of evolution depends on the number of germ-line replications per year, which is several
times greater for the short-generation rodents and grasses than for the long-generation
primates and palms. The rationale of the assumption is that the larger the number of
replication cycles, the greater the number of mutational errors that will occur.
From a theoretical, as well as operational, perspective, these and other supplementary
hypotheses have the discomforting consequence that they involve additional empirical
parameters, often not easy to estimate. It is of great epistemological significance that the
original proposal of the neutral theory is (i) highly predictive and therefore, (ii) eminently
testable. The supplementary hypotheses lead, nevertheless, to certain predictions that can
be tested. The ‘generation-time’, ‘population size’, and ‘biological properties’ hypotheses
uniformly predict that rate variations observed between lineages or at different times will
equally affect (in direction and magnitude) all genes of any particular organism, since
these attributes are common to all genes of the same species. The ‘slightly deleterious’
hypothesis predicts that the rate of evolution will be inversely related to population size,
and thus reduces to the ‘population size’ hypothesis (Ohta 1973).
In this chapter, we present an analysis of nine genes undertaken as a test of the four
supplementary hypotheses, as well as of the neutrality theory, the more general or ‘null’
hypothesis underlying the molecular clock hypothesis. We have, in the past, reported
results for three of these genes, showing that they exhibit overdispersed patterns of
molecular evolution that are incompatible with the proposed supplementary hypotheses
(Rodríguez-Trelles et al. 2001a, and references therein). The additional tests reported
here lead to the same conclusion. We surmise that inferences about the timing of past
events (and about phylogenetic relationships among species) based on molecular evolution
are subject to sources of error not altogether disparate from inferences based on anatomy,
embryology, or other phenotypic characteristics. Nevertheless, molecular investigations
have two obvious advantages over phenotypic traits, in degree if not completely in kind;
namely, that the number of ‘traits’ is very large, that is, every one of the thousands of
genes in the make-up of each organism, and that differences can be more precisely
quantified, measured as they are in terms of distinct units, such as amino acids or
nucleotides. There are many evolutionary issues concerning both timing and phylogenetic
relationships between species for which molecular sequence data provide the best, if not
the only, dependable evidence. The large-scale reconstruction of the ‘universal tree’ of
life is a case in point: the phylogenetic relationships among Archaea and bacterial
prokaryotes and between them and the eukaryotes have best been determined with DNA
sequences encoding ribosomal RNA genes. The multiplicity of genes opens up the
possibility of combining data for numerous genes in assessing the timing of particular
evolutionary events, or the phylogeny of species. Because of the time-dependence of the
evolutionary process, the multiplicity of independent results would probably tend to
converge (by the so-called ‘law of large numbers’) on average values reflecting, with
reasonable accuracy, the time elapsed since the divergence of species.
The nine genes and their protein productsWe have investigated the following nine nuclear genes: (1) alcohol dehydrogenase (Adh; E.C.
1.1.1.1), (2) aromatic-L-amino acid decarboxylase (Ddc; E.C.4.1.1.28), and its paralogue (3)
MOLECULAR CLOCKS: WHENCE AND WHITHER? 7