ASN presidential talk: Ricklefs disses the neutral theory of ecology

June 21, 2011 • 5:55 am

Bob Ricklefs, president of the American Society of Naturalists, spoke last night at the Evolution 2011 meetings. Although his talk was called “My life as a naturalist,” there was little biography or, indeed, natural history, though his theme was that natural history is essential to informing ecological theory.

I won’t go into all the details of Ricklefs’ talk, but wanted to point out that much of it was a critique of an ecological theory that has caused a lot of stir in the last decade: Steve Hubbell’s “neutral theory of biodiversity and biogeography.” This theory was made famous by his book:

(Ricklefs is an evolutionary ecologist at the University of Missouri at St. Louis, while Hubbell, also an ecologist, is now at UCLA.)

I’d like to make this post short, for the “neutral theory” is complicated and, in truth, I don’t understand all the mathematics.  It’s based largely on Hubbell’s famous work on tropical trees on Barro Colorado Island, in Panama, where he spent decades mapping distributions of the many individual trees (and the many species) that inhabit even a small patch of neotropical forest.

From his work, Hubbell proposed that in many communities (not just trees), species are ecologically equivalent to one another, and thus individuals in a group of related species can be regarded as simply selectively identical (i.e., “neutral”) entities whose abundance and identity drift around randomly over ecological and evolutionary time. (There’s an obvious parallel with the “neutral theory” of evolution, in which different alleles of a gene are selectively equivalent and are affected only by random processes.)  According to Hubbell’s theory, it doesn’t matter whether a few square meters of forest is occupied by an individual of species X or Y: the ecological dynamics will be the same.

Any “patterning” of ecological communities,then, is simply an artifact of individuals reproducing and competing with each other as full ecological equivalents.

This flies in the face of years of ecological theory (supported by data) maintaining that species are not ecologically equivalent, but compete with each other based on differential use of resources.  In his book, however, Hubbell claimed that assuming ecological equivalence of all species in a group (e.g., trees) could explain certain patterns of ecology, like species abundance curves showing that some species are very common and many others are quite rate.  It turns out that, with certain assumptions, the neutral theory predicts abundance curves very similar to those seen in nature. The conclusion: since the “neutral” predictions match the data, related species must really be ecologically equivalent.

This theory always puzzled me. I’m not an ecologist, but I knew that there is plenty of evidence from nature that different but related species do use different resources and, though they compete, they are not ecologically identical.  So how could a theory based on palpably false assumptions make accurate predictions about species distributions? My view was that this was simply a coincidence, but I stress again that I am not a trained ecologist and have watched the controversy as an outsider.

But Ricklefs is not an outsider: he’s deeply conversant with both theoretical ecology and natural history (he’s done a ton of field work), and so when he criticized the neutral theory in his talk last night, that was srs bzns.

He leveled several criticisms at the theory:

1. We know that species aren’t ecological equivalents.  As I said, there are plenty of data showing that, for example, plant species inhibit each other’s growth in different ways: if you surround a member of species X with plants of species Y, the inhibition of growth is different from what you see when you surround it with members of its own species, X.  Therefore the fundamental assumption that species are ecologically equivalent is empirically false. Ricklefs also cited a lot of work on birds showing ecological differences between species (Robert MacArthur’s warblers are a famous example) and differential competition.

2.  The nice fit of Hubbell’s predictions to empirical data relies on making untested assumptions about the size of parameters.  One of these is the rate at which new species arise.  If you use other values of this parameter, which is of course unknown, the fit between real data and Hubbell’s predictions isn’t so good.

3.  Species distributions change in regular ways along environmental gradients.  Ricklefs showed distributions of tropical tree species across the Isthmus of Panama (Barro Colorado Island sits in the middle of a lake in the Isthmus).  The array of tree species on the western (wet) side is very—and predictably—different from that on the eastern (dry) side. This is not expected if different tree species are ecological equivalents that respond identically to environmental differences.

4.  Over huge ranges of space in similar environments, the distribution of species in sub-areas remains similar.  If you look at patches of rain forest in similar environments thousands of kilometers apart, the distributions of tree species are pretty much the same. The neutral theory does not predict this: the great distances between those patches means that they should be attaining equilibrium more or less independently, and so shouldn’t show identical distribution of species’ abundance if all species are ecologically equivalent.

5.  Attaining ecological/evolutionary “equilibrium” under the neutral theory requires billions or millions of years.  Ricklefs said that the neutral theory predicts that distributions of species attain Hubbell’s equilibrium values only on times of the order of millions or even billions of years.  Since no area of earth, including the tropics, has been undisturbed for anywhere near that long, the distributions of species we see cannot be those predicted by the neutral theory. There must be some other non-“neutral” explanation for species abundance curves.

Ricklef’s talk included other stuff, but I was most intrigued by these criticisms of a popular and fashionable ecological theory.  I tend to agree with Ricklefs, though I’m not as qualified as he to judge these matters.  He predicted that the “neutral theory of biodiversity and biogeography” will be dead and gone in a decade.  I suspect he’s right.  And I guess I find that judgment congenial, for it leaves species with an individuality that we’ve long appreciated as evolutionary biologists.

31 thoughts on “ASN presidential talk: Ricklefs disses the neutral theory of ecology

  1. Thank you Jerry for bringing these posts to us!

    Thisis one of the reasons I am coming back regularly – to gain a balanced view with the background information and sound logic.

    All because Jerry can think logically and is not heaavily burdened by “beliefs” 🙂

    Thank you again.

    I do concurr on the conclusion about neutral theory to be dead in a decade.

    I particularly liked the power of point number 5.

    Well done to Jerry for bringing this to us and to Ricklef for tearing apart data mining and parameter fitting.

    And I am particularly delighted that there is some thought given to “deep time” by “professional scientists”

  2. A lot of the criticisms seem to read almost identical to some of the ones that have been leveled at the Neutral Theory of Molecular Evolution, and at the molecular level evolution IS largely neutral.

    It wouldn’t surprise me if the Ecological Neutral Theory had some truth to it, once you start factoring in near neutrality like we do in molecular evolution. Seems to me that there should be both stochastic neutral processes and competition going on in ecology and that formulating a theory in that light would fit the data much better.

  3. It sounds an awful lot like the adaptationist/pluralist debate. The various positions are probably …

    1. Almost everything in ecology is due to selection and competition at the species level (pure adaptationism).

    2. Almost nothing is due to selection and adaptation (pure neutral theory).

    3. Both adaptation and neutrality are important in ecology (pluralism).

    I’m betting on #3. There are probably thousands of examples of selection at work in ecology since field biologists have been specifically looking for it and documenting it for over one hundred years. It’s how you get papers published.

    There are probably very few examples of species equivalence since in the past this would have been largely interpreted as simply a failure to find the real selectionist cause of the distribution. Such studies weren’t published.

    The fact that species compete with each other is an argument against position #2 but it is not an argument against position #3.

    I doubt very much that proponents of the Neutral Theory of Ecology deny the existence of adaptation and species competition. In other words, they probably don’t hold position #2—that’s a strawman. They are pluralists.

    The real question is whether there are ecologists who believe that position #1 is correct.

    Jerry, are there ecologists who deny that some species may be equivalent in some communities?

    1. I’m with Larry. I’d add that my feeling is that with almost any complex biological system like this, there will be a neutral/stochastic component; it’s just a matter of how important it is.

      The fact that the equations only hold for certain values of certain parameters – such as species birth rate – is not a fatal flaw; it just indicates one way in which the relative importance of neutral vs. directional processes may vary in different systems.

    2. i am wit jerry and shortearedowl on this

      “pluralism” and “pure neutral theory”

      are a lot like “accomodationism” and ” pure god’s theory”

      the latter of course has served mankind as conjured explanations before mankind accumulated enough knowledge to progress to scientific method

      this is why both “neutral theory” and “religion” will eventually die- out

      “neutral theory” under peer pressure of biologists and ecologists within some decades and “religion” under the presures of genetic imperative to survive over “deep time” (evolutionary long time for mankind to feel the consequemce of overpopulation to the degree that it abandons “belif systems” and implements “pure science” as the basis of human condition)

    3. Soil communities are a promising ground for neutral theory-like views, because there is an outstanding number of species apparently performing very similar functions living together in small spatial scale – soil ecologists (especially regarding fauna) debate about the so-called “enigma of soil biodiversity”. This is mostly thought of, and studied, in terms of feeding habits. For instance, under a single meter square of soil in a temperate forest or pasture can live 50-60 species of mites feeding on the same resource (say, fungal hyphae).
      But is their coexistence an example of neutral theory in action? I’d say no, first of all because the actual spatial scale is “inflated” by habitat heterogenity, e.g. differences in soil texture, organic matter content and stage of decomposition, water retention, pore size etc. that occur at scales of mm or cm. So, competitive exclusion is likely to occur at very small (and dynamic) scales, so that it goes undetected by standard sampling, where you take, say, a 3 cm diameter soil core and see what’s in there
      Moreover, species with similar feeding habits may differ for other ecological trait. When the topic is directly addressed, such differences are usually found (e.g.

      I agree that there are probably few out there who are hard-core neutral theorists, and that a “pluralist” theory is likely to be closer to truth. I have however to say that few ecologists, in my experience, would subscribe to position #1 either

      1. All the above are very interesting!
        Question – isn’t competition between animals vastly different from competition between plants? Isn’t the latter more a case of getting there first, maybe out-sizing your competitors?

        1. Generally I’d say so, but some soil animals are so small and move so little that a plant (or microbial) scenario actually makes more sense for them, especially for those which rely on spore-like stages for dispersal

  4. Boy do I regret having to leave the meeting early. Travel budget issues…

    I will maybe say something after considering Jerry’s summary of Ricklefs’ talk.

  5. Way above my head, but I want to understand so I looked at the blurb for The Unified Neutral Theory of Biodiversity and Biogeography

    Can someone give me a start ? Please point me at a book for the generalist lay reader where the ‘biodiversity number’ is defined & discussed. From Jerry’s comments I suspect this is a heavily mathematical discipline though…

    Cheers !

    1. I’m sorry to be self promoting here but I just published an article in TREE with Hubbell and Etienne titled “The Unified Neutral Theory of
      Biodiversity and Biogeography at Age Ten”. I hope you find it helpful.

  6. “And I guess I find that judgment congenial, for it leaves species with an individuality that we’ve long appreciated as evolutionary biologists.”

    It also seems to make commonalities between species all the more interesting.

  7. As an ecologist I felt uneasy about the neutral theory too. I read some of the papers from Hubbell’s group and I have to admit that it didn’t clarify things. Ricklefs’ observation number 2 is new to me and it really makes my unease about the neutral theory even stronger

  8. I think there is a bit of misunderstanding here about the current interpretation of Hubbell’s theory. Today nobody thinks that the theory is precisely true–the points raised by Ricklefs are fatal for that interpretation, and many of those points were recognized from the beginning by the developers of the theory.

    The value of the neutral theory is that it provides a “null model” for comparison with real ecosystems. If we hope to detect patterns due to differences between species, we need to know exactly what patterns to expect when there are NO differences between species. This is what the neutral model provides us. It turns out that purely stochastic processes can produce strong patterns even when all species are equivalent.

    The neutral theory of genetic evolution plays a similar role. In fact, the two neutral theories are mathematically identical. The main difference is that many loci really are selectively neutral. But even for non-neutral loci, the neutral theory provides a “null model”, the backdrop against which natural selection must work. The neutral model is therefore quite important for understanding the nuts and bolts of things like speciation, and deviations from it are used for detecting natural selection.

    Incidentally, both the ecological and genetic neutral models are almost universally misinterpreted due to some mathematical misconceptions shared by most biologists. The evolutionary and demographic factors which determine the differentiation between regions or demes are quite different from the one usually identified (absolute number of migrants between regions or demes).

  9. Thanks to all for inspiring a very interesting debate about neutral theory. It is unfortunate that ‘neutral theory’ in ecology is taken to mean so many different things to different people. The ‘neutral theory’ presented in the original post will not be dead and gone in ten years – it was never alive. The world isn’t really neutral and no one thinks it is as far as I know.

    I support ‘neutral theory’ as a useful tool (just as Jost said above). It is a model that can be tested and can fail – being instructive when it does. It can be of use as an approximation to many macroecological patterns, or at the very least it shows us that (much to our disappointment) those macroecological patterns do not tell us much about what is really going on in terms of traditional ecology. Finally, Neutral theory can be used as a foundation and as inspiration for more advanced models. I agree with both Gaston and Moran that stochastic neutral processes and selection probably both have parts to play here.

    Problem 5 in the original post is the biggest issue and it certainly has played on my mind in recent years. ‘Protracted speciation’ is only part of the solution – I think it is most likely that the complete solution will be non -neutral in some way, but will derive from many insights that we learned from studying neutral theory first.

    I regret that I did not come to Evolution this year, but if Ricklefs’s talk was like the one he gave at the January IBS meeting, then it would have been outstandingly good. If I remember correctly, he said that “the world is not neutral, but appears neutral” (from certain viewpoints). In this scenario neutral theory is a very useful tool and that’s why people will continue to study it.

  10. Ricklefs left out what is to me the biggest difference between reality and ecological neutral theory. In the original neutral theory, speciation occurs in a single individual, just as mutation is treated in the genetic version. In the genetic version this makes sense, because a mutation arises in a single gene. But in the ecological version, this is probably very unrealistic, except occasionally in plants where accidental doubling of chromosomes can instantly create a new species reproductively isolated from its neighbors.

    A more realistic speciation scenario in ecology is the genetic divergence of two or more entire subpopulations, until some level of divergence is reached that results in genetic incompatibility between the subpopulations (not between individuals within a subpopulation). This more realistic scenario leads to different patterns than Hubell’s neutral model.

  11. I should add that the more realistic scenario described in my last paragraph is still a neutral model. It just isn’t Hubbell’s neutral model.

    1. Good point, I totally agree that speciation mode is a big problem too, and that it can be solved within a neutral context. There has been some work done on this sort of thing already by several authors including me. The topic is now deemed large enough that there is a whole review paper on just speciation within neutral theory.

      Kopp, M. (2010) Speciation and the neutral theory of biodiversity. Bioessays 32, 564–570

      1. Thanks very much for that, I had not seen the review paper. It is relevant to an article I am writing with Anne Chao on the population genetics of speciation (trying to correct the 70-year-old mathematical misinterpretation I mentioned in my first comment above). I’ll look up your articles!

        1. Sounds very interesting indeed, please send me a copy when you are ready to share it. If you look at my TREE paper (mentioned above) and look into any citations from that then you should be pretty covered. There is one article on arxiv by Gravel and his PhD student Desjardins-Proulx that seems germane to the speciation issues and was not cited in TREE

          Good luck with writing that paper

          1. I’ll be glad to send you a copy. The main point is that standard analyses of neutral theories (both genetic and ecological) have always used a measure of between-group “differentiation”(Fst or Gst in the case of genetics) that is not really a measure of differentiation. It can approach zero even if the subpopulations share no alleles at all, and it can equal unity even almost all subpopulations are genetically identical. Therefore its value gives no guidance about progress toward speciation. I have written about this here:

            Jost, L. (2008) Gst and its relatives do not measure differentiation. Molecular Ecology 88, 2427-2439.

            Incidenally, Jerry’s thesis supervisor Richard Lewontin was one of the first to introduce this mistake into ecology and anthropology, through his classic paper, “The apportionment of human diversity”. In that paper he uses entropy instead of heterozygosity (=Gini-Simpson index in ecology) and looks at similarity rather than differentiation, but the math error is the same: he uses the ratio of within-group diversity to total diversity to judge group similarity, but this only works if the diversity measure has certain mathematical properties, which entropy and heterozygosity lack. If Lewontin had applied his method to a locus with high diversity, he would always conclude that the groups were genetically very similar, even if the groups actually were completely distinct (no shared genes at all)!

          2. That is really interesting, thanks Jost. I will look at your paper, and send me the new work when it is done.

  12. Population ecology has established that species interactions (competition, predation, mutualisms, etc, and the catenation of interactions through communities) are rife. The Hubbell theory rests on the assumption that no species interact in biological communities.
    Since we know (read any ecology textbook) that species interact, the fact that the neutral model fits data from natural communities refutes the model.
    False positives may be due to the use of fitted parameters that rob degrees of freedom or simply an incorrect formulation of a complex problem.
    The analogy with neutral gene theory is invalid because gene neutrality is experimentally confirmed in the laboratory whereas species neutrality is inferred only through fits of the model.
    Steve Hubbell’s neutral theory relies on a hidden assumption, mystical at its base, that natural selection (or something) has somehow inexplicably adjusted interaction coefficients not only to be exactly balanced across all coexisting species, but also that the coefficients automatically adjust themselves along environmental gradients. An Intelligent Design proponent would be ashamed to make such an assertion.
    It is a mischievous theory that can be accepted only when the fundamental ecological assumption of ecological equivalence is confirmed. The success of the neutral model in predicting species diversity is no more evidence for its validity than was the success of Ptolemaic system of concentric spheres in describing planetary motion.

    1. Woody, as I mentioned in my first comment above, nobody thinks the world follows this model. But neutral models are very useful if we hope to detect and understand non-neutral ecosystems. We need a neutral baseline in order to know what sorts of patterns are due to purely stochastic effects. Deviations from this baseline will vary with the strength of species interactions—as the species of a guild become more and more equivalent, their patterns of abundances will be more and like a neutral model.

    2. I agree with Jost. To use a quote from the statistician George Box ‘All models are wrong, but some are useful’. No one believes the world is neutral, what surprises me is that people believe that people believe the that world is neutral! You may be interested to look at my TREE review just out with Hubbell as a co-author. You will see that we are very clear about the non neutrality of reality, and about the difficulty of confirming the model from data that Benson and others have mentioned.

      Proponents of neutral theory simply assert that neutral models are of value as another addition to the ecologist’s toolbox. The three basic headings are 1.) null model 2.) approximate model for some macroecological patterns 3.) a building block and a testing ground for more complex and non neutral theory. It is likely that there are some neutral forces and some non neutral forces acting together in nature. As Moran summarised so nicely above – no ecologist really believes the world is totally neutral – the real question is do some still believe the world is totally non neutral?

      One minor point is that neutral theory does in fact allow species to interact – only they do so in a neutral way. For one species to increase its abundance, another has to decrease – that is interaction because the species are not independent from one another.

      Thanks for an interesting post, I hope that my response has helped.

Leave a Comment

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