Kareiva and Marvier [1] allude to another limitation of the hotspot approach that Myers et al. [4] explicitly acknowledge and attempt to address. The hotspots Myers originally identified were based entirely on the distribution of vascular plants. Those identified in [4] are based on a more extensive database, but they are based on species lists of vascular plants, mammals, birds, reptiles, and amphibians - no fish, no insects, no arachnids, no fungi, no lichens, no mosses, no microorganisms. Unless the distribution of species in different major groups of taxa corresponds to a high degree, it's possible that the hotspots Myers et al. identified are hotspots for vascular plants and tetrapods, while the hotspots for insects or microorganisms would be found in very different places. Their data suggest a fairly high degree of congruence between vascular plants and tetrapods, but their analysis can't tell us how well their approach would work for the vast array of species not included in their analysis.
Data on the coincidence of invertebrate and plant distributions at a global level is hard to come by.10 Unfortunately, at smaller scales the coincidence doesn't appear to be particularly great. Consider, for example, an analysis of biodiversity changes associated with forest modification in the Mbalmyo Forest Reserve (south-central Cameroon). No one group is a good indicator for changes in species richness in other groups [2]. The groups were birds, butterflies, flying beetles, canopy beetles, canopy ants, leaf-litter ants, termites, and soil nematodes. Similarly, Prendergrast et al. [5] found a very low coincidence of species hotspots among five groups surveyed through the island of Britain.