Habitat loss and degradation are the most important threats that tropical forests and its associated fauna currently suffer (Primack, 2012). Consequently, primary forests tend to be rapidly replaced by the regeneration of secondary vegetation on previously deforested and degraded lands, which comprise five of the 11 million km2 of tropical forest coverage remaining today (Dent and Wright, 2009). The study of forest succession is important to understand the processes that control community assembly and generate diversity (Letcher et al., 2012), and to predict ecosystem responses to future anthropogenic disturbances (Prach and Walker, 2011).

One approach to study these processes is through the use of phylogenetic information. As phylogenies represent the accumulation of biological differences along branches, they may be used to explain or predict ecological and evolutionary processes assembling species into biological communities (Tucker et al., 2016). Based on this idea, many metrics have been developed to investigate patterns of community structure and diversity from a phylogenetic perspective (Vellend et al., 2011; Cadotte and Davies, 2016; Tucker et al., 2016). Phylogenetic structure represents nonrandom distribution of species within communities with respect to phylogeny (Webb et al., 2002). Some of the most commonly metrics used are net relatedness index (NRI) and nearest taxon index (NTI), which estimate how closely related are on average each pair of species within a community (Webb et al., 2002). From these metrics three outcomes can be obtained: phylogenetic clustering, overdispersion or random. Assuming that closely related species will tend to have similar ecological requirements, phylogenetic clustering suggests environmental filtering on shared physiological tolerances (Webb et al., 2002). Phylogenetic overdispersion suggests competition causing overdispersion of conserved traits or environmental filtering on ecologically important convergent traits (Webb et al., 2002). In a random pattern, no clustering or overdispersion can be detected. Therefore, the relative roles of niche-related (e.g., environmental filters and competition) vs neutral processes (e.g., random speciation and extinction, dispersal limitation) in community assembly can be inferred (Webb et al., 2002; Vellend et al., 2011; Cadotte and Davies, 2016).

Phylogenetic diversity is a measure of biodiversity incorporating phylogenetic differences between species (Faith, 1992). A commonly used metric is the phylogenetic diversity index (PD), which quantifies the total branch length spanned by the phylogenetic tree including all the species of a community (Faith, 1992). However, phylogenetic structure metrics can also give insights into phylogenetic diversity patterns (i.e., clustering = less phylogenetic diversity; Kellar et al., 2015). Phylogenetic diversity metrics thus explicitly incorporate species differences and not just the species richness and evenness (Vellend et al., 2011), and are often more useful because conserving the evolutionary diversity of a community would conserve their feature and functional diversity (Faith, 1992).

The tropical dry forest (TDF), one of the most biodiverse ecosystems around the world and with the largest extension in Mexico compared to the rest of America, is characterized by a pronounced seasonality in precipitation with a marked dry period during several months (Portillo-Quintero and Sánchez-Azofeifa, 2010). Despite its importance, the TDF is one of the most threatened ecosystems due principally to anthropogenic fragmentation and transformation because of farming and ranching (Trejo and Dirzo, 2000). Two of the most diverse and threatened groups that inhabit the TDF in Mexico are amphibians and reptiles (Ramírez-Bautista and Garcia, 2002). Around 77 (23%) amphibian and 243 (34%) reptile species inhabit this ecosystem in Mexico, representing 25% and 33% of all endemic species recorded in the country respectively (Ceballos and García, 1995). These organisms have important values as integral components of natural ecosystems and as indicators of environmental quality (Gibbons et al., 2000). Moreover, species of both classes are vulnerable to habitat loss and degradation (Gibbons et al., 2000; Gardner et al., 2007). Amphibians are more affected by deforestation and changes in humidity regimens, while reptiles are more affected by agriculture and the architectural change of their habitat (Gardner et al., 2007).

In this study, we assessed patterns of phylogenetic structure and diversity of amphibian and reptile communities along a successional gradient of TDF in western Mexico. We predicted a decrease in community relatedness across succession and a pattern of phylogenetic clustering (positive NRI and NTI) in pastures and first succession stages, compared to phylogenetic overdispersion or random patterns in later succession stages and primary forests (negative NRI and NTI for overdispersion and not-significant for random patterns; Fig. 1). This is expected as habitat filtering should be more important in newly opened areas because species are likely to experience environmental adversity (Chazdon, 2008). Similarly, we predicted an increase in phylogenetic diversity across succession, where primary forests and later succession stages will contain more distantly related species by having more niches available (Fig. 1). We also predicted higher values of PD and a stronger pattern of phylogenetic clustering during the dry rather the rainy season (i.e. higher NRI and NTI; Fig. 1). Studying phylogenetic structure and diversity across succession allow the evaluation of ecological forces that drive communities at different points in the assembly process (Letcher et al., 2012), and allow an indirect evaluation of communities’ functional diversity (Faith, 1992).

Fig. 1.

Two possible scenarios of changes in community phylogenetic structure and diversity expected during forest succession, and during two contrasting seasons. Points in the phylogeny indicate the presence of species in a community. Red branches represent the phylogenetic diversity index (PD), indicated by the total branch length spanned by the phylogenetic tree including all the species in a community. On the left, the community has four closely related species, representing low PD and positive values of the net relatedness index (NRI) and the nearest taxa index (NTI). On the right, the community has four distantly related species, higher PD and negative values of NRI and NTI.

(0.09MB).

Contrary to our predictions, we did not detect a consistent pattern of decrease in community relatedness and an increase in phylogenetic diversity across succession for both taxonomic groups. Due to the marked seasonality of the TDF and to the extreme environmental conditions that organisms can experience in deforested areas, differences in NRI between pastures and primary forests in reptiles during the dry season can be explained by environmental filters (Kraft et al., 2014). During the dry season, solar radiation is very intense and few places for shelter or to retain humidity are available. Therefore, species that coexist in pastures might share phenotypic features that allow them to persist in extreme conditions. As the succession progress, phylogenetic distance increases. Thus, unsuitable environmental conditions, such as those of the pastures, probably filter out species that lack adaptations necessary for survival and reproduction (Webb et al., 2002). For example, heliothermic lizards such as Holcosus undulatus, Aspidoscelis lineattissimus, and A. communis are abundant in forested areas, and in ecotones between highly modified and forested habitats, but are less common in pastures since this habitat could exceed their temperature tolerances (Casas-Andreu and Gurrola-Hidalgo, 1993). Primary forests had also four species of snakes (Imantodes gemmistratus, Leptophis diplotropis, Tantilla calamarina, and Crotalus basiliscus), and other species of lizards (H. undulatus and Phyllodactylus lanei), including an iguana (Ctenosaura pectinata), that were not detected in the pastures during the dry season. Conversely, Anolis nebulosus, Urosaurus bicarinatus, and Sceloporus utiformis are dominant species in trees surrounding pastures during the dry season. These species are habitat generalist disturbance-adapted, characterized by a combination of morphological (small size), physiological (thermoconformists), and life-history traits (high fecundity), as well as a fairly generalist diet that allow them to thrive in a wide range of environmental conditions. Differences in PD for reptiles, during the dry season, between pastures and primary forest could be the result of more and diverse microhabitats provided by complex habitats, such as primary and secondary forests under later succession stages, allowing the coexistence of more distantly related species when environmental adversity is higher.

Amphibians showed phylogenetically clustered patterns during the rainy season in the first three succession stages indicated by NRI and NTI. However, the pattern is clearer with NTI, as these succession stages showed significantly higher values than primary forests. This indicates that species are more related from their nearest taxon and therefore, the phylogenetic clustering occur towards the tips of the phylogeny, contrary to NRI which measures deeper divergences (Webb et al., 2002). Communities from the first three succession stages were dominated by species concentrated in Hylidae and Bufonidae, which congregate during the rainy season to mate in ephemeral or permanent water ponds formed mainly in the pastures and early forests. Such is the case of Incilius marmoreus, Triprion spatulatus, Smilisca fodiens, S. baudini, and Trachycephalus typhonius. Conversely, more distantly related species that inhabit intermediate and primary forests are more specialized to live on trees or in the use of forest leaf litter moisture during the rainy season, such as Eleutherodactylus nitidus, Craugastor mexicanus, Agalychnis dacnicolor, and other hylids tree frogs. Moreover, biotic and abiotic conditions required for amphibians to survive (higher levels of humidity, insect abundance, low solar radiation, etc.; Duellman and Trueb, 1994) offered by intermediate and primary forests would capture phylogenetically wider assemblages of species than early succession stages. During the dry season secondary and primary forests showed clustering patterns. It is likely that in pastures, the frequent permanent (artificial) water ponds for livestock promote the coexistence of more distantly related species, avoiding phylogenetic clustering.

Results from PD were not consistent with those of NRI and NTI, showing that primary forests had lower phylogenetic diversity than pastures and young forests. However, these results should be taken with caution because PD do not consider abundance or evenness among taxa, but only the presence or absence of species. Within both taxonomic groups, PD during the rainy season were significantly higher than during the dry season since many amphibians, and some reptiles like Heloderma horridum, several snakes and other species aestivate during the dry season (Secor and Lignot, 2010).

Only few recent studies have evaluated phylogenetic patterns of reptile or amphibian diversity in degraded ecosystems. Our results were similar to those studies as they have highlight the role of niche-based processes (i.e., environmental filtering), explaining structure and diversity patterns of amphibian or reptile communities along successional gradients in fragmented tropical rainforests (Hernández-Ordóñez et al., 2019), converted habitats across the globe (Nowakowski et al., 2018), agricultural landscapes (Ribeiro et al., 2017), seasonal tropical environments at a regional scale (de Araújo Martins et al., 2015), and across large-scale environmental gradients in reptile’s communities (Ramm et al., 2018).