Winner and loser effects: a meta-analysis

Aggressive interactions can strongly influence an animal's performance in subsequent contests. Winners of aggressive contests are more likely to win successive contests and losers are more likely to lose successive contests. Such winner and loser effects can significantly influence an animal's dominance status, ability to acquire resources and reproductive success. Thus, quantifying the magnitudes of winner and loser effects across taxa is important for our understanding of hierarchy formation, life history trade-offs and reproductive tactics in different species. Furthermore, it is unclear whether the magnitude of winner effects differ from that of loser effects. Finally, experimenters often employ one of two distinct methods for quantifying the strength of winner and loser effects: self-selection and random assignment. Due to selection bias, it is possible that self-selection protocols overestimate the magnitude of winner and loser effects. We therefore systematically searched the literature to conduct a comprehensive meta-analysis of winner and loser effects. We analysed a total of 168 effect sizes from arachnids, crustaceans, fishes, insects, mammals and reptiles. We found that prior winners tend to win approximately two-thirds of their subsequent fights, while prior losers tend to lose approximately two-thirds of their subsequent fights. While we did not find that studies using self-selection generated effect size estimates that significantly differed from random assignment protocols, future studies should still avoid self-selection protocols. Overall, our study highlights the ubiquity of winner and loser effects across the animal kingdom and suggests several avenues for future research to unravel the evolutionary origins and mechanistic underpinnings of such experience effects.