It is undeniable that we are experiencing a massive extinction event that surpasses the five previous Earth’s history mass extinctions (Barnosky et al., 2011). The current wave of biodiversity loss is mainly occurring due to a synergy among processes such as habitat loss, habitat modification, and climate change (Barnosky et al., 2011; Carmona et al., 2021). Several studies have reported that habitat loss leads to a diversity decrease in many taxa, including vertebrates, insects, and plants (Pardini et al., 2010; Rigueira et al., 2013; Spiesman and Inouye, 2013). The effects go beyond the loss of species, they also include evolutionary history and ecological processes reducing the ability of forests to sustain themselves in the long term (Mcdowell et al., 2020). However, the majority of these studies focused on species richness (SR), and there is a reasonable consensus that species richness per se often brings limited information on evolutionary history and function (Swenson, 2011). The integration of phylogenetic and functional diversity provides additional insights about ecological processes and might contain valuable information regarding ecosystem function (Cadotte et al., 2011). Thus, more recently, it is becoming common to include phylogenetic and functional diversity to capture important biodiversity dimensions that are also undoubtedly relevant to understand plant community assembly in human modified-landscapes.

In terms of plant phylogenetic diversity (PD) the outcomes are divergent so far. Negative effects of anthropogenic disturbances on phylogenetic relationships, either by reducing diversity or altering the structure by increasing phylogenetic clusters were recorded (Andrade et al., 2015). Otherwise, some studies detected an increase in PD or a phylogenetic overdispersed structure in disturbed areas, largely due to the incorporation of non-native species (Liu, 2016). However, there is an increasing number of studies finding that PD can be maintained in several tropical landscapes including highly fragmented forests (Santo-Silva et al., 2018). For functional diversity (FD), studies have reported that anthropogenic disturbances can cause either no effects (Flynn et al., 2009), negative (Suárez-Castro et al., 2020; Zambrano et al., 2020) or positive effects (Döbert et al., 2017) on plant community functional responses.

However, there is a clear bias towards studies focusing on adult tree assemblages. In this context, adult trees might reflect the accumulated responses to historical changes because they are prone to remain in the landscape for a longer time (Rigueira et al., 2013). On the contrary, recently established individuals, such as seedlings and juveniles, exhibit a higher sensitivity to habitat loss due to recent disruptive effects in reproductive, dispersal, and establishment processes influenced by deforestation (Rigueira et al., 2013). Therefore, distinct responses to anthropogenic disturbances are expected among life-stage groups, which can lead to divergent responses to richness, evolutionary history and ecological functions (Ernoult et al., 2006). Anthropogenic disturbances, can change local microclimatic conditions such as solar radiation available in the forest understory (Reis et al., 2021), which influence the establishment and survival of juvenile plants (Cerqueira et al., 2021). The younger assemblages represent the future of the forest, and can provide more accurate information on the consequences of forest loss in the long term (Mcdowell et al., 2020). Thus, although this knowledge is necessary for conservation decisions, these studies are still scarce.

Here we contrast adult and juvenile tree assemblage’s responses to forest cover loss on taxonomic, phylogenetic, and functional diversity. We predicted negative effects of forest cover loss mainly in taxonomic and functional diversity due to the increasing evidence suggesting maintenance of adult tree phylogenetic diversity in anthropogenic landscapes (Santo-Silva et al., 2018). We also predict that the impacts will be higher on juvenile than adult assemblages, because recruiting processes are impaired in the study area as forest cover is lost (Benchimol et al., 2017).

We recorded a total of 1950 adults from 507 species and 3397 juveniles from 441 species. After the exclusion of 51 species for which we did not find molecular information, we ended up with 1868 adults from 457 species and 52 families, and 3183 juveniles from 400 species and 54 families. The three dimensions of diversity (SR, PD and FD) were highly correlated, for both adults and juvenile assemblages (rs ≥ 0.7), thus we presented only SR results (Fig. S1 and S2). Phylogenetic metrics were in general not correlated with the equivalent functional metrics for both life-stage (PD × FD; pMPD × fMDP; SES PD × SES FD; SES pMPD × SES fMDP) (Figs. S1 and S2). For both adults and juveniles, FD was not correlated with the other functional metrics (SES FD, fMPD and SES fMPD), the same pattern was found for phylogenetic metrics (SES PD, pMPD and SES pMPD).

Specie richness (SR) differed between both ontogenetic stages that also showed an interaction with forest cover (Table 1). Both ontogenetic stages responded negatively to forest cover loss, with a stronger richness decrease for juvenile than adult assemblage (interaction term P < 0.001; adult slope: 0.42 and juvenile slope: 0.96; Fig. 1 and Table 1). In general, the phylogenetic and functional metrics were relatively similar between adult and juvenile assemblages (Table 2). Except for the mean phylogenetic distance (pMPD) that was higher in adult than juvenile assemblages, but neither ontogenetic stage responded to forest cover loss (Table 1 and Fig. 2). The average values of SES PD and SES pMPD were similar between adult and juvenile assemblages and did not respond to forest cover loss (Tables 1 and 2). Addressing functional metrics, mean functional distance (fMPD) was similar between ontogenetic stages and did not respond to forest cover, but juvenile and adult assemblages respond differently to forest cover loss (significant interaction, Table 1). Whereas, SES fMPD decreased as forest cover decreases at the landscape-scale, but only for adult assemblages (Table 1 and Fig. 2). Additionally, roughly 50% of the forest remnants presented a more clustered phylogenetic and functional structure than expected by chance (negative SES pMPD and SES fMPD values) in both life-stages (Fig. 2). Individual functional traits of wood density (CWM) were not affected by forest cover loss in both life stages (Table 3). However, the CWM for fruit diameter was different between adult and juvenile (Table 3). Also, they respond differently to forest cover variation (significant interaction, Table 3), with a greater decline in fruit diameter of species within the juvenile stage (adult slope: −0.003 and juvenile slope: −0.613).

Fig. 1.

Relationship between forest cover amount and species richness of adult and juvenile tree assemblages in 20 forest remnants from the Atlantic Forest, Bahia, Brazil.

(0.09MB).

Fig. 2.

Relationship between forest cover amount and phylogenetic and functional diversity metrics of adult and juvenile tree assemblages in 20 forest remnants from the Atlantic Forest, Bahia, Brazil. SES PD = standard effect size of phylogenetic diversity, SES FD = standard effect size of functional diversity, pMPD = phylogenetic mean pairwise distance, fMPD = functional mean pairwise distance, SES pMPD = standard effect size of phylogenetic mean pairwise distance, SES fMPD = standard effect size of functional mean pairwise distance.

(0.41MB).

Our results unveil differences in adult and juvenile tree responses to forest cover loss, which resulted in a marked decline in tree species richness in the Atlantic Forest, more sharply for juvenile than adult assemblages. Surprisingly, the phylogenetic and functional diversity, using metrics not correlated with species richness, were maintained along the gradient of forest cover for both ontogenetic stages. The only exception was the increase in SES fMPD with decreasing forest cover, but only for adult assemblage indicating over-dispersion of functional traits in deforested landscapes. We revealed that the evolutionary value of disturbed forests is high and possibly those forests are also partially maintaining ecosystem functions in present and future plant communities.

The pervasive effects of forest cover loss on plant species richness were expected and have been previously documented elsewhere (Benchimol et al., 2017; Montoya et al., 2008). The direct comparison of adults and juveniles showed distinct responses to landscape-scale forest cover loss with stronger decline in juvenile, probably due to the shorter life span of juvenile individuals. The drastic decrease in juvenile assemblage may be a reflection of demographic bottleneck effects occurring in some previous stages such as reproduction, dispersal and establishment. As previously shown in our landscapes, where forest cover loss impairs the regeneration process thus compromising the future tree community (Benchimol et al., 2017). We also showed that SR was a stronger predictor of PD and FD metrics (correlation ≥ 0.7), evidencing that these metrics are not adequate to measure the phylogenetic and functional diversity. Thus, we recommend using metrics standardized by effect size (SES) or Mean Pairwise Distance metrics. It is expected that in highly diverse regions, PD increases at a similar rate as SR, and thus, PD tends to be with SR (Fjeldså, 1994).

Conversely, phylogenetic diversity and structure (pMPD and SES pMPD) were maintained along the forest cover gradient. This result has been supported by previous studies on adult tree assemblages (Santo-Silva et al., 2018) and suggests that traits related to high susceptibility to disturbance are convergent or have a low phylogenetic signal. Some ecological traits that make species more susceptible to environmental changes, such as sensitivity to light incidence and drought tolerance, are not exclusive to phylogenetically related species (Niinemets and Valladares, 2006). Interestingly, the mean phylogenetic diversity was higher in adults than in juvenile assemblage and phylogenetic diversity and structure of juveniles were not affected by forest cover loss. Because juvenile assemblages might reflect current responses to anthropogenic disturbances we expected a greater phylogenetic impoverishment on this ontogenetic stage, as previously recorded for chronic disturbances (Ribeiro et al., 2016). On the contrary, our results revealed that there is a high variation in the values of phylogenetic diversity and structure between forest remnants, both for juveniles and adults. This variation is not influenced by forest cover loss, corroborating a recent study where plant phylogenetic diversity was little influenced by landscape conditions (De Pauw et al., 2021).

Regarding functional diversity, forest cover loss affected the SES fMPD only for adult assemblage. This indicates that after controlling for richness, an over-dispersion of functional traits is observed in deforested landscapes. Contrary to our expectation, this result shows that adult tree functional diversity is higher in species-poor fragments, with few species dispersed across a wider breadth of functions. Yet, functional divergence in tree assemblage has been previously reported (De Pauw et al., 2021; Döbert et al., 2017; Rocha-Santos et al., 2019). This pattern indicates that while tree assemblages selectively lose some functional traits due to forest cover loss, these traits are replaced by divergent traits favored by anthropogenic disturbances. Nonetheless, when the traits were analyzed individually, we show that forest cover did not influence wood density independent of the life-stage. However, forest cover affected fruit diameter, with the assemblage of juvenile species showing potentially smaller fruits than adults in more deforested landscapes.

It is important to highlight that trait selection can affect the functional metric results, therefore, we selected important traits that are related to some ecological processes (such as regeneration, dispersal, pollination, carbon stock). Also, a massive effort to collect trait data was performed, but we are still far from having a comprehensive coverage. Although we acknowledge that trait selection and its coverage could change part of the results reported here, we emphasize that this is an inherent problem of FD metrics. Indeed, although our forest fragments within highly forested landscapes were relatively well preserved, these forests suffered from disturbances such as selective logging and hunting, therefore they are unlikely considered as pristine. Thus, we suggest that additional studies with different traits and functions, and local levels of disturbance would be necessary to generalize to plant assemblages worldwide.

To conclude, our results reveal a high conservation value of disturbed forests. In terms of evolutionary history and functional trait maintenance, this is undoubtedly true. Although some diversity metrics differed between life-stages, they did not change as a function of forest cover loss, suggesting that extinction debt might not be masking long-term effects of deforestation on phylogenetic and functional diversity. However, forest cover loss in the southern Bahia has not only noticeable effects on tree assemblages, but also impacted on ecological processes such as fruiting, seed dispersal and herbivory (Benchimol et al., 2017; Dodonov et al., 2016; Hambuckers et al., 2020). These results combined can reinforce that the profound effects of deforestation already detected on richness are not mirrored by the phylogenetic and functional metrics evaluated here, regardless of life stage. Taxonomic diversity, due to its simplicity, was suggested as one effective index of biodiversity loss in anthropogenic landscapes, as thresholds in taxonomic diversity precede the loss of phylogenetic diversity (Boesing et al., 2018). In addition, we recommend using metrics not correlate with species richness to improve phylogenetic and functional conclusions. Thus, the noticeable conservation value of deforested landscapes found here in terms of function and phylogenetic diversity reinforces the need and importance to conserve every remnant of tropical forest. However, the combined approach integrating the different dimensions of biodiversity and different life-stages used here reinforce that increasing forest cover at the landscape scale is urgent to safe-guard tree taxonomic diversity.