MEASURING THE RATE AT WHICH COMMUNITY COMPOSITION CHANGES

Many active questions in paleoecology involve the rate at which community composition changes. For example: Do abundances change gradually, or are changes concentrated in specific taxa, intervals, or environments? Rates of community change can be measured as ecological distance divided by elapsed time, parallel to the traditional measurement of rates of phenotypic evolution. However, this approach has a similar drawback to traditional phenotypic rates: a strong negative scaling with interval length – rates will be slower when measured over longer periods of time. Here we develop an alternative approach to quantifying rate of community change using a simple model of ecological drift (i.e., the local component of Hubbell’s Neutral Theory). When species proportions are arc-sin transformed, the dynamics of species in a community can be approximated as independent random walks in the transformed space. The net result is a simple one-parameter model of community change, where the single parameter is a measure of rate: as it increases, greater changes in relative abundance are expected per unit time. Several approximations are required to implement this approach, but simulations suggest that it is possible to estimate reliably the rate parameter. Because this model fitting uses likelihood, we can apply the normal likelihood-based machinery for generating confidence intervals and testing models. We illustrate this approach by measuring rates of community change in pollen from Quaternary lake cores and in ostracodes from deep-sea cores, and by comparing models in which different subsets of taxa have different intrinsic rates of change.