Why Evolution Is True

There are no ring species

TRIGGER WARNING: Evolutionary biology.

A while back, when I said in the comments of an evolution post that there were no good “ring species,” a few readers asked me what I meant by that. “What about the salamander Ensatina eschscholtzii? Or seagulls in the genus Larus? Aren’t those good ring species?” My answer was that those had been shown not to be ring species in the classic sense, but there was still one species that might be a candidate: the greenish warbler Phylloscopus trochiloides around the Tibetan Plateau.

But now that one, too, has been struck off the list of ring species, leaving no good cases. Its removal from the class is documented in a new paper by Miguel Alcaide et al. in Nature (reference and link below), in a group headed by Darren Irwin, a professor at the University of British Columbia and including my next-door Chicago colleague Trevor Price.

But first, what is a ring species? Ring species constitute one big and supposedly continuous population in which the attainment of biological speciation (to people like me, that means the evolution of two populations to the point that they cannot produce fertile hybrids were they to live in the same place in nature) does not require full geographic isolation of those populations. Rather, speciation in that continuous population occurs through a gradual spread of the range of the animals, coupled with selection in different places that causes their genetic divergence.

It was long thought by many that for a single species to become two species—to undergo “speciation”—the populations had to be completely separated geographically, so they could evolve along divergent paths without the pollution of genetic interchange that would reduce their divergence.  We now know that that isn’t true, and speciation can result even though the populations becoming new species exchange some genes while diverging. Allen Orr and I discussed all this in our 2004 book Speciationthe book that I still consider my proudest accomplishment (see chapters 3 and 4 for the discussion of speciation with gene flow; “ring species” are discussed in chapter 3, pp 102-105).

So a ring species is one case of speciation that is supposed to occur without any geographical isolation.

It works like this: a species expands its range and encounters a roughly round geographic barrier like a valley, the Arctic ice cap, or an uninhabitable plateau. It divides and spreads around the edges of the barrier, so that its range becomes circular as it expands.  And as the range begins to form a circle, the populations within it begin to become genetically different as they respond to local selection pressures. But the circle is never interrupted, so while each part of the expanding species becomes genetically different, it still exchanges genes with adjacent populations.

What this causes is a group of populations in which adjacent areas are genetically similar, but become less similar as they become more distant. That’s because the more-distant populations supposedly experience more-different environments, and gene flow between distant populations is attenuated because genes have to flow through all the intervening populations.

At the end, the populations have expanded so far that the ring has “closed”: the species has completely encircled the barrier and the two most genetically diverged populations contact each other.  If they are so genetically diverged that they cannot form fertile hybrids, they then appear to be two biological species.

This is a bit of a conundrum because the two “good” species are connected by a chain of populations around the circle, and each population can exchange genes with the adjacent one.  This holds all the way around, so, in theory, every population in the ring really belongs to the same species.  It’s like breeds of dogs: the Chihuahua can exchange genes with a slightly bigger dog, and that one with a slightly bigger dog, and so on up to the Great Dane. Given this, are Chihuahuas the same species as Great Danes? If you put both species in a kennel together, they couldn’t form hybrids, so you might think “yes, they are different species.” But if you put all breeds of dogs in a kennel, they’d eventually, by mating with dogs of similar size, form a hybrid swarm of mongrels in which Chihuahua and Great Dane genes are found in the “hybrid swarm”.

So even though the populations at the end of the ring behave as two different species when they meet, they’re connected all around the ring by gene flow.

Such species are known as “ring species.” They’re interesting for two reasons. First, they show that you can get the evolution of complete reproductive isolation without geographic isolation. Second it’s a judgment call whether you call them one or two species. If you call them two species, where do you draw the line between the two, given that any adjacent pair of populations around the ring are clearly members of the same species?

There used to be several examples of “ring species” that were staples of evolution textbooks, the most famous being the salamander Ensatina eschscholtzii in California. This species was first worked on by Robert Stebbins but later and most intensively by David Wake of the University of California at Berkeley and his colleagues.  It was a classic example supposedly demonstrating all the principles I described above.

An ancestral population of salamanders from what is now Northern California or Oregon was supposed to have spread southward, and then split, with one group expanding its range down the Coast Range, and the other moving east and then expanding southward down along the foothills of the Sierra Nevada. The intervening “center” of the ring was the Central Valley of California, which is grassy, dry, and uninhabitable by these plethodontid salamanders, which need moist habitat.  The species thus formed a classic ring, differentiating genetically as both branches moved south. In fact, they became different in color and morphology, and were classified into seven subspecies, as shown in the diagram below.

The ring closed when the ranges encountered each other in southern California, where the subspecies E. eschsholtzii eschscholtzii encountered the long-diverged subspecies E. e. klauberii. These two did not interbreed in nature, and so behaved as different species. Genetic studies demonstrated a long divergence between these, attesting to the “move around the ring” scenario, but also to a lesser divergence between adjacent populations. The ring is shown in the following diagram, along with the distribution of subspecies:


This complex, then, was long regarded as the paradigm of ring species, and was (and is still in places) taught as an example of this form of speciation with gene flow.

Except it’s wrong. That is, it’s not a ring species in the classical sense. Why not? Because genetic studies, done by both Dick Highton at Maryland and then by Wake and his colleagues themselves (references below) also showed that in places around the ring there were sharp genetic breaks, suggesting not a process of continuous gene flow over the 5-10 million years it took to close the ring, but sporadic geographic breaks in the ring, so that the salamanders could differentiate without pesky gene flow from adjacent populations. Some adjacent populations showed very sharp genetic differentiation, implying geographic isolation in the past (Continuous gene flow would not produce such “breaks”.) Finally, geologic work has shown that it is very unlikely that there were two unbroken forest corridors for those millions of years required to produce a ring.

Based on these results, everyone has now concluded that the formation of this “ring” involved sporadic and important episodes of geographic isolation between populations, so it’s not the classic “continuous gene flow” scenario involved in making a ring species. As Wake himself said in his 1997 paper (reference below), “The history of this complex has probably featured substantial [geographic] isolation, differentiation, and multiple recontacts.” (You can read about the Ensatina story in greater detail at “Understanding evolution,” a great site produced by U.C. Berkeley.)

Well, that’s a bummer, but it still shows how geographic isolation by distance can promote reproductive isolation and speciation. Other putative cases of ring species, including gulls in the genus Larus encircling the Arctic, also fell victim to genetic studies, showing that it was very unlikely that they were ever a continuous ring that was geographically uninterrupted.

That left the greenish warbler, which we touted in our book as perhaps the one good case of a true ring species. This species was similar, in that ancestral populations south of the Tibetan plateau were said to have expanded around that uninhabitable plateau (these birds need forests!), meeting north of the plateau in Siberia. Here’s a diagram of the ring with its six present “subspecies” (in different colors), along with the sonograms of each population’s song. The place where the ring was supposed to have joined, and where the populations act as different species, is the crosshatched red/blue area at the top:

Here’s a greenish warbler from Sikkim:

Phylloscopus trochiloides

The original scenario, published by Irwin et al. (2001; reference below) and discussed in our book, suggested that as the species encircled the Tibetan Plateau from the south, the populations diversified in both color and song, with the songs becoming more complex—but in different ways—as the two branches headed north. The evolutionary impetus for song differentiation is not known, but probably involved sexual selection, possibly based on food availability.

By the time the populations met in the north, the male songs were so different that the two populations didn’t recognize each other as members of the same species (i.e., females of one population didn’t respond to the song of males of the other), and so there was no interbreeding, even though there was supposedly interbreeding all along the continuous ring. (The geographic “break” in China is probably due to recent deforestation.)

It was a classic example of ring species, though there was one worry. Mitochondrial DNA (mtDNA) studies in the southwestern part of the ring showed a sharp genetic break between two populations in Kashmir, suggesting that perhaps there was some ancient geographic isolation before the ring began moving. But that differentiation in what is, in effect one gene (the mitochondrial genes are all linked), could have other explanations.

Now, however, genetic analysis of 95 birds and more than 2,000 sites throughout the genome (not just in the mitochondrion), has revealed four genetic clusters around the ring with a sharp intergradation between them (there’s another cluster in the Caucasus, not around the ring). For those of you who know about Bayesian cluster analysis, here’s what the clusters look like around the ring. (Smaller clusters are in the supplementary information.)


Note especially the big break in the southwest, between the yellow and blue populations, and the plot showing the abrupt genetic break between them (lower left), indicating a hybrid zone between populations that were previously isolated geographically. This is in fact in the same place as the worrisome break in mtDNA reported earlier. There are also smaller breaks between Pakistan and Kyrgyzstan, and between Nepal and China (not shown above), but it’s not as clear that those reflect previous geographic isolation rather than simply sampling error or selection with some gene flow.

The authors also found, contrary to previous information, that there are indeed some hybrids (fertile ones, since they backcross) produced where the ring closes at the top, so they aren’t really acting like full biological species.

At any rate, the authors conclude that this case doesn’t correspond to the classical notion of a “ring species,” for that requires no geographic isolation while in this case there almost certainly was some. As the authors note:

Our results indicate that allopatric divergence played a strong role in shaping patterns of genomic divergence during the formation of the greenish warbler ring. Previous findings nonetheless support the idea that geographical isolation has not been the sole driver for the establishment of reproductive isolation; natural and sexual selection also appear to play major roles.

Well, yes, but geographic isolation is never the sole drive of any evolutionary divergence, for something has to make the populations diverge: either selection (natural, sexual, or both) or genetic drift.  But it’s good of them to pursue their initial finding more intensively, even if that effaced the cool result of a “ring species” that they reported earlier.  But nature is nature, and what happened is what happened. In the end, the authors conclude what I concluded before I saw their first paper: there are no good examples of ring species in nature:

Finally, this study provides meaningful information concerning the interpretation of ring species as prominent illustrations of speciation in the face of gene flow. Earlier analyses of AFLP markers supported the greenish warbler as the last known example in birds of an ideal ring species in which the terminal forms became reproductively isolated despite being connected by gene flow during their whole history of divergence. . It is possible that speciation-by-distance ring species could exist, but their extreme rarity may be explained by the rapidity of Earth’s climatic shifts compared to the time that reproductive isolation takes to occur; over these great spans of time, and as exemplified by the present study, populations are very likely to be temporarily divided.

In other words, perhaps it’s too much to hope for a “true” ring species, for that requires a species’ range to remain unbroken for millions of years—and yet climate and habitat change all the time.

Nevertheless, the results do show a “ring species” of a sort: isolation of two “end” populations of a ring that makes them look like two species, even though all through the ring you don’t see reproductive isolation of adjacent areas. And it shows that speciation can occur despite there having been some gene flow at some times. In nature, populations that form new species must often sometimes exchange genes if they’re not completely isolated by geography (i.e. the finch species that colonized the Galápago), so the dichotomy between “no gene flow” and “pervasive gene flow” may be artificial.

Oh, and there are no good ring species, so don’t go around saying that there are! Mayr concluded the same thing in his great 1963 book Animal Species and Evolution (this book was largely responsible for making me an evolutionary biologist), but he didn’t have genetic data, and he didn’t consider the greenish-warbler case. It’s no great loss, though, that we lack good examples, for ring species didn’t really demonstrate any new evolutionary principles. They showed something we already knew—that reproductive isolation is promoted by anything that reduces gene flow between populations. But they showed it in a cool and novel way.


Alcaide, M., E. S. C. Scordato, et al. (2014). “Genomic divergence in a ring species complex.” Nature 511: 83-85.

Highton, R. (1998). “Is Ensatina eschscholtzii  a ring species?” Herpetologica 54: 254-278.

Irwin, D. E., S. Bensch, et al. (2001). Speciation in a ring. Nature 409: 333-337.

Wake, D. B. (1997). “Incipient species formation in salamanders of the Ensatina complex.” Proc. Natl. Acad. Sci. USA 94: 7761-7767.

Wake, D. B. and C. J. Schneider (1998). “Taxonomy of the plethodontid salamander genus Ensatina.” Herpetologica 54: 279-298.