An evolutionary ecomorphological perspective on the assembly of a neotropical bat metacommunity

The evolution of the bat skull has been extensively studied at broad spatial scales. While ecomorphological partitioning of niches has been extensively analyzed in macroevolutionary studies, little is known about the interaction of wild phenotypes with novel ecological pressures to determine species co-occurrence. Here, we tested the influence of size, trophic guild, and foraging behavior on the cranium and mandible shapes of 32 co-occurring bat species. We used 2D geometric morphometrics and phylogenetic comparative methods for multivariate data to test the effects of foraging behavior, trophic guild, and size on the shape of cranium and mandible. We also tested for phylogenetic signal on the shape and size of cranium and mandible. Our results show that closely-related species were clustered together in the morphospace. Cranium allometry followed a common trajectory, probably related to olfactory-visual senses, and not trophic guild and foraging behavior. However, mandible allometry followed a unique trajectory for each group, suggesting differential pressures related to trophic guild and foraging behavior. Coexistence among frugivore stenodermatines is apparently achieved because they partition ecological niches by varying cranium and mandible size rather than their shapes. These findings suggest a convergence in cranium and mandible shapes for insectivorous bats, which may be related to the hardness of food resources.


Introduction
Studies integrating ecology and evolution at the level of metacommunities can improve our understanding about stochastic (ecological drift, dispersal) and deterministic (niche selection) processes that promote diversity (Urban et al. 2008;Vellend 2016).Ecomorphological approaches provide a link between functional morphology, which analyses scales (see Swartz et al. 2003), with few studies being conducted at regional scales focusing on form-resource relationship in species-rich communities, while also considering species shared ancestry.
Another key factor in morphological evolution is foraging behavior, because food processing depends on bite force (Nogueira et al. 2009;Santana et al. 2010Santana et al. , 2012)).For example, the evolution of skull shape in phyllostomid bats seems to be related to their trophic guild and foraging mode, because they modulate bite force in response to physical characteristics of prey (Dumont 1999;Santana and Dumont 2009).Nectarivore bats have neither to capture prey in movement nor process hard prey.As a result, they have long and narrow rostra with low bite force.Frugivores (e.g., Stenodematinae) that need to transport and process hard fruits have an equal length between the rostrum and braincase, and a robust mandibular ramus.Animalivorous or insectivorous that have to process hard items, such as small vertebrates or insects with a hard shell, need a strong bite, hence they have short rostra, robust braincase and mandibular ramus, usually with prominent sagittal crest (Nogueira et al. 2009;Santana et al. 2010;Ramírez-Fráncel et al. 2021).A previous study (Ramírez-Fráncel et al. 2021) found that obligate (Peropteryx macrotis and Molossops temminckii) and non-obligate (Noctilio albiventris) insectivore bats had different bite forces, which mainly depended on cranium size and shape.Therefore, where and how species capture prey potentially interact with their diet to determine cranium and mandible shape.Despite the importance of foraging behavior and trophic guilds in determining species membership in bat communities, few studies have evaluated both factors together using modern phylogenetic comparative methods.
Allometry is the relationship between shape and size (Klingenberg 2016).When analysed at macroevolutionary scales, allometry is useful to understand how selection might have shaped developmental processes to produce morphological novelty (Pélabon et al. 2014).If size differs between clades and this variation is related to ecological traits, like diet, it may indicate that clades are under differential selective pressures (Adams 2014a).However, allometries seem to have low rates of evolution, which in turn constrain phenotypic evolution to fixed trajectories (Pélabon et al. 2014).Natural selection may have contributed to the evolution of size in several instances, especially in functional characteristics (Voje and Hansen 2012;Pélabon et al. 2014).Nonetheless, few studies have analysed how allometric trajectories of the skull in groups of species with distinct ecologies vary at a regional scale, which might have profound consequences for species coexistence and eco-evolutionary feedbacks (Weber et al. 2017;Wilcox et al. 2018).
Many studies have analysed macroevolutionary processes in bats, especially addressing skull morphology and foraging behavior (Norberg and Rayner 1987;Salsamendi et al. 2012).Most often they have focused on either one genus (Artibeus, Hedrick 2021;Carollia, López-Aguirre et al. 2015;Myotis, Ospina-Garcés and de Luna 2017) or on Phyllostomidae, the family comprising the greatest variation in morphology and ecology (Freeman 1995;Monteiro and Nogueira 2011;Dumont et al. 2012;Santana et al. 2012).Recently, macroevolutionary trends were evaluated encompassing three large bat families: Phyllostomidae, Vespertilionidae, and Molossidae (Hedrick and Dumont 2018).These studies have greatly contributed to our understanding of' evolutionary patterns on a global scale, but the processes that take place on a regional scale have been little addressed under an ecomorphological perspective.
Here, we tested the influence of size, trophic guild, and foraging behavior on the evolution of the cranium and mandible of bats on a regional scale.We hypothesized that: (1) species will be distributed in the phylomorphospace depending on ecological factors and phylogenetic relationship, in a way that species with less evolutionary divergence will have greater morphological similarity (Blomberg and Garland 2002;Rossoni et al. 2017;Adams and Collyer 2019); (2) the allometry of the cranium will be under stabilizing selection; consequently, we expect a common trajectory, given that recent studies have shown significant influence of sensory factors, such as smell and vision on cranium evolution (Arbour et al. 2021); (3) the allometry of the mandible will be under differential selection; consequently we expect unique trajectories related to trophic guild, given that the mandible has a strong relationship with trophic function, as found in Phyllostomidae, Molossidae, and Vespertilionidae (Nogueira et al. 2009, Hedrick andDumont 2018); 4) the mean and variance of the cranium and mandible shape will differ between trophic guilds and foraging behavior.

Geometric morphometrics
We analyzed 233 specimens of 32 bat species recorded in the Serra da Bodoquena, a karstic region in the southern Brazilian Cerrado (Sallun-Filho et al. 2010).We included in this study all species known to occur in the region, with 28 species recorded during extensive samplings at 20 sites (Online Resource 1: Fig. S1) in the wet and dry seasons (in 2015, 2016, 2017, 2019 and 2021), which included > 2,700 mist-netted bat individuals.Members of the following families (and subfamilies) were analyzed: Phyllostomidae (Carolliinae, Desmodontinae, Glossophaginae, Lonchophyllinae, Micronycterinae, Phyllostominae, and Stenodermatinae), Emballonuridae (Emballonurinae), Molossidae (Molossinae), Noctilionidae, and Vespertilionidae (Myotinae and Vespertilioninae).The mean and standard deviation of the number of specimens per species was 7.31 ± 4.98 (range: 1-19) (Online Resource 1: Tables S1-S2).This number of specimens per species is usually adequate to describe interspecific variation in size and shape of biological structures (Cardini et al. 2015).All specimens were adult males obtained from the Zoological Collection of the Universidade Federal de Mato Grosso do Sul (ZUFMS) (Online Resource 1: Tables S1-S2), except for Pygoderma bilabiatum and Micronycteris microtis, for which we obtained voucher specimens from the Natural History Museum of Universidade Federal de Lavras, and Eptesicus brasiliensis for which we obtained high resolution pictures of the cranium and mandible from the Mammals of Espírito Santo virtual guide (https://lamab-ufes.wixsite.com/mamiferosdoes).For every specimen, we obtained 2D images with scale for the cranium and mandible using a Zeiss Discovery V.20 stereoscope microscope.
To obtain cranium and mandible shapes for each species, we placed landmarks and semi-landmarks onto these structures following previous studies with Phyllostomidae (Nogueira et al. 2009), Vespertilionidae (Ospina-Garcés and de Luna 2017), Molossidae (Freeman 1981), and Noctilionidae (Romero 2011) (Online Resource 1: Figs.S2-S3, Tables S3-S4).The positioning of landmarks was performed with the software TPSUtil v 1.83 and TpsDig v 2.32 (Rohlf 2015).When individuals did not have the structure where a given landmark should have been positioned, we estimated their locations from other conspecific specimens containing that structure, using the R package geomorph (Adams et al. 2022).Then, we used generalized Procrustes analysis (GPA) to remove effects of rotation, translation, and size.This analysis minimizes the Procrustes distance between corresponding landmarks on each specimen and computes a mean to each sample (Zelditch et al. 2012).We used centroid size (Zelditch et al. 2012) as a measure of cranium and mandible sizes for each species.

Trophic guild and foraging behavior
Species were categorized into five trophic guilds, following Kalko et al. (1996): animalivores (Ani), which include sanguivores and carnivores that prey mainly on vertebrates, but can occasionally prey on insects; insectivores (Ins), which exclusively or heavily prey on insects; frugivores (Fru), whose diet is based mostly on fruits; nectarivores (Nect) that feed mostly on nectar; and omnivores (Omn) that feed on a combination of several food sources.We combined the classifications of Kalko et al. (1996) and Denzinger and Schnitzler (2013) for habitat use, foraging behavior, and capture mode into three categories: active foragers (ACT), which are edge and open area insectivores, and animalivores (e.g., D. rotundus and N. leporinus); passive-active foragers (PAS-ACT), which are narrow space frugivores, nectarivores, and omnivores; passive foragers (PAS), which include narrow space animalivores and insectivores (Phyllostominae), the difference in relation to PAS-ACT is that these species uses sensory cues provided by the prey itself (Thiagavel et al. 2020).

Testing the effects of trophic guild and foraging behavior on cranium and mandible shape
We used the phylogeny of Shi and Rabosky (2015) for subsequent analyses.This phylogeny was constructed with a supermatrix of mitochondrial and nuclear sequences using Bayesian inference and contains data for 812 bat species (~ 57% of the currently known species).We pruned this tree to contain only the 32 species found in our study region.Four species absent in this phylogeny were replaced by phylogenetically equivalent ones (sensu Pennell et al. 2016): Lonchophylla dekeyseri was replaced by L. mordax (Woodman and Timm 2006), Micronycteris sanborni by M. minuta (Morales-Martínez et al. 2021), Artibeus cinereus by Artibeus bogotensis (Agnarsson et al. 2011), and Pteropteryx macrotis by Cormura brevirostris (Lim and Dunlop 2008) (Fig. 1).To visualize the species distribution in the morphospace of the cranium and mandible, we first performed a phylogenetic principal component analysis (pPCA) with the coordinates of the mean shape of each species (Revell 2009) in the R package geomorph (Adams et al. 2022).We built a phylomorphospace (Adams and Collyer 2019) using the first two PCs (Online Resource 2).
To test whether the cranium and mandible shapes are influenced by size (continuous predictor), trophic guild (five-level factor), and foraging behavior (three-level factor), we used a phylogenetic generalized least squares (PGLS) model in the R package RRPP (Collyer and Adams 2018).This method assumes that the phenotype evolved under a multivariate Brownian Motion model (Adams 2014a).Our model included the interaction between the two factors (shape ~ size + guild * behavior).As factors were significant, we performed a post hoc analysis to test differences between the mean distances and shape disparity for each factor level combinations (e.g., Ani/ACT, Ani/PAS, Fru.PAS/ACT, Ins/ACT).We acknowledge that some factor levels (e.g., omnivores and passive foragers) have few representative species.However, this seems to be a common pattern in Neotropical bat communities (e.g., Aguirre et al. 2002), which usually have a higher proportion of

Phylogenetic signal, rate heterogeneity, and model fitting for shape and size
To estimate and test for phylogenetic signal of the mean shape and size of the cranium and mandible, we used Blomberg's K statistic adapted for multivariate data (Adams 2014b).A value of Kmult lower than 1 indicates that species are less similar than expected under Brownian motion (BM), and greater than 1 indicates that species are more similar than expected under a BM model (Adams 2014b).Analysis was conducted in the R package geomorph (Adams 2014b).
To test for differences in evolutionary rates of cranium and mandible shapes between trophic guilds, we estimated σ 2 for each guild using a Maximum Likelihood method (Adams 2013).This method fits a Brownian Motion model of evolution to the shape data and then use permutation to test if the observed ratio between evolutionary rates of groups differs from a random expectation.This analysis both complement the phylogenetic signal measure and frugivores, nectarivores, and insectivores than carnivores and omnivores.The presence of species with distinctive phenotypes in a phylogeny is a common problem in comparative biology (Uyeda et al. 2018) and PGLS seems susceptible to high Type I Error when dealing with singular evolutionary events, such as rate heterogeneity and shifts in adaptive optima.Most solutions suggested by Uyeda et al. (2018) only apply to univariate phenotypes.However, this will only be problematic in our case when conducting post hoc hypothesis testing.When interpreting the results, we deemphasize significance, while focusing on effect size of each factor.To test for model adequacy, we diagnosed residuals using visual inspection of quantile-quantile and residuals vs. fitted plots.Residuals were normally distributed and only slightly deviated from homogeneity of variance, being a little more problematic in the model for mandible, than for the cranium (Online Resource 3).There seems not to be any outlier or leverage influencing parameter estimation.discolor, while species with elevated braincases and prominent sagittal crest had positive scores, as in Noctilio (Fig. 2).
Negative values of PC1 for mandible shape were associated with species with short and robust mandibles, such as D. rotundus, while positive values were related to species with long and gracile mandibles, as in nectarivores.PC2 was related to the height of the ascending branch, with negative values showing short ascending branches and robust and broad ramus, such as in animalivores Noctilio and Desmodus, and active insectivores that have strong muscular insertion.Positive values of PC2 were related to the variation of the coronoid process, which tends to be high in animalivores (Phyllostominae) and frugivores (Stenodermatinae and Carolliinae) (Fig. 3).

Effects of trophic guild, foraging behavior, and size on cranium and mandible shape
The interaction between trophic guild and foraging behavior was significant for both models, explaining about 6% of the variation in cranium and 5% of mandible shape (Table 1).Trophic guild alone explained a larger proportion of the cranium and mandible shapes (~ 40%; Z ≥ 5) than foraging behavior (~ 10%; Z = 3.3), while size had a slightly greater effect on the cranium (12%; Z = 3.4) than mandible (8%; Z = 3) shapes (Table 1).There were no differences in the mean (centroid position) between ecological groups for neither the cranium nor the mandible, except for the difference between active and passive animalivores for the mandible (Online Resource 3).However, we found significant differences between groups in terms of morphological disparity (Tables 2 and 3) for both the cranium and mandible.The groups with the greatest morphological disparity were active animalivores (0.003, 0.005) and active insectivores (0.002, 0.002) for the cranium and mandible shape, respectively (Online Resource 1: Fig. S4; Tables 2 and 3).

Phylogenetic signal, evolutionary rates, and model fitting
Closely-related species had similar cranium (K mult = 0.87; Effect size = 6.79;P < 0.001) and mandible shape (K mult = 0.99; Effect size = 7.84; P < 0.001).We also found a high phylogenetic signal for cranium (K mult = 1.07;Effect size = 4.21; P < 0.001) and mandible sizes (K mult = 0.80; Effect size = 2.75; P = 0.005).Evolutionary rates of the cranium and mandible did not differ between trophic guilds.The observed ratio for the cranium was 1.9446 (P = 0.7722) and 4.4884 (P = 0.2333) for mandible shape rates.The best fitting model to our data was BM for the cranium shape and OU or EB for the mandible shape.However, models for the mandible were not distinguishable based on ΔGIC provide a test of assumptions for the PGLS model, addressing rate heterogeneity in our tree.Analysis was conducted in the geomorph R package.
To further explore evolutionary patterns of cranium and mandible shape, we fitted alternative evolution models to the data using multivariate generalized least square models (Clavel et al. 2019).We included Brownian motion (BM), Ornstein-Uhlenbeck (OU), and Early burst (EB) models in this exercise.Then, we calculated the Generalized Information Criterion (GIC) for each model, which is similar to Akaike Information Criterion to compare model fit.Analysis was conducted in the R package mvMORPH (Clavel et al. 2015).

Evolutionary allometry
To test for evolutionary allometry between cranium and mandible shapes and size and if these differ between trophic guilds and foraging behavior, we built two models with interaction (shape means ~ size means * guild * behavior) and additive effects (shape means ~ size means + guild * behavior) that respectively test for unique and common allometry among ecological groups in the R package RRPP (Collyer and Adams 2018).We diagnosed model residuals using visual inspection of quantile-quantile and residuals vs. fitted plots.Residuals were normal and had no major deviance from homogeneity of variance (Online Resource 3).
These two models allowed us to test whether ecological factors constrained groups to have either unique (separate) or common trajectories in the shape-size relationship.Afterwards, we compared the fit of both models with a homogeneity of slopes test.If the model including the interaction is significant, groups do not share common slopes.Afterwards, we tested for differences in distance (length) and orientation of trajectories between trophic guilds and foraging behavior.Finally, we visually inspected trajectories using a method that calculates the predicted values from the shapeto-size regression and plots the PC1 obtained from the shape versus size regression (Adams and Nistri 2010).

Size and shape variation
The PC1 of the cranium was related to rostrum length.For example, nectarivores with elongated rostra had high negative scores along PC1, while species with robust crania and short faces were distributed in the positive side of PC1, such as Pygoderma bilabiatum and Desmodus rotundus.PC2 was related to braincase height, with negative values corresponding to flattened neurocrania, as in Glossophaginae or P. mandible, except for passive-active frugivores vs. passive animalivores, but the distance of allometric trajectories of active animalivores was different from all groups (Online Resource 3).
Interestingly, passive-active nectarivores had a positive slope when compared to other groups that follow a negative trajectory for the mandible shape, suggesting differential selective pressures on mandible size as a function of trophic guild or foraging behavior (Fig. 4).Active animalivores (Desmodus and Noctilio) had quite distinctive steep, positive allometric trajectories for the mandible and cranium, being positioned far from other groups (Fig. 4).These two species have certain traits, such as extremely short rostrum and robust neurocranium (e.g., D. rotundus) or prominent sagittal crest and robust neurocranium (e.g., N. leporinus).Active insectivores were not grouped together with other species with the same foraging behavior.The allometric trajectories for the cranium and mandible shape of passiveactive frugivore and passive-active nectarivore had less differentiation in length and angle.
(Table 4), likely due to low sample size.These results support the above results of PGLS models that assumed traits evolved under a BM.

Evolutionary allometry
Cranium allometry depended neither upon trophic guild nor foraging behavior (Table 5), since slopes for all ecological groups were similar and positive.This suggests that the hypothesis of a unique allometric trajectory for each group is not supported, implying a common allometric model (Fig. 4a; Table 5).Non phyllostomid insectivores and nectarivores had high negative scores along PC1 of cranium shape.There were no differences in trajectory correlation for the cranium, indicating that the orientation of allometric trajectory was similar for all groups.Conversely, we found a difference in distances (length) of allometric trajectories between active animalivores and all other groups (Online Resource 3).In contrast, we found a significant difference in slopes for the mandible, indicating support for the hypothesis of a unique allometric model for each ecological group (Fig. 4b; Table 5).As for the cranium, we did not find differences in the orientation of allometric trajectories for the   availability of food with similar hardness (Nogueira et al. 2005;Freeman and Lemen 2007).The similarity in cranium shape within Stenodermatinae (all passive-active frugivores in our sample) suggest it might represent an adaptive radiation in Phyllostomidae (Dumont et al. 2012), since members of this subfamily had only slight changes in cranium shape and size.Therefore, co-occurrence in stenodermatines with such a similar morphology might be possible by the exploration of fruit with different sizes (Dumont et al. 2012) and distinct forest strata (Munin et al. 2012).In addition, insectivores occupied central areas in the cranium and mandible phylomorphospace, indicating the absence of drastic changes in morphology.This pattern might be influenced by the ancient feeding habits in Chiroptera, given the early evolution of insectivory in the group (Gunnell and Simmons 2005).For example, early diverging species have small brains with a highly developed audition, specialized in detecting insects using sound clues.
Active animalivores, which corresponds to Desmodus rotundus and Noctilio leporinus, had the largest shape disparity in the cranium and mandible.Additionally, these species were distant from other groups, occupying a distinctive position in the morphospaces, due to their cranial morphology (robust crania with prominent sagittal crest or quite reduced rostrum), mandible (similar height between the coronoid and condylar processes, and broad mandibular ramus), and foraging behavior.Active animalivores were clustered together because they all have to process hard food items, such as small vertebrates or insects (Aguirre et al. 2003).This also applies to D. rotundus, which employs a hard bite to hook prey (Hermanson and Carter 2020).
We found that a common trajectory model was favored to explain the cranium allometry.This result suggests that the evolution of cranium allometry was constrained by stabilizing selection (Pélabon et al. 2014), probably related to the evolution of vision and echolocation, which evolved early in bats (Arbour et al. 2021) and are shared by all species in our study region.In contrast, the mandible allometry showed a unique trajectory related to trophic guild and foraging behavior (Nogueira et al. 2009; Hedrick and Dumont

Discussion
Our results showed that closely-related species had similar cranium and mandible shape and size.In addition, trophic guild and foraging behavior explained the distribution of species in the morphospace for the cranium and mandible, with trophic guild explaining a larger proportion of variance in the shapes of both structures than foraging behavior.We found greater morphological disparity in the cranium and mandible shapes for active animalivores, when compared to the other categories.The cranium had a common allometric trajectory for all groups.Conversely, mandible shape had unique allometric trajectories, with insectivores and frugivores having negative trajectories, while nectarivores and active animalivores had positive slopes.
The clustering in the cranium and mandible phylomorphospace driven by trophic guild and foraging behavior in our regional species pool shows that species with less evolutionary divergence present greater morphological similarity.The isolated distribution of specialist nectarivores or animalivores also supports the findings of Potter et al. (2021), which indicate that molecular-level adaptations to specialized diets are expressed in morphological novelties.For phyllostomids, Potter et al. (2021) found that specific genes in Glossophaginae and Desmodontinae are associated with nitrogenous waste excretion and glycolysis, respectively.This is an interesting finding because the same genes were related to craniofacial remodeling linked with dietary specialization in Phyllostomidae.We also recovered the differentiation of these two trophic groups in the phylomorphospace and in the allometric trajectories of both the cranium and mandible.Further studies should investigate other groups less explored at a genetic and morphological level, such as Noctilionidae.
The similarity in the cranium phylomorphospace between passive-active frugivores (all Stenodermatinae in our sample), some passive animalivores (e.g., T. bidens), and passive-active omnivores (e.g., Phyllostomus hastatus and P. discolor) suggests that their cranium shape might have evolved in parallel (convergence) in response to the

2018
).The common allometric trajectory of the cranium is likely constrained by the development of sensory functions, mainly related to echolocation (nasal and oral) that characterizes the Yangochiroptera suborder (Fenton et al. 2016).
Beyond being related to echolocation, cranium morphology is also related to the auditory system, and brain development also seems to influence foraging behaviors, and the visual development that did not accompany this auditory evolution (Fenton et al. 2016;Safi et al. 2005).Our results are consistent with previous findings for cranial evolution in Chiroptera, which seems more related to sound emission (Arbour et al. 2019), and consequently established early in the group evolution.The positive allometry of the cranium and mandible in passive-active nectarivores are likely influenced by the elongation of the rostrum, as also found by Bolzan et al. (2015).These morphological differences are related to the origin of nectarivory in Phyllostomidae (Datzmann et al. 2010;Baker et al. 2012).Glossophaginae and Lonchophyllinae pump nectar from flowers in a similar way, suggesting an evolutionary convergence at the functional level between these subfamilies (Tschapka et al. 2015).Interestingly, nectarivores of those two subfamilies were close to passive insectivores in the cranium and mandible phylomorphospace, which may be related to the fact that insectivorous bats occasionally consume nectar from flowers, as in the cactus Pachycereus pringlei (Frick et al. 2013).Natural selection likely shaped these specialized trophic habits, because the mandible is closely related to food processing, producing the allometric trajectories found here.Mandible size and sturdiness represent the potential of the bite to process or masticate food (Aguirre et al. 2003;Santana and Dumont 2009).For example, D. rotundus has a robust mandible, especially on the area of insertion of the incisors and canines, which may be the result of selection molded by foraging behavior, because D. rotundus holds its prey with the lower incisors and make suction holes with the upper incisors (Hermanson and Carter 2020).Conversely, nectarivores had thin and elongated teeth, which helps to support the tongue (Freeman 1995;Muchhala and Tschapka 2020).Bats provide several ecosystem services, such as pollination and insect suppression (Kunz et al. 2011).Given the spatial scale of our study, the ecomorphological diversity in region was relatively high.This diversity of forms is maintained by the heterogeneous landscapes of the Serra da Bodoquena and its surroundings.However, this region is expected to undergo drastic changes in land use and cover in the future (Cunha et al. 2021).Simulations of land use change for the Cerrado (Mendes and Marco 2018;Gonçalves et al. 2021) show that specialist bat species, such as frugivores and insectivores were more sensitive to habitat disturbance (see also Farneda et al. 2015).These were the same groups in which we found the greatest morphological disparity and foraging behaviors.Previous studies found that ecosystem services, such as pollination by bats, could be lost with land use and land cover change (Zamora-Gutierrez et al. 2021).Future studies evaluating the effects of habitat filtering and landscape disturbance (Peixoto et al. 2018) can improve our understanding of species response to landscape change and predict which bat species could be lost in our study region.

Implications for bat metacommunity assembly
Few species belonging to certain trophic guilds (e.g., omnivores) and foraging behaviors (e.g., passive foragers) occurred in our study region.We believe our results can shed light on a key question: Why are there so few omnivores and carnivores in Neotropical bat communities?By investigating macroevolutionary factors at a community scale, we could address which eco-evolutionary processes are involved in community assembly (HilleRisLambers et al. 2012;Mittelbach and Schemske 2015;Weber et al. 2017).Selection seems to favor specialization in loci related to blood and nectar-based diets in phyllostomids (Gutiérrez-Guerrero et al. 2020;Potter et al. 2021).This specialization in trophic resource use in bats is predicted by niche packing theory (Sexton et al. 2017;Carscadden et al. 2020) and is especially expected to happen in species rich communities, like Neotropical ones (Shi et al. 2018).Thus, microevolutionary processes, like those reported in Potters et al. (2021) seem to scale up to influence how communities assemble, by determining the proportion of co-occurring species specialized in a given resource.The number of generalist species is much lower than specialists in bats, both at the community (e.g., Aguirre et al. 2002;Giannini and Kalko 2004) and macroevolutionary scale (Dumont et al. 2012).Therefore, specialization seems to be adaptive (Monteiro and Nogueira 2011;Sexton et al. 2017), promoting diversification.This is a well-known pattern in some (Price et al. 2012;Rojas et al. 2018), but not all clades (Román-Palacios et al. 2019).Consequently, specialist species are more likely to be found in local communities.
Frugivorous species were clustered together in the morphospace and had similar cranium and mandible allometries.Assuming a tight form-function relationship between the cranium and mandible shape and trophic resource use, stenodermatines (Sturnira -Solanum, Artibeus -Ficus, and Cecropia, Saldaña-Vázquez et al. 2013;Platyrrhinus -Cecropia and Solanum, Silvestre et al. 2016) are supposed to exhibit a temporal or spatial segregation at fine scales (Carscadden et al. 2020) in order to coexist in this metacommunity.This is indeed what previous studies found, reporting differences in feeding peaks during the night in Artibeus and Platyrrhinus (Muller and dos Reis 1992), spatial differentiation in foraging strata (Findley et al. 1983;Munin et al. 2012), or even acoustic segregation (Denzinger et al. 2018).Conversely, we found a higher disparity in the cranium and mandible shapes of insectivores, which can promote niche differentiation and facilitate co-occurrence (Salsamendi et al. 2012;Shi et al. 2020).In contrast, nectarivores were the most similar group in terms of cranium and mandible shapes, which could partly explain their low richness in the metacommunity studied.

Fig. 1
Fig. 1 Community phylogenetic tree for the 32 bat species found in the study area classified into trophic guilds (colors) and foraging behavior (symbols).(modified from Shi and Rabosky 2015).Abbreviations: ACT, active foragers; PAS, passive foragers; PAS-ACT, passive-active foragers

Fig. 2
Fig. 2 Phylomorphospace of cranium shape in lateral view.Thin plate splines represent cranium shapes of species associated with maximum and minimum scores along each eigenvector.Abbreviations: ACT, active foragers; PAS, passive foragers; PAS-ACT, passive-active foragers

Fig. 3
Fig. 3 Phylomorphospace of mandible shape showing the distribution of the species in the reduced space, with the phylogeny superimposed.Thin plate splines represent mandible shapes of species with maximum and minimum scores along eigenvectors.Abbreviations: ACT, active foragers; PAS, passive foragers; PAS-ACT, passive-active foragers

Table 1
Results of the phylogenetic generalized least squares (PGLS) for the bat cranium and mandible shapes in lateral view.
Abbreviations: Df, degrees of freedom; F, Fisher statistic; MS, mean squares; P value, significance; R 2 , coefficient of determination; SS, sum of squares; Z, Z statistic (effect size)

Table 2
Results of the posthoc test of the PGLS model for differences in disparity (variances) between trophic guild and foraging behavior for cranium shape.Lower triangle shows P values and upper triangle shows test statistics (d, pairwise distances between variances).Significant P values are in bold.Abbreviations: ACT, active foragers; Ani, Animalivore; Fru, Frugivore; Ins, Insectivore; Nect, nectarivore; Omn, Omnivore; PAS, passive foragers; PAS-ACT; passive-active foragers

Table 3
Results of the posthoc test of the PGLS model for differences in disparity (variances) between trophic guild and foraging behavior for mandible shape.Lower triangle shows P values and upper triangle shows test statistics (d, pairwise distances between variances).Significant P values are in bold.Abbreviations: ACT, active foragers; Ani, Animalivore; Fru, Frugivore; Ins, Insecti-

Table 5
Results of the homogeneity of slopes test for comparing the two models for the influence of ecological groups on allometric trajectories of the cranium and mandible.Abbreviations: F, Fisher statistic; MS, mean squares; P value, significance; ResDf, Residual of degrees of freedom; SS, sum of squares; Z, Z statistic (effect size).Significant P values are in bold