RATES OF MOLECULAR EVOLUTION

Many patterns in the evolution of proteins and DNA are now well established and have been adequately reviewed in recent years (23, 37, 38, 52, 55, 61, 73). However, there are several gaps in our understanding of the process of molecular evolution, the most conspicuous of which is our ignorance of the forces responsible for most of the evolution of DNA sequences. Our ignorance is due, in part, to the lack of an adequate statistical description of molecular evolution. Whereas a few years ago modelling molecular evolution with a constant rate Poisson process seemed adequate (52, 76), today it is clear that the dynamics of molecular evolution are much more complex than those suggested by such a naive model. The statistical description of molecular evolution is hampered by the fact that we cannot observe the process directly. Instead, our only record is the accumulated changes in the sequences of species whose common ancestors lived many millions of years ago. With suitable data we can infer the number of changes that have occurred on a lineage, but we cannot infer the times when the changes occurred. Unfortunately, the times of the changes are needed to understand the forces responsible for molecular evolution. For this reason, much of the theoretical work in molecular evolution is directed toward the reconstruction of these past events. A critique of this work is presented in this review. The most compelling generalization to come out of molecular evolution studies is the concept of the molecular clock (76, 73). This is an abstraction from the common observation that the number of substitutions on a lineage is roughly proportional to the length of the lineage. (The constant of propor-