Wednesday, August 5, 2009: 2:30 PM
Dona Ana, Albuquerque Convention Center
Adam Tomašových and Susan M. Kidwell, Department of Geophysical Sciences, University of Chicago, Chicago, IL
Background/Question/Methods Multiple studies have found that long time series of fossil assemblages show relatively low temporal turnover in species composition within habitats. This observation contrasts with the high turnover observed in living assemblages and predicted by dispersal-limited neutral metacommunity models. However, null-model predictions for time series with community data usually do not take into account the reduced temporal resolution that arises from time averaging. To evaluate the degree to which temporal turnover in species composition is modified by decreasing temporal resolution, and whether time-averaged assemblages can distinguish neutral from non-neutral dynamics, we (1) model temporal dynamics of fossil assemblages in response to variation in time-averaging using individual-based simulations, and (2) compare them with observed patterns in mollusk and ostracod fossil records of yearly- and decadal-scale community history.
Results/Conclusions We find that (1) an increase in time-averaging increases community similarity and reduces temporal turnover, as represented by the slope of species-time relationships, regardless of migration rates, and (2) these changes are associated with a decrease in species dominance and shift in species-abundance distributions towards lognormality. The similarity values and turnover rates observed in mollusk and ostracod assemblages are statistically indistinguishable from the scaled similarity values and turnover rates that are predicted by the neutral model, using migration rates estimated from living assemblages. Outputs of neutral and trade-off models converge as temporal resolution decreases strongly. To evaluate community persistence using time series with time-averaged assemblages, ecological analyses must scale up model-predictions: because time-averaged species abundances are expected to have high stability, neutral models of unscaled time series will tend to be rejected.
