Predictability and Parallelism of Multitrait Adaptation.

Environments shape the traits of organisms. Environmental variation may rarely alter selection on only a few traits, but instead precipitate wholesale changes of the multidimensional selective regime-many traits might experience divergent selection across divergent environments. Such changes in selection can elicit multifarious evolution. How predictable (from theory) and how parallel (consistent occurrences) is multitrait divergence across replicated environments? Here, I address this question using the post-Pleistocene radiation of Bahamas mosquitofish (Gambusia hubbsi) inhabiting blue holes on Andros Island. These fish independently colonized numerous blue holes, some that harbor a major fish predator (bigmouth sleeper, Gobiomorus dormitor) and some that lack any major predators. I used 5 approaches to quantitatively explore the predictability and parallelism of multitrait divergence between predation regimes in Bahamas mosquitofish. Synthesizing data for 90 traits from 13 different types of character suites (e.g., body morphology, life history, genital morphology, coloration, mating preference, habitat use), I found widespread evidence for strong, predictable, and parallel divergence between predation regimes. Yet despite the great majority of traits showing predictable trajectories of change, and the majority of traits showing significant parallelism and strong magnitudes of predictable divergence, I uncovered that over half of the overall phenotypic variation among populations was not driven by variation in predation regime. Results suggest that focusing on few traits, or focusing on parallel aspects of divergence, can provide a misleading picture of adaptation, and nonparallel divergence appears widespread and warrants greater attention. Taking a multitrait perspective, and quantifying predictability and parallelism, can yield important insights. © The American Genetic Association 2017. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.