Whether the limits of species' ranges and their seasonal activity reflect physiological tolerance of climatic extremes is a long‐standing question in ecology and has implications for species' responses to recent climate change. We explored these associations in butterflies, using thermal tolerance traits and traits describing geographic distribution across 119 butterfly species, as well as adult flight season phenological traits across 87 species, accompanied by nearly 30 years of temporal population abundance trends. Butterflies with more poleward cold range edges and those that emerged earlier in the season were better able to tolerate low temperatures. By contrast, heat tolerance was unrelated to the equatorward warm range edge position and the timing of peak abundance across the flight season. Nevertheless, the difference between heat tolerance and high‐temperature extremes (warming tolerance) revealed differences in vulnerability of butterflies across large spatial extents from the tropics to the subarctic. Warming tolerances in the tropics approached zero or were in deficit for many species, whereas warming tolerances at higher latitudes were consistently large. Yet, even among butterflies at higher latitudes, there was substantial interspecific variation in warming tolerance. This variation in warming tolerance, including its components and correlates, had complex relationships with multi‐decadal population abundance trends. In some cases, our results directly implicated climate as a factor associated with population trends, as range‐wide coldadapted species had larger declines than more warm‐adapted species. In other cases, our results implicated indirect effects of ecological and demographic consequences of climate adaptation to seasonal variation in temperature, as species with earlier emergence and longer flight seasons (traits associated with better cold tolerance and worse heat tolerance) had smaller population declines than species with later emergence and shorter flight seasons. These results suggest caution when using physiological trait‐based analyses to forecast vulnerability without an explicit consideration of mechanism.
Diamond et al. (Sun,) studied this question.