We’d love feedback (e.g. any comments, predictions for predator communities, or datasets we should stress-test on). DMs welcome.

Moreover, we demonstrate that aposematic species have more similar dorsal & ventral patterns and reduced sexual dimorphism. These were long-standing hypotheses in this field which were really hard to test with conventional methods.

What we found? Across Nymphalidae, the aposematic axis is a dominant axis of variation in color patterns. It also separates aposematic vs. non-aposematic phenotypes in an out-of-distribution moth dataset.

What we did? Extract multivariate features from images → derive a continuous aposematic score → evaluate some long-standing hypotheses for the partitioning of variation in wing patterns in butterflies

Why aposematism? Warning signals are predicted to be conspicuous and convergent. If they’re pervasive, they could dominate the geometry of pattern space

Having the same issue