With natural forests increasingly threatened by human-caused disturbances (Laurance et al., 2012) and deforestation rates advancing quickly worldwide (World Resources Institute, 2021), biodiversity conservation cannot solely rely on protected areas. It also requires the implementation of biodiversity-friendly strategies in human-modified landscapes (Arroyo-Rodríguez et al., 2020; Melo et al., 2013). To this end, assessing species’ responses to landscape changes is critically needed to identify optimal landscape scenarios for biodiversity conservation (Arroyo-Rodríguez et al., 2020). Of particular importance is assessing the effect of forest loss on the abundance and diversity of different species, as our understanding on this topic is far from complete. Whereas several studies indicate that forest loss can be negatively related to the diversity of several vertebrate groups (Watling et al., 2020), in many other cases, forest loss can have weak or even positive effects on some species (e.g., Gestich et al., 2021; Pardini et al., 2009; Vallejos et al., 2020). The positive responses to forest loss can be related to the fact that populations of some species can survive the initial phase of deforestation and crowd in the remaining forest patches (the so-called ‘crowding effect’; Gestich et al., 2021; Vallejos et al., 2020). Therefore, additional studies on the effect of forest cover on biodiversity are needed to assist the implementation of adequate conservation strategies in human-modified landscapes.

Forest loss not only limits the availability of suitable habitat for forest species; it also causes an exponential increase in the isolation distance between forest patches (Fahrig, 2013, 2003). This can explain the decline in taxonomic diversity (Moreno-Opo, 2020), and the alterations in the structure of remaining communities in more deforested landscapes (Best et al., 2001). However, the effect of forest loss on species largely depends on their habitat requirements, and their ability to use resources from different land cover types (see “cross-habitat spillover hypothesis” Tscharntke et al., 2012). For instance, forest-specialist species are usually specialized in resources that can only be found within the forest (Miranda et al., 2021), and many of them have low vagility (Linnell and Lesmeister, 2019). Thus, forest loss can strongly impact forest-specialist species (e.g., Morante-Filho et al., 2015). In contrast, the so-called ‘disturbance-adapted’ species usually have high vagility in human-modified areas, where they can find food resources and space to reproduce (Fehlmann et al., 2021). Finally, many species are habitat-generalists; i.e., they can use resources from different land covers and adjust their diets to the resources found in various vegetation types, including both native and anthropogenic vegetation (Fehlmann et al., 2021). In addition, habitat-generalist species can change their vagility depending on the degree of landscape deforestation (Salinas-Melgoza et al., 2013). Therefore, separately assessing the effect of forest cover on different ecological groups is paramount for a deeper and more accurate understanding of species’ responses to deforestation.

Among vertebrates, birds comprise one of the leading groups affected by forest loss (Sekercioglu, 2012). However, as research on birds frequently focuses on forest-specialist birds (Boesing et al., 2018), or on the complete bird assemblage (Leyequien et al., 2010), the response of different ecological groups to forest loss remains to be relatively misunderstood (but see Matuoka et al., 2020; Morante-Filho et al., 2015). Furthermore, most landscape studies of birds use a patch-landscape design to assess how the bird community in focal sites is affected by the surrounding landscape (Bennett et al., 2006) and, usually the birds are sampled in a single land cover type (e.g., forest cover, Boesing et al., 2018; Carrara et al., 2015; Morante-Filho et al., 2015). Thus, we know relatively little about the effect of landscape changes on bird assemblages in different land cover types of the landscape mosaic.

As suggested by Bennett et al. (2006), mosaic-level sampling can be done by evaluating multiple sample points in a single type of land cover, or by evaluating several sample points in multiple land covers (the design adopted in our study). The first approach is particularly adequate when the research question involves a single but spatially heterogeneous land cover type, as it allows the researcher to capture the habitat heterogeneity (e.g. caused by different soil types) within the focal land cover. However, when species can use different land covers and habitat heterogeneity is smaller within land covers than among them, the second approach is more adequate, as it allows the researcher to better understand the impact of landscape forest cover on birds across the whole landscape mosaic (Bennett et al., 2006). Yet, as bird research is more frequently based on the first approach (Boesing et al., 2018; Matuoka et al., 2020; Morante-Filho et al., 2015), additional studies on the effect of the forest cover on bird communities in the whole landscape mosaic are needed, particularly in mountain regions, in which different land covers can provide habitats for different species (Mendenhall et al., 2016).

The “La Montaña” mountain region of Guerrero State, harbors a high percentage of endemic birds in Mexico (Navarro-Sigüenza et al., 2014). However, it has been subjected to a wide variety of land-use changes, resulting in heterogeneous landscapes with different-sized forest patches being surrounded by an anthropogenic matrix composed of small villages, low-input crops, and grasslands (Borda-Niño et al., 2017). Despite this, to our knowledge, only two studies have evaluated the response of birds to land-use changes in this region (Almazán-Núñez et al., 2018; Vázquez-Reyes et al., 2017). These studies demonstrate that anthropogenic disturbances negatively affect bird diversity. However, these studies use a patch-landscape design and do not evaluate the effect of landscape composition on bird diversity in other land cover types than forest.

Here, we assessed changes in species composition along a forest cover gradient and investigated how changes in forest cover affect the diversity and distribution of birds sampled in different land covers, including second-growth vegetation, annual crops, pastures, and villages. We conducted this analysis not only with the complete assemblage, but separately for species with different habitat requirements (i.e., forest-specialist, disturbance-adapted and habitat-generalist species). Specifically, we predicted that more forested landscapes would exhibit higher diversity of the bird complete assemblage, mainly by maintaining forest-specialist birds that depend more on forest resources (Morante-Filho et al., 2015). In contrast, we predicted that landscapes composed of low forest cover would present high diversity of disturbance-adapted birds. We also predicted a higher proportion of occupied sites by forest-specialist birds in more forested landscapes, whereas disturbance-adapted birds likely follow the opposite pattern. Finally, we predicted that the diversity of habitat-generalist birds and the proportion of sites occupied by these species are probably weakly affected by landscape changes.

We recorded a total of 497 individuals belonging to 56 bird species. Thryophilus sinaloa, Pheugopedius felix, Basileuterus rufifrons, and Turdus rufopalliatus – all habitat-generalist species – were the most abundant species in most landscapes (Fig. 2). However, forest-specialist species such as Setophaga graciae, Lepidocolaptes affinis, Myioborus pictus, and Piranga flava dominated the landscapes with higher forest cover (Fig. 2).

Fig. 2.

Rank-abundance curves showing the relative abundance of bird species recorded in 10 landscapes with different percentage of forest cover (indicated on each curve). We classified the species as forest-specialist (green diamonds), disturbance-adapted (red diamonds), and habitat-generalist (black diamonds). Dominant species: Aru = Aimophila rufescens; Abe = Amazilia berillina; Bru = Basileuterus rufifrons; Cpe = Contopus pertinax; Laf = Lepidocolaptes affinis; Mpi = Myioborus pictus; Pfe = Pheugopedius felix; Pfl = Piranga flava; Sco = Saltator coerulescens; Sgr = Setophaga graciae; Tas = Turdus assimilis; Tsi = Thryophilus sinaloa; Tru = Turdus rufopalliatus.

(0.15MB).

The species richness (°Dγ) and the diversity of common species (1Dγ) of the complete bird assemblage were positively related to forest cover (R2 = 0.62; P = 0.007, and R2 = 0.87; P < 0.001, respectively; Fig. 3a,b), but the diversity of dominant species (2Dγ) was not associated to forest cover (R2 = 0.10; P = 0.37; Fig. 3c). The relationship between species richness and forest cover differed among ecological groups (interacting term within the ANCOVA: F1;24 = 14.90, P < 0.001, Fig. 3d). In particular, the species richness of forest-specialist birds was positively related to forest cover, but the opposite pattern was observed when assessing disturbed-adapted species (Fig. 3d, Table S2). The species richness of habitat-generalist birds remained constant along the forest cover gradient (Fig. 3d, Table S2). The effect of forest cover on the diversity of common species also differed among groups (F1;24 = 19.76, P < 0.001, Fig. 3e), with the diversity of forest-specialist and habitat-generalist birds being positively related to forest cover, whereas the diversity of disturbance-adapted species being weakly related to forest cover (Fig. 3e, Table S2). Conversely, the effect of forest cover on the diversity of dominant species did not differ among ecological groups (F1;24 = 3.17, P = 0.06; Fig. 3f, Table S2).

Fig. 3.

Response of species richness (0Dγ, A and D), diversity of common species (1Dγ, Shannon's entropy exponential; B and E) and diversity of dominant species (2Dγ, Simpson's inverse concentration; C and F) to landscape forest cover, separately assessing bird complete assemblages and different ecological groups (forest-specialist birds, disturbance-adapted birds, and habitat-generalist birds). In all cases, we showed the accumulated alpha diversity in 16-point counts (i.e., gamma diversity per landscape). The shaded area is the 95% confidence interval of the linear models.

(0.32MB).

Regarding the effect of forest loss on birds’ distribution, we found that the proportion of occupied sites was not related to forest cover in the complete assemblage (R2 = 0.011; P = 0.77; Fig. 4a), but we found a strong relationship when analyzing ecological groups separately (Fig. 4b). This relationship differed among ecological groups (F1;24 = 20.99, P < 0.001), forest-specialist birds responded positively to forest cover (Fig. 4b, Table S2), but disturbance-adapted and habitat-generalist birds presented a negative trend (Fig. 4b, Table S2).

Fig. 4.

Association between the logarithm of proportion of sites occupied by birds and landscape forest cover separately assessing the complete assemblage (A) and different ecological groups (forest-specialist, disturbance-adapted, and habitat-generalist birds) (B). The shaded area is the 95% confidence interval of the linear model.

(0.17MB).

Our study indicates that the diversity of the complete bird assemblage, and particularly forest-specialist and habitat-generalist species, decreased in more deforested landscapes. In contrast, the richness of disturbance-adapted species increased in more deforested landscapes. Forest-specialists were also distributed in a lower proportion of sites in more deforested landscapes, while habitat-generalists and disturbance-adapted species tended to show the opposite pattern. Taken together, these findings support previous studies conducted with a patch-landscape design (Boesing et al., 2018; Morante-Filho et al., 2020). Therefore, forest loss not only affects the bird community within forest patches, but the complete community across different land cover types in the studied landscapes.

As expected, our results indicate that forest loss causes a strong impoverishment of the bird assemblage in a human-modified mountain region. The diversity of forest-specialist species decreased in more deforested landscapes. This finding is consistent with the habitat amount hypothesis (Fahrig, 2013), which predict that species richness in equal-sized sample sites should increase with the total habitat amount (forest in this case) in the local landscape surrounding the sample sites. Interestingly, and contrary to what it has been documented in previous studies (Carrara et al., 2015; Morante-Filho et al., 2015), forest loss also decreased the diversity of habitat-generalist species, likely because these previous studies only surveyed forest patches and did not assess the changes in the bird community across different land covers. Thus, a novel contribution of our research is that forest loss not only impact forest species in forest lands, but also habitat-generalist species in the whole landscape mosaic. This implies that forest cover is not only important for forest species, but for a larger number of generalist species that also depend on the resources offered by the forest (e.g., food, refuge, nesting sites, Cornelius et al., 2008; Neuschulz et al., 2011). In other words, forest-specialist and habitat-generalist species may be more reliant on forest cover than disturbance-adapted species.

In fact, the diversity of disturbance-adapted species slight increased in more deforested landscapes. Some of the disturbance-adapted birds that dominated these deforested landscapes were Saltator coerulescens and Thryophilus Sinaloa, species that are known to do relatively well in human-modified lands (Billerman et al., 2020). Although for many native species, some elements of the landscape such as villages, crops, living fences, among others, can be ecological traps (i.e., habitats that animals prefer in which their fitness is lower than in other available options, Hale and Swearer, 2016), disturbance-adapted birds can proliferate in deforested landscapes because of their high capacities to obtain resources from different human-modified lands (Escobar-Ibáñez and MacGregor-Fors, 2015; Fehlmann et al., 2021).

Regarding the distribution of species, our findings indicate that, as expected, the proportion of sites occupied by forest-specialist species decreased with increasing forest loss. However, habitat-generalist species were widespread across the landscape. In fact, many these species can move among different land cover types (natural and anthropogenic) in search for supplementary or complementary resources (Dunning et al., 1992). Following the cross-habitat spillover hypothesis (Tscharntke et al., 2012), this movement capacity drives the whole-landscape community structure and associated processes. In birds, the movement capacity greatly differs among ecological groups. In our study, the relatively narrow distribution of forest-specialist birds in deforested landscapes is consistent with previous studies (Miranda et al., 2021), and could be related to the low dispersal capacities of species within this group (Robertson and Radford, 2009), and its specialization on forest resources. In contrast, habitat-generalist birds have a relatively high vagility, which allow them to use resources from different land cover types (Cadavid-Florez et al., 2020), occupying a widespread distribution across the landscape. Finally, the less noticeable effect of forest cover on the spatial distribution of disturbance-adapted birds may be due to the high vagility of this group, which allows these species to occur in different environments (Fehlmann et al., 2021), regardless of the degree of landscape deforestation.

We conclude that the amount of forest cover in the landscape is of paramount importance for predicting the diversity of birds in a human-modified mountain region. As forest loss has negative impacts on both forest-specialist and habitat generalist species, our findings support the idea that preventing forest loss and increasing forest cover is a top priority for preventing biodiversity loss (Arroyo-Rodríguez et al., 2020). Therefore, increasing forest cover in the landscape surrounding villages is good for preserving birds and their ecological function in the ecosystem, and for improving human well-being.

This work was supported by project PAPIIT-UNAM IN300119.

The authors declare that there is no conflict of interest regarding the publication of this article.