Data from: Climate is a strong predictor of near-infrared reflectance but a poor predictor of colour in butterflies

Dataset includes all variables (climatic and colour) used for analyses in manuscript. Some variables were log-transformed in the models, details are given in the manuscript. Also includes the 4500 phylogenetic trees used in comparative analyses. File that matches tree tips and data rows is found as a tab in the complete dataset file.

Abstract

Colour variation across climatic gradients is a common ecogeographical pattern; yet there is long-standing contention over underlying causes, particularly selection for thermal benefits. We tested the evolutionary association between climate gradients and reflectance of near-infrared (NIR) wavelengths, which influence heat gain but are not visible to animals. We measured ultraviolet (UVA), visible (Vis) and NIR reflectance from calibrated images of 372 butterfly specimens from 60 populations (49 species, five families) spanning the Australian continent. Consistent with selection for thermal benefits, the association between climate and reflectance was stronger for NIR than UVA–Vis wavelengths. Furthermore, climate predicted reflectance of the thorax and basal wing, which are critical to thermoregulation; but it did not predict reflectance of the entire wing, which has a variable role in thermoregulation depending on basking behaviour. These results provide evidence that selection for thermal benefits has shaped the reflectance properties of butterflies.