RT Journal Article
SR Electronic
T1 Maximum entropy models elucidate the contribution of metabolic traits to patterns of community assembly
JF bioRxiv
FD Cold Spring Harbor Laboratory
SP 526764
DO 10.1101/526764
A1 Jason Bertram
A1 Erica A Newman
A1 Roderick Dewar
YR 2019
UL http://biorxiv.org/content/early/2019/01/22/526764.abstract
AB Aim: Maximum entropy (MaxEnt) models promise a novel approach for understanding community assembly and species abundance patterns. One of these models, the "Maximum Entropy Theory of Ecology" (METE) reproduces many observed species abundance patterns, but is based on an aggregated representation of community structure that does not resolve species identity or explicitly represent species-specific functional traits. In this paper, METE is compared to "Very Entropic Growth" (VEG), a MaxEnt model with a less aggregated representation of community structure that represents species (more correctly, functional types) in terms of their per capita metabolic rates. We examine the contribution of metabolic traits to the patterns of community assembly predicted by VEG and, through aggregation, compare the results with METE predictions in order to gain insight into the biological factors underlying observed patterns of community assembly. Innovation: We formally compare two MaxEnt-based community models, METE and VEG, that differ as to whether or not they represent species-specific functional traits. We empirically test and compare the metabolic predictions of both models, thereby elucidating the role of metabolic traits in patterns of community assembly. Main Conclusions: Our analysis reveals that a key determinant of community metabolic patterns is the "density of species" distribution, defined as the intrinsic number of species with metabolic rates in a given range that are available to a community prior to filtering by environmental constraints. Our analysis suggests that appropriate choice of of the density of species in VEG may lead to more realistic predictions than METE, for which this distribution is not defined, and thus opens up new ways to understanding the link between functional traits and patterns of community assembly.