Differences in sampling effort confound comparisons of species richness between sampling units.

Panels (a) and (b) show maps derived from a simulation of two 100 x 100 km sampling units, and the collection localities of specimens sampled from them. Both sampling units were simulated to have identical total number of species, species abundance distribution and geographic distribution of species. Details of the simulation are described in Methods. In panel (a), blue open symbols show collection localities for one set of specimens, and grey closed symbols for a second set of specimens. Both sets contain N1 = 3000 specimens, each constituting a random sample of individuals from the sampling unit. Therefore, the two sets of specimens differ in terms of the individuals and species included. Accordingly, the two sets of specimens produce different accumulation curves, shown in panel (c). Accumulation curves display the total number of species revealed as additional specimens are added to the pool of all previously collected specimens [11]. The differences between accumulation curves in panel (c) illustrate how sampling variation in accumulation curves derives from variation in the composition of samples (i.e., sets of specimens) in terms of individuals and species. Panels (b) and (d) show respective information for the second 100 x 100 km sampling unit, with two sets of N1 = 1000 specimens. Comparison of the number of species known to occur in each sampling unit would be invalid, because it would reflect differences in sampling effort and not differences in species richness.