Speciation, Natural Selection, and Karyotypes

I've been chatting up Wilkins about the role of natural selection in speciation (and when I say "speciation" I mean "reproductive isolation"). Wilkins listed a few cases where speciation would occur independently of natural selection. Amongst the mechanisms in Wilkins's list was speciation via karyotypic changes (polyploidy, inversions, fusions or fissions). I cried shenanigans, and this is why.

The karyotype refers to the organization of an organism's genome -- chromosome number, fusions/fissions of chromosomes, and gene order within chromosomes. One way to change the karyotype is to duplicate the entire genome (tetraploidization), which often results in reproductive incompatibilities between the tetraploid individuals and the diploid (normal) individuals. If the tetraploid individuals do find a way to reproduce (either by selfing, finding another tetraploid individual, or mating successfully with a diploid individual), there's a good chance that the new karyotype (the tetraploids) will be reproductively isolated from the diploids. In this case, the karyotypic mutation itself has led to reproductive isolation without natural selection acting on any other variation.

But what about other karyotypic changes? Do they result in "instantaneous" reproductive isolation?

As I have mentioned before, chromosomal fusions and fissions do not, in and of themselves, prevent matings between individuals with different karyotypes. Individuals with fused chromosomes should be able to successfully mate with individuals with unfused chromosomes producing heterokaryotypic progeny. Unlike the case with polyploidy, fusions and fissions do not induce instantaneous reproductive isolation.

But what I'm more interested in are chromosomal inversions. The meiotic effects of these karyotypic rearrangements are better studied than fusions and fissions. We also know about a lot of examples of inversions segregating within natural populations. It was once hypothesized that inversions could induce instantaneous speciation much like polyploidy can. But the fitness costs of carrying two different arrangements differentiated by an inversion are too low; this is inferred based on the amount of inversions segregating in natural populations.

Here's the twist: inversions can and do play a role in speciation, and it's in conjunction with natural selection (ie, Wilkins was wrong). Rather than acting as an intrinsic cause of reproductive isolation, inversions are the substrate upon which the causes of reproductive isolation arise and evolve. That's a mouthful and rather unintelligible without the details, some of which I'll provide for you below. Keep in mind that it's not the inversions themselves that allow for reproductive isolation, but without them speciation would proceed much more slowly.

First, let's lay out our assumptions. We will assume that two populations have diverged to some extent such that interpopulation hybrids (individuals produced by matings between individuals from different populations) are less fit than progeny produced by intrapopulation matings (we call this "post-zygotic isolation"). It follows that natural selection will favor pre-zygotic isolating barriers (ie, prior to the fusion of egg and sperm) between the populations to prevent the production of interpopulation hybrids. If the populations are not entirely isolated (they come in contact and can mate with each other) these barriers must be behavioral, and, if they are to be acted on by natural selection, must be heritable (ie, genetic).

Given that there is some gene flow between the populations, the genetic factors for prezygotic reproductive isolation within one population could be transmitted to the other population during the interpopulation matings that occur. Individuals carrying the alleles that prevent them from mating with the other population are more fit (because their progeny are more fit), but if those alleles move between populations due to interpopulation hybridization, the alleles will not be as effective at preventing interpopulation hybridization. How can the movement (or migration) of the alleles from within one population into the other be prevented? This is where inversions fit in.

Inversions suppress recombination in heterokaryotypic individuals. An individual with two different versions of a particular chromosome (they are differentiated by at least one inversion) tends to produce gametes with very few crossing over events along that chromosome (all the alleles within the inversion are tied together). Now, imagine our two populations not only have some postzygotic isolating barriers between them, but are also karyotypically differentiated by at least one inversion on at least one chromosome. The inversion does not prevent interpopulation hybrids, but what if the genes for reproductive isolation are located with the inversion? Rather than being transmitted as single, independent alleles between populations, they are locked within the inversion and must be transmitted as a unified group.

A similar scenario has been modeled by Navarro and Barton, and they found that inversions significantly accelerate the accumulation of postzygotic isolating barriers between populations. It's not the inversions themselves that lead to reproductive isolation; it's the genes contained with the inversions. Gene flow between the populations is drastically suppressed within the inverted regions because the inversions contain genes that decrease fitness in interpopulation hybrids, and they are transmitted as a block because of the inversion. And natural selection can act on the genes within the inverted regions because they either confer some intrinsic fitness benefit or they favor reproductive isolation between the populations.

Not only does the model work out well on paper, the model was tested by Mohamed Noor's lab using the model organisms from the Drosophila pseudoobscura subgroup. Noor's group mapped hybrid male sterility factors (genes that decrease fertility in interspecific hybrid males) between two species pairs. The first pair, D. pseudoobscura and D. persimilis, have overlapping ranges in the northwestern United States, while the second pair, D.p. bogotana and D. persimilis, do not (D.p. bogotana is located in northwestern South America, in and around Bogota, Columbia).

i-b53588284a49c414659ddeb17bf556ae-Dpse_speciation(sm).GIF

Navarro and Barton's model predicts that gene flow between populations will lead to genes for reproductive isolation mapping to within inverted regions. It makes no such predictions when there is no gene flow between populations. Consistent with the model, the hybrid male sterility factors between D. pseudoobscura and D. persimilis (the species pair with overlapping ranges) map to regions of the genome which contain inversions that differentiate the two species. In the non-overlapping species pair (D.p. bogotana and D. persimilis), the genes show no preference for inverted regions.

Additionally, there is some evidence that inversions may play a role in the reproductive isolation of a classic model organism in speciation research: Rhagoletis pomonella. These fruit flies appear to be speciating due to the recent introduction of cultivated apples into North America. Prior to the introduction of apples, the flies fed, mated, and laid their eggs primarily on hawthorns (another fruit), but they can now be found on both plants. The two plants fruit at different points in the season, and this has led to two separate host races that each prefer one fruit over the other. Jeff Feder and colleagues mapped loci for host plant preference and found evidence that they mapped to within inverted regions (identifying inversions in Rhagotelis is much more difficult than in Drosophila). They have also hypothesized that the inversions (and possibly the genes that predisposed the species for a speciation event) were introduced into the population via individuals migrating in from Mexico.

In these examples, the inversions themselves are not under natural selection and do not induce speciation. But the genes contained within are responsible for traits involved in pre- and post-zygotic reproductive isolation between populations. It's these genes that are under selection and responsible for the evolution of reproductive isolation (speciation). Therefore, to say that speciation via karyotypic mutations proceeds without selection is misleading. Yes, polyploid speciation occurs via the intrinsic reproductive isolation of tetraploids from diploids, but inversions play a much more intricate role in the speciation process, in concert with genes responsible for reproductive isolation and selection acting upon those genes.


Brown KM, Burk LM, Henagan LM, Noor MAF. 2004. A test of the chromosomal rearrangement model of speciation in Drosophila pseudoobscura. Evolution 50:1856-1860. doi:10.1554/04-174

Feder JL, Berlocher SH, Roethele JB, Dambroski H, Smith JJ, Perry WL, Gavrilovic V, Filchak KE, Rull J, Aluja M. 2003. Allopatric genetic origins for sympatric host-plant shifts and race formation in Rhagoletis. Proc. Natl. Acad. Sci. USA 100:10314-10319. LINK

Feder JL, Roethele JB, Filchak K, Niedbalski J, Romero-Severson J. 2003. Evidence for Inversion Polymorphism Related to Sympatric Host Race Formation in the Apple Maggot Fly, Rhagoletis pomonella. Genetics 163:939-953. LINK

Navarro A and Barton NH. 2003. Accumulating postzygotic isolation genes in parapatry: a new twist on chromosomal speciation. Evolution 57:447-459. doi:0014-3820(2003)057[0447:APIGIP]2.0.CO;2

More like this

You just know I have to respond to a post that has "Wilkins was wrong" in it, don't you?

So as I understand your argument, you are inferring from the fact that it can be shown that cases where inversions do not cause speciation (and did I really say "instantaneous" in the case of inversions? If I did that was very careless of me) but are subjected to selection, show that inversions never cause speciation, and hence I am wrong?

Here's my problem with that. Reproductive isolation can be caused by small or large differences, which for instance can cause inviability by interfering with developmental timing, and the like. Chung-I Wu has been studying these "speciation genes" now for a while. Now if small genetic changes can cause speciation, why not large scale ones. Yes, in most cases inversions do not affect pairing or gene function, but does that show that none do? What if the inversion or rearrangement crosses a crucial timing gene and breaks it?

I'm not saying that inversion always causes speciation. In plants, neither does polyploidy. But it can, and your response doesn't lead me to think it can't, just that it mostly doesn't.

An argument used by Darwin in the Origin against there being any fixed amount of change needed for speciation runs roughly like this: if change of magnitude 1-9 doesn't cause speciation, but 10- does, why? What leads us to think that ther ehas to be a set amount of (in this case molecular) change that will always and only cause speciation? The answer has to be that it is different in each case. What counts is that gene flow is iterrupted, by whatever means makes it so, and if inversions cause speciation, then they can. Showing that usually they don't is a non sequitur.

Sorry for making you the subject of so many entries, but you make a good foil. And that's meant in the most respectful way possible. Additionally, I don't think you said inversions cause instantaneous speciation (but we could always look that up).

As for inversions disrupting genes, causing instantaneous (my word, not yours) speciation: I'd think it's highly improbable. One nice feature of Navarro and Barton's model is that the alleles that are incompatible between species are advantageous. It's hard to imagine that a gene that is disrupted would be neutral or advantageous.

Of course, one example of such a case, and my hypothesizing would go down the drain. In this case, is absence of evidence evidence for absence?

We will assume that two populations have diverged to some extent such that interpopulation hybrids (individuals produced by matings between individuals from different populations) are less fit than progeny produced by intrapopulation matings (we call this "post-zygotic isolation"). It follows that natural selection will favor pre-zygotic isolating barriers (ie, prior to the fusion of egg and sperm) between the populations to prevent the production of interpopulation hybrids.

Hum. Would we also expect these barriers to be different in nature between males and females?

For both sexes, you'd predict changes in mate preference - to avoid wasting time on the "wrong" partner.

For females, you'd also expect some sort of changes between mating and fertilization - changes in egg membrane so the sperm can't penetrate, for example. That way, even if you do mate with the wrong guy, at least you're saving the egg for a better sperm.

For males, you wouldn't expect that - once you've mated and moved on, even a half-fit offspring is better than none

(The above of course assumes non-monogamy, and that males don't hang around to take care of the offspring.)

By Peter Ellis (not verified) on 05 Feb 2007 #permalink

Awesome post. You should post on speciation and speciation only from here on out. Who cares about simulations. Simulations aren't science.

Yeah, following Peter Ellis' comment, sounds like your argument rests on reinforcement being an important factor in speciation. Might there not be another process by which the genes for reproductive isolation get concentrated in the inverted regions?

Also, not related to the Noor et al. work, inversions might be important in speciation in a more drifty way, namely in founder effect speciation. Dobzhansky's early work on 3rd chromosome inversions and what might now be called conversion of epistatic variance during founder events seems particularly promising. Though this isn't as straightforward as polyploidization or other chromosomal phenomena causing speciation, ahem, reproductive isolation.

Yeah, following Peter Ellis' comment, sounds like your argument rests on reinforcement being an important factor in speciation. Might there not be another process by which the genes for reproductive isolation get concentrated in the inverted regions?

While my explanation was focused on reinforcement, Navarro and Barton's model deals with the accumulation of Dobzhansky-Muller postzygotic incompatability factors. The genetic factor contributing to postzygotic isolation in the model are beneficial mutations, but they are incompatible on the genetic background in the other population.

I'm not sure where you're going with Dobzhanksy's 3rd chromosome system. I'm quite familiar with the system, but I'm not clear what you're saying.

And Peter, you're introducing more biology into this than I intended to cover. Your comment brings up some important details regarding how exactly prezygotic barriers can develop and the asymmetry issues (and conflicting interests) that go along with them.

Well, I'm abandoning my idea about Dobzhansky's 3rd chromosome inversions for now. For what it's worth, I'm working on a paper about the history of Dobzhansky's experimental work on the founder effect in the 1950s and 60s and how it relates to work on founder effect speciation since the late 1970s, and in particular Rundle's 1998 and 2003. I'm interested in particular in how evolutionary causes are conceptualized in this work. But it is proving a little difficult to connect directly to the stuff you're writing about here, although I still think there's a connection.

But I have another question. It seems like the chromosome rearrangement is what causes persistence of sympatric species in Brown et al. (2004). Of course the genes for reproductive isolation contained in the inversion are necessary, but what makes the difference is the inversion, not the genes. In allopatric populations when the (presumably) same isolating genes are distributed unlinked (except by epistasis) throughout the genome, when the populations make contact the genes are broken up by meiosis and the populations fuse. The difference making cause here is the inversion because it is what differs between the two cases. Is this way off base?

If the inversions themselves cause reproductive isolation, then the 3rd chromosome polymorphism would not exist as a polymorphism -- there would be a Tree Line species and an Arrowhead species. Because the 3rd chromosome does not contain genes for reproductive isolation (that we know of, although this would be an interesting experiment) those inversion differences do not lead to speciation. But the X and chromosome 2 have QTLs that are associated with reproductive isolation and it's not simply because of the inversions (if it was, then the QTLs between Dper and Dpbog would also preferentially map to inversions). The reproductive isolation is due to genes located within the inversions, and the inversions prevent those genes from introgressing between species when gene flow is present.