Genetic analysis of hybrid zones

When two distinct gene pools meet and produce fertile hybrids, the outcome varies from gene to gene. At some loci a universally favorable allele has been established on one side. Such alleles soon spread through the whole population and hence differences are rarely observed. At other loci different alleles may be favored in different environments or genetic backgrounds; selection maintains these differences in the face of random mixing. At other loci—perhaps at most of those we observe in molecular surveys—different alleles may have been established by chance and may have no appreciable effect on fitness. These differences gradually fade away, at a rate that depends on the strength of selection against introgression at the other loci with which they are associated. The frequencies of the various genotypes found in a hybrid zone tell us about the overall strength of the selection, the number of genes involved, the rate of individual dispersal, and the ease with which alleles cross from one gene pool into the other. The aim of this chapter is to explain how data on discrete markers and on quantitative traits can be used to estimate such parameters. We illustrate the methods using examples from some of the hybrid zones that are discussed in more detail elsewhere in this book and use computer simulations to show that the estimates do not depend on exactly how selection maintains the differences between the hybridizing populations. Previous reviews have considered the wider questions of what hybrid zones can tell us about species and speciation and what role they themselves might play (Barton and Hewitt, 1985, 1 989; Hewitt, 1988; Harrison and Rand, 1989; see also Ch. I ). We concentrate instead on the practical issues involved in the genetic analysis of hybrid zones. A systematist, whose aim is to classify organisms, sees hybrid zones as boundaries between distinct types. A population geneticist, on the other hand, views them as sets of geographic gradients (i.e., of clines) in allele frequencies or quantitative traits. Both extreme views are misleading. Classification of individuals into parental, F,, F2, and backcross types wastes much information and, moreover, depends on which markers are used: an individual who is heterozygous for diagnostic alleles at five loci might be classified as an F, and yet be homozygous at the sixth locus. If even a small proportion of hybrids reproduce, all the individuals in the vicinity of the hybrid zone eventually carry introgressed alleles in some of their genes. However, describing a population