D3157: Simulating and understanding trait evolution using insect phylogeny case studies

Understanding the mechanisms that have shaped the staggering diversity of insects and their biology is a fundamental goal of evolutionary biology. Statistical models of trait evolution hold great promise for elucidating the processes the shape insect traits while considering the phylogenetic relationships of target organisms. Yet many analytical and computational challenges nonetheless await those seeking to study the insect trait evolution.We developed TraitTrainR, a new software application designed as a streamlined, user-friendly framework for large-scale, flexible, and fast simulations of trait evolution under complex processes. TraitTrainR includes a number of user-defined options for customized applications and improved capabilities, including flexible output formats, variability in model parameters across replicates, models of trait measurement error, and multi-trait scenarios.To demonstrate the utility of TraitTrainR, we applied our approach to several phylogenetic case studies focusing on beetles, invasive insects, and the greater Arthropod phylogeny. Within seconds, TraitTrainR generated millions of replicates under canonical models of trait evolution that have been used previously to study insect diversity.While we envision many applications for our approach, here we applied TraitTrainR specifically to illustrate the task of correct evolutionary model selection using insect phylogenies as case studies. Our demonstration revealed both power and pitfalls for predicting the correct model using traditional procedures, suggesting challenges for analyzing trait data sampled from nature, particularly for certain trees and when traits are imprecisely measured. Future work seeks to leverage these advances to better understand model expectations and their adequacy for studying evolution in insects and beyond.