In my classes that I teach at Pace University, the theme of parallel and convergent evolution is a recurrent topic. For example, we’ve discussed parallel evolution of lactase persistence in human populations from northern Europe and parts of Africa, and convergent evolution of the camera eye in vertebrates and cephalopods. Stickleback fish are a classic textbook case of parallel evolution, adapting repeatedly to freshwaters from marine ancestors, diverging into benthic and limnetic forms along similar lines in multiple lakes, and adapting along similar trajectories in lake and stream habitats throughout much of the northern hemisphere.
I first began thinking about parallel evolution as a master’s student at McGill when I was practicing analyzing microsatellite data. A question came to my naïve graduate student mind: Couldn’t we compare microsatellite and morphological data among populations to determine the proportion of within-population variation that is due to constraints versus adaptive evolution? I expected that a high level of “exchangeability” of neutral genetic markers would reveal a higher level of morphological “exchangeability” between lake and stream populations from the same watershed (due to gene flow or shared ancestry), whereas genetic (microsatellite) distinctiveness would result in morphological traits in lake habitats that are more similar to fish in other lake habitats than in the stream habitats (and vice versa) due to adaptation. I proposed that we could use classification techniques to directly compare genetic markers and morphological traits.
Years later, my ideas have culminated in a study on stickleback from parapatric lake and stream habitats from each of six watersheds on Vancouver Island. We used discriminant analysis to classify individuals to populations for each of several measures, including diet ecology (stomach contents and stable isotopes), trophic morphology (body shape, and gill raker number and length), armor traits (plates and spines), and microsatellites (6 neutral loci, and 6 loci linked to QTL). This approach differed from traditional analyses that compare means among populations in that “misclassified” individuals could inform us as to which populations were more “exchangeable”; that is, it could tell us whether a lake fish would be a better fit into another lake population than into any stream population, based on any of the traits or loci (and vice versa for the stream fish).
We found that populations within watersheds were most exchangeable with respect to genetic markers, which would make sense if gene flow occurred between lake and stream habitats. This was less likely to be the case with diet and body shape, however, for which fish were more likely to be classified into a similar habitat in a different watershed than into a different habitat. Why are these results important? In addition to shedding light on the deterministic nature of habitats in shaping parallel patterns of evolution, these results might provide insight to conservation managers planning to relocate individuals to new environments, for example when the native range is under threat or to enhance genetic variation in a bottlenecked population.
Hendry AP, Kaeuffer R, Crispo E, Peichel CL, Bolnick DI. In press. Evolutionary inferences from the analysis of exchangeability. Evolution. DOI: 10.1111/evo.12160