A new study suggests that decorative traits used to attract mates evolve more quickly than the physical weapons males rely on in combat, based on an analysis spanning 531 species across 10 animal groups. The researchers combined a mathematical model with broad comparative data and found ornaments generally outpace weapons in evolutionary speed, a pattern they attribute largely to shifting mating preferences across generations.
Sexual selection operates via two main routes. Direct competition within the same sex—most often male rivals fighting for access to mates—favors the evolution of weapons such as tusks or enlarged canine teeth. By contrast, mate choice—typically female selection of partners—favors ornaments, including vibrant plumage or distinctive vocal traits.
Biologists have long noted that traits shaped by sexual selection can change faster across evolutionary time than ordinary body parts. The new work builds on an idea proposed by Ronald Fisher and colleagues in 1981, which suggested a positive feedback between female preferences and male ornament size could drive rapid, self-reinforcing evolutionary bursts. The current study sought to test whether this dynamic translates into different speeds for ornaments versus weapons.
“The initial idea came from an excellent paper by Erin McCullough, Christine Miller, and Doug Emlen, ‘Why sexually selected weapons are not ornaments,’” said William L. Allen of Swansea University, one of the joint senior authors. “Among its several important contributions was the proposal that the evolutionary dynamics of weapons should differ fundamentally from those of ornaments. In particular, because the Fisher process — positive feedback between female mating preferences and male ornament traits — does not operate in the same way for weapons, the authors predicted that weapons should evolve more slowly than ornaments. They also noted that this prediction had never been tested.”
Allen and colleagues further explained that their aim was to “develop a theoretical framework for asking whether ornaments and weapons should evolve at different rates, and then test that prediction using comparative data from across the animal kingdom.”
In the theoretical portion, the team extended the 1981 coevolution model to track three inheritable traits across generations: male ornament size, male weapon size, and female preference strength for the ornament. The model assumed an effective population size of 1,000, roughly the number of individuals contributing genes to each new generation. All traits carried survival costs, meaning larger ornaments or weapons or paying the costs of being a picky chooser reduced survival fitness.
Under the model, weapons consistently provided a mating advantage, while ornament size depended on the average female preference in the population. The researchers also incorporated genetic drift—the random shifts in gene frequencies that occur in finite populations—so that trait averages could wander around a long-term balance point. They measured short-term evolutionary speed by the width of those fluctuations.
Across a wide set of simulated conditions, the model predicted that male ornaments evolve faster than male weapons. Weapons remained relatively stable near their balance point because their payoff did not hinge on shifting female tastes, whereas ornaments fluctuated whenever drift moved average female preferences.
When female preferences were artificially held constant in the model, ornaments slowed to the pace of weapons. Yet even without strong genetic coupling between ornaments and preferences, drifting preferences alone could accelerate ornament evolution. The researchers also found that if weapons and their correlations with ornaments and preferences were introduced, ornaments still tended to outpace weapons, unless certain conditions reversed the advantage—such as strong, survival-heavy preferences limiting drift or weapons greatly cheaper to produce than ornaments.
To test whether real-lineages mirror the theoretical dynamics, the team assembled published trait measurements and time-calibrated evolutionary trees, covering 531 species in 10 data sets. They restricted analyses to groups where a trait primarily functioned as a weapon or as an ornament, excluding multi-purpose structures such as fiddler crab claws. Five paired comparisons linked related ornament- and weapon-bearing taxa with similar generation times: leg spine length in leaf-footed bugs paired with eye-stalk span in stalk-eyed flies; bite force in anole lizards paired with throat fan support bone length in agamid lizards; leg spur length in landfowl paired with neck-feather colour; tusk length in large mammals paired with vocal pitch in carnivorans; and canine tooth length in primates paired with vocal pitch in primates.
After standardising trait sizes for body size, the researchers ran evolutionary regressions under two models of trait change on the family trees: Brownian motion (a random-walk process) and Ornstein-Uhlenbeck (incorporating stabilising selection toward an optimum). The Ornstein-Uhlenbeck model provided a better fit for five of the 10 datasets and performed similarly to Brownian motion for four others; one dataset (anole bite force) diverged.
When ornaments and weapons were analysed using the better-fitting Ornstein-Uhlenbeck framework, ornaments evolved faster than weapons in four of the five matched pairs, with the insect comparison standing out as an exception where leg-spine spines evolved faster than stalk-eyed fly eyespans. Combining all five pairs, ornaments appeared to evolve at roughly twice the pace of weapons, with standardized rate estimates of about 0.129 for ornaments versus 0.064 for weapons.
Overall, estimated evolutionary speeds varied widely across the 10 datasets, ranging from 0.02 to 0.19 under the stabilising model. “We observed large differences in the rate of trait evolution in different groups of animals, only some of which was explained by whether the trait was a weapon or ornament,” Allen noted.
The study found that the difference between ornaments and weapons in the strength of the stabilising pull toward an optimum was not statistically significant. The authors suggest that shifts in optimal ornament size driven by drifting female preferences may be absorbed into the observed fluctuations in ornaments. They also caution that removing body size from analyses can introduce biases into rate estimates, particularly when using regression residuals to standardise across species.
When the data were evaluated under Brownian motion or when body size was not adjusted, the overall rate difference between ornaments and weapons lost statistical significance. The authors argue that keeping body size in the measurements may dilute differences between sexual traits themselves, since body size is not expected to evolve at different rates between weapon-bearing and ornament-bearing groups.
Allen emphasised that the work illustrates how combining theoretical and comparative approaches can reveal mechanisms difficult to identify from either method alone. “The theory generates a prediction, while the comparative analysis allows us to ask whether that prediction is reflected in patterns of evolution across real animal groups,” he said. He added that the study shows how evolution acts on the two main forms of sexually selected traits—in weapons and ornaments—in different ways, shaping the rate of trait evolution more broadly.
He also offered a cautious note on interpretation. “There are a few things to keep in mind when interpreting these findings,” he said. “Most notably, the overall twofold rate advantage for ornaments in the cross-species data emerged only under the stabilising model and only after statistically removing body size using regression residuals, a common comparative method that the authors acknowledge can introduce statistical biases into evolutionary rate estimates.”
The authors recognise that their sampling excluded multipurpose traits and noted that some comparisons spanned vast evolutionary timescales, which could influence results. They also acknowledge that the simplified model—tracking a single ornament and a single weapon within a well-mixed population—may not capture the full complexity of natural systems. Future work may explore multiple signals, local mate availability and historical episodes of trait loss to better understand how sexual selection shapes trait diversity.
Allen said there are several directions for future research. “One is to understand in more detail why sexually selected traits evolve at different rates in different groups, and how ecological and behavioral factors influence these dynamics.”
The study, Ornaments evolve faster than weapons, was authored by Mohammadali Dashtbali, Louisa Hutton-Adams, Megan Watts, Jonathan M. Henshaw, and William L. Allen.
