Investigating species incidence over habitat fragments of different areas—a look at error estimation

The use of incidence functions and their error structure is explored as a means of interpreting patterns present in fragmented systems. Incidence functions describe the probability of a species' presence on a fragment. The only information required for an incidence function is presence/absence data for a number of fragments of known size. Methods are developed for establishing prediction bounds on the incidence function. The example developed uses data from mammals on isolated mountaintops. Distributions of number of species expected on a fragment predicted the actual number of species well, but prediction of identities of species on small fragments was poor. Although the number of species expected on an assembly of small fragments compared to a single large fragment of the same total area was nearly always equal, the identity of species differed. Species with large area requirements were never found on any number of small fragments; species which occurred with a medium probability over most fragments had a higher probability of being present in an assembly of small fragments than on a single large fragment.