The composition of local communities reflects both colonization from external sources, and local extinction. Two areas of ecology—island theory (together with metapopulation and metacommunity ecology) and succession—focus on how extinction influences community structure. In their purest forms, these theories provide contrasting predictions for extinction of early succession species in a fragmented landscape. Island biogeography theory, for example, suggests that smaller patches have smaller populations and experience higher likelihoods of extinction. By contrast, if we assume competition/colonization trade-offs, replacement of early succession species will happen where succession occurs more rapidly. In our experimental landscape, we know that succession occurs faster on larger patches, so it seems reasonable to predict that early-succession species will disappear there first. In this study, our primary objective was to assess the relative contribution of island/metapopulation versus successional processes for governing extinction of early-succession species in an experimentally fragmented grassland in northeast Kansas. We used a new metric, Rank Occupancy-Abundance Profiles (ROAPs) and re-sampling procedures to compare the occupancy and abundance patterns of species that decline in abundance during succession. We also used ROAPs to compare patterns of a species’ decline (within a patch size) over successional time, and related observed patterns to three conceptual models that propose different relative shifts in occupancy versus abundance during extinction.

Results/Conclusions

Patch size affects plant extinction patterns in our landscape.  As succession progressed, we observed statistically significant differences in the distribution of abundance for declining species in large versus small habitat fragments. By contrast, distance from the source had no effect on spatial patterns of extinction. The influence of patch size varied among species. Clonal species, in particular, declined faster on small patches. Additionally, occupancy and abundance were decoupled for many declining species; collectively, these intraspecific patterns influence the interspecific occupancy-abundance relationship we observed over the course of succession at our site. We conclude that detecting the influence of time and space on extinction requires a species-by species approach that incorporates both occupancy and abundance.  Further, we propose that ROAPs provide a useful comparative tool for assessing differences in distribution of abundance among landscape types, years, or species.