Document: JAM-3-8-4

Using estimates of species richness to test ecological theory .

NICHOLS, J.D.*, J.R.SAUER, E.CAM and J.E.HINES

USGS Patuxent Wildlife Research Center, Laurel, MD, U.S.A. ¹

Abstract:
Confronting theory with information from the real world is healthy, but often very difficult. Biodiversity studies frequently sample species richness, but most sampling methods are surprisingly flawed. Superficial use of survey data that use these methods can lead to biased estimates because probabilities of detecting species are < 1. Fortunately, capture-recapture estimators can be applied in many ecological studies to estimate species richness. We use these estimators to address questions of temporal variation in species richness, species-area relationships, and nested subset analyses. With respect to species-area relationships, we show that positive relationships of form similar to those frequently reported in the literature are a natural consequence of species detection probabilities < 1. Use of species richness estimates that account for detection probabilities produces species-area relationships with smaller slopes than those based on raw counts. The latter relationships confound sampling and ecological hypotheses, whereas the former are restricted to ecological hypotheses and should be used for testing and estimation. Nested subset analyses are based on species presence-absence data for multiple sites. Current methods do not account for the possibility that absences represent presence, but not detection, of the species. We present estimators for the probabilities of species sharing, and show what values these probabilities should attain in the case of nested and non-nested communities. These ideas lead to a test for nestedness and an index of nestedness when detection probabilities are < 1.

Keywords: estimation, nestedness, species area curves, species richness

This abstract is being presented at: 2:15 PM in session:
Symposium # 18: The Role of Theoretical Ecology in Biodiversity Conservation and Management.