upward bias, and that distances based on concatenated sequences may help reduce this
bias.
Pairwise distance methods have commonly been used to estimate divergence times (e.g.
Kumar 1996; Ayala et al. 1998; Gu 1998). Later studies have used a maximum likelihood
(ML) framework for branch length estimation (Bromham et al. 1998; Cutler 2000;
Hausdorf 2000; Yoder and Yang 2000).
Calibration pointsThe calibration of genetic distance to produce an estimate of molecular change per unit
time is arguably the most serious difficulty with divergence time calculations. This is likely
to be particularly acute when deep divergences, such as the basal Cambrian are
considered. The incomplete nature of the fossil record means that it is unlikely that the
very earliest specimens following a divergence will be preserved. Fossils used to date
divergences are seldom claimed as direct ancestors of the taxa in question, but must always
be regarded as members of a plesion (Figure 3.4). Such fossil dates will tend to
underestimate the true date of the divergence, thus causing evolutionary rates to be
overestimated. This in turn causes molecular date estimates to be too young (Rambaut
and Bromham 1998).
Additionally, since allelic divergence generally predates species divergence, appearance
of any taxon-defining morphological-taxonomic features (including those seen in the
‘earliest’ fossils) will post-date actual genetic divergence by a significant period of time
(Smith and Peterson 2002). Thus, when these earliest fossils are used as calibration points
in divergence time estimation (even assuming a perfect molecular clock) they will still
underestimate genetic divergence times.
Despite such problems associated with calibration from the fossil record, a surprisingly
large number of studies use only one or two calibration points (Hedges et al. 1996; Kumar
1996; Gu 1998; Wang et al. 1998). The argument for using few calibration point estimates
Figure 3.4 The ideal case is shown by fossils wherein well-dated fossils fall just before and after a
branching event enabling tight constraint on the true divergence time. More realistically, fossils may
be sister taxa, (shown as ) and will instead provide an underestimate of the true divergence date.
50 RICHARD A.FORTEY ET AL.