It has been a longstanding goal to find simple to measure metrics that can readily discriminate between communities that are different. In particular, it is desirable to find metrics that are simple but able to make nuanced descriminations between different degrees of human impact on communities. Two general tools have been commonly used towards this goal: diversity metrics and species abundance distributions (SADs). These have been popular because of their simplicity and ease of use. However, relatively few general principles have emerged.

I identify a new way of plotting species abundance distributions. I also performed an extensive analysis of resampling from communities to identify which traditional diversity and SAD parameters are efficient and unbiased (behave well in small samples). I then performed multivariate analysis on many empirical datasets to identify families of closely related parameters.

Results/Conclusions

I show that the new SAD plot allows for effective comparison between communities of different species richness (something the traditional rank-abundance plot does very badly). I give an example where simple species richness fails to detect differences that this new plotting method make clear.

I show that most metrics perform very poorly and suggest that sample sizes typically need to be about 1000 individuals (much larger than usually used today). I identify a few metrics that perform well on smaller samples. I identify 3 distinct families of diversity metrics and SAD parameters and 4 additional subgroupings that appear to be robust. Using these data I describe three different scenarios depending on the data in hand and make specific recommendations about which metrics will work best in each scenario.