User:Stephen Thorpe/Global Biodiversity Estimation

From Species-ID
Jump to: navigation, search

Global Biodiversity Estimation

How many species are there in the world? A simple enough sounding question. But, is it so simple to answer?

Costello et al. (2011) have taken an interesting approach to the estimation of global biodiversity, and have arrived at by far the lowest ever estimate of only around 2 million species. They also conclude that all remaining undescribed species can be described within the next 50 years, with only a modest increase in current rates of description. The lack of quantification of confidence levels on the claimed figures is a little puzzling.

Why is the Costello estimate so low? The data that they based the estimate on is data about known described species, gathered from WoRMS, for marine species, and CoL, for nonmarine species. To be absolutely clear about this, they have based an estimate of total biodiversity (i.e., known and unknown) on data about known biodiversity. This is not necessarily a problem. However, most of the uncertainty about the extent of global biodiversity is about tropical rainforests. So, there is a potential ecological/geographical bias in the gap between known and actual biodiversity.

WoRMS is a very good database for known marine species. It is not perfect, but there are no big gaps in the data. I have nothing negative to say, in the present context, about WoRMS. However, most biodiversity is arguably nonmarine, and there is no nonmarine equivalent of WoRMS.

CoL is also a very good database for known marine species. The reason for this is that it gets most/all its data on marine species from WoRMS. But, because there is no nonmarine equivalent of WoRMS, CoL just has to make do with "bits and bobs" of nonmarine data, harvested from various specialist databases. There are many big gaps. Many of the probably most megadiverse groups are represented in CoL by significant gaps, e.g. Coleoptera, Hymenoptera, Acari, and Nematoda. Also, in these megadiverse groups, the known diversity is probably considerably less than the actual diversity, compared with other groups. This is another potential source of data bias.

One example which illustrates the magnitude of CoL gaps is the jewel beetle genus Agrilus. There are currently about 2886 recognised valid species (see here for details), of which CoL lists just 1 species! There is no reason to think that this is an isolated anomaly.

Bebber et al. (2007) have already found that [quoting abstract] unless the inventory of a group is nearly complete, estimating the total number of species is associated with very large margins of error. The strong influence of unpredictable variations in the discovery process on species accumulation curves makes these data unreliable in estimating total species numbers[unquote]. The margins of error associated with the estimate by Costello et al. (2011) is unspecified and unknown, but, curiously, they state that [quoting abstract]Bebber et al. (2007) correctly concluded that using description rates was problematic because of variable taxonomic effort and because lack of flattening out of the cumulative description rate curve. In this paper, we analysed trends in authorships of species descriptions over time to see if there were indications of increased or decreased taxonomic effort that may influence the rate of description, and we calculated confidence limits to quantify the variation in the rate of description. Therefore, this paper provides the most comprehensive and statistically rigorous approach to date that predicts global species richness while considering sampling effort[unquote]. This does not seem to address the reason that Bebber et al. (2007) gave for the large margins of error, which was [quote again]the strong influence of unpredictable variations in the discovery process on species accumulation curves makes these data unreliable in estimating total species numbers[unquote], and not [quote again]variable taxonomic effort and because lack of flattening out of the cumulative description rate curve[unquote]! Surely, the true source of the problem is the highly fragmentary and incomplete dataset for nonmarine species, especially from the most critical areas like the tropical rainforests? In other words, the fallacy of Costello et al. (2011) amounts to an "overreliance on the Catalogue of Life" ...

Given the above considerations, it seems to me that Costello et al. (2011) have demonstrated at best only that *if* CoL is a good source of data on nonmarine species (including those from the critical tropical rainforests), then there may be no more than 2 million species on Earth. Regrettably, this is not how the conclusions are being reported, e.g. here. It is regrettable for the following reason: if the Costello et al. estimate was within (or even nearly within) the range of current estimates, then there would be no problem. It would merely be one estimate amongst many. But it is by far the lowest estimate. This needs to be backed up by solid science and statistics. Otherwise, we run the risk of commercial industry, of the kind that would turn tropical rainforest into palm oil plantation for profit, for example, choosing the lowest estimate that looks to be from a reputable source, as the one to "go with". They could conveniently disregard any subsequent debate, as "normal in science", and focus on the fact that it was accepted and published in a high impact journal, after having gone through peer review, and was written by reputable scientists from reputable institutions. Although biodiversity conservation issues are clearly not preventing deforestation entirely, they may be helping to stem the tide of deforestation to at least some degree. It appears that citation rates and impact factors may be promoting the publication of "controversial" conclusions in high impact journals, a phenomenon which might be termed "tabloid science" ...

Stephen Thorpe 00:57, 13 October 2011 (CEST)


  • Bebber, D.P. et al. 2007: Predicting unknown species numbers using discovery curves. Proceedings of the Royal Society (B), 274(1618): 1651-1658. doi
  • Costello, M.J.; Wilson, S.; Houlding, B. (online, 2011): Predicting total global species richness using rates of species description and estimates of taxonomic effort. Systematic biology, doi