The rise of morphological diversity in butterflies and moths

Abstract As one of the most speciose animals groups 1,2 , butterflies and moths (Lepidoptera) are well known for their distinctive and colorful wings 3,4 . Yet, due to difficulties in quantifying differences in appearance 5-7 , we have a surprisingly poor understanding of how interspecific differences in wing appearance and function (morphological disparity) evolved within this diverse group. Here, we use generative artificial intelligence (AI) to analyse more than 32,000 images from over 17,000 species in order to quantify wing morphological disparity and evaluate the underlying macroevolutionary processes. We show that only three major morphological groups existed 140 million years ago, but this increased to six groups 110 million years ago, a time when the terrestrial angiosperm revolution provided new ecological opportunities 8 but also 30 million years after the explosion of Lepidopteran taxonomic diversity and the first emergence of flowering plants. The evolution of wing shape during this period, transitioning from forms adapted for stable flight to those designed for agile maneuvering, is a product of gradual change governed by morphological functional constraints intrinsic to primitive flyers 9 . Accelerated evolution of wing shape and color also occurred 66 million years ago, following the Cretaceous-Paleogene (K-Pg) mass extinction event that further expanded ecological opportunities by amplifying habitat diversity. Thus, Lepidoptera wing morphology is shaped by both gradual evolution and ecological opportunity, resulting in a decoupling of morphological disparity and taxonomic diversity. Ultimately, tools like generative AI will help revolutionize the study of adaptation and the macroevolutionary processes that influence biological diversity 10-12