mean dissimilarity due to species turnover (β_-3) was not different for the three grain sizes (Figure 3b).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Components of dissimilarity at three spatial grain sizes.

The figure shows mean total dissimilarity (A), dissimilarity derived from species turnover (B), and dissimilarity derived from richness differences (C) for terrestrial vertebrates on the Isthmus of Tehuantepec. Error bars are standard errors. ANOVA results are included, equal low case letters indicate no difference in pairwise comparisons.

https://doi.org/10.1371/journal.pone.0082905.g003

Dissimilarity and Geographic Distance

Using the finest grain, total dissimilarity (β_cc) is positively and significantly correlated with geographic distance for all groups of vertebrates, though correlation coefficients are low in some cases (Table 1). The correlation of β_cc is strongest for amphibians, and it was the only group for which geographic distance was significantly correlated with richness differences (β_rich, Table 1).

Using the data from the intermediate grain, total dissimilarity and dissimilarity due to turnover were positively and significantly correlated with geographic distance for most of the groups, with the strongest correlation for the total dissimilarity of amphibians. However, none of the correlations with dissimilarity due to richness differences was significant (Table 1).

Finally, using the coarse grain grid total dissimilarity was positively and significantly correlated with geographic distance for amphibians and nonflying mammals (Table 1). For this latter group we also found a significant correlation between distance and dissimilarity derived from species turnover, but for none of the biological groups did we find a significant correlation between distance and richness differences (Table 1).

The Isthmus of Tehuantepec: a Region with Notably High Beta Diversity

The results confirm our first prediction: beta diversity is very high in this region, with values above 0.77. And species dissimilarity is clearly derived from both species turnover and richness differences. This may be influenced by several factors. First, the biogeographic history of the isthmus had an effect when the highlands were considerably altered by tectonic activity, and became flatlands [16,19]. Also, given its location in tropical latitudes, the area did not experience drastic climate changes during glaciations, as those recorded in temperate regions. This favored the speciation of some vertebrates. In this biogeographic context, the current environmental conditions include a highly heterogeneous landscape, which shapes the distribution of the vegetation. Important climate factors in the region include environmental humidity and the flow of air masses from the coasts of the Gulf of Mexico and the Gulf of Tehuantepec, across the region in the absence of geographic barriers, as well as the strong north winds that cross the plains of the isthmus, resulting from gradients of atmospheric pressure [14]. Also, as a tropical region the study area has high productivity, and in areas with large energy availability (measured as annual potential evapotranspiration) the beta diversity of terrestrial vertebrates is higher [42]. All these factors contribute to climatic and environmental heterogeneity, resulting in a fine subdivision of the region into three physiographic subprovinces [15].

It is possible that speciation and the constant movement of vertebrates in the region hampered the assemblage of nested faunas to some extent, and thus the dissimilarity due to richness differences is not too high. This is quite different from the patterns reported for northern Europe and North America, where glaciation caused the phenomena of extinction and colonization, influencing the structure of current beetle and vertebrate faunas [6,10], but see 45, who also discusses the influence of non-climatic drivers in nestedness.

The Highest Beta Diversity is Detected Using a Fine Spatial Grain

Our results corroborate the trend of increasing dissimilarity as grain size decreases. This is one of the most consistent of the patterns of beta diversity that have been reported for several groups, including insects and terrestrial vertebrates [5,8,24,43]. This is because as grain size increases, environmental differences between sampling units decrease, and more species are shared between the sample units. Moreover, bigger areas have more species and the degree of aggregation in species distribution is lower. Besides, there might be an influence of sampling biases: coarse grain may imply less undersampling, and therefore less beta diversity [44].

Also, with a coarse grain biogeographic history may explain much of the variation in diversity patterns and in the change in species composition [2,8,9]. In contrast, with a fine grain size, environmental heterogeneity is more important as a driver of changes in species composition at local scales [2,5,7,9,10,45].

Dispersal Ability Does Not Determine the Dissimilarity for Vertebrates on the Isthmus of Tehuantepec

Contrary to the results of several terrestrial vertebrate studies that report higher beta diversity in organisms with low dispersal ability [4-6,8,43], in this study birds had high beta diversity (except for the dissimilarity due to species richness at the coarse grain size) while amphibians had low dissimilarity. Thus, dispersal ability does not seem to be a strong determinant of the beta diversity value of terrestrial vertebrates on the Isthmus of Tehuantepec. For birds, the isthmus is an obligatory route during their annual migration [18]. Of course, this has an impact on the high bird species richness in the area, and may be an important driver of the high beta diversity. To further explore this possibility, it would be interesting to perform separate analyses for different subgroups of birds, such as residents, migrants, and the assemblages that coexist during different seasons of the year. Our analyses were gross in the sense that we treated all bird species as a single group. However, we predict that if migratory birds were excluded, a more reliable pattern of beta diversity could have appeared. Thus, we suggest that further studies should focus on resident species, and assess the potentially different responses of functional groups.

For amphibians, the low total dissimilarity is probably related to the low species richness recorded in the majority of the cells, because rare species are occasional. Most of the cells include species with widespread distributions, and rare species are the key elements that may increase beta diversity. Amphibians are particularly difficult to survey (as compared to other vertebrates) because their detectability is biased by the experience of researchers and particular climatic conditions. Thus, a more thorough survey requires trained personal and extended sampling periods. Moreover, in our study area the records of species with restricted distributional ranges occur in the northern part of the isthmus, mainly in the region of Los Tuxtlas (see Figure S1). Thus, we need to improve amphibian inventories in the central and southern part of the Isthmus of Tehuantepec to determine whether low beta diversity is a generalized characteristic, or these results are due to sample biases.

For nonflying mammals and bats the differences in beta diversity may be related to the latitude of our study area. Latitude has a different effect on the beta diversity of North American mammals [45]: beta diversity of nonflying mammals decreases as latitude increases, while beta diversity of bats remains constant. Thus, at the latitude of the Isthmus of Tehuantepec beta diversity of bats is higher than that of nonflying mammals [46], as occurs with total dissimilarity at the intermediate and coarse grains in our study. Therefore, the effect of dispersal ability among flying and nonflying mammals may be obscured by the differential effects of latitude on beta diversity.

It is worth mentioning that the beta diversity studies on the dispersal ability of terrestrial vertebrates have been done on global, continental or national scales [4-6,8,43]. At these scales, changes in energy availability and in the environment are more evident. But for the regional scale of our work, energy availability does not appear to have an important effect because this is the narrowest region of the country and lacks geographic barriers. In spite of this, the relationship between dissimilarity and geographic distance may indicate some degree of influence by the organisms’ dispersal ability (discussed in the next section).

Dissimilarity Increases with Geographic Distance

Total dissimilarity and dissimilarity derived from species turnover are positively correlated with geographic distance, and these relationships are more evident at the intermediate and fine grain sizes for almost all biological groups. These results reflect the phenomenon of distance-decay in similarity: similarity in species composition between sites decreases as the distance between the sites increases [47]. Three main mechanisms, which are not exclusive, have been proposed to explain this pattern [47,48]. First, environmental conditions change as the distance increases. This implies a spatial separation of species with different physiological requirements. The second mechanism depends on the configuration of the environment (both in spatial and temporal terms), because it influences the movement of species. With more barriers, similarity decreases more abruptly than in a topographically open and homogeneous site. Finally, the third mechanism is related to limitations in the dispersal ability of species, because the relationship between similarity and distance occurs even when the environment is completely homogeneous. On the Isthmus of Tehuantepec these three mechanisms may influence dissimilarity and its positive correlation with distance. However, the high environmental heterogeneity and the convergence of three physiographic subprovinces may be the most important factor.

Despite the lack of direct evidence of the influence of dispersal ability on dissimilarity, correlations between dissimilarity and geographic distance were weaker for vagile groups (for bats the three dissimilarity measures at the three grain sizes, and for birds β_cc on the coarse grain grid, β_-3, and β_rich at fine and intermediate grain sizes).

Geographic distance was not related to the dissimilarity derived from richness differences for any of the analyzed grain sizes (except for amphibians at the fine grain size). This pattern was also recorded in Europe for nonflying mammals, where dissimilarity due to richness differences was not related to distance in any geographic direction (north-south, east-west), or for specific regions within the continent; species turnover was so high that even the remote cells did not represent nested subsets [45]. At small scales (less than 250 000 km²) high vertebrate species turnover and low richness differences results from topographic and environmental heterogeneity [44,49], while over larger scales (continents) biogeographic history plays a more important role than environmental heterogeneity does [5,7,9,10,45]. Thus, the high environmental heterogeneity of the Isthmus of Tehuantepec may influence dissimilarity due to richness differences despite the geographic distance between faunas. However, we think that this idea would benefit from an analysis carried out within each physiographic subprovince, where cells of different sizes may be nested subsets of the total fauna in the subprovince.