Heterogeneity test

As extrapolation to estimate the total species complement for the habitat, Smax , is only possible if the species accumulation curve is derived from a homogeneous (stable) community, the first task is to look for heterogeneity. If species abundance rather than presence-absence data is available, Colwell & Coddington (1994) suggest comparing the mean randomised species accumulation curve with the curve expected if all the individuals caught over all the samples were randomly assigned to the samples. If the expected curve rises more steeply from the origin then heterogeneity is greater than could be explained by random sampling error alone. The expected curve calculated by Species Diversity and Richness is that described by Coleman (1981) & Coleman et al. (1982). The increase in species number over a series of samples, Sa, is calculated as:

where STOT is the total species number recorded, ni the total number of individuals belonging to species i, a the sample number and α the fraction of the total sampling effort undertaken by sample a.

The graph below was created from a very large data set on estuarine fish. It can be seen that the Coleman curve is above the observed species acquisition curve, suggesting some sample heterogeneity. This might be expected as the samples were collected over a 25 year period and there were considerable changes in the community over the period of study.