A complete inventory of North American butterfly occurrence data: narrowing data gaps, but increasing bias

Abstract

Aggregate biodiversity data from museum specimens and community observations have promise for macroscale ecological analyses. Despite this, many groups are under‐sampled, and sampling is not homogeneous across space. Here we used butterflies, the best documented group of insects, to examine inventory completeness across North America. We separated digitally accessible butterfly records into those from natural history collections and burgeoning community science observations to determine if these data sources have differential spatio‐taxonomic biases. When we combined all data, we found startling under‐sampling in regions with the most dramatic trajectories of climate change and across biomes. We also used multiple methods with each supporting the hypothesis that community science observations are filling more gaps in sampling but are more biased towards areas with the highest human footprint. Finally, we found that both types of occurrences have familial‐level taxonomic completeness biases, in contrast to the hypothesis of less taxonomic bias in natural history collections data. These results suggest that higher inventory completeness, driven by rapid growth of community science observations, is partially offset by higher spatio‐taxonomic biases. We use the findings here to provide recommendations on how to alleviate some of these gaps in the context of prioritizing global change research.