How many species are there?

July 2, 2010 • 10:01 am

Since I work on speciation, I’m often asked how many species there are on Earth. You’d think that we have a pretty good idea of this, but we don’t. Most species of viruses and bacteria—indeed, if there are such things as viral and bacterial “species”—can’t be easily seen in the field or grown in the lab, and most other species are nematodes or insects hidden in the forests of the tropics.

According to Robert May, who took up this issue without resolution in a 1988 paper in Science, biologists have named roughly 1.5 million species, with 15,000 more named each year.  This, of course, is only a fraction—perhaps a small one—of all species living on earth.  And, given the predominance of arthropods among all species, this effort is patchy.  As May points out in a note in this week’s Science, one third of taxonomists work on vertebrates, which are only about 1% of all species, another third work on plants (around 10% of all species), and the remaining third work on the other 89% of taxa.

Estimates have varied between 5 and 50 million species on Earth, and that’s a big range.  At the upper end of this range come calculations from biologist Terry Erwin, who, in a short but famous paper in 1982 in The Coleopterists Bulletin, tried to estimate the number of species of beetles in the world’s tropics.  (One third of the world’s species are insects, and of these about a third are in the order Coleoptera—beetles.  There’s a possibly apocryphal story about biologist J. B. S. Haldane, who was once asked what one could infer about the nature of the Creator from his creation.  “An inordinate fondness for beetles,” Haldane supposedly replied).

Erwin fogged  (sprayed with toxic chemicals) 19 individuals of the tropical tree Luehea seemannii in Panama, killing everything in their canopies, and carefully counted all the beetles he found.  There were 1143 species represented, most of them leaf-eaters.  Erwin then made some calculations.  Assuming that a certain percentage of these beetles were “host-specific,” found only on that species of tree (this percentage ranged from 5% for predatory and scavenging beetles to 20% for herbivores), he calculated that this species of tree harbored 163 species of host-specific beetles.

Erwin then parlayed these data into an estimate of the total number of tropical, host-specific arthropods in the world.  This involved making assumptions about not only the proportion of beetles that are host-specific, but the number of trees in the tropical forests (50,000), the proportion of arthropods that are beetles (40%), and the number of total beetle species per hectare of tropical rain forest canopy (12,448). Erwin came up with an astounding figure: thirty million species of tropical arthropods in the world!  And that’s just arthopods!  This suggested to many that there were far more species in the world than previously thought.  Erwin later (1988) produced an even larger estimate–one hundred million species of arthropod.

To be sure, Erwin recognized that this estimate was very rough, based on untested assumptions.  As he said, “I hope that someone will challenge these figures with more data.”

Well, someone has—not just with more data, but with better statistical analysis, too.  As May reports in this week’s Science, Andrew Hamilton and his colleagues have, in a new paper in The American Naturalist, used new data on species numbers, and have varied the parameter estimates that Erwin saw as fixed, to come up with new estimates of the number of tropical arthropods.

I w0n’t belabor the methods, which themselves involve assumptions, but they yield estimates of arthropod species much lower than that found by Erwin. Hamilton, using probability distributions for various estimates (such as the proportion of beetle species that specialize on a certain tree), come up with a range of species numbers, each having a relative likelihood.   The two separate models give a median species number of 2.5 million (90% confidence interval: 1.1-5.4 million) and and 3.7 million (c.i. 2.0-7.4 million) tropical arthropod species in toto; Erwin’s estimate of 30 million or more has a probability of less than 0.001% of being true.

Using these medians, Hamilton et al. estimate that, since only 855,000 species of arthropods have been described, 66%-77% of the world’s species are still unknown.  As they say, “This represents an enormous amount of work for taxonomists that will take hundreds of years to complete at the current rate that species are described, taxonomists are trained, and funding is allocated for invertebrate taxonomy.

They leave out one consideration: at the current rate of deforestation, we won’t have any rainforest left in a hundred years, and most of those species will vanish without ever having been seen by humans.

So how many species are there on Earth?  It’s still murky, but a fair back-of-the envelope estimate, including all those nematodes and the unknown parasites of living species (but not including bacteria), would involve tripling the estimate of total tropical arthropods, to make about ten million species.  And that’s what I’ll tell people if they ask, making sure to add that it’s a very rough estimate.

Fig. 1. The goliath beetle, Goliathus orientalis, one of many tropical insects. Males can reach 4 inches in length and weight 3.5 oz.


Erwin, T. L.  1982.  Tropical forests: their richness in Coleoptera and other arthropod species.  Coleopterist’s Bull. 36:74-75.

___________.  1988.  The tropical rain forest canopy: the heart of biotic diversity.  pp. 123-129 in E. O. Wilson and F. M. Peter, eds.  Biodiversity. National Academy, Washington D. C. (Note: I haven’t read this paper).

Hamilton, A. J.  Yves Basset, Kurt K. Benke, Peter S. Grimbacher, Scott E. Miller, Vojtech Novotný, G. Allan Samuelson, Nigel E. Stork, George D. Weiblen, and Jian D. L. Yen. 2010.  Quantifying uncertainty in estimtion of tropical arthropod species richness.  Amer. Natur.  176:90-95.

May, R. M.  1988. How many species are there on Earth? Science 241:1441-1449.

May, R. M. 2010.  Tropical arthropod species, more or less?  Science 329:41-42.

37 thoughts on “How many species are there?

  1. As we astronomers like to say, ‘what’s an order of magnitude between friends’? In that sense, 10 million works for me — it’s almost definitely between 1 and 100 million.

  2. I spent some years working with the Anopheles gambiae species complex – if you go down the route of including cryptic species and sibling species that might not be easily recognised on morphological criteria, estimates of number of arthropod species would go through the roof.

  3. So maybe someone can explain this to me… I recognize that the term “species” has always been pretty arbitrary, but at least in the case of sexual organisms, there is an objective referent (can they/do they reproduce and make viable offspring). If one wanted, one could even conceivably come up with an objective metric, e.g. if two chosen populations of organisms were to reside in the same geographical area, in order to call them two separate species then the percentage of successful matings resulting in viable offspring that had one parent from each population would have to be below a certain threshold (ignore the implausibility of ever measuring this — my point is merely that an objective metric is possible in principle, and to use that as evidence that species classification in sexual organisms is at least partially objective)

    But what the hell criteria do you use in asexual organisms? A certain amount of DNA divergence? A judgment call on morphological or behavioral differences?

    I just don’t understand how one could even begin to say that this asexually reproducing bacteria is of species A, and this other similar one is of species B, though in the same genus. It doesn’t just seem partially arbitrary, it seems completely arbitrary.

    Just live with it, I guess?

    1. Allen Orr and I discuss this issue at length in our book Speciation (Sinauer, 2004). One solution that people have devised (not universally accepted) is to use the criterion of ecological equivalence. If two bacteria inhabit the same niche (in practice, this means if seeing if an adaptive mutation in one would displace the other), then they’re considered members of the same species. But there are other solutions, all of which are more or less arbitrary.

    2. One layman observation is that you could say that of any clustering problem; yet data can cluster.

      So yeah, it may be arbitrary, contingent on pathways (trajectories) et cetera, but still a possible characterization.

      1. I think it’s more profound than that — what are variables are you even searching for clusters in the case of asexual organisms? (Jerry gave the suggestion of “ecological equivalence” — hrm, might be another book I need to read now)

        I think your answer applies very well in terms of hand-waving the arbitrary nature of classifying sexual organisms into species… yeah, there’s messiness at the boundaries, but “you could say that of any clustering problem”.

        In the case of asexual organisms, it’s not even clear to me what the data is…!

        1. I’m not describing variable search (but of course there could be variables) or handwaving or messiness, but clustering or as Ken Pidcock say “organization is discontinuous”.

          This in turn means that there _are_ classifications on the measure of whatever space you are observing statistical clustering in. Down below Pete Moulton claims that shouldn’t be possible, but obviously PCA is a useful method.

          1. Without going into too much detail (I’m very opinionated about the issue of bacterial species), look at it from the other end. Saying that E. coli and any Cyanobacterium aren’t different species borders on insanity.

            Personally, I find the biological species concept (species are reproductively isolated populations) an often useful but sorely inadequate definition. It doesn’t work for so many animal species (North American hummingbirds, for example), and plants…!

    3. Although the biological species concept cannot be applied to bacteria, it is nevertheless clear that there are bacterial species; that is, genomic organization is discontinuous. In practice, species are defined by DNA-DNA hybridization measures.

      As big a problem as defining bacterial species is getting bacteriologists to be disciplined about taxonomy. They are notoriously sloppy. For example, we have a genus, Shigella, that everyone knows doesn’t exist, but everyone talks about it anyway.

      1. Well, indeed, when you get to unicellular organisms, most of the categories that taxonomy traditionally relies on just aren’t there. And then there’s the issue of infection with bacteriophage (as in the lethal E coli of hamburger fame – it’s coli infected with a specific phage that are the real problem).

        And then there’s the further problem of bacteriphage taxonomy. Here’s a paper where anyone interested can get their toes wet:

        And then a page showing the calculations that support the statement that if all the bacteriphages on the planet were laid end-to-end, they’d span 10exp8 light-years:

    1. Oh please, you aren’t even up on the latest advances in the study of baraminology?

      That’s the trouble with you evolutionists… you criticize and criticize, but you never bother to read up on the latest advances in creation science! How can you dismiss Creationist ideas unless you have studied every single weird rationalization that has been put forth for it? Hmmm?

      1. Rggg, I put some faux-HTML tags in that last comment to make it clear I was kidding — and it decided those were real tags and ignored them! Next time, I will remember to use my <s and &rt;s…

      2. So you are saying that I can’t reject Yule Claus because I haven’t studied all possible rationalizations for how he manage to cheat physics (knows wishes; needs to go faster than light)?

        Well, I guess that a yule theologian could say that Claus “can hear souls” (whatever that is) and uses a time machine, but then again time machines are impossible. (Computers employing time machines would break down the algorithmic complexity classes of computer science and make all physics equally ‘easy’, so for example no chaos would be possible – which we observe.)

        A yule apologist could even say that it is the deistic claus that counts, the mere principle of yule, despite that physics say it couldn’t happen according to yulism. A pantheistic yulist would say that it is the spirit of the thing, never mind that there are actually persons doing the deed.

        But, sorry, no. For good physics reason I have to reject your yule fantasy as just fantasy and superstition.

        1. Actually, the “professional” creationists do have answers for fitting all those species on Noah’s ark. Not that they aren’t answers directly contrary to biological evidence… Check out Answers in Genesis, if you must know the answer.

    2. Anybody remember the old PGC Scientifics catalogs? There was one that had Noah standing by a cart with flasks of bacteria and cell cultures, looking quizzically up to the sky.

  4. “There’s a possibly apocryphal story about biologist J. B. S. Haldane, who was once asked what one could infer about the nature of the Creator from his creation. “An inordinate fondness for beetles,” Haldane supposedly replied.”

    I read an article a year or so ago by someone who decided to chase this one down. He found that, far from being apocryphal, Haldane was very proud of his aphorism and trotted it out on several occasions.

    But I can’t for the life of me remember where I read this. Anyone out there with a better memory than me?

    1. Sorry, no can do; nature doesn’t seem to have an inordinate fondness for memories.

      But your recount seems plausible.

    2. Steven Jay Gould tracked this down in one of his essays in Nature. It is in one of the published collections, but I don’t have access to my library at the moment and can’t remember which one.

      1. Googling “Haldane beetles Gould” found me this:

        Hi Wendee,

        Apparently, Haldane had a “special fondness” for the quip but did not
        publish it.

        I looked into this a while back and found a nice summary by Gould
        (Gould, S. J. 1993. A special fondness for beetles. Natural History
        102(1): 4-9) which may be of some use to you. It seems that a good
        portion of the confusion surrounding Haldane’s famous phrase stems
        from the fact that he never actually wrote it down. None of Haldane’s
        published work contains the statement. In fact, the only written
        record of the statement comes from an address to the British
        Interplanetary Society in which, according to the society’s secretary,
        Haldane said, “…the fact there are 400,000 species of beetles on
        this planet, but only 8000 species of mammals” suggests that “the
        Creator, if He exists, has a special preference for beetles.” Gould
        (1993) mentions, humorously, that Haldane did have an inordinate
        fondness for wittier version of the statement and repeated it often.

        Hope this helps.

        Best regards,


        Tyler Cobb, PhD
        Curator, Invertebrate Zoology
        Royal Alberta Museum

      2. And it occurs to me that Gould topped him in “Full House” with his reference to the Modal Bacter. From God’s point of view, we vertebrates are unknowingly serving His first and greatest creation.

    3. See Stephen Jay Gould , “A Special Fondness for Beetles,” Natural History, January 1993, 4-11, reprinted as pages 377-387 of Dinosaur in a Haystack.

  5. It strikes me that papers like this are really just a form of mental masturbation, unless and until there’s some consensus on what actually constitutes a species. About 5 years ago, James Mallet published a paper detailing no fewer than a dozen separate and distinct species concepts, each of which has adherents and detractors, and each of which would give a different total number of species. Hamilton and his cohorts have just tried to estimate the undefined. That seems like a major waste of their time.

    1. I think you dismiss this paper extraordinarily too easily. First, your point about species concepts would be valid, except that that most biologists accept the biological species concept for animals (less the asexual rotifers). Most species concepts outside the biological are attempts to create a general species concept for asexual species that is consistent with the BSC. We have to be able to quantify diversity somehow and even if the quanta is arbitrary. Perhaps there will be an agreed upon continuous measure of diversity, but those are similarly arbitrary. Ecologists/evolutionists can’t just sit on their hands in terms of measuring diversity until some consensus of which measure is the best.

      1. Yes, but what exactly are you measuring, James? My position is that if you can’t even define something how can you expect to measure or otherwise quantify it?

  6. A study of tropical parasitoid wasps revealed that host specificity was more common that had been assumed on the basis of morphology. That seems about right.

  7. I wonder if I could manage to hold that giant beetle? I think it helps that it looks like a giant sunflower seed;I’d be more afraid if it was all dark. I bet more people would hold it if it was lady-bug colored or it looked like the opalescent green Japanese beetles.

  8. I was just discussing this last week with a friend while walking in the Lake District mountains and looking at beetles — and I was somewhat off on what I remembered about the estimated number of arthropod species. Thanks to your post, we’re now both up-to-date on this.

  9. Regarding 150 million “prokaryotic” species – we are working now on a new estimate of this number but I think this is actually an underestimate. Now – there is a separate issue as to what a species is for bacteria and archaea, but using some simple definitions, there are certainly many many many.

Leave a Reply