Estimating Genetic Similarities Within and Between Populations

Similarities, as measured by the average proportion of DNA restriction fragments shared by pairs of individuals from the same or different populations, are now commonly used to assess the levels of genetic relatedness and levels of genetic variation within and between populations. Similarity means generally involve the use of the profile of each individual more than once with consequent difficulties in estimating the required standard errors. Analysis of variance procedures are shown to be appropriate and to avoid the need to discard some data or to use complicated resampling techniques. The best allocation of individual samples from different populations to the lanes of a gel when only neighboring lanes can be compared is also discussed. The approach is illustrated by an analysis of some unbalanced similarity data for a single population.

Key Words
DNA fingerprinting; Population structure; Similarity matrices.

R. N. Curnow is Emeritus Professor, Department of Applied Statistics, The University of Reading, P.O. Box 240, Earley Gate, Reading RG6 6FN, United Kingdom (E-mail: r.n.curnow@reading.ac.uk).