Choice of species-area function affects identification of hotspots

I tested the reliability of species-area curves for use in identifying hots pots, political or geographical regions of high species richness. On a spec ies-area plot, hotspots are points (regions) that appear above the curve to a greater extent than other points. Because several different curves can b e fit to species-area data, identification of hotspots may differ depending on the curve-fitting function used. I tested this hypothesis by comparing hotspots identified by the power function, the extreme value function, a li near function, and the exponential function. I examined several species-are a data sets varying in size and in the presence of endemics. I defined hots pots as the highest 25% for small data sets and highest 10% and 25% for lar ge data sets of standardization residuals from each function fitted to each data set. For some data sets, the functions agreed in identification of ho tspots in that they identified 75% or more of the same hotspots. The extrem e value function tended to identify hotspots not identified by the other th ree functions. For most data sets, the functions did agree completely in id entifying hotspots. Therefore species-area curves should not be used as the sole means of identifying hotspots of species richness, although they can be used to examine the effect hotspots area has on richness for hotspots id entified by other methods.