Modeling the sampling effect in the species--time--area relationship

Recent empirical work in numerous systems has demonstrated the interdependence of spatial and temporal accumulation of species in the species--time--area relationship (STAR). The purpose of this study was to develop a process-based stochastic model for the STAR that assumes species neutrality and to compare the model's expectations to data collected on plant species in a tallgrass prairie. We varied two important aspects of the neutral species assemblage: evenness in the species pool and individual replacement rate (R). When R is larger than ~0.5 and evenness is intermediate to high, the neutral STAR generates patterns qualitatively similar to the empirical STAR. Our model also indicates that space and time were not symmetrical in their effects on species accumulation, except in the special case of R = 1.0. We observed both positive and negative time-by-area interactions in the sampling model, which indicates that nonzero interactions are not necessarily evidence of ecological processes. Furthermore, as accumulated richness approaches the size of the species pool, the time-by-area interaction becomes increasingly negative in our model. This suggests that negative time-by-area interactions should be expected a priori in empirical systems if rates of species accumulation decrease due to increasing rarity of unique species. Given the wide range of STARs that the sampling model generated, the difficulty in estimating key parameters, and the complexity of assessing the relative abundance distribution and scale of the species pool, we cannot refute the sampling effect, and we suggest caution in accepting ecologically oriented explanations of empirical STARs. Key words: biodiversity; grassland; rarefaction effect; sampling effect; scale dependence; scaling; species richness; species--time--area relationship (STAR); species turnover.