COS 107-6 - Climate variability, not predation pressure, drives interspecific variation of butterfly eyespots globally.

Background/Question/Methods

How biotic and abiotic environments shape the adaptation of species and generate diverse functional traits is fundamental to evolutionary biology. Interspecific variation of butterfly eyespots exhibits a study model but has been focused mainly on the biotic aspect. High contrast is hypothesized to intimidate predators while more eyespots deflect attacks. Nevertheless, whether climatic factors across the globe drive the eyespot variation remain unexplored. We explored the pattern in Nymphalidae butterflies by obtaining approximately 110,000 digitized specimens representing 2560 species from Global Biodiversity Information Facility (GBIF). We applied deep learning algorithms, SOLOv2 and U-Net to detect and segment eyespots and butterflies. The models filtered 45,000 specimens representing 838 species and processed the images accordingly. We quantified four traits dominating eyespot effectiveness, including an averaged number of eyespots, area proportion of a maximum eyespot, and contrasts of brightness and hue between eyespots and the entire surface. We employed insectivorous bird richness for predation and the 19 variables from WorldClim 2.1 and three vegetation indexes for environmental variables as predictors. The traits and the predictors were summarized and analyzed using 0.5*0.5 degree grid cells across the globe. Contributions of the predictors were assessed by SHapley Additive exPlanations (SHAP) with a gradient-boosting algorithm, XGBoost.

Results/Conclusions

In contrast to conventional hypotheses, we found that temperature and its variability, rather than predation pressure, largely determine the interspecific variation of eyespot traits. Among the conspicuous features, higher brightness and color contrast are associated with higher mean annual temperatures and the dominant diurnal temperature, respectively. Specifically, warmer temperature leads to a brighter wing surface, reflecting the needs of thermal regulation; meanwhile, the eyespots get darker resulting in stronger brightness contrast. Tropical climates, including high diurnal temperature range and low annual temperature range, lead to smaller and fewer eyespots, respectively, which is against what has been implied by previous predation hypotheses. Although these climatic gradients have often been referred to as proxies of predation pressure toward the tropics, the predation pressure itself only better explained the color contrast, suggesting that color contrasts may be a more general strategy for coping with predation pressure. Our results explain how climate gradients and variability act as proximate mechanisms for interspecific eyespot variation. The phenomena provide insights into the mechanisms of macroevolution and call for new hypotheses to test the diverse adaptation of eyespots across a global scale.