Reply from C.D. Jiggins and J. Mallet

Abstract

Schilthuizen's comments1xBimodal hybrid zones and the scale of a snail. Schilthuizen, M. Trends Ecol. Evol. 2000; 15: 469Abstract | Full Text | Full Text PDFSee all References1 highlight two important aspects of bimodality in hybrid zones. First, perhaps we did not emphasize enough that, when identifying bimodality and measuring linkage disequilibria within a zone, it is critically important that samples represent locally panmictic populations2xBimodal hybrid zones and speciation. Jiggins, C.D. and Mallet, J. Trends Ecol. Evol. 2000; 15: 250–255Abstract | Full Text | Full Text PDF | PubMed | Scopus (326)See all References2. It is a well known result of standard population genetics that mixing geographic samples with divergent gene frequencies will create artificial heterozygote deficits and linkage disequilibria; it is the local bimodality and concomitant deviations from Hardy–Weinberg or linkage equilibrium that are of special interest in the study of speciation. Of course, in practice, ‘local populations’ might be difficult to define. If present, bimodality should be evident at the smallest spatial scale at which it is feasible to sample, provided this is small relative to the dispersal distance of the organism concerned. We would probably consider the Albinaria zone to be unimodal throughout because ‘unimodality would appear at extremely small sampling areas’.Bimodality can also be difficult to detect because many loci might be undifferentiated. If there are few loci examined, each with small frequency differences between taxa, many heterozygotes and pairwise recombinant genotypes might be present locally, thus leading to unimodal distribution on a hybrid-index plot. Adding more loci, especially if more strongly differentiated, will give greater statistical power to detect bimodality. On a plot of hybrid index, the two peaks might slowly pull apart as more loci are added. Part of the problem is the loss of resolution involved in representing a multidimensional (multilocus) property – bimodality – on a two-dimensional hybrid-index plot. Likelihood analysis of multilocus genotypes3xSee all References3 is a more appropriate statistical method and should detect bimodality with greater sensitivity; hybrid-index plots2xBimodal hybrid zones and speciation. Jiggins, C.D. and Mallet, J. Trends Ecol. Evol. 2000; 15: 250–255Abstract | Full Text | Full Text PDF | PubMed | Scopus (326)See all References2 are merely a useful means of data display.Second, there is often considerable variation between populations within a single zone. Given the ubiquity of ecological differentiation across hybrid zones, it is not surprising that environmental variation, such as that described by Schilthuizen, should affect the genetic structure of hybrid zone populations. In the case of Albinaria, it appears that the width of the hybrid zone is primarily affected, thus leading to difficulty in choosing a suitable sampling scale. However, the degree of bimodality can also vary. As with sympatric sibling species4xSee all References4, the local strength of disruptive natural selectionmight vary in space and time. Alternatively, habitat and population structure, leading to differences in dispersal pattern, can have similar effects: in Bombina hybrid zones, most populations are unimodal but some are nearly bimodal where the habitat is more patchy2xBimodal hybrid zones and speciation. Jiggins, C.D. and Mallet, J. Trends Ecol. Evol. 2000; 15: 250–255Abstract | Full Text | Full Text PDF | PubMed | Scopus (326)See all References2. Although in individual cases it is often difficult to determine whether divergence between hybridizing taxa occurred in situ or in allopatry, such variation in bimodality is exactly the pattern expected if hybrid zone populations represent transitional stages in a process of gradual parapatric speciation.