Exploring male and female mate preferences using genetic algorithms and digital phenotypes

Abstract During mate choice, both males and females can exhibit choosiness when selecting a partner. The expression and measurement of preferences is typically constrained by the traits presented during mate-choice trials, which generally represent a subset of the phenotypic trait space and thus may not capture the full range of potential preferences. Here, we present a proof of concept for a novel method building on the commonly used experimental protocol of screen-based stimuli for examining mate preferences. We used genetic algorithms to allow males and females to iteratively select for virtual mates displaying a range of phenotypes, using green swordtail fish (Xiphophorus hellerii). We found that males actively selected for digital females with larger body size. By contrast, female-driven searches did not converge on one particular phenotype, despite evidence that females preferred their final iterated phenotype. Simulated preference models confirmed that these outcomes could not be explained by random drift alone. This suggests the possibility of dispersed or individual female preference structures rather than one population-level optimum. Our results demonstrate that iterative search methods can reveal nuanced preferences, including some that are difficult to detect with standard paired designs, offering a flexible and powerful approach for studying mate choice across complex trait spaces.