A cladistic analysis of phenotype associations with haplotypes inferred from restriction endonuclease mapping. II. The analysis of natural populations.

Genes that code for products involved in the physiology of a phenotype are logical candidates for explaining interindividual variation in that phenotype. We present a methodology for discovering associations between genetic variation at such candidate loci (assayed through restriction endonuclease mapping) with phenotypic variation at the population level. We confine our analyses to DNA regions in which recombination is very rare. In this case, the genetic variation at the candidate locus can be organized into a cladogram that represents the evolutionary relationships between the observed haplotypes. Any mutation causing a significant phenotypic effect should be imbedded within the same historical structure defined by the cladogram. We showed, in the first paper of this series, how to use the cladogram to define a nested analysis of variance (NANOVA) that was very efficient at detecting and localizing phenotypically important mutations. However, the NANOVA of haplotype effects could only be applied to populations of homozygous genotypes. In this paper, we apply the quantitative genetic concept of average excess to evaluate the phenotypic effect of a haplotype or group of haplotypes stratified and contrasted according to the nested design defined by the cladogram. We also show how a permutational procedure can be used to make statistical inferences about the nested average excess values in populations containing heterozygous as well as homozygous genotypes. We provide two worked examples that investigate associations between genetic variation at or near the Alcohol dehydrogenase (Adh) locus and Adh activity in Drosophila melanogaster, and associations between genetic variation at or near some apolipoprotein loci and various lipid phenotypes in a human population.