This recent article in the American Naturalist has taken a second look at some of the famously inflated species estimates, some going high as 100 million (Erwin, 1988). Estimates conducted by the authors indicate that projections above 30 million have probabilities of <0.00001. Their estimated range is more likely to be between 2.5 and 3.7 million species (with 90% confidence). This seems somewhat reasonable given that these extraordinary estimates were based heavily on extrapolation. There are clearly many difficulties in assessing diversity based on tropical arthropod surveys – this paper again uses phytophagous (plant-eating) beetles for estimates. They are careful to point out that these methods do not account for non-phytophagous insects, but assume that they will follow traditional biogeographic patterns of diversity. This is somewhat of a new concept given that when I was in college I was taught that parasitoids are counterintuitively not more diverse in tropical regions. This hypothesis is more often than not being proven false in the light of more precise modern taxonomic methodology. Rather proudly I helped play a role with the parasitoid project at the UIUC. In short, host specificity is more extreme in tropical environments with hundreds of cryptic species hidden amongst rapidly radiating groups such as the microgastrine Braconids (Hymenoptera) – the same has held true across similar taxa.
One interesting note about the paper is their inclusion of a secondary estimation based on Lepidoptera canopy assemblages. They assumed that a) all Lepidoptera can be found in the canopy and b) that all leps are phytophagous. This is clearly a very conservative estimation given that not all Lepidoptera are found in the canopy and not all are phytophagous. While I do not have the numbers on hand, a certain percentage of lep diversity must have been excluded from these estimates. I will also go out on a limb and assume that the authors (Novotny 2002) did not include microlepidoptera morphospecies – and most likely estimated abundances with our current taxonomic understanding. However I do not have access to this 2002 paper, so I may be incorrect. Using these Lepidoptera numbers (from the same survey as the Coleoptera) a global diversity was estimated by Hamilton et. al. at around 8.5 millions arthropod species.
While I agree that extraordinary estimates of tens of tens (or hundreds) of millions of arthropod species are probably ridiculous; I am of the camp that current research is indicating that estimates of the lower tens of millions of species are possible. The authors have failed to include research that counterbalances their premise that tropical species exhibit a lower beta diversity (Novotny 2002, 2007). In the same journal, Nature 2007, Dyar et. al. have indicated that the American tropics exhibit a higher beta diversity than previously assumed. Either it can be said that estimates of beta diversity in the australasian tropics are incorrect, or they are incompatible with species assemblages of neotropical forests. All of this speaks to the difficulty in extrapolating estimations of species across all tropical regions. These estimates are based on comprehensive insect surveys of New Guinea, perhaps they do not accurately reflect the true diversity of American tropical forests, and these number ranges are low.
As a final thought, most assesments are focused on tropical arthropods. It seems all too possible that the total number of all species, including bacteria and archaea, can easily exceed tens of millions. But extrapolating those numbers is even more precarious than arthropods, given the extreme lack of knowledge we have.
