SPECIES NUMBER, GENERATION LENGTH, AND THE MOLECULAR CLOCK.

Abstract

The fundamental tenet of the molecular clock hypothesis is that evolutionary rates of homologous proteins are regular, so that the interval separating living species from common ancestors is reflected in the degree of protein dissimilarity between them. In general, the difference between species in numbers of individuals is accorded a relatively insubstantial role as a determinant of protein divergence measures, a neglect implicit in the concept itself. The postulation of clock-like divergence presupposes protein divergence rates to be uniform despite this (or other) outstanding differences. Indeed, the appeal of one explanation of such stochastic regularity, the neutral mutation-random drift theory (Kimura, 1968, 1969; King and Jukes, 1969), lies in its assignment of the rate of amino acid replacement to the nucleotide mutation rate alone, assuming the latter to be constant. In order that an equilibrium between mutation and eventual replacement by drift be completely realized, however, the theory requires that species number remain unchanged over indefinitely extended intervals. The difficulty in evaluating empirically the theory's central thesis, that the majority of such replacements are indifferent (or nearly so) to natural selection, is therefore greatly compounded by ordinary recurrent circumstances in the natural history of species-stochastic variation in species number (see, e.g., Haigh and Maynard Smith, 1972; Nei et al., 1975; Chakraborty, 1977). Perhaps in partial consequence, the validity of the neutrality hypothesis (or more properly, the extent of its applicability) is problematic after a decade of debate. But does the molecular clock principle depend upon validation of the neutral mutation-random drift theory? Several among the proponents of the clock thesis have either rejected the neutrality hypothesis (see, e.g., Langley and Fitch, 1974; Fitch, 1975; Fitch and Langley, 1976, 1977) or disengaged their own claims for regularity of rates from its confirmation (see, e.g., Sarich and Wilson, 1967; Sarich, 1974, 1977a, 1977b; Sarich and Cronin, 1977); where the former find evidence for departures in nucleotide substitution rates from the theory's predictions, the latter argue that the self-consistency of their own measurements renders explanation superfluous. It is the purpose of this paper, however, to demonstrate that species number and generation length may be important determinants of the degree to which proteins diverge over time, and that variability in these parameters might easily lead both to the impression that molecular evolution rates are nonconstant and to the systematic underestimation of species divergence times. My argument follows mainly from the assumption of selective neutrality. Insofar as this assumption is correct, these effects constitute a component of inter-specific protein variation; so long as this assumption remains unfalsified, these effects contribute uncertainty to the clock's dates.