How many species are there? My latest for the New York Times

In 1833, John Obadiah Westwood, a British entomologist, tried to guess how many species of insects there are on Earth. He extrapolated from England to Earth as a whole. “If we say 400,000, we shall, perhaps, not be very wide of the truth,” he wrote. Today, scientists have found over a million species of insects and keep finding more every year.

The question of how many species there are on Earth has been a tricky one ever since Westwood’s day. I’ve written a story for the New York Times about a new estimate that was published today: 8.7 million.

What makes the paper particularly interesting is that it introduces a new method for estimating biodiversity. The method is based on Linnean taxonomy. While we have lots of new species left to find, we may have found most of the classes, orders, and phyla. It turns out that for a number of groups–mammals, birds, and so on–the numbers of each of these rankings rise as you descend the hierarchy.

Here’s a diagram that summarizes this striking pattern (courtesy of the Census of Marine Life). I couldn’t fit it into the story, so I thought I’d show it here:

The scientists reasoned that we’re probably closer to having found most kingdoms, classes, and other high level groups. So they used this relationship to estimate how many species there are in well-studied groups like mammals and birds. They found this method got them a number close to the actual number of species. So they applied to other groups, such as plants and fungi.

As I write in the article, some experts love this method, and some don’t think much of it. I couldn’t get into deep details in a 1,000 word piece. Here’s part of a long email I got from Lucas Joppa, an ecologist at Microsoft Research in Cambridge, England. Joppa thinks the new method is important and intriguing. And he added some interesting thoughts about why knowing this number matters–aside from just being a very basic question that’s worth answering because we can–

I do think that it matters that we try to estimate this number, although given that we are talking about millions, I don’t really think it changes our daily perception of how many species there are (the human mind has problems with any number larger than a few hundred!). Moreover, I’m not quite so sure it matters if we are able to put an exact figure of how many species there really are, as when you look at the scope of the problem (2 million currently described, likely 7 million more!) it is unlikely that we will ever reach a full census of life on earth.

That said, the goal of coming up with a sensible estimate is not only noble, but worthy from a conservation perspective…the species currently unknown to science (at least in a terrestrial sense for well-known groups such as flowering plants) are likely to have ecological traits that are correlated with extinction risk (small ranges, rare within those ranges, etc.). Because of this, putting a number on the total number of species gives us insight into the number of missing species, and thus insight into the increase in the estimated numbers of species threatened with extinction around the world. In a recent paper in the Proceedings of the National Academy of Sciences, Stuart Pimm and I, along with collaborators, show that at least for one taxonomically important group (flowering plants), those species currently “missing” (ie, undiscovered) are most likely to be found in places that are already identified as global conservation priorities.

So, the good news is that even without having a full catalogue of life, the global conservation community is already actively engaged in protecting those places where species are most at risk (ie, Biodiversity Hotspots, locations with high number of species found nowhere else, but with extensive (>70%) of natural habitat loss). The bad news is that most new species will come from places around the world most at risk! As you can see from that direct example, while knowing every single species on earth is not a likely scenario, estimating information about those species, as Mora et al. do, can drastically change the way we view current estimates of species extinction risk around the world.

12 thoughts on “How many species are there? My latest for the New York Times

  1. “The scientists reasoned that we’re probably closer to having found most kingdoms, classes, and other high level groups.”

    I’ll be impressed, as soon as someone can define “kingdom”, “class”, etc. for me. Any takers? And the figure seems to assume that “species is to genus as genus is to family, etc.”. What is the justification for this? We already know that “families” of plants are not comparable to “families” of insects or vertebrates, neither in age, nor diversity, nor disparity…so what exactly are we counting?

    Down with Linneaus!

    [CZ: Nick–the scientists don’t think such a definition is necessary to use the hierarchy to get these estimates. As artificial as its origins may be, Linnean classification today captures something about the nature of diversification, as evidenced by the validation the scientists carried out on well-studied groups. At least, that’s what they say.]

  2. Linnaeus’ genius was pretty much an accident, but its importance is hard to deny when one considers the facts. Even though he was himself a creationist, Linnaeus showed that relationships between biological organisms were hierarchical: Species within one genera are (ideally) more similar to each other than to members of other genera within the same family, genera within one family are more similar to each other than to members of other families etc. This pattern, which always was suspected, was formalised through Linnaean nomenclature and was thus made subject of constant testing.
    The pattern of groups within groups corresponds very well to the modern paradigm of cladism, in which groups originate when their parent species branch in two or more.
    Whereas Nick is right in that a group of one given Linnaean rank can not be directly compared to another group at that rank, often the groups that have Linnaean ranks are better supported than other groups in terms of monophyly, breadth of evidence etc. “Phylum Arthropoda”, “Phylum Mollusca”, “Class Insecta”, “Class Mammalia” etc. are supported as monophyletic both by genetic data and by the distribution of morphological characters in space and time, whereas most clades are supported by much smaller numbers of characters.
    The Linnaean classification thus ensures both accuracy (the groups in the classification corresponds to actual groups generated by natural evolutionary processes) and stability (the groups in the classification do not change much over time).

  3. I completely agree with Nick Matzke that this kind of counting and estimating is rather worthless considering the fact that there is no non-arbitrary criterion at all for assigning ranks or to compare unrelated taxa of equal rank.
    I do not agree with Gunnar that groups with Linnean rank are better supported as monophyeletic. Obsolete paraphyletic taxa like Protozoa, Invertebrata, Apterygota, Hemimetabola, Anamnia, Reptilia and Pisces all had well-established Linnean ranks. Furthermore, taxa like Mammalia or Aves are only well supported if they are restricted to the recent crown group and compared to their recent sister group, but they get more and more watered down in support when fossil stem group representatives are included.

  4. “may have foud”- foud should be “found”.

    More substantively, I was thinking along lines similar to Nick’s. It is interesting that the lack of precise definitions doesn’t hamper this method too much. I’m also wondering if it might not handle the issue that some phylums and families are just tougher to find or detect distinct species. This might cause this method to undercount somewhat. I’m not sure. They’ve probably thought about this issue.

    [CZ: Type fixed. Thanks. The authors said to me that their suspicion is that modern taxonomy is so closely tied to real phylogeny that it preserves a signal of how lineages diversify–on average.]

  5. As a comparison, in physics effective theories (or models) do have some predictive ability.

    Not being a biologist, it seems to me here taxonomy compare to cladistics as newtonian gravity compare to general relativity as it “preserves a signal”. (And even general relativity is known to be effective, non-quantum. Similarly I assume cladistics gets modified by HGT, say.)

    And at least in physics you don’t use the more complex methods if the simpler works. Though you may well want to know _why_ they work.

  6. So they applied their method to well known groups like birds and mammals and found it fitted their relationship and then they conclude it works because these groups are well studied. But surely the ‘well studied’ part also includes modifications to the groupings called species, genera etc which doesn’t happen to the less well studied.

    Some nice observations but hardly supportive of a predicted figure precise to 2 significant places.

  7. Ken Ham and his Answers in Genesis group are well along with their real life sized Noah’s Ark theme park project in Kentucky, USA. They have announced that the ark will have between 2000 and 4000 individual species in it. Assuming no insects or fish, that puts the creationist upper limit on what constitutes a valid species. This should save scientists a lot of time in making classifications from now on. We can only hope that the dinosaurs are in a secure cage.

Leave a Reply

Your email address will not be published. Required fields are marked *