Cameroon lake cichlids probably did not speciate sympatrically: Part 2

June 26, 2015 • 10:45 am

Yesterday I gave the background necessary for understanding a new paper in Evolution by Christopher H. Martin et al. (reference and link below). Today I’ll briefly describe the paper’s findings—findings that cast doubt on one of our premier examples of sympatric speciation.

That example was the existence of assemblages of cichlid fish in small volcanic crater lakes in Cameroon. Because genetic evidence by Schliewen et al. (1994) showed that each assemblage was monophyletic, that is, appeared to descend from a single common ancestral species that invaded the crater lakes some time ago (between 1 and 2.5 myr for Lake Barombi Mbo and 100,000 to 2.5 myr for Lake Bermin), this gave evidence that the formation of 11 species in Barombi Mbo and 9 in Bermin had occurred sympatrically—without geographic isolation of populations.

Read yesterday’s post for background. Here are the two radiations at issue, showing the location of the crater lakes.


The authors re-did the genetic analysis of the two radiations shown above (and others from different lakes as well), producing more than 350 million DNA sequences. (This is the level of analysis that new techniques permit us.) From this they then re-did the phylogenetic analysis. Here’s what they found.

  • The “monophyly” (descent from a common ancestor) of the two radiations shown above is weaker than previously thought. The authors conclude that this monophyly, which is crucial evidence for sympatric speciation, is not “strongly supported.”
  • Further, there appears to have been substantial introgression (movement of genes) between not only some of the species in each of the two radiations, but also between the species in the lake and different “outgroup” species in rivers outside the lakes. This suggests that there have been multiple invasions of the lakes by ancestors of the crater-lake fish, and this again militates against a single common ancestor forming multiple species in the lake.

How did Martin et al. detect this introgression? By looking at the asymmetry of genetic similarity between species within the radiations, and also between those species and others outside the lake. As they explain:

Information about the directionality of introgression comes from the asymmetries in the relationships among populations given the tree. For example, if we imagine a tree ((A,B)(C,D)) with subsequent introgression from B into C, population C would show unusually high covariance with A, but B would not with D.

In other words, if the tree shows two sister pairs of species (species A closely related to B, and species C closely related to D, and the (A + B) group more distantly related to the (C + D) group, gene movement between B and C, which reside in different groups, could be detected when some genes in C could also be found in A. This shows a resemblance between species of different groups not seen in the species B and D, also residing in different groups. This pattern could only occur if there was movement of genes between species in different groups, either before or during speciation.

I won’t go into the gory details, and, truth be told, the complete details of the methodology elude me, but the results are clear: the species in the lakes are not convincingly monophyletic, and also exchanged genes not only with each other, but with some species outside the lake.

How could this happen? In several ways, all of which violate the notion that speciation (the generation of largely reproductively isolated entities within the lake after an invasion by a single ancestor) occurred with free gene flow. First, there could be multiple invasions of the crater by fish from a single lineage, with that lineage changing genetically between invasions.  That could mean that some reproductive isolation could have evolved between the invasions, and that could promote speciation between the fish in the lake.

That reproductive isolation wouldn’t be complete, and the incipient species could exchange genes but gradually become full species due to the process of reinforcement. In this process, species that already have evolved some reproductive isolation—in the from of hybrid problems like hybrid sterility or inviability—then experience natural selection for increased sexual isolation, because those individuals in each incipient species that mated with their own kind would leave more genes than those which mate with the other kind. This could eventually lead to reproductive isolation, but also to the misleading appearance of monophyly among species due to gene exchange. This is not sympatric speciation because a period of geographic isolation would be necessary to evolve the “reproductive isolation” genes.

An alternative scenario involves multiple colonizations of the lake by either different ancestral species or, as above, different segments of a single ancestral lineage at different times. The multiple colonists could then form a hybrid swarm, with copious mixing of genes, and then that swarm could sort itself out into different species in the crater lake. Again, some allopatry is required.

In both cases, the gene exchange that occurred while new species were forming in the lake gives them the appearance of all descending from a single ancestor that invaded only once, but in reality the speciation involved an allopatric (geographically isolated) phase.

Whatever happened, the signal of gene exchange is clear, and pretty much eliminates these two crater-lake radiations as a result of purely sympatric speciation. (I hasten to add that some of the genetic divergence between species surely evolved within the crater, but the reproductive isolation also required an allopatric phase for at least part of the process.)

The authors note too that the crater lakes do have some different ecological niches, and, if fish tend to mate where they live, that would facilitate speciation within a small area, but an area in which there is still some spatial isolation between the speciating populations. Traits in animals or plants that produce an automatic association between location and time of mating are called “magic traits” because they have a double effect. For example, a single pregnant female fish might like to hang around at the bottom of the crater lake, and if its offspring are produced there, and they show a tendency to stay there, as well as mating with others that are nearby, this would lead to a kind of geographic isolation based on restriction of gene flow between fish that like to live in other parts of the lake. All that is required is that fish tend to mate and produce offspring in certain areas, and offspring mate with those nearby. This would eventually yield selection to adapt to your habitat, and the appearance of reproductive isolation as a byproduct. In support of this, the authors note that many of the species in the two radiations above are “habitat specialists”, found most often in one particular part of the lake.

So the existence of “magic traits” can facilitate sympatric speciation, but that’s not sympatric speciation in its strictest sense, which involves the generation of new species without any physical separation. The authors conclude:

Available evidence suggests that all crater lake cichlid radiations speciated with the help of double invasions (Schliewen et al. 2006; Geiger et al. 2013) or remain stalled as incipient species complexes (Elmer et al. 2010b; Martin 2012, 2013). To our knowledge, all compelling examples of sympatric speciation besides crater lake cichlid radiations involve some form of au- tomatic linkage between ecological divergence and mating time or location, known as “automatic magic traits” (see review in Servedio et al. 2011).

What are those compelling examples of sympatric speciation beyond the crater lake cichlids? One of them are the two sister species of palm trees that diverged on Lord Howe Island. (See my review of this situation in my News and Views “Speciation in a small space“; reference below).  In that case, a few palms from a colonizing species found themselves in dry soil, which automatically makes them flower earlier than plants on wetter soil (in dry soil you must to get your seeds out ASAP because soil drying over time can reduce your ability to reproduce). Thus those palms, producing pollen and ovules at the same time, were more likely to mate with each other than with their conspecific palms on wetter soils that flowered later. This could eventually produce differential adaptive evolution and reproductive isolation. Living on dry soils is one example of a magic trait in which habitat and reproduction are automatically associated.

Martin et al. cite three other examples of sympatric speciation that they consider convincing. One is the case of two groups of pea aphids that live on alfalfa and tomato in the U.S., and are said to have speciated in sympatry (tomato fields often lie next to alfalfa ones). But Allen Orr and I showed in Speciation that this is not a good case of sympatric speciation, since the divergence probably occurred in Europe, where the aphids live on multiple plants, and was probably followed by double colonization of the U.S.

The other two cases involve mole rats and spiny mice in Evolution Canyon, a canyon in Israel with ecologically distinct sides. I haven’t yet read those papers, so I can’t comment on them. But what’s clear is that, after the Martin et al. paper, perhaps the most convincing example of speciation we had is no longer so convincing.

What remains? The Lord Howe plants are a good case of speciation on a very small island, but need further investigation to see if “magic traits” could have been involved. Besides those plants, there are a few other cases (other fish and fig wasps) we mentioned in Speciation, but the evidence is not very strong.  All the data to date suggest that while sympatric speciation with free gene flow is a theoretical possibility, there is little evidence for it occurring in nature. There are certainly some cases (everything possible happens at least once in evolution), and of course getting good evidence for those cases would be hard. What we can say, though, is that there are not enough data to support the frequent assertion that sympatric speciation is common.

Coyne, J. A. Speciation in a small space. Proc. Nat. Acad. Sci. USA 108:12975-12976.

Martin, C. H., et al. (2015). Complex histories of repeated gene flow in Cameroon crater lake cichlids cast doubt on one of the clearest examples of sympatric speciation. Evolution 69(6): 1406-1422.

Schliewen, U. K., et al. (1994). Sympatric speciation suggested by monophyly of crater lake cichlidsNature 368: 629-632.

42 thoughts on “Cameroon lake cichlids probably did not speciate sympatrically: Part 2

  1. At what point do we move beyond the idea that something is theoretically possible but never actually observed to the conclusion that it’s impossible?

    It’s theoretically possible for the random motions of the air molecules in the room to simultaneously converge in a corner, thereby asphyxiating all within. But the nature of statistics is such that there’s no meaningful way to use “possible” as a description of such a phenomenon.

    Could sympatric evolution be a similar phenomenon? It seems to me it almost must be, because about the only possible mechanism would be that individuals in a population “randomly” sorted themselves into two groups without any influential factors. But without some sort of barrier, simple bell-curve deviations would preclude that sort of thing, no?


  2. What about Rhagoletis pomonella, the apple maggot fly? Is that still considered an example of incipient sympatric speciation?

    1. A lot of insect genera with host-specialized sister species should fit the “magic trait” hypothesis — each potential host has its own micro-habitat and phenology, which should assort insect predators. Add to that any trait leading to adult chemical attraction to larval host, and you have a recipe for differentiation of species within a landscape.

    2. This question immediately occured to me too. My quick reread of Coyne and Orr (pgs 159-162) suggests that the R. pomonella example may suffer the same sort of general difficulty as the cichlid one – some of the “magic traits” may have evolved allopatrically. Also, the broad distribution of these flies makes general conclusions about the evolution of the different host races tricky. Obviously, Jerry can correct me if I’m wrong – after all, he wrote the book.

      1. No, you’re generally right. Even the Rhagoletis people now admit that an important part of the reproductive isolation between apple and hawthorn races evolved in allopatry–in Mexico. Thus it’s not an example of sympatric speciation: it’s sympatric host divergence (there’s still a lot of gene flow between races) that was promoted by an initial allopatric phase.

  3. Another great article simplified to a layperson’s level. However, both of the two articles raised a couple questions for me that wiki & UC Berkley didn’t have answers to.

    1) Is it possible that some genetic mutation, that while neither harmful nor helpful, could cause sexual selection to lead to speciation (the genes are distributed strictly by random statistics)?

    2) In the case flowering plants, is it possible that the prevailing wind direction, at the time of pollination, could cause a genetic drift? Even in the absence of actual geographical isolation?

    As a non-biologist, I realize these questions may have already been answered in other places, but I just couldn’t find them. I’d appreciate any help more informed readers could give.

    1. “1)Is it possible that some genetic mutation, that while neither harmful nor helpful, could cause sexual selection to lead to speciation…”

      yes, that is possible

      “…(the genes are distributed strictly by random statistics)?”

      I’m not sure what you mean here, it seems to contradict the first part of your question; sexual selection, a form of natural selection, would be a nonrandom process.

      “2) In the case flowering plants, is it possible that the prevailing wind direction, at the time of pollination, could cause a genetic drift? Even in the absence of actual geographical isolation?”

      Can you elaborate?

    2. Not bad questions. I cannot answer in detail, but for #1..
      Instances that I know of involve female preference for male colors in African cichlids. Here, very closely related species of cichlids in the same lake seem to have reinforced speciation b/c the females prefer different colors in their males. Maybe due to genetic differences in color vision in females? I am not sure. There are lots of papers about it, & <a href= Y>here is one. Like others it shows that you can experimentally break down the species barrier by having them in tanks with colored lights to cancel out the colors of the two male species.

        1. Neat article!

          Last I heard, I believe female color morph preference was the leading hypothesis for the maintenance of so many different morphs of the dart frog Oophaga pumilio, no?

      1. Thanks Mark!

        The article had me convinced that speciation within that group didn’t happen. But my biological naivete prompted the question.

        You went above and beyond.

  4. So the original generalization in your book, that sympatric speciation, if it occurs, is very rare, is yet further supported.

  5. I read that sympatic speciation was first identified in 1904 by Poulton. This is obviously the rarest type of speciation, and I’m wondering if Poulton attributed it back in 1904 to many more species that (like the cichlids) turned out to be a false assumption. Wikipedia will need an update as it mentions cichlids as being an example of sympatic speciation.
    Interestingly, it also mentions Orcas:

    A rare example of sympatric speciation in animals is the divergence of “resident” and “transient” Orca forms in the northeast Pacific.[26] Resident and transient orcas inhabit the same waters, but avoid each other and do not interbreed. The two forms hunt different prey species and have different diets, vocal behaviour, and social structures. Some divergences between species could also result from contrasts in microhabitats. A population bottleneck occurred around 200,000 years ago greatly reducing the population size at the time as well as the variance of genes which allowed several ecotypes to emerge afterwards.

    As always, thanks for the continued learnings PCC. I’m going to miss your prolific posts in the coming weeks.

    1. Maybe. But of course the test for sympatry is very stringent and we do not know if these orcas speciated while in isolation.

  6. Very interesting. I did my master’s on a system with apparent sympatric genetic divergence, so this topic is near and dear to me. I need to read up more on the current status.

  7. Thanks so much for this – I always love the science articles.
    One question though about this sentence(last line, second last paragraph): ‘But what’s clear is that, after the Martin et al. paper, perhaps the most convincing example of speciation we had is no longer so convincing.’
    Should that read ‘the most convincing example of sympatric speciation’ ?

    1. I read right past that, due to context, but now that you point it out, that does seem terribly attractive to quote miners.

    2. “Should that read ‘the most convincing example of sympatric speciation’ ?”

      You are right… JC simply forgot to add ‘sympatry’ there…

      Note: I wonder, if there is any case of population sub-division and then evolution of reproductive isolation, that happened fully under overlapping locations. I do not think we will find any such case.
      There is always some non-overlap needed to initially diverge populations to different trajectories, and then, sympatry might work…

  8. Not for the first time, an ex-chemist (and non-biologist) says VMT for this lucid and fascinating couple of articles.

    Time after time, Jerry elegantly shows how and why science works, and how it leads to advancement of knowledge and understanding, even (especially?) when it provides evidence against hypotheses as well as for them. For me, this is almost more important than the intrinsic interest of the phenomena themselves.

  9. “…producing more than 350 million DNA sequences…”


    Very neat story, thanks for the cogent explanation! FWIW, I really like the breaking up of a long subject like this into two posts.

    “Magic trait” was a new one on me–at first thought it seems like an unfortunate name, but I suppose one gets used to it.

  10. This is completely anecdotal, but once when I was playing around with the wonderful evolution simulator Gene Pool, I observed something that might qualify as sympatric speciation. Each game consists of a population of sexually reproducing “swimbots” that pass on their traits to their offspring, with their ability to reproduce being determined by the number of randomly generating food pellets they’re able to eat. Normally, when I simulate a pool in game, I end up with one single species (that is, a group of critters of very similar appearance and behavior), but in one case, for reasons that I don’t really understand, I got three. Note that there was no strict reproductive isolation or temporal isolation, here; they all occupied the same tank, and were all constantly on the hunt for food and mates. The difference between them was the ecological niches they had evolved to fill.

    One was a rapid open-“water” generalist, very similar to the kinds of swimbots that usually evolve in the game; it lived by swimming around freely in the pool at a medium rate of speed, searching for food and mates. A second organism was far faster than the first, but ONLY when it was swimming along the wall of the tank in a counterclockwise direction; it had evolved an asymmetrical body that prevented it from swimming well in open water. There was nothing actually preventing it from reproducing with the first group, as the free swimmers would collide with the walls fairly frequently, and yet the two managed to maintain distinct populations. Finally, a third group of free swimmers evolved that lived a life much like the first group, except at a far slower pace; they used energy much more sparingly, and consequently had to eat far less often, and could afford to drift aimlessly around and literally wait for a food pellet to materialize directly in front of them. These, again, frequently encountered and mated with the other two populations.

    This state of affairs actually lasted for quite a long while in game (I couldn’t give an estimate for the number of generations, unfortunately, but it was likely in the hundreds–maybe thousands, but I kind of doubt that), but eventually collapsed when the two rapid-swimming species managed to consume the food around them too effectively, and both starved out of existence–leaving the critters with slow metabolisms as the sole inheritors of the tank.

  11. For me, I’m disappointed. Sympatic speciation in volcanic lake cichlids seemed like the best way to explain the diversity of cichlids isolated for so long. I guess the missing link was they weren’t isolated after all which only became clear using modern DNA analysis.
    Thanks Jerry. If I had to get this info from the the other article, I’d still be in the dark. To the layman, that seemed like an amazing bit or research.
    As an aside, the rapid extinction of Lake Victoria cichlids is a crime and these amazing animals could be mostly gone in our lifetime.

  12. Well, I stil think my Teagueia radiation in Banos, Ecuador is a potential example of sympatric speciation. It is a monophyletic lineage with no (living) close relatives, with multiple distinctive morphological features. They all live (today) in an area of about 30 km x 90 km. A deep dry valley divides them into two sets of species; no species (currently) has a range that crosses the valley. The phylogeny shows that species did cross that gap a couple of times over geological time spans, but most of the species appear to have diverged sympatrically on their side of the valley. Many species currently grow together, 16 species on one mountain between the elevations of 3100m and 3800m, though there is (currently) some slight elevation segregation, so that maybe about six species grow together in any one spot.


    for more info.

    1. Seems logical that plants would evolve some different speciation strategies, being far less motile than most animals. (Polyploidy of course is another example.)

    2. And great fun to revisit that blog post. I highly recommend it to everyone, not only for the cool biology but also the lyrical and inspiring writing.

  13. Is there a suggestion to how multiple invasions could have reached the crater lakes previously thought to be isolated enough for sympatric speciation to be evidenced (ie flood?)

    1. Episode 1 mentioned that there were previously rivers flowing from the craters, so these lakes were connected to other bodies of water.

  14. Thanks for posting this and the explanations. Very interesting. (I read Speciation a while ago.) Sorry to be late about this, but I’ve been busy. I do want to remember to thank you for the science posts, though.

Leave a Reply