Population genetics takes the “co” out of snake-newt coevolution (maybe)

Taricha granulosa, photographed in Northern California. (Wikimedia: Don Loarie)

A textbook example of predator-prey coevolution could need revision, if the conclusions of a recently posted pre-print hold up more broadly. The manuscript, lead-authored by Michael Hague with Amber Stokes, Chris Feldman, and Ed and “Butch” Brodie, calls into question whether poisonous rough-skinned newts (Taricha granulosa) and the garter snakes (Thamnophis sirtalis) that prey on them truly exert reciprocal selection on each other. The data in the manuscript are consistent with newts creating selection for greater toxin resistance in the snakes — but not with the snakes selecting for more toxic newts.

Rough-skinned newt populations in Western North America are distinguished by one of the most over-the-top defenses against predation seen in a vertebrate: they secrete tetrodotoxin, the same neurotoxin produced by pufferfish and blue-ringed octopuses. Tetrodotoxin disables the molecular channels that allow nerve cells to generate electrical signals, which paralyzes just about any predator with nerves. Some populations of garter snakes (and other snakes that feed on tetrodotoxin-defended amphibians) have mutations to the channels that let them resist the paralyzing effect — and this should set the stage for a coevolutionary arms race between newts’ production of the poison and snakes’ ability to cope with it.

Across much of the Pacific coast, where the snakes and the newts encounter each other, garter snakes’ resistance to tetrodotoxin tracks or exceeds the quantity of poison found in the skin of local newts. We’ve generally understood this to suggest an arms race, with complications — some cost to the production of tetrodotoxin, or some additional source of selection on the trait, that counterbalances selection from snake predation to create situations in which the newts produce less toxin than the snakes can tolerate. Newts have other predators besides garter snakes, and although snakes’ tetrodotoxin resistance is understood down to the molecular level, the physiology behind the poison’s production is still something of a mystery.

Hague et al. tackle the question of snake-newt coevolution with a population genetic dataset from a transect of sites from the Olympic Peninsula down the Oregon coast, along which toxin production and tolerance are well matched. They sequenced the garter snake gene for the molecular channel targeted by tetrodotoxin to track the frequency of resistant variants, and used ddRADseq to generate genome-wide SNP datasets to describe the population structure of both snakes and newts.

The dataset recapitulates the phenotype matching result seen in this region — snake populations that coexist with more toxic newts have higher frequencies of resistance mutations. Snakes’ quantitative resistance to tetrodotoxin generally tracks the frequency of resistance mutations, which makes sense. However, the results for predator and prey diverge when the authors add population structure to the comparison: newts’ production of tetrodotoxin closely tracks their overall population structure; while snakes’ resistance does not align with the snakes’ population structure. The authors performed a formal cline analysis with their phenotypic and genetic data, which identified clinal centers for newt toxin production, newt population structure, snake toxin resistance, and snake resistance allele frequencies within a few dozen kilometers of each other; but the clinal center for snake population structure lies hundreds of kilometers to the south.

Newts’ tetrodotoxin production matches newts’ overall population structure, while snakes’ tetrodotoxin resistance does not match theirs. Instead, snakes’ quantitative resistance, and the frequency of resistance alleles, form a cline with a center aligned to the clinal center for newts’ tetrodotoxin levels. (Hague et al. Figure 3)

Hague et al. interpret this to mean that snakes’ tetrodotoxin resistance is under selection created by the newts, but the newts’ tetrodotoxin production is shaped not by selection from snakes, but by the demographic forces described by the population genetic data. For the region they’re sampling, then, that would mean the two species are It’s a sign of selection if a gene shows a spatial pattern of variation at odds with overall population structure, which reinforces the conclusion drawn from the original matching with newt phenotypes. The opposite pattern, then — the one seen in the newts — is consistent with a lack of spatially varying selection, or selection strong enough to overcome the homogenizing effect of migration among populations. That would mean the snakes are under selection by their deadly prey, but don’t manage to exert reciprocal selection on the prey’s toxicity, so the two species aren’t coevolving.

As the authors point out, this doesn’t mean snakes and newts aren’t coevolving everywhere; they’re looking at a relatively small portion of the range where these species interact. Even for that range, this conclusion is complicated by the fact that, while it’s possible to track the frequency of specific mutations that contribute to snakes’ resistance, newt toxicity is only tracked at the phenotypic level. It’s not clear either what proportion of variation in tetrodotoxin production is genetic — increased toxicity might be a plastic response to predation stress, apparently! — or what other selective forces could be acting on it. Still, it’s hard to construct an explanation involving plasticity or selection that results in tetrodotoxin levels neatly tracking the newts’ population structure. (Maybe assortative mating by tetrodotoxin production?)

So there is, as we so often say, a lot to learn from future data collection, and possibly more analysis of the current data — this is, after all, a preprint rather than a peer-reviewed paper. In the meantime, we might have to rethink how we frame the snake-newt interaction in undergraduate biology courses. Even if it’s not a textbook example of coevolution, it’s still a fascinating case study in the evolutionary outcomes of predator-prey interactions.


Hague MTJ, AN Stokes, CR Feldman, ED Brodie Jr., ED Brodie III. 2019. Genetic structure of prey populations underlies the geographic mosaic of arms race coevolution. bioRxiv 585851; doi: 10.1101/585851

Hanifin CT, ED Brodie Jr, ED Brodie III. Phenotypic mismatches reveal escape from arms-race coevolution. PLOS Biology. 2008 6(3):e60. doi: 10.1371/journal.pbio.0060060

McGlothlin JW, ME Kobiela, CR Feldman, TA Castoe, SL Geffeney, CT Hanifin, G Toledo , FJ Vonk, MK Richardson, ED Brodie Jr, ME Pfrender. 2016. Historical contingency in a multigene family facilitates adaptive evolution of toxin resistance. Current Biology. 26(12):1616-21. doi: 10.1016/j.cub.2016.04.056


About Jeremy Yoder

Jeremy Yoder is an Assistant Professor of Biology at California State University, Northridge. He also blogs at Denim and Tweed, and tweets under the handle @jbyoder.
This entry was posted in Coevolution, natural history, population genetics, selection and tagged , . Bookmark the permalink.