phylogenetic and body size structure within ... - Wiley Online Library

Download PDF
9 downloads 101 Views 948KB Size Report
Comment
Haya, 91070 Xalapa, Veracruz, Mexico,. 3Departmento de Biologeıa y Geologeıa, Feısica y. Queımica Inorgeanica, Universidad Rey Juan. Carlos, Meostoles ...
Journal of Biogeography (J. Biogeogr.) (2017) 44, 136–146

ORIGINAL ARTICLE

Global patterns of mammalian co-occurrence: phylogenetic and body size structure within species ranges  Olalla-Tarraga3, Marcus V. Cianciaruso1, Fabricio Villalobos1,2,*, Miguel A. 1 Thiago F. Rangel and Jose Alexandre F. Diniz-Filho1

1

Departamento de Ecologia, ICB, Universidade Federal de Goias, CxP. 131, 74001-970 Goi^ania, Goias, Brazil, 2Red de Biologıa Evolutiva, Instituto de Ecologıa, A.C., Carretera antigua a Coatepec 351, El Haya, 91070 Xalapa, Veracruz, Mexico, 3 Departmento de Biologıa y Geologıa, Fısica y Quımica Inorganica, Universidad Rey Juan Carlos, Mostoles 28933, Madrid, Spain

ABSTRACT

Aim To analyse the geographical co-occurrence among mammal species based on their complete geographical distributions, considering their phylogenetic relationships and body size data. We describe species-level patterns and test the relative effects of ecological and evolutionary processes in determining species co-occurrence under the phylogenetic field framework. Location Global. Methods We gathered distributional, phylogenetic and body size information for 3697 mammal species. We defined phylogenetic fields of species by estimating the phylogenetic structure of species co-occurrence within a focal species’ range. Likewise, body size structure within focal species’ ranges was defined as body size fields. We applied a spatial-phylogenetic statistical framework to evaluate geographical variation on species fields. Also, we tested the significance of phylogenetic and body size fields based on biogeographically informed null models. Analyses were done for all mammal species as a whole and within particular taxonomic orders. Results Phylogenetic and body size fields of mammal species showed significant geographical patterning beyond their spatial and phylogenetic dependence. Phylogenetic fields were strikingly different between the New and Old World, with mammals co-occurring with more closely related species in the New World and more distantly related species in the Old World. Clustered phylogenetic and body size fields showed geographically congruent patterns. Similar findings were obtained within particular mammalian orders.

*Correspondence: Fabricio Villalobos, Red de Biologıa Evolutiva, Instituto de Ecologıa, A.C., Carretera antigua a Coatepec 351, El Haya, 91070 Xalapa, Veracruz, Mexico. E-mail: [email protected]

Main conclusions Geographical co-occurrence among mammal species reveals the imprint of historical origins and dispersal of mammalian lineages. Phylogenetic and body size structure within mammalian ranges is driven by the distinct histories among biogeographical regions and mainly between the New and Old World. We demonstrate the usefulness of a new protocol integrating species’ distributional, phylogenetic and body size information for linking evolutionary and ecological approaches to understand geographical patterns of biodiversity. Keywords biodiversity gradients, body size, community phylogenetics, geographical range, historical processes, phylogenetic fields

INTRODUCTION Geographical variation in species richness results from differential co-occurrence of species in distinct regions of the globe (Hillebrand, 2004), emerging from the overlap of 136

http://wileyonlinelibrary.com/journal/jbi doi:10.1111/jbi.12826

species ranges that differ in size, shape and location (Gotelli et al., 2009). Thus, processes defining species ranges and their overlap ultimately drive species co-occurrence patterns at broad spatial scales (Guisan & Rahbek, 2011). Such processes are varied, including ecological, historical and ª 2016 John Wiley & Sons Ltd

Global mammalian co-occurrence evolutionary factors (Warren et al., 2014). Traditionally, these processes have been dichotomized according to their supposed scale of action, with ecological processes acting at local scales and evolutionary processes acting at regional scales (Ricklefs, 2004; Fine, 2015). However, both kinds of processes may equally act at any spatial scale and thus should be considered together when inferring the causes of biogeographical patterns (Warren et al., 2014). Evaluating species co-occurrence in relation to their morphological characteristics can help assessing the relative roles of ecological and evolutionary processes in shaping such cooccurrence (Davies et al., 2007). The logic rests on the following assumptions: closely related species share ecological traits owing to common ancestry (Harvey & Pagel, 1991), species’ morphology reflects ecology (Ricklefs & Miles, 1994), and species’ ecological similarity precludes coexistence (MacArthur & Levins, 1967). Following this sequential reasoning, species with similar morphologies may have broadly overlapping ranges owing to their similar habitat requirements or little overlap as a result of competitive exclusion (Bowers & Brown, 1982; Letcher et al., 1994). Conversely, species with dissimilar morphologies may greatly overlap owing to reduced competitive interactions or different habitat requirements (Brown & Wilson, 1956; Davies et al., 2007). In addition, species co-occurrence is also a key factor in the study of speciation given that closely related species tend to be geographically separated (Barraclough et al., 1998). It is, therefore, expected that combining geographical, morphological and phylogenetic data will provide a better understanding on the causes of biogeographical patterns (Barraclough et al., 1998). Recently, the ‘community phylogenetics’ approach, integrating ecology and phylogenetics, has highlighted the importance of both ecological and evolutionary processes from local to continental scales (Cavender-Bares et al., 2009). This approach has mostly concentrated on using phylogenies as a proxy to infer community assembly processes (Gerhold et al., 2015) and focused primarily on sites, either single site, multiple sites or regions (Cardillo, 2011). Consequently, when studying species co-occurrence at large spatial scales, two caveats may arise from the application of the community phylogenetics approach. First, it is commonplace to only use phylogenies instead of species’ traits assuming that species trait similarity is proportional to their divergence time, which is often an unsupported assumption (Gerhold et al., 2015). Second, focusing on sites may fail to represent species-level patterns if all sites and co-occurring species within complete species’ ranges are not considered (Villalobos & Arita, 2010; Villalobos et al., 2013). Hence, a combination of phylogenetic and trait information across geographical domains encompassing all species and their ranges is required to provide an integrated perspective on the ecological and evolutionary drivers of species co-occurrences (Villalobos et al., 2013; Barnagaud et al., 2014). Throughout its range, a given species co-occurs with closely or distantly related sets of species. To evaluate this Journal of Biogeography 44, 136–146 ª 2016 John Wiley & Sons Ltd

pattern, Villalobos et al. (2013) recently introduced the ‘phylogenetic field’ concept to study species co-occurrences under a species-oriented phylogenetic framework. Focusing on species ranges as the study unit, instead of sites, they proposed the evaluation of ‘phylogenetic fields’ (PFs) – the phylogenetic structure of species co-occurrence within a focal species’ range – to infer evolutionary processes at more appropriate biogeographical scales. Phylogenetic fields have the advantage of being a species’ property, portraying the overall phylogenetic structure of the assemblage with which a given species co-occurs (Villalobos et al., 2013; Barnagaud et al., 2014). Hence, PFs can be associated with traits influencing species co-occurrence (e.g. body size; Barnagaud et al., 2014). Analogous to PFs, an ecologically relevant trait such as body size, could be used to describe the structure of species co-occurrence within focal ranges (e.g. a ‘body size field’). This would provide a means to evaluate the role of biotic interactions in determining large-scale species cooccurrence (Davies et al., 2007), in a manner akin to community-wide character displacement at local scales (Dayan & Simberloff, 2005). Here, we apply for the first time a species-oriented approach considering phylogenetic and body size structure within species ranges to evaluate the geographical co-occurrence among the world’s terrestrial mammals. Mammals represent an ideal group for this purpose given the availability of distributional, phylogenetic and body size data as well as our current knowledge on their geographical and evolutionary patterns (Jones & Safi, 2011; Hawkins et al., 2012). Current explanations of mammalian biodiversity patterns favour historical processes such as geographically non-stationary diversification rates (Davies & Buckley, 2011; Rolland et al., 2014) and phylogenetic niche conservatism (PNC; Buckley et al., 2010), with some studies also suggesting the imprint of ecological processes, namely competition, on assemblage structure (Cooper et al., 2008; Cardillo, 2011). Most of these studies were based on assemblage- instead of species-level patterns. Thus, remaining questions are whether the abovementioned processes determine species-level patterns and what is their relative influence. Bearing the above in mind, we derive and test the following predictions about phylogenetic and body size structure within mammalian ranges. First, under the joint effect of diversification rates and PNC leading to the largest accumulation of species in the Tropics (Buckley et al., 2010; Rolland et al., 2014), we would expect clustered PFs in low-richness temperate areas and overdispersed PFs in high-richness tropical areas (Qian et al., 2014). However, we are currently unaware of the phylogenetic structuring of species co-occurrence within recently colonized regions. For instance, Villalobos et al. (2013) suggested for a major mammalian family (Phyllostomidae, Chiroptera) that overdispersed PFs could emerge at low-richness regions from the historically recent colonization of few distantly related lineages. Accordingly, we expect that this pattern will be general for mammals with clustering of PFs decreasing from tropical to temperate regions. Second, 137

F. Villalobos et al. because different biogeographical regions reflect distinct evolutionary histories of major taxa including mammals (Holt et al., 2013), we hypothesized that patterns of PFs will differ among species confined to different biogeographical realms. Third, under a strong phylogenetic signal on body size (Cooper & Purvis, 2010) and local ecological processes (i.e. competitive exclusion) not affecting its variation at broad spatial scales (Brown & Nicoletto, 1991), species co-occurring with closely related species will also do so with species of similar body sizes. Therefore, those areas where species show clustered PFs are also expected to harbour species with body size clustering within their ranges. MATERIALS AND METHODS Distributional and phylogenetic data We gathered distributional data (i.e., extents of occurrence) for terrestrial mammal species from the IUCN database (2009) and body size measurements from PanTHERIA (Jones et al., 2009) and other sources (Safi et al., 2011). Phylogenetic relatedness among mammalian species was established from a species-level supertree (Fritz et al., 2009). This supertree is the most comprehensive mammalian phylogeny available, but it is not fully resolved, having low resolution at the species level. Nevertheless, the lack of resolution at the tips of broad-scale trees does not affect phylogenetic structure measures (Swenson, 2009). We only considered species that had all three kinds of data, distributional, body size and phylogenetic, totalling 3697 species of terrestrial mammals. However, depending on the analysis (see below), a smaller set of species was considered owing to the minimum withinrange richness needed for the calculations. Although we did not consider all currently recognized mammal species owing to the paucity of data, we trust that our observed patterns are still meaningful since most data-lacking species belong to and concentrate in species-rich groups and regions (Safi et al., 2011). We overlaid an equal-area grid of 200 9 200 km resolution onto the distributional data to obtain a presence–absence matrix of 4462 sites 9 3697 species. Species presence within a cell was determined if its range intersected with that grid-cell centre. The geographical range size of each species was the number of grid cells overlaying its range map, and its geographical position was its latitudinal midpoint. Also, we defined the membership of species to particular biogeographical realm(s) using the zoogeographical regionalization recently proposed by Holt et al. (2013). These authors proposed 11 realms based on a cross-taxon, distributional and phylogenetic analysis, which are appropriate to elucidate the historical events driving the formation of evolutionarily distinct areas, instead of smaller regions that may compound more recent time influences (Rueda et al., 2013). In addition, we used the total number and identity of species co-occurring within each species’ range to describe its diversity field (Villalobos & Arita, 2010). 138

Phylogenetic and body size fields We described species phylogenetic fields to characterize the overall phylogenetic structure contained within each species range, including the focal species. Phylogenetic fields assess whether particular species co-occur with either closely or distantly related species. Such assessment is represented by a single value of phylogenetic structure comprising all species co-occurring within a focal range, rather than an aggregate measure (e.g. mean) from individual sites composing that range (Villalobos et al., 2013). Despite its dependence on the geographical co-occurrence among species, PFs can be better interpreted as describing the assemblage, in terms of history and composition, with which a given species cooccurs – its phylogenetic ‘neighbourhood’. We estimated PFs by calculating the phylogenetic species variability (PSV) index (Helmus et al., 2007) of the subtree representing the species co-occurring within a focal species range, including itself (PSVsp) (Villalobos et al., 2013). Species in our data set that co-occurred with only one species were not assigned a PSVsp value, leaving a final data set of 3597 species with PF values. We described body size structure within each species range – body size field – by calculating the variance in body size ratios (VSRsp) among its co-occurring species. VSRsp values were calculated as the variance in the difference in log-transformed body size between adjacent pairs of species ranked by their trait values (Davies et al., 2012). We excluded 50 species in our data set that co-occurred with less than three other species within their ranges and thus were not amenable to calculate VSRsp values, leaving a final data set of 3547 species with VSRsp values. The variance in body size ratio is a metric favoured when testing for community-wide character displacement (Dayan & Simberloff, 2005), as it deals with discontinuities in trait distributions and has good statistical properties (Pleasants, 1994). It informs about the distribution of trait values across all co-occurring species to identify groupings around similar values (‘trait clustering’) or equal spacing between species (‘trait overdispersion’). Large VSRsp indicates trait clustering whereas low VSRsp indicates trait overdispersion, which is an evidence for interspecific competition (Davies et al., 2012). Integrated phylogenetic and spatial analysis We evaluated the relationship between phylogenetic/body size fields and diversity field properties (within-range species richness and latitudinal midpoint), as well as the variation in these fields among biogeographical realms and between the New and Old World. Owing to the phylogenetic and spatial non-independence of our species-level metrics, we applied a generalized least-squares (GLS) approach that jointly accounts for phylogenetic and spatial effects (Freckleton & Jetz, 2009). This approach allows estimating individual effects of phylogeny and space, as well as a component independent of both effects, on trait variation. Accordingly, three Journal of Biogeography 44, 136–146 ª 2016 John Wiley & Sons Ltd

Global mammalian co-occurrence independent parameters can be obtained along with those of predictor variables (e.g. regression coefficient). More specifically, a parameter / accounting for space is included in the model along with the k parameter accounting for phylogeny. Both parameters are estimated by maximum likelihood based on spatial and phylogenetic matrices, describing geographical and phylogenetic distances among species, respectively. Simultaneously estimating phylogenetic (k) and spatial (/) parameters allows calculating trait variation independent of these effects as c = (1/)(1k). Finally, once the effects of space are accounted for, the relative contribution of phy0 logeny can be obtained as k = k(1/) (see Freckleton & 0 Jetz, 2009 for details). These parameters (/, k , c) can then be interpreted as the net relative effects (i.e. independent contributions) of space, phylogeny and variation independent of both on trait variation (Freckleton & Jetz, 2009). In models relating phylogenetic and body size fields with diversity field properties, we used the original PSVsp value and the log-transformed VSRsp as response variables and within-range species richness and the absolute latitudinal midpoint as predictor variables. To evaluate the variation of PSVsp and log-transformed VSRsp among biogeographical realms and between the New (i.e. the Americas) and Old World (i.e. the rest of the world), the latter were codified as dummy variables using a presence–absence (1,0) matrix of species within realms and hemispheres, respectively. Owing to different evolutionary origins and dynamics of distinct mammalian clades (Springer et al., 2011), we also fitted the GLS models described above within single orders with more than five species in our data set (see Table S1 in Appendix S1 in Supporting Information). Null biogeographical models and benchmark test Statistical significance of species phylogenetic and body size fields was evaluated under a null model approach. In this approach, species were randomly sampled (without replacement) from a species pool with a probability proportional to their range sizes. This ‘range-based’ null model considered independent distributional patterns of species with large-ranged species being more likely to co-occur with other species than small-ranged species (Villalobos et al., 2013). Accordingly, we generated 1000 random diversity fields for each species from which phylogenetic fields and body size fields were estimated. For each focal species, the species pool was defined by all species occurring at the same biogeographical realm(s) occupied by it. By using biogeographical realms to inform species pools, we avoided the inclusion of historically different biotas in the null models. We considered our ‘range-based’ null model as appropriate for testing species field patterns owing to the effect of species frequencies on co-occurrence and phylogenetic structure metrics (Gotelli, 2000; Kembel, 2009). To detect the probability of false positives or type I error in our null model (e.g. finding significant patterns when none exists), we used an artificial data set of similar Journal of Biogeography 44, 136–146 ª 2016 John Wiley & Sons Ltd

properties than our data (e.g. number of species and rangesize frequency distribution). We were interested in knowing if significant patterns of phylogenetic fields and variance in body size ratios would arise from simple random data. We generated 50 random scenarios of 3697 species distributed across the globe under a spreading-dye model. For each of these scenarios, we generated a random phylogeny of all species by reshuffling the tips in the original phylogeny while keeping their original body size values. Then, we applied the above methods to describe phylogenetic and body size structure within species ranges and their statistical significance in relation to 50 realizations of the ‘range-based’ null model. Although we observed significant phylogenetic fields and variance in body size ratios, a chi-square test of independence showed that the numbers of significant species in the random scenarios were significantly lower than the observed ones (v2 = 540.45, d.f. = 2, P < 0.001; v2 = 46.18, d.f. = 2, P < 0.001; for phylogenetic fields and variance in body size ratios, respectively). Therefore, we are confident that our procedures are robust for correctly identifying the patterns of interest. We depicted the geographical patterns of observed phylogenetic and body size fields of species by mapping the mean value of these indices for all species occurring at each grid cell (mapping median values showed similar results – not shown to conserve space). In addition, we mapped the proportion of species showing significant phylogenetic and morphological structure within each grid cell. Finally, we correlated the proportional richness of species with significantly clustered phylogenetic fields against those with significantly clustered body size fields. To obtain unbiased estimates of significance tests, we used Clifford’s method to calculate geographically effective degrees of freedom and control for spatial autocorrelation using the modified t-test function of the SpatialPack library (Osorio & Vallejos, 2014) in R. RESULTS Phylogenetic fields of mammal species Phylogenetic structure of species co-occurrence within mammalian ranges did not follow a clear latitudinal but a strong longitudinal gradient (Fig. 1a). Observed PF patterns showed a clear aggregation of species around two different PSVsp values (see Fig. S1 in Appendix S2). Such groupings roughly corresponded to differences between Old and New World species, rather than tropical and temperate species. Overall, Old World mammals co-occur with more distantly related species, whereas New World mammals co-occur with more closely related species. Similarly, there was great variation on PSVsp values of species from different biogeographical realms. For instance, mammals within the Afrotropical, Madagascan, Oriental, Palearctic, Saharo-Arabian and SinoJapanese realms showed higher PSVsp values, whereas mammals inhabiting the Neotropical, Australian, Oceanian and 139

F. Villalobos et al.

Figure 1 Global pattern of mammalian phylogenetic fields. (a) Observed phylogenetic fields (PSVsp) gradient of mammal species. Values were averaged for species occurring within each grid cell. Large values suggest phylogenetic overdispersion and low values phylogenetic clustering. (b) Proportional species richness within grid cells of species with significantly clustered phylogenetic fields. (c) Proportional species richness within grid cells of species with significantly overdispersed phylogenetic fields. Data are depicted using equal intervals. Maps are in Mollweide equal-area projection.

Panamanian realms showed lower PSVsp values. Owing to the clear aggregation of PFs around two PSVsp values, we did not evaluate the statistical relationship between species PF and within-range richness. GLS models accounting for both phylogeny and space showed no significant effect of species latitudinal position (midpoint) on PF variation across all mammals (b = 0.00007, SE = 0.0001, R2 = 0.0001, P = 0.527). Similar results were obtained when analysing individual mammal orders (see Table S2 in Appendix S3). The independent contribution of space was higher compared to the purely phylogenetic effect and the variance independent of phylogeny and 0 space (/ = 0.782, k = 0.052, c = 0.165). Such contributions of space, phylogeny and variance independent of both varied among mammal orders. Some orders showed higher contribution of space (e.g. Carnivora and Rodentia), whereas other orders showed higher contribution of phylogeny (e.g. 140

Afrosoricida and Eulipotyphla) or a strong component of variance independent of phylogeny and space (e.g. Cingulata and Peramelemorphia). Mammalian PFs showed significant variation among biogeographical realms when accounting for the effects of space and phylogeny, across all species (F(11,3585) = 103.6, P < 0.001) as well as for most orders (see Table S3 in Appendix S3). For all species, independent contribution of space was moderate but higher than that of phylogeny and similar to the variance independent of both phylogeny and 0 space (/ = 0.483, k = 0.024, c = 0.493). For most mammalian orders (10 of 17), the variance independent of space and phylogeny was higher than the pure contributions of these factors, whereas for some orders space (Carnivora, Dasyuromorphia and Rodentia) or phylogeny (Afrosoricida, Cetartiodactyla and Primates) showed higher contribution (see Table S3 in Appendix S3). Similarly, when comparing Journal of Biogeography 44, 136–146 ª 2016 John Wiley & Sons Ltd

Global mammalian co-occurrence New and Old World, mammalian PFs across all species showed significant differences (F(2,3594) = 274.3, P < 0.001). This was also true for all the orders that occur on both hemispheres (see Table S4 in Appendix S3). Across all mammals, the independent contribution of space was higher than the effect of phylogeny and the variance independent of both 0 (/ = 0.646, k = 0.033, c = 0.321). Such independent contributions varied among mammalian orders (see Table S4 in Appendix S3). Some orders showed higher contribution of space (Carnivora, Chiroptera and Rodentia), whereas other orders showed higher contribution of phylogeny (Cetartiodactyla, Eulipotyphla and Primates) or a strong component of variance independent of phylogeny and space (Lagomorpha and Perissodactyla). Body size fields of mammal species Conversely to PFs, body size fields did not show a clear Old– New World structure, but instead followed a more common latitudinal gradient with low values over the tropics and higher values in temperate regions (Fig. 2a). GLS models accounting for phylogeny and space showed a strong negative relationship between the variance in body size ratios (VSRsp) and species richness within mammalian ranges (b = 1.306, SE = 0.01, R2 = 0.815, P < 0.001), with higher VSRsp (trait clustering) corresponding to mammals co-occurring with a low number of species, whereas lower VSRsp (trait overdispersion) corresponded to mammals co-occurring with a high number of species (Fig. 3). Results within individual mammalian orders followed those of the whole class (see Table S5 in Appendix S3 and Fig. S2 in Appendix S2). The independent contribution of space was higher compared to the purely phylogenetic effect and the independent effect of phylogeny and space (/ = 0.614, 0 k = 0.156, c = 0.229). These variance effects differed among mammal orders (see Table S5 in Appendix S3). Most orders showed higher contribution of the variance independent of space and phylogeny (e.g. Cingulata and Scandentia), whereas other orders showed higher contribution of phylogeny (e.g. Peramelemorphia and Primates) or space (e.g. Afrosoricida and Pilosa). Similar to phylogenetic fields, body size fields did not show a significant effect of their latitudinal position (b = 0.001, SE = 0.001, R2 = 0.0001, P = 0.569). All individual orders showed similar results (see Table S6 in Appendix S3). The independent contribution of space, phylogeny and the variance independent of both were low and 0 similar (/ = 0.396, k = 0.338, c = 0.266). These effects varied among mammal orders (see Table S6 in Appendix S3). Most orders showed higher contribution of the variance independent of space and phylogeny (e.g. Afrosoricida and Diprotodontia), whereas other orders showed higher contribution of phylogeny (e.g. Chiroptera and Perissodactyla) or space (e.g. Dasyuromorphia and Pilosa). Body size fields of species also showed significant variation among biogeographical realms when accounting for the Journal of Biogeography 44, 136–146 ª 2016 John Wiley & Sons Ltd

effects of space and phylogeny, across all species (F(11,3535) = 95.61, P < 0.001) as well as for most orders (see Table S7 in Appendix S3). For all species, independent contribution of phylogeny was moderate but higher than that of space and similar to the variance independent of both phy0 logeny and space (k = 0.476, / = 0.119, c = 0.405). For most mammalian orders, the variance independent of space and phylogeny was higher than the pure contributions of these factors (e.g. Carnivora and Cetartiodactyla), whereas for some orders, phylogeny (e.g. Perissodactyla and Scandentia) or space (e.g. Pilosa) showed higher contribution (see Table S7 in Appendix S3). Significant differences in body size fields were also observed when comparing between the New and Old Worlds across all species (F(2,3544) = 5.945, P = 0.003). Only two orders showed such differences between hemispheres (Carnivora and Perissodactyla), whereas the other orders did not (see Table S8 in Appendix S3). Across all mammals, independent contributions of space, phylogeny and the variance independent of 0 both were low and similar (/ = 0.380, k = 0.352, c = 0.268). Some orders showed higher contribution of space (Cetartiodactyla, Eulipotyphla and Rodentia), whereas only one showed higher contribution of phylogeny (Chiroptera) and the rest showed a strong component of variance independent of phylogeny and space (Carnivora, Lagomorpha, Perissodactyla, Primates) (see Table S8 in Appendix S3). Null models When applying the range-based null model, we found that most mammals have random PFs, with only few species exhibiting significantly clustered or overdispersed phylogenetic (12.88% and 17.15% of all mammals, respectively) or body size structure (10.39% and 5.9%) at the global scale, with similar findings within biogeographical realms (see Tables S9–S12 in Appendix S3). Still, species with significant phylogenetic or body size structure within their ranges showed considerable geographical patterning with different regions around the world being preferentially occupied by these species (i.e. representing a high proportion of the total number of species present in those regions) (Figs 1b,c & 2b,c). These patterns were spatially congruent between species with significantly clustered phylogenetic and body size fields (r = 0.564, P = 0.002). DISCUSSION Our species-oriented approach, focusing on complete geographical ranges of species and their co-occurrence patterns while considering phylogenetic and body size structure, illustrates the pervasive effect of broad-scale historical processes in determining mammalian species co-occurrence across the globe. We have empirically documented an Old to New World dichotomy in phylogenetic and body size structure within mammalian species ranges that highlights the 141

F. Villalobos et al.

Figure 2 Global pattern of body size structure within mammalian ranges. (a) Observed gradient of the variance in size ratios (VSRsp) within geographical ranges of mammal species. Values were averaged for species occurring within each grid cell. Large values suggest morphological clustering and low values morphological overdispersion. (b) Proportional species richness within grid cells of species with significantly clustered morphological structure. (c) Proportional species richness within grid cells of species with significantly overdispersed morphological structure. Data are depicted using equal intervals. Maps are in Mollweide equal-area projection.

influence of lineages’ origins and dispersal, over that of competition, in shaping niche partitioning among species and their phylogenetic neighbourhood at broad spatial scales. Inferring the relative roles of evolutionary and ecological processes on species co-occurrences can benefit from the integration of distributional, phylogenetic and trait data (Davies et al., 2007; Cooper et al., 2008). Seminal studies by Bowers & Brown (1982), Letcher et al. (1994) and Davies et al. (2007) evaluated the degree of range overlap and morphological similarity across species pairs, concluding that competition determined the distribution of closely related mammal species. Cooper et al. (2008) obtained similar results, but without considering trait data and focusing on assemblages instead of species pairs. In contrast with these studies, we did not find a strong role of competition on species co-occurrences and instead support the role of evolutionary processes. However, we do highlight the need for 142

integrating different kinds of data to disentangle between ecological and evolutionary processes. Instead of evaluating the degree of range overlap and morphological similarity across species pairs, we described the phylogenetic and body size structure of co-occurrence within species ranges (phylogenetic and body size fields, respectively). Such species fields varied geographically, with species occupying different regions showing distinct fields. This pattern, however, may be expected given that species occupying different regions would almost necessarily co-occur with distinct sets of species. Species fields are, thus, phylogenetically and geographically non-independent, requiring an appropriate control for these effects. Applying a framework that simultaneously controls for phylogenetic and spatial effects on species trait variation (Freckleton & Jetz, 2009) supported our findings on geographical variation in species fields. Although we did not find a significant effect of latitude (i.e. Journal of Biogeography 44, 136–146 ª 2016 John Wiley & Sons Ltd

Global mammalian co-occurrence

Figure 3 Relationship between (log) within-range species richness and (log) variance in size ratios (VSRsp).

species midpoints) on species fields, against our first expectation, we did find significant species field (phylogenetic and body size) variation across biogeographical realms and hemispheres supporting our second expectation. Moreover, species richness also influenced the variation in body size fields. Such species field variation implies differential effects of evolutionary and ecological processes across regions. The most striking pattern we found was the difference between the Old and New World species co-occurrences. Such difference was significant for both species fields across all mammals and individual orders, but particularly striking for phylogenetic fields (Fig. 1a). Accordingly, Old World mammals co-occur with more distantly related species, whereas New World mammals do so with more closely related species. This pattern reflects the distinct histories of both hemispheres (Davies & Buckley, 2011). Most mammalian orders probably originated in Laurasia and only a few did so in Gondwanan regions, mainly in Africa (most Afrotherian orders) and less so in South America (Xenarthra) (Springer et al., 2011). Consequently, the dominance of basal, distantly related lineages in the Old World seems to underlie the overdispersion of mammalian PFs. In contrast, the New World and historically isolated realms such as the Australian and Oceanian are composed by more closely related lineages, hence showing a more clustered pattern of mammalian PFs. Indeed, Davies & Buckley (2011) found support for the distinct histories between Old and New World defining contemporary patterns of mammal species richness and phylogenetic diversity. Our results support their findings, highlighting a differential effect and timing of evolutionary processes between distinct hemispheres of the world. This confirms that the signature of evolutionary processes is evident at the scale of complete species ranges and not only at level of grid-cell assemblages. Journal of Biogeography 44, 136–146 ª 2016 John Wiley & Sons Ltd

Predictions on the effects of evolutionary processes are not specific about broad-scale patterns in the distribution of morphological traits among co-occurring species. Instead, ecological processes are usually deemed responsible for structuring such traits at local but not regional scales (Dayan & Simberloff, 2005; Cardillo, 2011). Thus, it is unclear whether such local processes can in fact influence species assembly at broader spatial scales (Barnagaud et al., 2014). Here, we found similar geographical patterns between phylogenetic and body size fields of mammals, with no latitudinal gradient but a clear biogeographical and hemispheric (east–west) differentiation. In addition, for body size fields, we found that mammals co-occurring with high numbers of species do so with more dissimilar species (i.e. overdispersed body size fields) than those mammals co-occurring with low numbers of species and doing so with more similar species (i.e. clustered body size fields). This relationship between richness and body size similarity within ranges may support stronger competition within ranges of mammals occupying speciesrich regions like the Tropics. However, as with phylogenetic fields, the significance of body size fields is dependent on the species range sizes and their biogeographical pools. In fact, we found more complex patterns when looking at the geographical distribution of significant species fields. Taking into account biogeographically defined source pools, we found that most mammal species do not show significant phylogenetic or body size structure within their ranges. Nevertheless, species with significant phylogenetic or body size fields showed considerable geographical patterning and represented an important component of species assemblages in different regions of the world. This is the case for the Australian and southern Nearctic, as well as in the western Neotropics, where high proportions of species showed clustered phylogenetic fields. Conversely, mammals with overdispersed PFs represented high proportions of species present at the Neotropical, Palearctic, Sino-Japanese and Oceanian realms. In the same vein, geographical patterning of species with clustered body size fields dominate in the Nearctic region, whereas species with overdispersed body size fields do so in the Oceanian and southern Australian realms. Thus, our findings suggest additional effects structuring cooccurrence among mammals in different regions of the world beyond those already attributed to the differences between mammalian assemblages from the Old and New World. Two main conclusions can be derived from these findings. First, the fact that most species did not exhibit significant fields does support the effect of different evolutionary histories between the Old and New World mammalian biotas (Davies & Buckley, 2011), most likely because these histories are already taken into account in our null models. Second, after considering such different evolutionary histories between landmasses, species with clustered and overdispersed phylogenetic, as well as body size, structure are still found in distinct regions of the world. This suggests the action of evolutionary processes in particular locations and/or lineages within the larger biogeographical realms. That is, even within 143

F. Villalobos et al. regions where species are on average more closely (or distantly) related, closer inspection of their co-occurrence patterns reveals that some species do not conform to the regional pattern and co-occur with distantly (or closely) related species. It is important to note that this observation could not have been derived from estimating phylogenetic diversity or structure at the level of individual sites without considering complete geographical ranges of species. Further studies are, thus, required to account for these particular patterns. As with global species richness gradients, evolutionary processes such as diversification rates and niche conservatism could also explain patterns within specific regions. For instance, high proportions of species with clustered PFs in certain regions could result from higher net diversification rates of some (proportionally) dominant lineage(s) and/or the accumulation of species through time within their region of origin (e.g. rodents in the southern Nearctic; Riddle et al., 2000; Stevens et al., 2012). Regarding the high proportions of species with overdispersed PFs, processes related to dispersal events and/or the diversification of distantly related lineages within specific regions could explain such patterns. This may be the case, for example, in regions of the Neotropics. The asymmetrical mixture of North and South American faunas through the Great American Biotic Interchange (Woodburne et al., 2006), with more influence of the northern to the southern fauna than the reverse (MoralesCastilla et al., 2012), could explain the high proportion of species with overdispersed PFs in the east of the Neotropical realm. Such faunal mixtures and colonization by different lineages could also explain the similar pattern in the southern Paleartic and Sino-Japanese realms. Another potential explanation can be related to the preponderance of largerange species in those Old World regions, which may cooccur with species from different lineages across their ranges. These widespread species, therefore, show overdispersed PFs and contribute significantly to the proportional richness of some of the regions they occupy. Interestingly, the geographical patterning of species with significantly clustered phylogenetic fields corresponded to the one presented by species showing significantly clustered body size fields. Such spatial congruence supports our third expectation that mammals co-occurring with close relatives are also doing so with species of similar body sizes, therefore, producing co-occurrence patterns that are both clustered phylogenetically and in body size structure in the same geographical regions. Thus, at the broad scale of our study, no clear evidence of competitive exclusion driving the distribution of body sizes among co-occurring species could be detected. Instead, the evolutionary history of mammalian lineages, including their origin in different regions, dispersal among these regions and diversification history, has greatly contributed to the observed phylogenetic and body size structure of species co-occurrence. Finally, the use of an integrated framework jointly considering phylogenetic and spatial effects on species trait 144

variation reinforced the relevance of our species-oriented approach. Such framework allowed us to demonstrate that species fields are indeed informative beyond their phylogenetic and spatial non-independence, supporting geographical variation on evolutionary processes driving species co-occurrences. In fact, most of our results from this modelling framework showed that the spatial effect was stronger than the phylogenetic one. This could be interpreted as species fields being driven by where species live instead of their particular evolutionary history (Machac et al., 2011), perhaps highlighting the historic legacy of speciation (Cardillo, 2015). This interpretation does not mean that evolutionary history was not important but rather imply its effect in determining the origin and dynamics of mammalian lineages within particular regions that consequently define the spatial distribution of species. Such insights emphasize the relevance of a species-oriented approach linking evolutionary and ecological data that, along with the traditional assemblage-oriented approach, can foster our understanding on species geographical co-occurrences driving broad-scale biodiversity patterns. ACKNOWLEDGEMENTS We thank B.A. Hawkins, M.A. Rodrıguez and L. Jardim for discussions and G. Thomas for suggestions on an earlier version of this article. We also thank Nick Gotelli, Holger Kreft, Kamran Safi and an anonymous reviewer for their comments that help improved our article. F.V. was supported by CNPq PDJ and BJT ‘Science without borders’ grants. M.V.C., T.F.R. and J.A.F.D.-F. have been continuously supported by CNPq productivity grants. M.A.O.-T. was supported by an AEET 2012 Young Investigator Grant.

REFERENCES Barnagaud, J.Y., Daniel Kissling, W., Sandel, B., Eiserhardt, W.L., Sßekercioglu, C ß .H., Enquist, B.J., Tsirogiannis, C. & Svenning, J.C. (2014) Ecological traits influence the phylogenetic structure of bird species cooccurrences worldwide. Ecology Letters, 17, 811–820. Barraclough, T.G., Vogler, A.P. & Harvey, P.H. (1998) Revealing the factors that promote speciation. Philosophical Transactions of the Royal Society B: Biological Sciences, 353, 241–249. Bowers, M.A. & Brown, J.H. (1982) Body size and coexistence in desert rodents: chance or community structure? Ecology, 63, 391–400. Brown, J.H. & Nicoletto, P.F. (1991) Spatial scaling of species composition: body masses of North American land mammals. The American Naturalist, 138, 1478–1512. Brown, W.L. & Wilson, E.O. (1956) Character displacement. Systematic Zoology, 5, 49–64. Buckley, L.B., Davies, T.J., Ackerly, D.D., Kraft, N.J.B., Harrison, S.P., Anacker, B.L., Cornell, H.V., Damschen, E.I., Grytnes, J.A., Hawkins, B.A., McCain, C.M., Stephens, P.R. Journal of Biogeography 44, 136–146 ª 2016 John Wiley & Sons Ltd

Global mammalian co-occurrence & Wiens, J.J. (2010) Phylogeny, niche conservatism and the latitudinal diversity gradient in mammals. Proceedings of the Royal Society B: Biological Sciences, 277, 2131–2138. Cardillo, M. (2011) Phylogenetic structure of mammal assemblages at large geographical scales: linking phylogenetic community ecology with macroecology. Philosophical Transactions of the Royal Society B: Biological Sciences, 366, 2545–2553. Cardillo, M. (2015) Geographic range shifts do not erase the historic signal of speciation in mammals. The American Naturalist, 185, 343–353. Cavender-Bares, J., Kozak, K.H., Fine, P.V. & Kembel, S.W. (2009) The merging of community ecology and phylogenetic biology. Ecology Letters, 12, 693–715. Cooper, N. & Purvis, A. (2010) Body size evolution in mammals: complexity in tempo and mode. The American Naturalist, 175, 727–738. Cooper, N., Rodrıguez, J. & Purvis, A. (2008) A common tendency for phylogenetic overdispersion in mammalian assemblages. Proceedings of the Royal Society B: Biological Sciences, 275, 2031–2037. Davies, T.J. & Buckley, L.B. (2011) Phylogenetic diversity as a window into the evolutionary and biogeographic histories of present-day richness gradients for mammals. Philosophical Transactions of the Royal Society B: Biological Sciences, 366, 2414–2425. Davies, T.J., Meiri, S., Barraclough, T.G. & Gittleman, J.L. (2007) Species co-existence and character divergence across carnivores. Ecology Letters, 10, 146–152. Davies, T.J., Cooper, N., Diniz-Filho, J.A.F., Thomas, G.H. & Meiri, S. (2012) Using phylogenetic trees to test for character displacement: a model and an example from a desert mammal community. Ecology, 93, S44–S51. Dayan, T. & Simberloff, D. (2005) Ecological and community-wide character displacement: the next generation. Ecology Letters, 8, 875–894. Fine, P.V. (2015) Ecological and evolutionary drivers of geographic variation in species diversity. Annual Review of Ecology, Evolution, and Systematics, 46, 369–392. Freckleton, R.P. & Jetz, W. (2009) Space versus phylogeny: disentangling phylogenetic and spatial signals in comparative data. Proceedings of the Royal Society B: Biological Sciences, 276, 21–30. Fritz, S.A., Bininda-Emonds, O.R. & Purvis, A. (2009) Geographical variation in predictors of mammalian extinction risk: big is bad, but only in the tropics. Ecology Letters, 12, 538–549. Gerhold, P., Cahill, J.F., Winter, M., Bartish, I.V. & Prinzing, A. (2015) Phylogenetic patterns are not proxies of community assembly mechanisms (they are far better). Functional Ecology, 29, 600–614. Gotelli, N.J. (2000) Null model analysis of species co-occurrence patterns. Ecology, 81, 2606–2621. Gotelli, N.J., Anderson, M.J., Arita, H.T. et al. (2009) Patterns and causes of species richness: a general simulation model for macroecology. Ecology Letters, 12, 873–886. Journal of Biogeography 44, 136–146 ª 2016 John Wiley & Sons Ltd

Guisan, A. & Rahbek, C. (2011) SESAM–a new framework integrating macroecological and species distribution models for predicting spatio-temporal patterns of species assemblages. Journal of Biogeography, 38, 1433–1444. Harvey, P.H. & Pagel, M.D. (1991) The comparative method in evolutionary biology. Oxford University Press, Oxford. Hawkins, B.A., McCain, C.M., Davies, T.J., Buckley, L.B., Anacker, B.L., Cornell, H.V., Damschen, E.I., Grytnes, J.A., Harrison, S., Holt, R.D., Kraft, N.J.B. & Stephens, P.R. (2012) Different evolutionary histories underlie congruent species richness gradients of birds and mammals. Journal of Biogeography, 39, 825–841. Helmus, M.R., Bland, T.J., Williams, C.K. & Ives, A.R. (2007) Phylogenetic measures of biodiversity. The American Naturalist, 169, E68–E83. Hillebrand, H. (2004) On the generality of the latitudinal diversity gradient. The American Naturalist, 163, 192–211. Holt, B.G., Lessard, J.-P., Borregaard, M.K., Fritz, S.A., Ara ujo, M.B., Dimitrov, D., Fabre, P.-H., Graham, C.H., Graves, G.R. & Jønsson, K.A. (2013) An update of Wallace’s zoogeographic regions of the world. Science, 339, 74–78. IUCN (2009) IUCN red list of threatened species version 2009. 1. Available at: http://www.iucnredlist.org. Jones, K.E. & Safi, K. (2011) Ecology and evolution of mammalian biodiversity. Philosophical Transactions of the Royal Society B: Biological Sciences, 366, 2451–2461. Jones, K.E., Bielby, J., Cardillo, M., Fritz, S.A., O’Dell, J., Orme, C.D.L., Safi, K., Sechrest, W., Boakes, E.H. & Carbone, C. (2009) PanTHERIA: a species-level database of life history, ecology, and geography of extant and recently extinct mammals: Ecological Archives E090-184. Ecology, 90, 2648–2648. Kembel, S.W. (2009) Disentangling niche and neutral influences on community assembly: assessing the performance of community phylogenetic structure tests. Ecology Letters, 12, 949–960. Letcher, A.J., Purvis, A., Nee, S. & Harvey, P.H. (1994) Patterns of overlap in the geographic ranges of Palearctic and British mammals. Journal of Animal Ecology, 63, 871–879. MacArthur, R. & Levins, R. (1967) The limiting similarity, convergence, and divergence of coexisting species. The American Naturalist, 101, 377–385. Machac, A., Zrzavy , J. & Storch, D. (2011) Range size heritability in Carnivora is driven by geographic constraints. The American Naturalist, 177, 767–779.  Purvis, A., HawMorales-Castilla, I., Olalla-Tarraga, M.A.,  (2012) The imprint of Cenokins, B.A. & Rodrıguez, M.A. zoic migrations and evolutionary history on the biogeographic gradient of body size in New World mammals. The American Naturalist, 180, 246–256. Osorio, F. & Vallejos, R. (2014) SpatialPack: Package for analysis of spatial data. R package version 0.2-3. Available at: http://cran.r-project.org/package=SpatialPack. Pleasants, J.M. (1994) A comparison of test statistics used to detect competitive displacement in body size. Ecology, 75, 847–850. 145

F. Villalobos et al. Qian, H., Wiens, J.J., Zhang, J. & Zhang, Y. (2014) Evolutionary and ecological causes of species richness patterns in North American angiosperm trees. Ecography, doi:10.1111/ecog.00952. 38, 241–250. Ricklefs, R.E. (2004) A comprehensive framework for global patterns in biodiversity. Ecology Letters, 7, 1–15. Ricklefs, R.E. & Miles, D.B. (1994) Ecological and evolutionary inferences from morphology: an ecological perspective. Ecological morphology: integrative organismal biology (ed. by P. Wainwright and S.M. Reilly), pp. 13–41. University of Chicago Press, Chicago. Riddle, B.R., Hafner, D.J. & Alexander, L.F. (2000) Phylogeography and systematics of the Peromyscus eremicus species group and the historical biogeography of North American warm regional deserts. Molecular Phylogenetics and Evolution, 17, 145–160. Rolland, J., Condamine, F.L., Jiguet, F. & Morlon, H. (2014) Faster speciation and reduced extinction in the tropics contribute to the mammalian latitudinal diversity gradient. PLoS Biology, 12, e1001775.  & Hawkins, B.A. (2013) IdentiRueda, M., Rodrıguez, M.A. fying global zoogeographical regions: lessons from Wallace. Journal of Biogeography, 40, 2215–2225. Safi, K., Cianciaruso, M.V., Loyola, R.D., Brito, D., ArmourMarshall, K. & Diniz-Filho, J.A.F. (2011) Understanding global patterns of mammalian functional and phylogenetic diversity. Philosophical Transactions of the Royal Society B: Biological Sciences, 366, 2536–2544. Springer, M.S., Meredith, R.W., Janecka, J.E. & Murphy, W.J. (2011) The historical biogeography of Mammalia. Philosophical Transactions of the Royal Society B: Biological Sciences, 366, 2478–2502. Stevens, R.D., Gavilanez, M.M., Tello, J.S. & Ray, D.A. (2012) Phylogenetic structure illuminates the mechanistic role of environmental heterogeneity in community organization. Journal of Animal Ecology, 81, 455–462. Swenson, N.G. (2009) Phylogenetic resolution and quantifying the phylogenetic diversity and dispersion of communities. PLoS ONE, 4, e4390. Villalobos, F. & Arita, H.T. (2010) The diversity field of New World leaf-nosed bats (Phyllostomidae). Global Ecology and Biogeography, 19, 200–211. Villalobos, F., Rangel, T.F. & Diniz-Filho, J.A.F. (2013) Phylogenetic fields of species: cross-species patterns of

146

phylogenetic structure and geographical coexistence. Proceedings of the Royal Society B: Biological Sciences, 280, 20122570. Warren, D.L., Cardillo, M., Rosauer, D.F. & Bolnick, D.I. (2014) Mistaking geography for biology: inferring processes from species distributions. Trends in Ecology and Evolution, 29, 572–580. Woodburne, M.O., Cione, A.L. & Tonni, E.P. (2006) Central American provincialism and the Great American Biotic Interchange. Advances in late tertiary vertebrate paleontology in Mexico and the Great American Biotic Interchange (ed. by O. Carranza-Casta~ neda and E.H. Lindsay), pp. 73– 101. UNAM, Mexico City. SUPPORTING INFORMATION Additional Supporting Information may be found in the online version of this article: Appendix S1 Species numbers and mammalian orders used. Appendix S2 Relationship between species fields and within-range richness. Appendix S3 Spatial and phylogenetic models and null model comparisons. BIOSKETCH Fabricio Villalobos is a full researcher at Instituto de Ecologıa, A.C. in Mexico and a collaborating professor of the Universidade Federal de Goias, Brazil. His research focuses on geographical patterns of biodiversity integrating tools and concepts from macroecology and evolutionary biology. He is also interested in theory development and testing. Author contributions: F.V., M.A.O.-T. and J.A.F.D.-F. conceived the idea; M.V.C. compiled and provided data; T.F.R. developed software and analyses; F.V. performed most analyses and led the writing with substantial input from the other authors.

Editor: Holger Kreft

Journal of Biogeography 44, 136–146 ª 2016 John Wiley & Sons Ltd