When two or more characters are measured in members of a species, it is often found that they are correlated. Taking data from a series of populations, it may be (for example) that two measurements both tend to be larger in the same populations. Treating each population as an independent piece of evidence is, however, not correct. Gene flow due to migration may have made neighboring populations have the same phenotypes, so that they cannot be treated as independent sources of evidence. It is proposed to develop methods of correcting for this lack of independence. Using an estimate of the migration rates among populations, it is possible to compute quantities (contrasts) that would vary independently. These contrasts can then be treated as independent data points and used to make valid statistical statements about whether the characters are actually correlated. These methods will be made available in the PHYLIP package of free phylogeny computer programs. Other projects involve seeing whether correlations of characters can be inferred while simultaneously estimating the evolutionary tree, seeing whether models with developmental thresholds can allow us to use all-or-none traits, and seeing whether models of adaptive peaks can be used.
