Bayesian hierarchical models

Frugivore traits predict plant-frugivore interactions using generalized joint attribute modeling
Under an adaptive hypothesis, the reciprocal influence between mutualistic plants and frugivores is expected to result in dispersal syndromes comprising both frugivore and plant traits that structure fruit consumption. How frugivore traits and within-species variation contribute to structuring dispersal syndromes is, however, often ignored. To address both gaps, we analyze traits for the mutualistic ecological network comprising Carollia bats that feed on and disperse Piper seeds. We used generalized joint attribute modeling (GJAM), a Bayesian modeling approach that simultaneously accounts for multiple sources of variance across trait types. In support of the adaptive hypothesis and indicating niche partitioning among Carollia bats, we find differential consumption of a suite of Piper species influenced by bat traits such as body size; however, Piper morphological traits had no effect on bat consumption. Slow evolutionary rates, dispersal by other vertebrates, and unexamined fruit traits, such as Piper chemical bouquets, may explain the lack of association between bat Piper consumption and fruit morphological traits. We have identified a potential asymmetric influence of frugivore traits on plant-frugivore interactions, providing a template for future trait analyses of plant-animal networks. As intraspecific trait variation is rarely included in studies on trait matching, this paper contributes to closing that important knowledge gap.