New paper on speciation from our lab

March 28, 2010 • 7:52 am

My hot-dog student, Daniel Matute, has published a new paper in PLoS Biology, and although I say the paper is from “our lab,” it’s entirely Daniel’s work.

The paper is about a phenomenon whose importance in speciation is contested: reinforcement. Reinforcement describes a form of natural selection that can actually increase the reproductive barriers between a pair of species that descend from a common ancestor.

As I’ve mentioned on this site, most evolutionists define “species” as “groups of populations separated from other such groups by reproductive isolating barriers (RIBS, an appropriate Chicago acronym).” Those barriers are all the genetically-based traits of organisms that prevent them from exchanging genes where they co-occur in nature.  RIBS include things like mate discrimination, a preference for living in different niches (so that members of different species don’t encounter each other), different mating seasons, or the sterility or lethality of any hybrids that are formed.

Most of us think that these barriers are simply the byproduct of evolutionary divergence that occurs when geographically isolated populations of a single species begin to evolve in different directions. I give examples of how this works in WEIT, but the discussion is too long to repeat in this short post.

In this view of life, species are simply evolutionary accidents. That is, nature doesn’t somehow favor the production of species, nor does natural selection favor the erection of reproductive barriers between groups per se.  Rather, physically isolated populations evolve in different directions due to natural selection or random genetic drift, and when that divergence has proceeded to a certain extent, reproductive barriers appear simply as a byproduct. For example, if two species adapt to two different niches, their physiologies may diverge to the point that the combination of both species’ genes in a single hybrid makes it unfit or inviable.

But there is one form of selection that can favor the evolution of reproductive barriers, and that’s reinforcement.  Reinforcement begins with two physically isolated populations that have been apart long enough to evolve some barriers to gene flow, but barriers that are not quite complete—so that they can’t qualify as “full” species.  For example, two populations of flies might produce hybrids that are partially (but not completely) sterile, so if they meet there is still the possibility of some gene flow between them through the bridge of semi-fertile hybrids.

Now assume that these populations expand their ranges and, though previously isolated physically, now encounter each other. (This happens all the time in nature as geographic barriers like rivers, forest, and the like disappear or change their positions.)  Members of the two populations will encounter each other, and some of them will mate.  But if you mate with members of the “wrong” population, some of your offspring will be sterile or partially sterile.

There is thus an evolutionary disadvantage to mating with the wrong group, and an advantage to mating with the “right” one. Under these conditions, natural selection will favor individuals with genes that favor their mating with the right group, because those genes leave more copies than genes that don’t promote discrimination—”nondiscrimination” genes tend to wind up in sterile hybrids!

Thus, over time, a partially isolated species can become a fully isolated one through natural selection, and that’s how reinforcement works.  It’s a process that directly favors the erection of reproductive barriers, and can, we think, put the finishing touches on speciation.

The problem is that though this sounds nice, and can indeed work if you make mathematical models, it’s hard to demonstrate in nature.  To do so, you’d need to show that, in a pair of species that have some geographical overlap, the degree of mate discrimination, or of other reproductive barriers that prevent the formation of hybrids, is higher between populations in the area of overlap than it is between populations taken from areas where they don’t overlap (these tests are done in the laboratory).

There are a few promising examples of this (one was published in 1995 by my ex-student Mohamed Noor), but the frequency of reinforcement remains unclear. (When I was young, I was taught, for example, that it was always the final step in speciation. That of course can’t be true, because speciation can go to completion between groups that never contact each other in nature, for example species like the Galápagos tortoise that evolved after an ancient colonization of a remote island.)

Daniel noted that the species of Drosophila on the small African island of São Tomé, where we’ve worked for some years, have a distribution that looks promising for studying reinforcement.  One species, Drosophila yakuba (“yak”), lives not only on the island but also on the African mainland.  The other species, D. santomea (“san“) is endemic only to the island, and presumably evolved after the island was colonized by the ancestor of yak (we estimate this happened about 400,000 years ago).

The situation is propitious for studying reinforcement because, on the island, which is basically a volcanic mountain, the species have overlapping distributions: yak lives at sea level and up to about 1400 meters on the mountain, while san is found between 1100 and 1400 meters. There is thus an area of overlap between 1100 and 1400 meters, and in this area you can find a few hybrids (about 5% of the individuals).  These hybrids are at a disadvantage, evolutionarily, because half of them—all the males—are sterile.

Thus we have two species with overlapping ranges, some hybridization in the area of overlap, and the possibility of gene flow through the fertile female hybrids.  This is just the situation we need to look for reinforcement.  So, is the mating discrimination between yak and san stronger when they come from the area of overlap than from the areas where they don’t overlap? (Remember, we do these tests in the lab and so can examine any pair of strains.)

Surprisingly, the answer, which we got eight years ago (Coyne et al. 2002), was no: mating discrimination, though strong, was equally strong no matter where you took your samples from. There didn’t seem to be any reinforcement.  We don’t know why, but there are several explanations (perhaps, for example, there simply weren’t any mutations that sharpened mate discrimination.)

Daniel, however, went further, and found that there is indeed a form of reinforcement between these species, but not reinforcement of mate discrimination.  Rather, it’s reinforcement of what we call postmating, prezygotic isolation or gametic isolation: barriers to gene flow that act after mating but before fertilization. One example is females of one species not using the sperm of males from the “wrong” species if they are inseminated by them.

This is what Daniel found:

1.  Yak females from the area of overlap produce many fewer eggs (and progeny) when mated with san males than do yak females derived from outside the area of overlap.  Thus, the reproductive barriers between the species were higher from places where they encountered each other. The effect was very strong and highly statistically significant. (Daniel’s Figure 2 shows this clearly).  Daniel didn’t find this effect, however, when he looked at san females.

2.  But what is the evolutionary advantage of producing fewer eggs/offspring when you’re inseminated by a male of the “wrong species”? For that would have to be the selective pressure that impelled the evolution of the “fewer-egg” syndrome in yak females who encounter san males.

One possibility is that if you live in the area of overlap, and sometimes mate with the wrong species, you have an evolutionary advantage if you can “recognize” this (I’m speaking of course of physiological recognition), dump those foreign sperm that produce semisterile offspring, and thus get a chance to mate sooner with the right species. (Species in this group tend to mate multiple times over their life.)

Sure enough, Daniel showed that when yak females were mated in the lab to san males, those females from the area of overlap lost sperm much faster than did yak females from outside the area of overlap.  And he also showed that those females remated more quickly to their own males (yak) than did females who hadn’t used up their “wrong” store of sperm. In other words, there was an evolutionary advantage to dumping sperm when you’re inseminated by the wrong species, because by so doing you will, in the long run, produce more offspring with members of your own species. (We don’t know the mechanism by which females from the area of overlap get rid of the wrong sperm more quickly).

3.  So, we find a pattern suggesting reinforcement and show that this pattern comports with how selection should operate.  But can we mimic this whole process in the lab?  Surprisingly, Daniel did, and it took only a few generations.

Daniel simply forced strains of the two species (taken from areas where they do not encounter each other) to live together in glass bottles in the lab. There were seven different pairs of strains to serve as evolutionary replicates.  Each generation he removed the hybrids, mimicking complete sterility of the hybrids (of course, they’re not completely sterile in nature, but we wanted to avoid gene flow between the species, which makes it harder to tell them apart).

And, in only four generations (about two months), he observed that yak females had indeed evolved the “reinforcement” trait seen in the lab: when inseminated by san males, those females laid fewer eggs than did “control” yak females who were reared at the same density, but without the presence of the other species.  Moreover, the degree of mating discrimination also increased, something that we didn’t observe as a pattern in the wild.

CONCLUSION:  Well, the reason the study got published in PLoS Biology, I think, is because it’s not only a rare, pretty-well-documented case of reinforcement, but also because it’s the first case in which reinforcement in nature strengthened not mating discrimination, but a reproductive barrier acting after mating: gametic isolation.  Most people hadn’t realized that postmating barriers could be also strengthened by selection.  But they can be, and cases like this may be common.  The reason why they might have been missed is, of course, because detecting the presence of a gametic barrier is much harder than simply measuring increased mate discrimination.

The next step is to allow some gene flow between the captive species, and see if we can still observe the evolution of gametic (and sexual) reinforcement under more realistic conditions.  After all, in nature some fertile female hybrids are produced in the area of overlap, and presumably mate back to yak or san. That study is underway.  And we’d like to understand why the evolution of higher mating discrimination occurs so readily in the lab, but isn’t seen in nature.

And a h/t to Daniel here, a terrific student who promises to be a star.


Matute D. R. (2010) Reinforcement of gametic isolation in Drosophila. PLoS Biol 8(3): e1000341. doi:10.1371/journal.pbio.1000341

Coyne J. A, Kim S. Y, Chang A. S, Lachaise D, Elwyn S (2002) Sexual isolation between two sibling species with overlapping ranges: Drosophila santomea and Drosophila yakuba. Evolution 56: 2424–2434. (Our earlier study failing to find increased sexual isolation between the species in the area of overlap).

Noor M. A (1995) Speciation driven by natural selection in Drosophila. Nature 375: 674–675. (One of the best examples of reinforcement in nature, done in my lab by Mohamed Noor fifteen years ago.)

21 thoughts on “New paper on speciation from our lab

  1. “although I say the paper is from “our lab,” it’s entirely Daniel’s work”

    Ah, another point for you, Jerry Coyne. Nice to see you’re not one of those senior scientists who force their students to add them as co-authors to a paper regardless of whether they’ve contributed the slightest bit to it or not.

    1. Yes, I see this practice as verging on unethicality! Some day I’ll write a post on the ever-increasing and nefarious practice of gratuitous co-authorship by senior scientists.

      1. Please do, Professor Coyne.

        When it gets to these things, I often have to think of the professor who had himself added to Jan-Hendrik Schön’s physics papers in one of the greatest science frauds of all times. He was happy to partake in co-author glory — but when the fraud came out, he had the nerve to justify himself by claiming he really couldn’t have known!

      2. I hear a story about some senior scientist who complimented Dick Lewontin on Marty Kreitman’s 1983 single author adh paper in Nature: “I liked YOUR paper in Nature.” It seems like even despite one’s best efforts, that redistribution of credit still gets accomplised.

        At the very least, it would be nice if pis would surrender corresponding author more often.

        Out of curiosity, how do you think tenure review would go for a young pi who followed the Lewontin/Coyne philosophy and half of the papers conducted in the lab lacks the pi’s name?

          1. I praise Dr Coyne and others that bring forth now and before, this vastly understated, underreported problem that impacts with overwhelming consequences many budding scientists. This form of abuse reflects of course deeper problems of silent, widespread intellectual dishonesty and bullying, difficult to tackle but of unarguable urgency , nonetheless.

            1. The issue is one of culture. Everyone knows it exists, and the people that do it tend not think they do it in a “bullying” way (though exceptions abound, of course). In fact, many students would feel naked without a P.I. slapped onto a paper because they judge (rightly so in some fields) that their papers will be ignored by other labs unless their boss is there to get attention.

              The whole culture stinks in general and changing it isn’t simply a matter of telling the big-bad meanie P.I.s to stop usurping the downtrodden serfs/grad students. Too many people buy into it. It is people like Jerry (and his adviser Dick Lewontin for two familiar examples) that show how wrong the assumptions that that culture is based on are.

              As I understand it, not being on a paper doesn’t hurt a P.I. in terms of grant renewals. As long as the grant is acknowledged on the paper, that’s all that matters. (If you’ll look at the funding statement of Daniel’s paper, Jerry’s NIH grant gets its due: “This work was funded by National Institutes of Health grant R01GM058260 to Jerry A. Coyne.”)

              However, what I’m not so sure about is how tenure review goes if the author isn’t on the paper. NSF and NIH are happy if they appear in a funding statement. But when your peers are reviewing your work for tenure, are they so easily mollified if they come from the culture that requires a P.I. to be on every paper? And if the review is being done in an oddly structured university where MD/PhDs can vote on tenure for speciation geneticists, would they even understand why a P.I.’s publication record doesn’t even have the P.I.’s name for half of the papers?

              Anyway, I’m still curious about what advice Jerry would give to a young pre-tenure P.I. in that situation.

  2. Nice research. Now, of course, you enticed us lay people with a tantalizing post to force ourselves to go read the entire paper. Oh, you nefarious, evil scientists, making us learn things!

    Were the subjects of the project released back into their habitat in the wild upon project completion? 🙂

  3. Very interesting indeed.
    I think the creationists have something to be jealous about. Science is thrilling.
    But I am very curious to know how they manage to tell apart the “healthy sperm” from the “toxic sperm”. After all the insight might have potential for the banking industry…

  4. Sometimes I wish I could turn the clock back about 20 years. Then I come to my senses and just enjoy watching these young people (and not so young people) write new chapters in our story.

  5. How is a hybrid “semi-fertile”?

    Sounds like being a little bit pregnant.

    Or is it a statistical distribution of fertility over the hybrid offspring of the two (semi-)separate species?

    (This is from a meteorologist.)

    1. All there has to be is some fertility problem in the hybrids. In this case it’s that half of the hybrids are completely sterile (males), while females are fertile. In other cases either sex can be fertile, but not as fertile as pure species (think of those human males who produce fewer motile sperm and thus have a reduced chance of fathering a baby).

      1. Are some ‘reinforcement” mrchanims under the umbrella of sexual selection? (question out of ignorance)

        1. The interesting thing is, they would begin within the natural selection domain, then end up squarely under sexual selection.

          It’s not hard to imagine a small change which helps a female recognize a male of the same species (i.e. the gene pool that provides maximum fertility to offspring, in this case), which becomes amplified later, long after it’s utility has been maximized.

          That would be a case where Fisher’s explanation of sexual selection applies (despite so many biologists implying that there’s a correct answer, there certainly is not – there are, rather, many correct answers). Genes for both creating a larger X and preferring to mate with partners having a larger X would find themselves together, which means subsequent generations would make still larger X’s, and desire still larger X’s.

          X, in that example, could be a bump that turns into a giant horn on a beetle, a longer or more colorful feather on a bird, etc.

  6. I demand that I be allowed to do some science and the following is a demonstration of my unrivaled ability 😉

    is greater in the area of overlap that it is between populations take from areas where they don’t overlap (these tests are done in the laboratory).

    I think should be:
    …overlap than it is between populations taken…

  7. Nice study! Pass on my regards to Daniel! It is always nice to see a fascinating evolutionary study from one of the ‘lesser known’ but equally amazing island systems.

    I am now wondering how to find the perfect model system to look at this for pollination interactions – along the lines suggested by my colleague Timo van der Niet a few years ago, for South Africa… (van der Niet et al 2006, Evolution, 60: 1596–1601). Must go back and drool some more in Namaqualand, methinks…

    By the way, if you and/or Daniel return to Sao Tome and/or Principe, keep an eye out for the endemic collared fruit bat & let me know what you see it feeding on 😉

  8. Hi,

    Very cool result! I am very interested in these types of questions. You mention the next step is to allow gene flow between the species which interesting but I also wonder about gene flow between the sympatric and allopatric populations within species too. Given that the yak population size was smaller in the sympatric region and the seemingly high likelihood for gene flow (I’m guessing based on the map) this finding seems to suggest the possibility of some sort of cryptic barrier between the yak populations as well does it not? Is there a mechanism that could maintain the trait in the face of ongoing (swamping?) gene flow or is strong selection enough?


  9. I seem to remember reading something about birdsong in neighbouring species of birds in the American southwest. At the centre of the range of each species there was a considerable variation in birdsong, but at the edge of the ranges where the two were likely to encounter each other, there was little variation in song, and the songs were quite distinct.

Leave a Reply