Florida State University Tallahassee, FL, United States
Abstract: Relationships between climate variables and the timing of life-history events such as emergence or flowering are key to understanding the responses of organisms to climate variability and change. Many studies assume that organisms have similar responses to spatial and temporal variability in climate, allowing estimates derived from spatial variability (e.g., observations across mountain slopes or across latitudes) to be used in forecasting future phenologies in a changing climate. Such “space-for-time substitution” has been shown to hold for some groups of organisms at some spatial and temporal scales, but numerous processes could cause spatial and temporal responses to differ, including adaptation to local climates and divergent patterns of co-variation in climate variables across space and through time. Here we provide the first rigorous test of space-for-time substitution in the effects of climate on the reproductive phenology of plants. Our work combines two large phenology datasets which track a diverse group of montane and subalpine species inhabiting a single landscape in western Colorado, USA. Both datasets track similar species across a similar range of climatic conditions with a consistent methodology, but one is extensive in time and the other is extensive over space. This makes these datasets ideally suited for testing the assumption that space-for-time substitution holds for climate impacts on flowering phenology.
Using hierarchical Bayesian regression models, we found that flowering time responded differently to spatial and temporal variation in key climate variables for many species in our focal landscape. Specifically, fewer than 50% of species showed similar spatial and temporal responses to the most important climatic variable in our study system—the timing of seasonal snowpack disappearance. These results are robust to variation in both the statistical estimation method and the inclusion of other important climatic covariates. Furthermore, the differences in estimated climate-phenology relationships between spatial and temporal models are large enough to drive up to 12-day variations in predicted mean flowering time under a future global warming scenario (2080s, RCP8.5). Although our work does not provide conclusive evidence for the ecological or evolutionary processes driving differences in climatic responses over space and through time, it highlights the potential importance of those factors in modifying phenological responses to climate. In addition, our findings should provide pause for those relying primarily on spatial variation to forecast decadal-scale responses to climate change and their downstream effects on important processes such as pollination and carbon sequestration.