Curator, Department of Entomology, National Museum of Natural History, Smithsonian Institution, Washington, D.C.
A few years ago in a short paper in the Coleopterists Bulletin, I hypothesized that instead of the current estimate of 1.5 million species on Earth, there were 30 million species of insects alone (Erwin, 1982). This hypothesis was based on collections of beetles from tropical forest canopy samples in Panama (Erwin and Scott, 1980), rather than on the catalog counts of taxonomic names used in all the earlier estimates. I used simple arithmetic based on actual numbers of beetle species in my samples, estimated numbers of tropical forest tree species given me by the leading botanists, and a conservative estimate of the host specificity of tropical forest canopy insects. Host specificity in this sense means that a species in some way is tied to the host tree species and cannot exist without it.
This reestimation of the magnitude of life on Earth got a lot more coverage than I anticipated and began the usual controversy of right or wrong. Those engaged in the controversy, most of whom never read this obscure paper in the Coleopterists Bulletin, in a way actually missed the point of the paper. Consequently, I now want to take the opportunity to clarify the situation.
Science, at least in natural history, proceeds from casual observations, usually in the field or on museum specimens, to the erecting of hypotheses and finally to the testing of those hypotheses. Repeated failure to prove a hypothesis false lends support to the possibility that it may be true. For the 30 million species of insects hypothesis, which was based on a brand new set of observations never before available to scientists, I suggested that testing must begin by refining of our knowledge about host specificity of insects in tropical forests.
In a subsequent paper, analyzing data from the canopies of four different forests