Phenological research based on natural history collections: practical guidelines and a Lepidopteran case study

Background/Question/Methods

Natural history collections (NHCs) have been indispensable to understanding longer-term trends of the timing of seasonal events. Massive-scale digitization of specimens promises to further enable phenological research, especially the ability to move towards a deeper understanding of drivers of change and how trait-environment interactions shape phenological sensitivity. Despite the promise of NHCs to answer fundamental phenology questions, use of these data resources presents unique and often overlooked challenges requiring specialized workflow steps, such as assembling multisource data, accounting for date imprecision, and making decisions about trade-offs between data density and spatial resolution. Here we provide a set of key best practice recommendations and showcase these via a case study that utilizes NHC data to test hypotheses about spatiotemporal trends in adult Lepidoptera (i.e., butterflies and moths) flight timing across North America. Our case study is a worked example of these best practices, helping practitioners recognize and overcome potential pitfalls at each step, from data acquisition and cleaning, to delineating spatial units and proper estimation of phenological metrics and associated uncertainty, to building appropriate models.

Results/Conclusions

We confirm and extend previous work documenting the critical importance of voltinism and diapause strategy for predicting Lepidoptera phenology spatiotemporal trends. Our case study also showcases the unique power of NHC data to test existing hypotheses and generate new insights about temporal phenological trends. Specifically, species that fly in spring, but not fall, are lengthening the durations of their flight periods in more recent years, even after accounting for climate context. These results may reflect the ability of facultative and obligate multivoltine species to add generations, especially earlier-fliers. Additionally, we found a temporal trend of earlier arrival of migratory species, even after accounting for climate warming. We close by noting the value of partnerships between data scientists, museum experts, and ecological modelers to fully harness the power of digital data resources to address pressing global change challenges. These partnerships can extend approaches for integrating multiple data types to fully unlock our understanding of the tempo, mode, drivers, and outcomes of phenological changes at greater spatial, temporal, and taxonomic scales.