Geometric morphometrics meets metacommunity ecology: environment and lineage distribution affects spatial variation in shape

Patterns of univariate trait variation across metacommunities are widely explored, as are searches for their underlying causes. Surprisingly, patterns of multivariate shape remain unknown, and the search for drivers of functional traits of communities often neglect the biogeographical distribution of phylogenetic clades. Our aim was to investigate multivariate shape distribution across metacommunities and to determine the main environmental drivers of shape beyond/taking into account the phylogenetic distribution of lineages. We obtained mean skull and mandible shape for 228 species of Neotropical sigmodontine rodents through geometric morphometrics (GM), and then calculated mean shapes for 1 degrees x 1 degrees cells across the Neotropics based on the incidence of sigmodontines. We investigated the effects of lineage distribution on mean trait variation by using phylogenetic fuzzy weighting to calculate principal coordinates of phylogenetic structure (PCPS). Effects of environmental variables on shape variation incorporating phylogenetic composition were realized through redundancy analysis. We found that the different distributions of major lineages throughout the Neotropics were responsible for much of the mean shape variation. The association of landscape features with tribal groupings (Oryzomyini with Amazonia and Phyllotini and Abrotrichini with the Andes) were standouts. Environmental variables and lineage distribution explain the same (i.e. shared) portion of shape variation, suggesting phylogenetic niche conservatism at the metacommunity level. Seasonality in temperature and land cover were the best environmental predictors of mean shape: larger tympanic bullae, incisive foramina, and check teeth are all associated with highly seasonal and less vegetated areas. Our new approach of using GM shape across metacommunities was demonstrably useful in understanding large-scale biogeographical patterns of shape variation and identifying its underlying causes. The overlap between environmental variables and phylogenetic lineage distribution suggests that a process of niche conservatism is likely: the phenotype-environment correlation is mediated by the differential biogeographical distribution of the main clades.