Estimates of the Near‐Imponderable: A Reply to Erwin

Abstract

Erwin ( 1991 ) apparently finds little with which he can agree in my recent study of the question of how many extant species of insects there are (Gaston 1991). His comments in major part reflect, however, a somewhat narrow view of the kinds of information that may or may not contribute to debate on this subject. The central objective of my paper was to ascertain how well taxonomists' understanding of the levels of species richness attained by the groups they work upon support assertions that there are thirty million or more species of insects (Erwin 1982, 1988). The intention was neither to generate a definitive estimate of how many insects there might actually be, nor to maintain that this approach to the problem was without its limitations. Rather, it was to provide a further viewpoint on global species numbers. Erwin has a number of criticisms. Foremost of these is that the database is nonscientific. Here, we must be aware of some of the special problems pertaining to estimates of global insect species richness. Estimates are invariably extreme extrapolations from very limited data sets, and in this sense most methods are perhaps straining the limits of what is and is not treated as science. Thus, whilst Erwin regards his own method of producing estimates (Erwin 1982) as the antithesis of that based upon taxonomists' opinions, I doubt whether extrapolations to a global fauna from samples of the insect assemblages of 19 trees, albeit tropical ones, are any more firmly in the realms of science. Of course, the precise basis of taxonomists' estimates is often difficult to ascertain, and in this sense they are rather different from other forms of evidence that may be used. I see no reason, however, why this should entirely invalidate their consideration as Erwin proposes. Indeed, Erwin himself relies upon such opinions in his calculations of the global numbers of insect species in using an estimate that there are 50,000 species of tropical trees and that beetles make up 40% of all arthropod species (Erwin 1982). Such estimates might perhaps best be regarded as working figures which may stand until improved information demonstrates them either to be too small or too large. The point remains that the current working figures for the species richness of most insect groups do not come close to supporting the idea that there are thirty or fifty million species of insects. This leads us to Erwin's second major criticism, that taxonomists' estimates are unreliable. I have made no claims that the estimates are the final word on species numbers, nor to hide the enormous inconsistencies there are between some of them; one must invoke some implausible gradients in richness to explain differences in the sizes of the major insect orders determined on the basis of these estimates. One has to maintain, however, that estimates of the sizes of the more speciose taxa are, with a few notable exceptions, out by a factor in the region of ten (in some cases much more) to attain totals of thirty million species. As yet, for the overwhelming majority of families, there is no evidence that this is the case. In proposing that estimates of the richness of various higher insect taxa tended to support overall figures of less than ten million, a substantial margin for error in their formulation was already being invoked.