COS 132-4 - CANCELLED - Using herbarium data to predict the timing and duration of population-level flowering displays: advancing collections-based research towards practical application

Abstract: Forecasting the impacts of changing climate on the timing of critical life-history events within and among conspecific populations is essential for understanding potential ecological disruptions to biotic communities caused by climate change, and to conservation projects targeting herbivorous or florivorous taxa that depend on floral or fruit resources being synchronous with their activity period. Herbarium specimens have become an essential resource for assessing plant phenology across broad spatial and temporal scales, and represent a critical resource for forecasting plant phenology due to their unrivaled spatial and taxonomic coverage. However, specimens are often temporally (and geographically) biased because they are not collected randomly, and may also be preferentially collected during specific portions of individual flowering displays. This may limit the accuracy of phenological predictions derived from such data, particularly when predicting population-level onset and termination dates, which can be more susceptible to sample biases than mean flowering dates.

Using simulated data, we conducted the first systematic examination of the accuracy with which the onset and termination of phenological displays (in this case, of flowering) can be predicted from natural-history collections data, and the degree to which collection biases and species-level traits influence the accuracy or temporal biases of such predictions. We determined that model accuracy declined with increasing flowering duration of individual plants, but was independent of both the rate at which species’ phenologies shifted in response to climate conditions and the magnitude of intrapopulation phenological variation. The amount of data required to model population-level phenological displays is not impractical to obtain; the mean absolute error declined by less than 1 day as sample sizes rose from 300 to 1000 specimens. Our analyses of simulated data indicate that, when underlying phenological responses to climate are linear and drivers of phenological change are captured by climate data, specimen data can predict the onset, termination, and duration of a population’s flowering period with similar accuracy to estimates of median flowering time.