Foraging strategies, craniodental traits, and interaction in the bite force of Neotropical frugivorous bats (Phyllostomidae: Stenodermatinae)

Abstract Bats in the family Phyllostomidae exhibit great diversity in skull size and morphology that reflects the degree of resource division and ecological overlap in the group. In particular, the subfamily Stenodermatinae has high morphological diversification associated with cranial and mandibular traits that are associated with the ability to consume the full range of available fruits (soft and hard). We analyzed craniodental traits and their relationship to the bite force in 343 specimens distributed in seven species of stenodermatine bats with two foraging strategies: nomadic and sedentary frugivory. We evaluated 19 traits related to feeding and bite force in live animals by correcting bite force with body size. We used a generalized linear model (GLM) and post hoc tests to determine possible relationships and differences between cranial traits, species, and sex. We also used Blomberg's K to measure the phylogenetic signal and phylogenetic generalized least‐squares (PGLS) to ensure the phylogenetic independence of the traits. We found that smaller nomadic species, A. anderseni and A. phaeotis , have a similar bite force to the large species A. planirostris and A. lituratus; furthermore, P. helleri registered a bite force similar to that of the sedentary bat, S. giannae. Our study determined that all the features of the mandible and most of the traits of the skull have a low phylogenetic signal. Through the PGLS, we found that the diet and several cranial features (mandibular toothrow length, dentary length, braincase breadth, mastoid breadth, greatest length of skull, condylo‐incisive length, and condylo‐canine length) determined bite force performance among Stenodermatiane. Our results reinforce that skull size is a determining factor in the bite force, but also emphasize the importance of its relationships with morphology, ecology, and phylogeny of the species, which gives us a better understanding of the evolutionary adaptions of this highly diverse Neotropical bat group.


| INTRODUC TI ON
The skull of vertebrates is a complex structure that is closely associated with resource collection, food processing, and behavior of the species (Bels & Herrel, 2019). Examining the patterns and mechanisms that lead to cranial variation, including size and shape, enables an understanding of the morphology, ecology, and general fitness of animals (Santana et al., 2010). Variation in cranial morphology of bats is attributed to evolutionary processes of ecological specialization, which results in an ecomorphological niche division between similar species . This variation responds mainly to functional requirements related to nutritional performance and the sensory system (Thiagavel et al., 2018). Among the bats of the New World, rostral length is the main morphological feature that has allowed morphological differentiation, diversity of food, and ecological niches (Hedrick et al., 2020;Shi et al., 2021).
Among chiropterans, New World leaf-nosed bats (Phyllostomidae) represents one of the largest and most morphologically diverse mammal families (Rossoni et al., 2017). Phyllostomids have the highest diversity of bats in the Neotropics with more than 70 species that can be found in sympatry (Giannini & Kalko, 2004;Reid et al., 2015). Ecological diversification in Phyllostomidae is related to bite performance and mechanical demands of different diets, including frugivorous, insectivores, nectarivores, carnivores, and sanguinivores (Dumont, 2007;Manhães et al., 2017;Nogueira et al., 2009). Dietary differences require specific mechanical modifications, including variation in the rostral length and height of the skull (Santana et al., 2010). However, there is a lack of understanding about the patterns in the variation of shape and performance (López- Aguirre & Pérez-Torres, 2015). The remarkable specializations seen in these bats provide a unique opportunity for studying the relationship between cranial morphology, K E Y W O R D S ecomorphology, frugivores, morphometry, skull, Stenodermatinae, trophic performance F I G U R E 1 Geographic distribution of the locations in Colombia of the stenodermatine bat samples used for the biometric analyses TA B L E 1 The craniodental, mandibular, and external measurements used in this study of stenodermatine bats from Colombia

Main trait
Description Abbreviation Unit

Key supporting reference
Bite force Head Maximum bite force produced by molars BF N/g Shi et al. (2020) Forearm length Body Distance from the olecranon process to anterior surface of carpals in the folded wing  feeding performance, foraging strategy, and dietary ecology (Aguirre et al., 2002;Rossoni et al., 2017;Soriano, 2000).
Within Phyllostomidae, the Stenodermatinae subfamily contain >43% of all described New World leaf-nosed bat species (Shipley & Twining, 2020 et al., 2002;Dumont et al., 2009;Santana et al., 2012;Santana & Miller, 2016), the evaluation of these variables has been carried out individually (Shi et al., 2020); there are existing information gaps that make it difficult for us to understand the morphological features associated with the diet of fruit bats. According to Soriano (2000), the Stenodermatinae subfamily are the only ones that present two foraging strategies for fruit consumption. Nomadic bats present a wide range of home, presenting preferential consumption of figs (Ficus), hard fruits, while the members of the sedentary strategy (only the genus Sturnira) prefer soft fruits (Solanum), reflecting specialization toward a specific fruit or group of fruits (Santana et al., 2010.
The objectives of our study were to identify the functional traits associated with bite force and the effects of foraging strategies in seven representative species of Stenodermatinae occurring in Colombia to address trophic ecology in a phylogenetic context.
We hypothesize that nomadic frugivorous species will have a greater biomechanical advantage of stronger bite force irrespective of size in relation to sedentary frugivorous bats, because they have to carry their food further to their roosts. While the sedentary diet has fixed feeding points.

| Field collection and laboratory analysis
We followed two procedures for collecting data. Mammalogists guidelines for the use of wild animals for research purposes (Sikes et al., 2016). After capture, the age, sex, and reproductive status were evaluated, and only adult males and adult nonpregnant, nonlactating females were used for measurements.
Age was based on the degree of ossification of the wing joints.
Reproductive status in females was determined by examining the nipples and palpation of the abdomen. Forearm length and body mass were recorded before euthanasia. Specimens were deposited in the biological collection of the University of Tolima CZUT-M (Ibagué, Colombia), and skulls were cleaned for craniodental morphometry (Table S1).
The bite force of bats was measured using a portable digital fruit hardness tester Lutron FR 5120 (made in Taiwan) with a capacity of 196.10 Newton and precision ±0.05 that corresponded to in vivo measurements of the maximal force. Bite force was recorded at the molars, and measurements were repeated five times for each bat with a trial interval of at least 5 min following the method of Freeman and Lemen (2008). The maximum value of the five measurements was considered as the maximum bite force produced by that individual. The bite force of the species was calculated by averaging the maximum bite force of each individual.
The second procedure for collecting data consisted of measurements of 16 craniodental traits and two body traits (Table 1,  Note: All measurements are presented in millimeters (mean ± SD) except for BS which are presented in Newton and mass in grams. Abbreviations as in Table 1 and Figure 2. The # symbol that precedes BS indicates the number of individuals that were used for bite force. The values inside the parentheses correspond to ±SD.
on the ossification of the phalange epiphyses in the wing (Dietz et al., 2007).
Among frugivorous bats, two foraging categories adapted to fruit consumption were distinguished based on Soriano (2000): nomadic frugivorous species whose strategy consists of feeding on trees with massive production of short-lived fruit, while sedentary bats have search itineraries more or less fixed every night and focus the consumption of fruits in plants with continuous production throughout the year. This classification involves a specialization toward a group of plants, reducing competition and generating a coevolution with plants and bats (Fleming, 1986;Giannini & Kalko, 2004).
Stenodermatinae is the only subfamily with species that are either nomadic sedentary; therefore, these organisms constitute an excellent model to evaluate the skull traits involved in the consumption of contrasting fruits. We classified in our data set large-sized nomadic species with weight 55-65 g and medium-to small-sized nomadic with weight 10-18 g. Although our size limit is arbitrary, it may have some biological relevance because Artibeus planirostris and A. lituratus are large bats with high dispersal capacity (Trevelin et al., 2013).
Individuals captured in mist nets were placed in a clean cloth bag for up to 4 hr to obtain fecal samples, which were analyzed at a later date in the laboratory for seeds. The seeds were washed, examined with a dissection microscope, identified at the lowest possible taxonomic level (Table S2), and supplemented with bibliographic information from the registry (García-Herrera et al., 2019b).
Fruits known as food were collected from Neotropical bats, and their hardness was recorded by a portable digital fruit hardness tester Lutron FR 5120. These measurements were made on 10 fruits taken in the field, and we averaged the values to estimate the hardness per fruit species. Fruits with hardness recorded as between 5 and 7 N were classified as hard, 4 N as intermediate, and 1 to 3 N as soft (see Table S3 for more details).

| Statistical analyses
To analyze the intra-and interspecific morphological variation, the mean ± SD was calculated for all morphometric variables per species. The assumptions of normality and homoscedasticity were corroborated with Shapiro-Wilk's test and Levene's test, respectively.
The preliminary analyses showed that our data fit a normal distribution and had homogeneity of variances, so we used a two-way multivariate analysis of variance (MANOVA) to test for differences between species and sex. Post hoc comparisons between sexes were made within species to identify dimorphism.
A multiple linear regression was performed to evaluate the effects of body parameters on changes in bite force. In the models, the averages of bite force and body size (forearm length, greater skull length, and mass) of each species were used. The linear model was as follows: Ln (y) = β0 + β1x1 + β2x2 + β3x3 + ε, where y corresponds to the bite force (dependent variable); x1, x2, and x3 are the length of the forearm, the greatest length of the skull, and the mass, respectively (independent variables); β0 is the intercept; β1, β2, and β3 are the regression coefficients (for x1, x2, and x3, respectively); and ε is the random error term. A principal component analysis (PCA) of averages for species was used to visualize how they occupy morphospace. The correlation values showed a high association between the forearm length trait and mass (0.99; Table S4). Therefore, the rest of the morphometric features were used for the analysis canonical variate analysis (CVA) to establish the major axes of discrimination among individuals and between groups, to find the linear combinations of the initial variables with the maximum discriminating power between the groups, to test whether the means of these groups are significantly different, and to study the dimensionality of the data.
To determine the existence of significant variation between species and morphological traits, an analysis of covariance was performed using a generalized linear model (GLM). We use bite force as the response variable and cranial and body features as covariates (to correct for preexisting variation among species/sex). The model used was as follows: Yijķ = µ + Ţi + δj + Ωķ + εijķ, where Yijķ represents the bite force response at the j-th sex and the i-th species; µ general average, Ţi effect produced by the i-th species, δj effect produced by the j-th sex, Ωķ effect due to the R-th trait, and εijķ the random error.
A multiple linear regression model was applied to study the relationship between bite force (Y) and the variables identified with the greatest influence in the previous models as explanatory variables.
For the detection of masked variability, an intuitive and qualitative procedure based on graphic representation was used; then, post hoc tests were performed using Fisher's least significant difference (LSD) pairwise comparison procedure. Statistical significance for all tests was p ⩽ 0.05. All analyzes were performed in R 3.5.3 (R Core Team, 2019).

| Phylogenetic signal
We tested the phylogenetic signal because it can create a confounding effect as taxa that are more closely related to exhibit similar trait values, and the similarity of traits decreases as the phylogenetic distance increases (Losos, 2008). For phylogenetic analyses, we used the phylogeny of Phyllostomidae corresponding to the maximum clade credibility tree used in Rolland et al. (2014).
We trimmed the tree to only include the 40 frugivorous bats from Stenodermatinae. We estimated Blomberg's K with the function phylosignal, package picante (Blomberg et al., 2003;Kembel et al., 2010). Blomberg's K is the most appropriate metric because it is not significantly affected by polytomies, and is less sensitive to small sample sizes (20 taxa) than other indices (Münkemüller et al., 2012).
Values of K-statistics lower than 1.0 indicate that related species resemble each other less than expected according to the Brownian motion model of trait evolution, while values greater than 1.0 mean that more related species are more similar, for the trait, under study, which was predicted by this model (Blomberg et al., 2003;Kamilar & Cooper, 2013).

| Bite force measurements, cranial traits, and diet
Bite force was corrected for body size (forearm length, greater skull length, and mass), and the residuals were obtained. To find differences between species, the cranial features and bite force were
The PCA showed that the first two principal components (PC) explained 91.8% of total variation (PC1 87.4% and PC2 4.4%, respectively). All variables had high positive eigenvectors on the first component indicating overall size (Table S6). The variables that were highly correlated with PC1 were forearm length ( and Table S5).

| Variation in the bite force associated with craniodental traits
The GLM found that the bite force of stenodermatine bats was significantly influenced by all the cranial features (p < .001; Table 3).
It was identified by multiple comparison analysis that 15 traits are significantly associated with bite force, and the traits with the highest correlation are MTRL, MANDL, M1-M1, and DENDL regardless of sex ( Figure 5, Table S7).

| Interspecific bite force variation
The bite force varied between species; large bats (A. lituratus and A. planirostris) had the highest bite force, followed by medium-sized bats (P. helleri, U. convexum, and S. giannae). The lowest force was recorded for the small-sized bats of the species A. anderseni and A. phaeotis (p < .001). Significant intersexual difference between species was detected with females having a greater bite force compared with males (p < .04; Figure 6).

Although the analysis of variation of the interspecific bite force
showed that the bite force is directly proportional to the size of the bat, the results of the GML determined that the bite force varies differently from the size. Obviously, the large species showed F I G U R E 3 Distribution of body and craniodental traits for seven species of stenodermatine fruit bats from Colombia based on a principal components analysis (PCA) significant differences in bite force in relation to the other species.
However, for the couples who presented p > .05, it is accepted that the null hypothesis is true; therefore, the bite force of A. phaeotis is similar to that of P. helleri, but greater than that of S. giannae and U. convexum, while A. anderseni registered a bite force similar to that of P. helleri and S. giannae, but greater than that of U. convexum (Table 3).

| Differences in bite force between bat species
Significant differences in skull morphology and bite strength were observed between species (

| Phylogenetic signal
Traits had a K-value between zero and unity; only the CIL, CCL, MTRL had K > 1, indicating that this trait is phylogenetically con-  Table 1 (PL), and de la mandible, dentary length (DENL), mandibular toothrow length (MANDL), and coronoid height (COH) presented a had a significant K < 1. significant result according to the p-value (Table 5). Evidenced as a result that these traits have evolved independently of the phylogeny.

| Bite force measurements, cranial traits, and diet
Our PLS result with size-corrected bite force reveals that residual bite force differs indistinctly from size, for example, large A. lituratus and small A. phaeotis groups on negative side of the plot, whereas large A. planirostris and small A. anderseni groups on the positive side ( Figure 7).
The first CP of the association between bite force, cranial shape, and diet explains 70.6% of the total variation and shows that DENL, MANDL, BB, MB, GLS, CIL, and CCL are associated with bite force.
The second CP explains 15% and is negatively associated with the PB and positively with diet, M1-M1 and M2-M2. High bite force values are associated with the DENL and MANDL, traits that could explain the better performance of the A. lituratus and A. phaeotis species with respect to the other species of the same genus studied here. Although A. anderseni and A. planirostris also had high bite forces, this was associated with diet, which could indicate resourcedependent modulation of bite force. Diet was also a determining factor for bite force for U. convexum, S. giannae, and P. helleri species, but skull characteristics were not associated with residual bite force, linking diet with a weaker bite. The associations of the species with diet and skull traits were regardless of sex (Figure 8).

| D ISCUSS I ON
Neotropical leaf-nosed bats (Phyllostomidae) are an ecologically diverse group of mammals with distinctive morphological adaptations associated with specialized modes of feeding (Camacho et al., 2019).
In particular, Stenodermatinae have a strict nomadic frugivorous diet, related to the consumption of hard and soft fruits , which is linked to a strong bite that is associated with a short and wide skull (Santana et al., 2010. Fruit feeders are morphologically diverse, exhibiting cranial and mandibular morphologies that overlap with other guilds . This morphological diversity reflects the variety of physical properties represented among fruits and the tendency of frugivorous species to specialize in particular fruits (Rossoni et al., 2017;Santana et al., 2010). These specializations are determined by the functional traits that allow them to exploit different fruits.
In our study, we found differences in skull, forearm, and mass variables between males and females in the seven species of stenodermatines studied here. This variation may be associated with the diet of bat species, possibly as a result of differences in energy requirements during the reproductive season (de Camargo & de Oliveira, 2012). Although sexual dimorphism has been relatively well documented in evening bats (family Vespertilionidae), with females larger and heavier than males (Bornholdt et al., 2008), in phyllostomid bats this information is less documented despite being the most diverse family and distributed in the Neotropics (Gardner, 2008).
One exception is López-Aguirre and Pérez-Torres (2015) identified that Artibeus lituratus females in Colombia had greater fluctuating asymmetry in the splanchnocranium, resulting in a differential bite force between the sexes.
Bite force has been established as an important performance trait for vertebrates that is associated with both cranial morphology and trophic ecology (Santana et al., 2010). Our study shows, for the first time, strong quantitative evidence of such a correlation by using  , which is reflected in the variety of foods, as well as in the tendency of these species to specialize in a group of particular fruits (Rojas et al., 2012;Rossoni et al., 2017). Several craniodental features contribute to generating a greater bite force, and the differences between these particular features have shown a clear segregation between species .
The association between craniodental features and bite force by species can be explained by the type of fruit consumed. For example, A. phaeotis, P. helleri, and S. giannae formed an independent group in the least significant difference pairwise comparison test.
These species have a short skull and a shorter distance from the teeth to the jaw joint, which allows them to consume hard and soft fruits (Arias & Pacheco, 2019;Dumont et al., 2012;García-Herrera et al., 2019b;Santana et al., 2012). Our study did not consider the phylogenetic correction of the species; therefore, it presents limitations to understand how evolutionary changes in the diet of bats are correlated with skull traits, and how they modulate bite force. Clearly, our understanding focuses on morphological analyzes and on the low phylogenetic signal found among the traits studied here for Stenodermatinae, concluding that the changes presented in bite force are modulated by greatest length of skull, braincase breadth, zygomatic breadth, palatal length, and de la mandible; dentary length, mandibular toothrow length, and coronoid height. Traits that have diverged as a result of environmental pressures (Pitnick et al., 2006;Santana et al., 2012). Murillo-García (2018) found several adaptive changes across the phylogeny of neotropical fruit bats (Phyllostomidae: Stenodermatinae and Carollinae), indicating divergence in skull and jaw morphology. Stenodermatins have the additional ability to consume both soft and hard fruits (Aguirre et al., 2003;Dumont et al., 2009 Tables S2 and S3). Species scores are represented by italicized species acronyms. Diet variables are represented as vectors identified by the variable names in bold. Abbreviations as in Figure 2 F I G U R E 9 Phylogeny of stenodermatine fruit bats used in our study rooted with two other subfamilies of Phyllostomidae. Sedentary and nomadic foraging strategies are parsimoniously mapped onto the tree based on Soriano (2000) and Rolland et al. (2014) Solanum), such that larger bats can process larger and harder foods (Aguirre et al., 2003;Arbour et al., 2019;Santana et al., 2012 The frugivorous foraging strategies in Phyllostomidae based on Soriano (2000) and Rolland et al. (2014) suggest that sedentary feeding is the ancestral behavior for the family and nomadic feeding in Stenodermatinae is a derived behavior (Figure 9). This indicates that there is a phylogenetic basis to foraging in Neotropical fruit-eating bats.
Our study shows strong quantitative evidence of the relationship between diet, foraging strategy, skull traits, and bite force in stenodermatine bats. Our results, however, are limited by the number of species included in the analysis. This was a consequence of the difficulty in catching many of the potentially rare frugivorous species in tropical dry forest of Colombia. Additionally, information on bite force available for phyllostomids bats in the literature is limited.
Future studies should focus on collecting bite force data for other species of bats to get a better understanding of functional morphological variation.
Our findings illustrate that skull size is a determining factor in the bite force, but emphasizes that the use of functional traits is relevant for establishing the feeding performance of bat species. This highlights the importance of studying the relationships between morphology, bite force, ecology, and phylogeny of the species to get a better understanding of evolutionary adaptions of highly diverse Neotropical bat groups.

ACK N OWLED G M ENTS
The authors wish to thank the staff of the Research Group in Zoology (GIZ), Universidad del Tolima (Ibagué, Colombia Autoridad Nacional de Licencias Ambientales-ANLA). We thank Dr Jairo Pérez for allowing access to the voucher specimens of the Museo Javeriano de Historia Natural "Lorenzo Uribe, SJ".

CO N FLI C T O F I NTE R E S T
None declared.

O PE N R E S E A RCH BA D G E S
This article has earned an Open Data, Open Materials and Preregistered for making publicly available the digitally-shareable data necessary to reproduce the reported results. The data is available at https://datad ryad.org/stash/ share/ N6vW4 UGTQ Z bdh13 2I0ga PR2GY sa4iD Cy5Yd sEeNaerM.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data are available at https://datad ryad.org/stash/ share/ N6vW4 UGTQ Z bdh13 2I0ga PR2GY sa4iD Cy5Yd sEeNaerM.