Experimental hybridization as a tool for studying selection in the wild

Abstract

Selection experiments in the wild have greatly aided our understanding of how selection shapes phenotypic diversity in natural populations. However, these experiments have been hindered by the fact that the traits most important to adaptation and speciation may be fixed within populations. Experimental hybridization offers a means to circumvent this problem since crosses can be made between populations or species that differ for the trait of interest. Here, we discuss the advantages and limitations of this approach, review results from published studies, and suggest strategies for future experiments. Advantages associated with this approach include the ability to "generate" variation for traits of interest, as well as the increased sensitivity of selection assays because of the wider range of trait values characteristic of segregating hybrids. Moreover, experimental hybridization can be extended to crosses between near-isogenic lines, allowing the effects of major quantitative trait loci (QTLs) to be dissected in a well-defined genetic background. Limitations include widespread linkage disequilibrium created by hybridization, possible cosegregation of hybrid incompatibilities, and the large number of variable traits likely to affect hybrid fitnesses. Even with these limitations, this basic approach has been remarkably successful. We now know, for example, that many morphological differences between divergent populations or species are under selection and that the selection gradients may be surprisingly large. Also, small genetic changes in premating barriers may have large effects on assortative mating. With respect to discussions about the role of hybridization in evolution, experiments indicate that some hybrid trait combinations may be advantageous, particularly when the hybrid populations are placed in a new environment. Future studies seem likely to combine selection measurements with genetic mapping of the QTLs underlying the selected traits. Natural hybrid zones hold particular promise for these kinds of studies because the only experimental manipulation required is the removal of small amounts of tissue for genotyping.