Illuminating patterns of firefly abundance using citizen science data and machine learning models

As insect populations decline in many regions, conservation biologists are increasingly tasked with identifying factors that threaten insect species and developing conservation strategies. One insect group of global conservation concern are fireflies (Coleoptera: Lampyridae). Researchers have hypothesized that North American firefly populations are most threatened by habitat loss, pesticide use, and light pollution, but the importance of these factors in shaping firefly populations has not been rigorously examined at broad spatial scales. Using data from >24,000 surveys (spanning 2008–16) from the citizen science program Firefly Watch, we trained machine learning models to evaluate the relative importance of a variety of factors on bioluminescent firefly populations: pesticides, artificial lights at night, landcover, soil/topography, short-term weather, and long-term climate. Our analyses revealed that firefly abundance was driven by complex interactions among soil conditions, climate/weather, and land cover characteristics. Given the significant impact that climactic and weather conditions have on firefly abundance, there is a strong likelihood that firefly populations will be influenced by climate change, with some regions becoming higher quality and supporting larger firefly populations, and others potentially losing populations altogether. Collectively, our results support hypotheses related to factors threatening firefly populations, especially habitat loss, and suggest that climate change may pose a greater threat than appreciated in previous assessments. Thus, future conservation of North American firefly populations will depend upon 1) consistent and continued monitoring of populations via programs like Firefly Watch, 2) efforts to mitigate the impacts of climate change, and 3) insect-friendly conservation practices.