Fish Designed Their Own Ideal Mates, And Males Wanted Bigger Females

Male Swordtails Chose Bigger Digital Mates in a Fish Dating Game, but Females Never Agreed on a Type

Green swordtail fish played matchmaker in a video dating game, deciding which digital mates survived, and the males steered their ideal partner toward bigger bodies. After nine rounds of choosing among computer-generated mates, the digital females that males had shaped averaged about 24% larger in body area than the random group the experiment began with. Female swordtails, given the same setup with digital males, never settled on any shared favorite.

Published in Royal Society Open Science, the study borrowed a technique from computer science called a genetic algorithm, a program that mimics breeding. Real fish chose among digital mates, the favored ones produced digital offspring, and the rest were dropped. Kathryn Bullough, the study’s first author, said the approach “allowed fish to ‘design’ their ideal partner.” Traits that won approval carried into the next batch of candidates.

Mate-choice experiments usually offer an animal two options picked by scientists, which means a preference for something unexpected can slip past unnoticed. This approach let the fish’s own choices guide the search instead. Females supplied the surprise, because earlier research had found that female green swordtails generally favor larger males.

Fish Made 1,242 Swordtail Mate Choices Across Nine Rounds of Digital Breeding

Researchers at the University of Exeter in England photographed one male and one female green swordtail, a species that responds to video images much like it responds to live fish. Video lets scientists control exactly how each mate looks. Each photo was split into parts, including the eye, the body, and, for males, the stripes and the long sword-shaped tail. Software recombined the parts into animated fish that swam across a screen, and male versions added a tail-wiggling courtship move.

Each digital fish was built from 16 numerical settings, called genes, that controlled traits such as body size, body height, color, and sword length. Those settings stretched beyond the ranges seen in real swordtails, opening up far more possible looks than a scientist would think to test.

Twelve sexually mature males and twelve females started the study, all lab-bred descendants of fish collected in Mexico. One female was euthanized after the fourth round because of an illness unrelated to the study, leaving eleven females who finished. Each fish watched pairs of digital mates on tablet screens at opposite ends of a tank, and whichever one it spent more time near won that matchup. Winners were randomly paired and combined, with a small chance of a random tweak, to create the next batch. Nine rounds meant 54 pairs per fish and 1,242 choices in all. Fish completed two rounds per week, so a full run took around five weeks. Co-author Laura Kelley noted that “A study like this using real fish would take years.”

Computer simulations added a point of comparison. These versions picked winners strictly by size, strictly by color intensity, or by pure luck, letting the team judge whether the fish’s results looked like real selection or chance.

Male Swordtails Steered Toward Larger Females but Showed No Color Shift

That size shift was real. Simulated picking by luck produced nothing similar, and color did not budge from the starting group, so males as a whole showed no detectable color preference.

A follow-up test backed up the pattern. Each male saw six top digital females from the final round paired against six females rejected in the first round. Males averaged about 54 seconds near the final-round females versus about 37 seconds near the rejected ones.

Bigger females tend to carry more eggs in related fish species, a link that could explain why males favored size. That choosiness stands out in a group of fish where males compete hard for mates, but the authors suggest that green swordtails live in rivers where a male is likely to meet many females, so being picky carries little risk.

Female Swordtail Mate Choice Never Settled on One Type of Male

Females told a different story. After nine rounds of choosing digital males, the final group of male images differed from the starting group by too little to rule out chance, with average body area only about 5.7% smaller.

Yet females as a group were not choosing at random. Their results differed from the luck-based simulation, and in the follow-up test, females averaged about 50 seconds near final-round males versus about 37 seconds near the rejected ones.

Why the group never settled on one type remains unclear. The authors list relaxed female preferences for male traits, no single best combination of traits, or individual females favoring different combinations. Female preferences in general tend to vary more than male preferences, which the authors link to males gaining a direct payoff from choosing fertile mates. Individual tastes could be behind the pattern, but this experiment cannot confirm that.

A larger group of females tested over more rounds would show whether a group-wide pattern emerges.

Male swordtails, as a group, followed a simple rule, and the rule was size. Female swordtails did not, which leaves the question of female taste wide open. The authors call their method a “proof of concept,” and a dating game for fish may be what eventually answers it.

---