Complex species’ responses to climatic variability in butterfly populations monitored by a statewide citizen science program

Background/Question/Methods

A key goal for ecologists is predicting species’ responses to climate change to assess vulnerability. In addition to the effects of mean temperatures on population dynamics, three additional considerations are often added to models of species’ responses: temperature extremes and variability, nonlinear responses to temperature, and different sensitivity to temperature across life stages. Citizen science monitoring programs provide an opportunity to understand these complex responses to climate, since they observe many species across a wide range of environmental conditions. We use 15 years of weekly butterfly counts collected by volunteers across Ohio to test how seasonal climatic differences influence population growth in 42 species. Comparing climatic variability across space and time permits observations across a nearly 6-degree Celsius gradient in mean annual temperature. To account for collinearity in mean and extreme climatic variables, a principal components analysis was used to characterize variation in climate. For each species, population growth was modeled as a function of the previous year's total count, principal components of seasonal climatic variation (linear and quadratic terms), and population trend over time. An information-theoretic approach was used for this exploratory analysis to visualize the range of species' responses to climatic variability.

Results/Conclusions

Mean temperature and the frequency of extreme temperatures group into the first principal component, making it difficult to separate the effects of the correlated variables. This principal component explained 45% and 52% of the variation in climatic conditions for winter and summer, respectively. Precipitation and temperature fluctuations characterize the second principal component, describing 25% of the variation. Nonlinear effects of seasonal temperatures are common (18 species), but occur in different seasons for different species, highlighting the need for experiments to derive mechanisms behind species’ responses. Summer climatic conditions affect population growth in opposite ways depending on when they occur. Generally, hot and dry summers decrease population sizes in the following year but increase population sizes in the year they occur. This counterintuitive result could be explained by detection probability, with surveys more likely to occur in drier years. Species interactions could also play a role, with a lagged effect of wet summers improving host plant quality for larvae. We will discuss whether an analysis of species’ life-history or physiological traits may help explain the variation in species’ responses. We conclude by questioning how predictions of complex responses to climate change might be useful for the conservation of vulnerable species.