How do you measure species richness? Our ongoing construction of model networks for Greater Antillean coral reefs relies heavily on data from the REEF project to document species occurrences on local scales. The overall picture, as reported in earlier postings on this blog, suggests that the Cayman Islands are richer in vertebrate (bony and cartilaginous fishes, and sea turtles) species than both Jamaica and Cuba. This would perhaps make some sense given an impression of lesser anthropogenic impacts, such as fishing, on Cayman reefs, as well as the number of protected areas there. The REEF database, however, also documents the number of times that a specific locality has been sampled, and a casual browse indicated that the Caymans have been sampled far more heavily than either Jamaica or Cuba. We therefore wanted to account for the effect of sampling effort.

A simple plot (see figure) of the number of species discovered per sampling event shows that indeed the Caymans (blue) have been sampled more heavily than Jamaica (red), and that the number of species (species richness; one measure of biodiversity) is greater in the Caymans. But what if Jamaica was sampled as much as the Caymans? There is no definitive way to answer this, of course, without actually increasing sampling effort in Jamaica. We can, however, ask if our conclusion of greater diversity in the Caymans is a fair one. The solid lines plotted through the data are nonlinear regression functions of the form
$y = b_{0} + b_{1}x + b_{2}\ln x$
The pale-coloured ribbons behind each line represent 95% bootstrap confidence intervals. It should be fairly obvious that if sampling effort was equal between the two systems, that Jamaica would be at least as diverse as the Cayman Islands! I can’t explain why Jamaica is under-sampled relative to the Caymans without becoming speculative about socio-political matters and tourist tastes, so I’ll leave that up to others.

I did not include Cuba in this analysis for several reasons. First, Cuba is very clearly under-sampled when compared to the other two systems, and our data do not include data from important near-pristine areas such as Los Jardines de Reina. Second, Cuba is so much larger in area (coastal) than either other system, that perhaps some consideration of spatial coverage would be needed. And third, it is entirely possible that there are compositional differences, at least among invertebrates, between the northern and southern coasts of Cuba (see Roopnarine and Vermeij, 2000).

The simple message here is that mere counts of species within an area are insufficient measures of biodiversity. This has long been known among ecologists and has led to very interesting work on measuring biodiversity, the spatial distributions of species, and the composition of biological communities. But it’s a message worth repeating every now and then.