coevolutionary genetics

Nuismer SL, Gomulkiewicz R & Ridenhour BJ 2010 When is correlation coevolution? Am Nat 175:525-537.

  • coevolution is neither a necessary nor a sufficient condition for the evolution of spatially correlated traits between two species
  • coevolutionary selection fails to consistently generate statistically significant correlation
  • non-coevolutionary processes can readily cause statistically significant correlations to evolve
  • at least three non-coevolutionary mechanisms could explain correlations between the traits of interacting species across sites
  • positive correlations will arise if, for instance, long-tongued pollinator individuals congregate in regions where plants tend to have, on average, long corollas but short-tongued pollinator individuals tend to congregate in regions where plants have, on average, short corollas
  • traits may become correlated any time one species evolves to match the phenotype of an interacting species that fails to evolve in response either because it experiences only weak selection from the interaction or because it lacks heritable variation (evolutionary commensalism)
  • correlated traits could evolve if the abiotic environment favors similar traits in both of the interacting species
  • a number of models now support the basic predictions of the geographic mosaic theory
  • none of these models predicts the distribution of correlations expected to evolve as a result of coevolutionary and non-coevolutionary processes
  • correlations between traits of interacting species observed across populations are not sufficient evidence for inferring a coevolutionary process
  • a failure to demonstrate correlated traits is not evidence for an absence of coevolution
  • despite these arguments, studies of trait correlations across populations continue to be used as partial evidence either for or against a coevolutionary hypothesis
  • we assume that random genetic drift occurs at the end of each generation
  • by randomly sampling ni individuals independently at each location of species i to form the next generation
  • coevolutionary selection can produce significant correlations between the traits of interacting species in some cases, they also reveal that s
  • coevolutionary selection per se is not required for substantial correlations to evolve
  • traits become correlated even if the fitness of only one species is affected by interactions (evolutionary commensalism)
  • previous theoretical studies have found that gene flow and local population size can have important effects on coevolution under some circumstances (Gandon et al. 1996; Gandon and Michalakis 2002; Gandon and Nuismer 2009)
  • but not others (Ridenhour and Nuismer 2007; Gavrilets and Michalakis 2008)
  • our results provide a sobering view of what can be inferred from even precise estimates of trait correlations
  • correlations between species readily evolve in the absence of reciprocal biotic selection
  • this can occur if interactions cause selection on only one of the specie
  • or if the trait optima favored by stabilizing selection on the two species are strongly correlated
  • reciprocal selection often fails to generate statistically significant correlations between trait means even when it occurs
  • our results also bear on the increasingly popular use of trait correlations as a means of testing the geographic mosaic theory of coevolution
  • the GMTC predicts that coevolutionary selection will not always lead to strong correlations between traits of interacting species
  • the coevolutionary process—due to the action of drift and gene flow (among other forces)—actually leads to considerable "trait mismatching" that we formally interpret as imperfect correlations between traits of interacting species
  • this rather loose prediction has spurred a proliferation of studies that estimate the population mean trait values of interacting species across broad spatial scales and interpret the results (typically an imperfect correlation) as support for the GMTC
  • although these studies have provided valuable data and yielded interesting insights, reported correlations neither support nor refute the GMTC because there is no a priori expectation for the value of the correlation expected in the absence of a geographic mosaic process
  • the distribution of correlation coefficients that arises through a geographic mosaic process (some combination of spatially variable coevolutionary selection, gene flow, and finite populations) cannot be distinguished from a distribution resulting from a nongeographic mosaic process (i.e., no spatially variable coevolutionary selection)