Predictability and conceptual repeatability of the predator-associated burst speed ecophenotype in western...

Predation exerts strong selection pressure on morphological traits and is often studied in freshwater fishes. A common morphological ecophenotype driven by predation from pursuit predators is the predator-associated burst speed (PABS) ecophenotype. This ecophenotype, characterized by a smaller head, smaller body, and larger caudal region, is commonly found in western mosquitofish (Gambusia affinis) residing in environments with sunfish (family Centrarchidae) predators. However, the repeatability and transferability of the PABS ecophenotype across populations have not been tested. The four objectives of this study were to (1) assess the repeatability of methods to confirm the presence of the PABS ecophenotype in independent populations, (2) test whether morphology is correlated with predator abundance, (3) assess the repeatability of utilizing a discriminant function analysis (DFA) to accurately classify fish to predator status, and (4) use a DFA to conduct a cross-validation test of the PABS model using previously studied populations of G. affinis to predict predator status of newly studied populations. There was consistency in the occurrence of the PABS ecophenotype among populations with predators. Permuted MANOVA (PMANOVA) models fit to each dataset revealed a significant effect of collection site nested within predator status (i.e., presence or absence of predator), suggesting location-specific variations of the PABS ecophenotype. In addition, a PMANOVA model fit to body shape as a function of predator abundance (i.e., 0, 0.1-1.0, or >1.0 predators per minute of electrofishing) revealed a significant effect of predation level in the newly studied populations, suggesting morphology differs among populations with varying predator abundances. Cross-study validation of the DFA revealed low between-study accuracy compared to within-study accuracy, but elongation of the caudal region in the presence of predators was consistent across studies. Our findings ultimately suggest that although the PABS ecophenotype at studied locations provides partial predictive capacity at unstudied locations, the nature and magnitude of the PABS ecophenotype depend on environmental settings, encounter histories with predators, level of abundance of predators, or other spatially structured mechanisms.