Study Area

We worked in a 225 km² area in the upper Balsas River basin (18°03′–17°54′N, 98°49′–98°59′W) in northeastern Guerrero, Mexico. The area presents an elevational gradient from 650 to 1,950 m. The climate is warm and semiarid with a summer rainy season (‘Aw' according to the Köppen climate classification) and mean annual precipitation of 780 mm (Meza and López García 1997). Dominant vegetation types are tropical deciduous forest (650–1,200 m) and oak forest (1,400–1,950 m; Rzedowski 2006), with a transition zone around 1,300 m (Figure 1).

FIGURE 1.

(A) Tropical dry forest and oak forest study sites in northeastern Guerrero, Mexico. Colored areas indicate the approximate locations of habitat patches surveyed at different perturbation levels in tropical deciduous (TF, orange areas) and oak forests (OF, dark green areas). p = primary forest; sg = secondary growth; and hs = human settlement. Gray lines indicate elevation. (B) Photograph illustrating the topography of the region (Photo credit: L.D.V.-R.).

Local human communities affect both forest types by clearing natural vegetation for subsistence agriculture, livestock grazing, firewood, and timber for building material. These activities define 3 levels of anthropogenic perturbation: (1) Primary forest, in which the original forest structure is practically unaltered. Human activity is limited to occasional human and livestock movements. Tropical deciduous primary forests are located at elevations between 726 and 994 m. Primary oak forests are found between 1,742 and 1,930 m. (2) Second-growth forest, in which the forest has been transformed by human activities within the last ∼20 yr. Some vegetation has regrown, although cattle grazing frequently continues. Tropical deciduous secondary forests are located between 735 and 887 m. Oak secondary forests occur between 1,578 and 1,780 m. Our habitat characterization of primary and second-growth forests showed that primary forest had smaller vegetation clearances and more vertical strata than second-growth forest (Appendix Table 1). (3) Human settlements, in which most trees of the original forest have been replaced by houses, streets, dirt roads, orchards, and paddocks for cattle and poultry. Human settlements that have replaced tropical deciduous forest are located between 640 and 735 m. In former oak forest, human settlements are located between 1,600 and 1,800 m.

Bird Counts

We performed surveys of bird communities along the perturbation gradient described above (Figure 1). For both tropical deciduous forest and oak forest, we selected 3 primary forest sites (a total of 175 ha in tropical deciduous forest, 196 ha in oak forest), 3 second-growth sites (355 ha and 211 ha), and 2 human settlements (97 ha and 109 ha). The size of individual habitat patches varied from 31 to 199 ha (mean = 71.5 ± 40.5 ha). The distance between adjacent stands of the same forest type was always >300 m, and between tropical deciduous and oak forest stands was 8–21 km.

We used 50-m fixed-radius point counts to record all species and the number of individuals of all species seen or heard during a 10-min sampling period. Observations were conducted during the first 4 hr after sunrise, avoiding rain. To ensure the independence of point counts in each survey event, points were located ≥200 m apart whenever possible, as recommended by Ralph et al. (1996). Our survey considered 100 point count stations, at which we performed a total of 342 point counts distributed across 4 sampling events between February and September, 2014, so that seasonal changes in habitat use by permanent resident birds (Navarro-Sigüenza et al. 2014; see full species list in Appendix Table 2) were taken into account (Vega-Rivera et al. 2010). In each survey event, we conducted between 6 and 9 point counts in each of the 3 primary and second-growth forest patches, and between 4 and 7 point counts in each of the 2 human settlements.

We based our analyses only on permanent resident species, excluding winter residents, summer residents, transients, and vagrants (Komar 2002). Residency status was assigned based on the seasonal occurrence of each species across the study area from surveys completed in 2009–2014 (L. Vázquez-Reyes personal observation) and on the existing literature (Howell and Webb 1995, Navarro-Sigüenza et al. 2014, Berlanga et al. 2015).

Data Analyses

Biotic Homogenization Assessment

The cluster analysis based on all birds recorded in the 3 habitats indicated 2 general clusters, each with at least 1 subdivision (Figure 3). The first cluster was composed of a group that included all human settlements (in both forest types), and a group corresponding to tropical deciduous forest, which in turn separated into a cluster corresponding to second-growth tropical deciduous forest and a cluster corresponding to primary tropical deciduous forest. The other general cluster included all remaining oak forest patches, both primary and secondary (Figure 3).

FIGURE 3.

Classification analysis of habitat patches along the human perturbation gradient in seasonal forest in the Balsas River basin, west-central Mexico. The Sørensen dissimilarity matrix and complete linkage method were used. Colors and boxes in the graph indicate the clustering arrangement of each habitat patch within the analysis. From top to bottom: Human settlements (gray clusters); tropical secondary (yellow cluster) and primary forests (red cluster); and both secondary and primary oak forests (green clusters). Vertical dashed line shows the average value of the dissimilarity matrix. Analysis was based on occurrence data of all resident bird species recorded in each habitat patch.

Assessing the biotic homogenization hypothesis, we found low values for the nestedness component of dissimilarity (βnestSor), and we did not find significant differences for βnestSor values in any comparison (Figure 4), either for habitat patches (H_2,21 = 0.69, P = 0.70) or for count events (H_2,15 = 3.5, P = 0.17). However, we found a statistically significant decrease of the dissimilarity turnover component (βturnSor) along the human perturbation gradient. This pattern was consistent when bird data were separated by habitat (H_2,21 = 12.48, P = 0.002, Figure 4A) and by count event (H_2,15 = 8.95, P = 0.011, Figure 4B). Turnover was highest in bird communities in primary forests, and lowest in human settlements. Mean βturnSor among habitat patches (84% confidence interval) was 0.66 (0.76–0.96) for primary forests, 0.49 (0.29–0.69) for second-growth, and 0.32 (0.27–0.37) for human settlements (Figure 4A). βturnSor for count events was 0.66 (0.59–0.71) for primary forest, 0.54 (0.44–0.64) for second-growth, and 0.38 (0.30–0.46) for human settlements (Figure 4B).

FIGURE 4.

Mean values of taxonomic dissimilarity along the perturbation gradient (primary forest, secondary forest, and human settlements) in the Balsas River basin, west-central Mexico. We considered values for: (A) habitat patches and (B) count periods. Open circles indicate the nestedness component of dissimilarity, which reflects dissimilarity owing to changes in species richness when one community is a subset of the other. Filled circles indicate the turnover component, which is associated with species replacement between communities, regardless of species richness in each. Error bars and horizontal dotted lines show 84% confidence values, equivalent to significant differences at α = 0.05.

Loss of Forest Specialists

The loss of bird diversity as a function of intensification of human activities as found in our study has been previously documented in other tropical deciduous forest and oak forest systems, driven by agricultural activities (MacGregor-Fors and Schondube 2011), cattle grazing (Maya-Elizarrarás and Schondube 2015), and land-use changes associated with human settlements (Ramírez-Albores 2007, MacGregor-Fors and Schondube 2012). Anthropogenic landscape transformation can promote an environmental filtering effect that reduces the availability of resources and changes ecological conditions. This filtering decreases the ecological success of bird species that have traits associated with the specialized use of natural environments (e.g., large body mass, low fecundity, limited ecological plasticity), eventually leading to local extirpation (Croci et al. 2008, Newbold et al. 2013, Paz Silva et al. 2016). In this way, habitat perturbation can drive a selective process of erosion of bird diversity in highly disturbed environments (McKinney and Lockwood 2001).

In our study, forest specialists were concentrated in primary habitats, occasionally occurred in second-growth habitats, and were never found in human settlements. In tropical deciduous forest, these specialist species included the West Mexican Chachalaca (Ortalis poliocephala) and Military Macaw (Ara militaris), both large-bodied bird species that feed on fruits in the canopy (Gurrola 2002, Contreras-González et al. 2009). Other relevant species included the Balsas Screech-Owl (Megascops seductus), Elegant Trogon (Trogon elegans), and Pale-billed Woodpecker (Campephilus guatemalensis), which are dependent on large trees in mature forest for nesting and foraging (Monterrubio-Rico and Escalante-Pliego 2006). In oak forest, a bird species dependent on large trees for nesting is the Mountain Trogon (Trogon mexicanus; Monterrubio-Rico and Escalante-Pliego 2006), and a species dependent on forest trees for foraging is the White-striped Woodcreeper (Lepidocolaptes leucogaster). Other bird species associated with this habitat were the Woodhouse's Scrub-Jay (Aphelocoma woodhouseii), Gray Silky-Flycatcher (Ptiliogonys cinereus), and Painted Redstart (Myioborus pictus; Ramírez-Bastida et al. 2015).

Anthropogenic influences can act on bird species through different mechanisms. Extinction risk owing to habitat loss is higher for habitat specialists. Extinction risk for large-bodied birds is often better explained by direct human persecution (Owens and Bennett 2000, Kattan et al. 2016). The West Mexican Chachalaca, Pale-billed Woodpecker, and Military Macaw are large-bodied birds that suffer direct human pressure (L. Vázquez-Reyes personal observation), and their exclusion from human-disturbed habitat can be related directly to human hunting.

Loss of Endemic vs. Widely Distributed Species

Proportional losses of endemic species and widely distributed species were similar in human settlements (39% compared with 38%; Figure 2). Nine of 39 (23%) species in human settlements were endemics. Eight bird species were recorded at all levels of perturbation in both forest types (full species list in Appendix Table 2), including 2 species endemic to the Balsas River basin, the Dusky Hummingbird (Cynanthus sordidus) and Black-chested Sparrow (Peucaea humeralis), as well as the Russet-crowned Motmot (Momotus mexicanus), a Mexican near-endemic associated with tropical deciduous forest (Gordon and Ornelas 2000, Berlanga et al. 2015). The fact that restricted-range species did not show a greater rate of loss differs from the typical expectation of increasing vulnerability with decreasing range (McKinney and Lockwood 1999).

Habitat Shapes Avian Community Composition

Communities in primary and second-growth oak patches were grouped separately from all other habitats. This result was driven by species found only in oak forests in the study region, including the Acorn Woodpecker (Melanerpes formicivorus), Woodhouse's Scrub-Jay, Bridled Titmouse (Baeolophus wollweberi), Painted Redstart, and Hepatic Tanager (Piranga flava; Ramírez-Bastida et al. 2015). Tropical deciduous forest patches were clustered through shared species with tropical affinity, such as the Balsas Screech-Owl, Golden-cheeked Woodpecker (Melanerpes chrysogenys), Pale-billed Woodpecker, Orange-fronted Parakeet (Eupsittula canicularis), and Military Macaw (Vega-Rivera et al. 2010).

However, human settlements within oak forest were clustered with other human settlements and tropical deciduous forest, rather than with other oak habitat patches. This pattern is evidence of the reduction in taxonomic differences in bird communities in the face of human perturbation (Figure 4). Six species occurred only in human settlements but in both forest types (full species list in Appendix Table 2), including 2 invasive species, the Rock Pigeon (Columba livia) and House Sparrow (Passer domesticus), increasing the number of species shared among human settlements and contributing to the homogenization of bird communities (Lockwood et al. 2000). As well as exotic species, native bird species capable of using anthropogenic habitats in both forest types, including the range-restricted Dusky Hummingbird, Russet-crowned Motmot, and Black-chested Sparrow, further reduced biotic differences between environments under anthropogenic pressure (Davey et al. 2012).

Erosion of Beta Diversity Defines Biotic Homogenization

The major consequence of biotic homogenization is the loss of biodiversity between ecosystems (Olden et al. 2004, McGill et al. 2015), and our results confirm this pattern for both taxonomic dissimilarity and beta diversity. In dissimilarity comparisons between tropical deciduous forest and oak forest at the same perturbation level, the nested component showed that differences in species richness were not significant (Figure 4). It may be that species loss affected species richness in the 2 habitats similarly, such that species were lost at similar rates with perturbation in both forest types (Baselga 2012). Several studies have documented reduced taxonomic differences between biotas in response to agricultural management (Luck and Smallbone 2011, Karp et al. 2012), land-cover change, and urbanization (Blair 2001, La Sorte and Boecklen 2005, Clergeau et al. 2006, Olden et al. 2006, Sorace and Gustin 2008). Our results, which showed a reduction in species turnover associated with perturbation intensification in tropical seasonal forests, are consistent with those of studies in other Neotropical seasonal forests, where intensification of agriculture reduced bird species turnover at local and landscape scales (Karp et al. 2012). However, in birds of the Atlantic forest of Brazil, homogenization processes showed a different path (Villegas Vallejos et al. 2016). In this case, loss of dissimilarity was due to the loss of the nested component of dissimilarity, instead of species turnover, because of the floristic and ecological differences between compared environments. Recent studies at the continental scale in North America have also found homogenization in urban birds due to species turnover loss (Murthy et al. 2016). Results are different in tropical wet forests, where turnover has been found to increase between ecosystems (Solar et al. 2015, Morante-Filho et al. 2016). Hence, environmental filters may act in different ways (Morante-Filho et al. 2016), emphasizing the need for further investigation in other Neotropical forests (Socolar et al. 2016).

Expansion of Tropical and Exotic Birds as Homogenization Drivers

Biotic homogenization is relevant to modern conservation planning, so it is necessary to identify and understand the patterns and processes behind this phenomenon (Rooney et al. 2007). Our results provide information toward understanding the taxonomic homogenization process. Reduction of taxonomic dissimilarity between tropical deciduous and oak forests was driven by: (1) directional expansion of species representative of tropical deciduous forest into oak forest areas; and (2) invasion of nonnative, human-commensal species that occurred exclusively within human settlements.

Some birds of tropical deciduous forest (e.g., Dusky Hummingbird, Russet-Crowned Motmot) were able to exploit the whole perturbation gradient, expanding their upper elevational ranges by >1,200 m into the highlands, where human settlements in former oak forest were located. This pattern is consistent with previous studies that have shown that some bird species can expand their ranges in the face of land use change, especially when they are locally abundant and express ecologically plastic traits (La Sorte and Boecklen 2005, Tabarelli et al. 2012).

Changes in avian community composition seem to have a directional pattern. While some tropical deciduous forest birds were consistently recorded in the disturbed patches of oak forest, no distinctive oak forest species were found in tropical deciduous forest (Ramírez-Bastida et al. 2015). This may indicate a pattern behind the process of biotic homogenization, in which tropical deciduous forest birds expand their altitudinal distributions up to 1,200 altitudinal meters. This upward, but not downward, altitudinal shift in geographic range may be explained by increasing temperatures (Şekercioğlu et al. 2012). Recent studies based on coarse-scale distributional data of Mexican endemic birds suggest that changes in their geographical distribution over the past century have been driven by temperature increases due to climate change (Peterson et al. 2015). As additional supporting evidence, studies of land-cover change have shown that temperature in the region has increased by 0.6°C during the last 50 yr, probably as a result of deforestation for cattle grazing (Cruz Hernández et al. 2007). The occurrence of invasive Eurasian species (Rock Pigeon, House Sparrow) in all human settlements but not in any forest patch is evidence that human transportation contributes to break down biogeographical barriers, a principal driver in the homogenization process (Hulme 2009, Winter et al. 2010, Baiser et al. 2012).

Implications for Use and Management of Dry Forests

Our results demonstrate that agriculture, cattle grazing, and human settlement have eroded the biodiversity of the Alto Balsas region, a process that has probably been repeated in other tropical deciduous forests and oak forests (e.g., Ramírez-Albores 2007). At present, primary tropical dry forest covers only 34% of its original area; primary oak forests and related temperate forests retain only 64% of their original extent (Sánchez Colón et al. 2009). Annual rates of deforestation for tropical and temperate forests represent losses of 0.76% and 0.25% of coverage, respectively (Mas et al. 2004). Even so, it is unrealistic to expect that these activities can be stopped, as they are fundamental for human subsistence in the region (Good-Eshelman and Barrientos-López 2004).

Because habitat loss is a general pattern across Latin America (Wright 2005), homogenization threatens biotic distinctiveness across the whole region. Biotic homogenization may affect ecological and evolutionary processes beyond the loss of species (Olden 2006, Şekercioğlu 2006, Şekercioğlu et al. 2012, Morelli et al. 2016). To confront these threats, it is necessary to consider mitigation strategies (Olden 2006, Rooney et al. 2007). One possible strategy is to maintain scattered native trees and shrubs throughout cultivated plots, as was practiced in traditional agriculture in west-central Mexico. This management practice maintains species richness similar to that of native primary forests (Vallejo et al. 2014). Leaving scattered native trees also improves connectivity among forest patches, favoring the maintenance of functional ecosystem integrity (Manning et al. 2006). Such a strategy, instead of total clearance of original vegetation, should be considered to preserve the biodiversity of forest ecosystems in the Balsas River basin, and other areas in Latin America.