Genetic structure of prey populations underlies the geographic mosaic of arms race coevolution
Reciprocal adaptation is the hallmark of arms race coevolution, but the symmetry of evolutionary change between interacting species is often untested, even in the best-studied battles of natural enemies. We tested whether prey and predator exhibit symmetrical local co-adaptation in the example of a geographic mosaic of coevolution between toxic newts (Taricha granulosa) and resistant garter snakes (Thamnophis sirtalis). Prior work showing a tight correlation between levels of newt toxin and snake resistance is regarded as textbook evidence of the intense arms race between natural enemies. Here, we similarly found that toxin and resistance are functionally matched in prey and predator populations, further suggesting that mosaic variation in the armaments of both species results from the local pressures of reciprocal selection. Contrary to conventional wisdom, however, we found that local variation in newt toxin is best predicted by neutral population divergence rather than the resistance of co-occurring predators. Snake resistance, on the other hand, is clearly explained by local levels of prey toxin. Prey populations seem to structure variation in defensive toxin levels across the geographic mosaic, which in turn determines selection on predator resistance. Exaggerated armaments suggest that coevolution occurs in certain hotspots, but our results imply that neutral processes like gene flow--rather than reciprocal adaptation--structure the greatest source of variation across the landscape. This pattern supports the predicted role of \"trait remixing\" in the geographic mosaic of coevolution, the process by which non-adaptive forces dictate spatial variation in the interactions among species.\n\nSIGNIFICANCE STATEMENTWhen the weapons of natural enemies like prey toxins and predator resistance are matched across the geographic landscape, they are usually presumed to result from arms race coevolution. In the textbook example of an arms race, matched levels of newt toxin and garter snake resistance have long been regarded as evidence of such local co-adaptation. To the contrary, we found that local variation in newt toxicity is best explained by the neutral geographic structure of newt populations. This spatial variation of prey in turn dictates local selection on garter snakes, structuring the geographic pattern of predator resistance. These results demonstrate how landscape patterns of phenotypic variation are determined by a mixture of natural selection, historical biogeography, and gene flow that comprise the geographic mosaic of coevolution.