Criteria for the determination of the direction of character state changes

Phylogenetic reconstructions cannot be adequately assessed except in terms of probability models which represent the processes of change. An tionary tree model of bifurcating splits, together with a Brownian motion model of genetic drift have been shown to allow successful reconstruction of phylogeny from data relating to gene frequencies of blood groups from human populations. Changes in state cannot be dealt with by the Brownian motion model, and no adequate models have been proposed for character state changes from the many possible sources. So, while the tionary tree model is applicable to these problems too, only heuristic methods of analysis of the state data are available, and these are known to be unsatisfactory under certain conditions. The concepts of homology, polarity and homoplasy have been developed out of the attempt to describe the nature of morphological state data, which is unpredictably related to genetic state data. The experimental study of comparative functional morphology, at any developmental stage of an organism, is considered to be the only valid tool for investigating the resulting character state tree hypothesis. However, the speculative nature of such investigations is admitted.