How many species are there on Earth? This is one of those simple questions that rapidly unravel when you try to pin down an answer. Despite over 250 years of the Linnaean system being available as a "filing cabinet" for humanity's systematic exploration of life on Earth, the answer is still not clear.
That we lack an answer is far from astonishing - there's plenty of Earth (and plenty of Ocean) to explore. Taxonomists still argue about what exactly is a species and how to recognize one when you see it.
A paper just published in PLoS Biology by Camilo Mora and colleagues has put forward the figure of 8.7 million species. This has been widely reported in the mainstream press (BBC, The Guardian, Today Online), though most omit mentioning that this figure has a generous margin, a standard error of +/- 1.3 million.
What I'm intrigued by is why there is so much attention being paid to this piece of news. After all, estimating the total species count of the planet is not a new endeavour.
The ecologist Robert May, known for his contributions to ecological theory, already asked this question in a classic 1988 paper titled "How Many Species are there on Earth?" (Science, 16 Sep 1988, p. 1441). Fittingly, he has contributed a commentary article to the recent paper, which builds on his earlier contribution.
There have been numerous tries before to get an estimate. One famous attempt (cited by both May and the recent paper) was by Terry Erwin, an entomologist with the Smithsonian Institution in Washington DC who specialized in beetles. He traveled to Panama and fogged trees with biodegradable insecticide, while assistants held out sheets below to catch the "rain" of dying insects. By laboriously identifying all these bugs, and making a few educated guesses, he surmised that each tree species in the tropical forest had 160 species of beetle that were specific to the canopy of that host species alone. He then multiplied this by the number of tree species, the proportion of beetles to the rest of the arthropods, and the ratio of canopy-dwelling to ground-dwelling species, he arrived at the figure of 30 million arthropod species in total.
It's easy to question his figures and assumptions. What's difficult is to come up with a better guess, and that's what the current authors claim to have done (view their table of other estimates and methods).
Like all previous estimates, Mora and colleagues extrapolated from what we know to guesstimate what we don't. They began with the observation that higher taxa - that is to say, families, orders, and phyla, are much better "sampled" than species. That's straightforward enough - it's harder to discover a new phylum of animals than to discover a new species (and new phyla are often "discovered" by shuffling already-known species around).
By modeling the accumulation curve for each taxonomic rank (the number of taxa of a certain rank, e.g. families, that have been discovered by a certain date), they extrapolated the asymptotic value, i.e. the projected number of taxa that would be discovered if the existing processes of discovery extend onward to completion.
At the level of species, however, the accumulation curve is nowhere near an asymptote. There, a different sort of extrapolation was necessary. They observed that taxa at different levels in the Linnaean hierarchy tended to have consistent ratios between them. If the higher levels are relatively well-sampled, then we can extrapolate downwards to the species level to estimate how many species there are in total. That is how they obtained their estimate of 8.7 million. By comparing species counts estimated by this method with other estimates in well-studied groups (like birds and mammals), they showed that their estimates were not too far off.
As Robert May stated: "This is higher than my earlier “best guess”, but I like the simplicity of this new method."
There are limitations, of course, and the authors are quick to point them out. For example, some taxonomic groups like the protozoa are still poorly known at the species level, and the scattered literature is not well aggregated. For the prokaryotes (Bacteria and Archaea), even the validity of having "species" is in doubt, because they do not practice sexual reproduction. There are remarkably few (only a few thousand) formally-described species of bacteria, despite their small size and ubiquity. This is almost certainly because the requirements for naming a new species (isolating the strain, establishing a pure culture, storing the culture in a depository) are much more arduous per species than the comparable procedures for plants and animals, and the proportion of microbiologists who are interested in taxonomy alone has always been much smaller than among other biologists.
Which brings us back to the original question: why has the media picked up on this paper and reported it so widely? After all, it's only a number. Although it's good to know, it doesn't have much of a direct impact on conservation (which would require more detailed and locality-specific data). Perhaps it might be an important parameter in some theoretical ecological models.
I think its appeal and wide reportage owes much to the desire for hard numbers. The natural world has always seemed infinite in its diversity and extent. Being able to pin down how many species exist could act as an anchor, a point of reference for thinking about nature. 8.7 million, after all, seems like a big number but isn't really so. If we accept the figure, then there are almost a thousand people for each species of life on the planet.
It also acts as a springboard for discussions about diversity and the on-going extinction of species because of human activity. Species discovery has not yet reached an asymptote, this line of commentary goes, but we humans are moving that asymptote downwards as we speak.
Reporters are happy when they have figures to cite. This species count is a sort of talisman. Complex ecological or climate models are hard to explain, and even their creators may not be fully conversant in the details or behavior. Counting the animals as they board the ark: now that's something that people can easily understand and that a journalist can write about in 500 words or less.
This suggests that scientists who are looking for more publicity for their work should try to encapsulate it in a telling figure or result. The obvious criticism is that this is merely trafficking in soundbites, but I think it can be done wisely. If out of a hundred people who read about this paper in the news, only one went to look up more details online, a significant contribution is still being made to public knowledge of science. Science reporting is most valuable when it serves as an invitation to learn more, than when it just goes "gee-whiz isn't that amazing?" without further comment.