SPACE-FOR-TIME SUBSTITUTION ACROSS TEMPORAL SCALES

Contemporary spatial patterns of biodiversity are often used in ecology to predict future temporal changes in diversity. This method assumes that the drivers of community turnover in space are the same as those that drive turnover through time. Paleontological data has been used to test this assumption, but at a single temporal scale. Here we use time series of planktonic foraminifera assemblages ranging from decades to million years to test empirically at which temporal scales space can substitute time. We fitted a generalised dissimilarity model to contemporary spatial data (core tops) and used it to predict community turnover through time based on local temperature time-series. We find that longer time scales (deglaciation, ice-ages) are more predictable by contemporary spatial variation than shorter time scales (years to decades). At shorter time scales, stochastic processes and dispersal limitation decouple the relationship between planktonic foraminifera assemblage composition and temperature, but time-averaging reduces these effects at longer time scales. In summary, our results show that the space-for-time substitution is less valid at temporal scales commonly used in ecological research.