Habitat loss is the leading cause of the decline in biodiversity and ecosystem functioning worldwide (Brooks et al., 2002) as species diversity is dependent on the amount of habitat available (Watling et al., 2020). Around one million species may become extinct by the end of the 21st Century (IPBES, 2019), with half of these extinctions occurring in tropical forests (Pimm and Raven, 2000). However, which species will be lost remains unclear, and the understanding of this process, known as community disassembling, is a central theme in community ecology (Murphy and Weiss, 1992; Ostfeld and LoGiudice, 2003). Species may not be eliminated randomly, and some may be more vulnerable to extinction than others (Owens and Bennett, 2000), depending on their specific functional attributes (Almeida et al., 2018; Suding et al., 2005). A possible consequence of this non-random loss of species is the removal of unique ecological functions from the ecosystem, such as pollination or seed dispersal, leading to an impoverishment of the functional characteristics of ecological communities (Cadotte et al., 2011).

Traditional ecological indices used to characterize communities, based solely on the numbers of species and individuals, are not sufficient for a systematic understanding of the impacts on ecosystem functioning (Gagic et al., 2015). A more suitable approach would include functional traits that describe the ecological functions of the species (Violle et al., 2007). The concept of functional diversity, based on the ecological, morphological, and physiological traits of the species, has been a fundamentally important advance in the description of the variation found in ecological communities (Ding et al., 2013; Tilman, 2001; Valdivia et al., 2017). The functional structure of a community reflects the ecological and biogeographic processes that support the formation of assemblages, including biotic interactions and the abiotic factors that influence the characteristics of regional pools of species (Cadotte et al., 2011; Mouillot et al., 2013). These aspects of species diversity and functional structure also reflect the response of ecological communities to anthropogenic pressures on an ecosystem, and are thus essential for the effective conservation of habitat remnants (Mouillot et al., 2013).

The dominant processes that structure ecological assemblages involve a combination of abiotic, biotic and stochastic factors (Webb et al., 2002). Abiotic factors select for species and lineages with specific adaptations (i.e., functional traits) that favor their survival under certain conditions, which act as environmental filters. The result of this process is the formation of local assemblages with species that are more similar to one another (i.e., clustered) regarding their functional traits (Kraft et al., 2015). Biotic factors affect the structure of assemblages through biotic interactions, in particular, competition. Species with similar resource requirements (i.e., taxa that are functionally similar) will compete, leading to the exclusion of those with a lower competitive capacity or reduced population size, a process known as competitive exclusion by limiting similarity (MacArthur and Levins, 1967). Assemblages in which this process predominates tend to comprise species with more distinctive functional traits (i.e., over-dispersed; Kraft et al., 2015). Finally, neutral theory posits that stochastic processes (e.g., dispersion and ecological drift) will prevail over trait-dependent processes in determining community structure (Hubbell, 2001).

Most research that has focused on the relative influence of competition and environmental filters on community assembling has dealt with broad spatial scales. Most of the studies have identified a predominance of environmental filters over limiting similarity (Kraft et al., 2015; Mouchet et al., 2013; Weiher and Keddy, 1999). However, the findings of some studies indicates that the process of communities assembling is scale-dependent, with environmental filters being more influential on community structure at a regional scale, with species interaction becoming more important at the local scale (Mouchet et al., 2010; Smith et al., 2013). Local-scale studies that encompass the functional aspects of the species are necessary to shed light on the interplay of ecological processes that structure ecological communities. This perspective can provide profound insights into the processes that underly the disassembling of local ecological communities, which may be potentially valuable for the effective management of biodiversity (Almeida et al., 2016; Cadotte and Tucker, 2017; Ding et al., 2013).

Bird communities are an essential component of terrestrial ecosystems, where they provide critical ecological functions, such as pollination, seed dispersion, nutrient cycling, and the control of insect populations (Mahendiran and Azeez, 2018; Morante-Filho and Faria, 2017). Local bird diversity is related primarily to the distribution of feeding and nesting resources (Whelan et al., 2008). As habitat is lost, these local resources decrease (Watling et al., 2020) and new conditions are created (e.g., edge habitats or open areas), which may attract opportunistic species, increasing competition for the limited resources (Gascon et al., 2000; Pimm and Raven, 2000). The size of a forest patch has a positive effect on bird species richness (Watling et al., 2020), which is one of the parameters most affected by the reduction of the area of forest in tropical regions (Betts et al., 2019).

The understanding the effects of habitat loss on forest bird assemblages at a local scale requires data on multiple aspects of the diversity of functional traits, especially the traits linked to the essential factors that determine community assembly, such as the exploitation of different feeding resources and microhabitats by the bird species (Cadotte et al., 2011; Mazel et al., 2014). In the present study, we investigated different aspects of the taxonomic (i.e., species richness and uniformity) and functional (i.e., richness, uniformity, divergence, and dispersion) diversity of bird assemblages along a gradient of fragment size in the Atlantic Forest of northeastern Brazil. We tested whether the functional and species richness varied at the same rate along the fragment size gradient. We expected that smaller fragments have lower taxonomic and functional richness, given that these two parameters are naturally correlated (Mouchet et al., 2010). However, we hypothesize that, if the species that are absent from the smaller fragments have similar functional traits to the species that persist in these fragments (i.e., they are functionally redundant), the slope of the functional richness–area relationship (FAR) will be lower than that of the species–area relationship (SAR). This would indicate the limiting similarity is the leading process of community disassembling along the gradient of fragment size (Mouchet et al., 2010). We further predict that one consequence of this effect would be an increase in functional dispersion, given that, as the size of the fragments decreases, taxonomic and functional evenness will both increase, together with the functional divergence in the smaller fragments.

To test whether the estimates of FRic differed from the values expected by chance, we compared the observed values with the values determined by a matrix-swap null model (Manly and Sanderson, 2002; Mason et al., 2013), using the standardized effect size (SES) of Gotelli and McCabe (2002). The simulated assemblages were created by randomizing (999 permutations) all the species recorded in all the fragments, while maintaining the abundance and species richness observed in each fragment. We calculated the SES of functional richness (SESFRic) for each fragment, which measures the number of standard deviations that the observed metric is above or below the mean value obtained for each simulated community, based on the formula: SESFRic = Obs-Exp/SDexp, where Obs = the observed FRic, Exp = the mean of the 999 simulated values, and SDexp = the standard deviation of this mean (Gotelli and McCabe, 2002).

Under a normal distribution, 95% of the SESFRic values will be between -1.96 and 1.96 standard deviations of the mean. Values outside this confidence interval are considered to be significantly different from the null model, at α = 0.05 (Wittman et al., 2010). When SESFRic is significantly lower than the null model (SESFRic < -1.96), the assemblage is characterized by functional clustering, which indicates that environmental filters are the dominant assembling rule. When functional diversity is higher than expected (SESFRic >1.96), by contrast, the assemblage has species that are functionally more distinct from one another, which indicates that limiting similarity is the predominant force in the community (Mouchet et al., 2010).

We obtained 1896 records of 92 bird species in a total of 192 McKinnon lists. The observed species richness corresponded to 91% (∼99sp. ± 6.43SD), of that estimated by the Chao 1 procedure, with a total of between 29 and 77 species being recorded per fragment. Sample completeness was similar in all fragments, at ∼83–89% (Table 2).

The models of linear regression between the diversity indices and the area of the fragments revealed that both the SAR and the FAR were positive and significant (Fig. 2a). By contrast, uniformity (J’) and functional dispersion (FDis) were related negatively with the area of the fragment (Fig. 2c, d). Neither functional divergence (FDiv) nor functional uniformity (FEve) were related significantly to fragment area (Fig. 2e, f). The ANCOVA used to compare the slopes of the SAR and FAR models revealed that functional richness decreases more slowly with fragment size than species richness, i.e., with a significantly lower slope (F = 9.7302, p = 0.0066; Fig. 2a).

Fig. 2.

Linear regression plots with 95% confidence intervals (shaded areas) showing the predicted relationship between the diversity indices and the forest spot area gradient. S = Species Richness, J = Pielou’s evenness, FRic = Functional Richness, FEve = Functional evenness, FDiv = Functional divergence, FDis = Functional dispersion, SESFRic = Standardized Effective Size of Functional Richness. * = p < 0.05, ** = p < 0.01, *** = p < 0.001.

(0.41MB).

Seven assemblages had a SESFRic higher than the upper limit of the confidence interval, which indicates that their FRic values were significantly higher than those predicted by the null model (Fig. 3, Table 2). The linear regression between the SESFRic values and the area indicate a significantly negative relationship (R² = 0.67, p = 0.0022, Fig. 2b). That is, although larger fragments had more species than smaller ones, they are functionally less diverse than expected according to their species richness. This reflects the accumulation of redundant species in larger areas, as also indicated by the result of the ANCOVA, above (Fig. 2a).

Fig. 3.

The FRic density curves recorded for each study fragment (null distribution model) in comparison to the observed functional richness (vertical dashed line). The light gray area below each curve represents the confidence interval of the null model (-1.96

(0.13MB).

We found that fragment size affects both the taxonomic and functional diversity of the bird assemblages surviving in remnants of the Atlantic Forest in northeastern Brazil. More notably, we found that, as the area of the fragment decreases, functional richness decreases at a rate slower than that of species richness, indicating that the species that are absent from the smaller fragments share functional traits with those that persist in these fragments. A high proportion (70%) of the assemblages had significantly higher functional richness than expected according to their species richness, i.e., their SESFRic. This metric was also related negatively with fragment area, indicating that the dissimilarity among species tends to increase in smaller fragments. Although other features of a forest fragment may affect the local diversity, including its shape, isolation, and the characteristics of the matrix (Barros et al., 2019; Watling and Donnelly, 2006), the effects of these factors on bird diversity was determined by their collinearity with the area of the fragment (Supplementary Material). Future studies should investigate these effects on bird diversity more systematically while controlling for the dominant effect of fragment area.

These findings suggest a high level of interspecific competition in the community (Cadotte and Tucker, 2017; Sobral and Cianciaruso, 2015), which is consistent with the hypothesis that limiting similarity is the dominant assembling rule in these bird assemblages (MacArthur and Levins, 1967; see also Almeida et al., 2016; Mouchet et al., 2010). This conclusion is also supported by other results of the present study. For example, the intensification of competition among species will be expected to exclude the less competitive taxa, while leaving the more abundant ones, which will tend to reduce the disparities in the relative abundance of the more persistent species (Cadotte and Tucker, 2017; Hillebrand et al., 2008; MacArthur and Levins, 1967). We also recorded a significantly higher species evenness (J’) in the smaller fragments, which indicates that the less abundant species are more likely to be excluded, which is consistent with the role of competition as the principal driver of local diversity (Cotgreave, 1994; Hillebrand et al., 2008). One other important finding is the significant negative relationship found between functional dispersion (FDis) and fragment area, which indicates that the few species that persist in smaller fragments are ecologically more different from one another than expected by chance (Karadimou et al., 2016; Mouchet et al., 2010).

Functional evenness (FEve) and divergence (FDiv) were not associated with fragment area, although these metrics do provide some useful insights. Functional evenness expresses how evenly the abundance of individuals is distributed in functional space, and this metric decreases as the number of functionally unique and rare species increases (Sitters et al., 2016). The generally low values of FEve recorded in the present study indicate a lack of functionally rare species in the study area, even in the larger fragments (Mason et al., 2005; Villéger et al., 2008). At less than 250 ha, in fact, even the largest fragments surveyed in the present study may be too small to support a full local bird assemblage, which implies that local extinctions may already have occurred at all sites. The FDiv, in turn, describes how the traits of the most abundant species differ from those of the other taxa (Villéger et al., 2008). Low FDiv values reflect the recurrence of common species, which are usually the more abundant in the fragments. In the present study area, these species include Coereba flaveola (Linnaeus, 1758), Columbina talpacoti (Temminck, 1810), and Tyrannus melancholicus (Vieillot, 1819), which were the most abundant species in five, six, and seven fragments, respectively.

The predominance of limiting similarity as a result of habitat size is consistent with that observed in other groups of organisms (Mason et al., 2013, 2007; Slingsby and Verboom, 2006; Valdivia et al., 2017) including birds from islands (Ding et al., 2013), dense rainforest (Almeida et al., 2016), different vegetation types (Lee and Martin, 2017), and at varying spatial scales (Sobral and Cianciaruso, 2015). Even so, little is known about how habitat loss affects the functional structuring of tropical forest bird communities (Bovo et al., 2018; Si et al., 2017; Ulrich et al., 2016). Previous research has emphasized the assembling, i.e., the formation, of ecological communities, with substantial theoretical and empirical developments (e.g., Diamond, 1975; Morin, 2011). However, as habitat modification and other threats intensify, it remains to be seen whether analogous processes – i.e., limiting similarity and habitat filtering – drive the disassembling of ecological communities in an equivalent way. In other words, does the subtraction of species from an assemblage following impact follow a similar sequence to that observed when the community is formed?

Perhaps even more importantly, the results of the present study have a number of implications for the conservation and management of the local bird communities and the Atlantic Forest remnants they inhabit. Firstly, our results underscore the well-known negative effect of forest loss on bird species diversity (Brooks et al., 2002; Watling et al., 2020). The preservation of both species diversity and the redundancy of the ecological functions of ecologically-similar species tends to require the maintenance of much larger tracts of forest (Cadotte et al., 2011). However, our findings do indicate that the presence of a few larger fragments in the landscape may not reduce the chances of local extinction in the smaller remnants. Local extinctions in the smaller fragments might nevertheless be avoided by adopting measures that ensure a reduction in competition in these remnants. These measures may include not only an increase in the connectivity between the fragments within the landscape, but also between landscapes, that is, by ensuring greater species dispersal, which would lower local extinction risk (Newmark et al., 2017). Birds also have essential ecological roles in forest ecosystems, including seed dispersal, pollination, and the control of insect populations, which means that the preservation of bird diversity is fundamental to the maintenance of forest dynamics, including plant reproduction and recruitment, which are essential for the regeneration of the habitat (Bovo et al., 2018; Mahendiran and Azeez, 2018; Morante-Filho and Faria, 2017).

Overall, then, we found that the reduction in species diversity with decreasing fragment size in a Atlantic Forest landscape involves primarily the loss of species with similar ecological functions, which indicates the influence of interspecific competition as the process driving the local community disassembling. In this context, we would point out that the analysis of species richness alone may not be sufficient to guarantee the understanding of shifts in bird species diversity in response to habitat impacts, in particular the loss of taxa (Gagic et al., 2015). Given this, a functional perspective would appear to provide a more conclusive perspective on the interactions among species and their environment (Mason et al., 2013; Tilman, 2001; Violle et al., 2007). Finally, our study underscores the need for the maintenance of large and interconnected fragments of Atlantic Forest, at both local and regional scales, in order to guarantee the preservation of bird species diversity, the functional diversity of the community, and the stability of ecosystem functions.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

This study is part of HSO's master degree, which was financed in part by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - Brasil (CAPES) - Finance Code 001. SFG and SFF by Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq 302854/2019-3 and 310852/2017-0, respectively). This project was also supported by CAPES/FAPITEC grants (#88881.157451/2017-01 and #88881.157961/2017-01) provided to SFF and SFG. SFG is a member of the National Institute of Science and Technology - Ecology, Evolution, and Conservation of Biodiversity (INCT-EECBio, CNPq/FAPEG, grant 380733/2017-0).

The authors confirm that the data supporting the findings of this study are available within the article [and/or] its supplementary materials.

This article does not contain any studies with human participants or animals performed by any of the authors.

Helon Simões Oliveira: Conceptualization, Methodology, Formal analysis, Investigation, Data curation, Writing - original draft, Writing - review & editing. Sidney F. Gouveia: Data curation, Validation, Formal analysis, Project administration, Writing - review & editing. Juan Ruiz-Esparza: Methodology, Writing - review & editing. Stephen F. Ferrari: Validation, Writing - review & editing, Supervision.