One must address the large disparity between ages derived from fossil and molecular
clock evidence. Either the fossils or molecular clock (or both) are wrong. I am of the
opinion that the fossil record is more likely to provide an accurate time (see discussion
above), although it is clear that fossils can only provide minimum ages. There are a
number of potential pitfalls with the techniques adopted by Heckman et al. which may
explain the anomalous result. These include: (i) their technique assumes that the proteins
of animals, fungi, and plants evolved at the same rates (many workers dispute this); and
(ii) calibration is indirect (essentially using the animal-plant-fungus, nematode-chordate
and nematode-arthropod, and arthropod-chordate divergences, obtained from the
previous analysis of Wang et al. 1999), and ultimately calibrated using the vertebrate fossil
record—in fact a datum at 310 Ma (indirect calibration is likely to enhance errors, such as
range extension, when applied to different datasets, particularly when applied over vast
tracts of geological time as is the case in this analysis). I consider that the molecular clock
estimate for the colonization of land by plants provided by Heckman et al. (2001) is far too
old, primarily as a consequence of the factors outlined above.
Molecular clock techniques are consistently providing divergence estimates far greater
than expected based on fossil evidence, and concern has been expressed that many of the
currently utilized molecular clock techniques are flawed, and indeed that there may be no
molecular clock (e.g. Rodríguez-Trelles et al. 2001). There are a number of potential
sources of error when using molecular clock techniques to estimate divergence times, and
these are often enhanced when considering increasingly older events. One of the major
problems is rate variation, which is now an acknowledged fact and has a number of
different causes (see Muse 2000; Sanderson and Doyle 2001). None the less, attempts are
being made to compensate for rate variation. For example, Sanderson (1997) has
developed an approach termed ‘non-parametric rate smoothing’ (NPRS). This method
accepts rate variation, as opposed to rate constancy, but assumes that rate changes are
autocorrelated (i.e. immediate descendants inherit rate change from the ancestral
lineage). It should be noted, however, that there is little evidence to support the
assumption that rate variation is autocorrelated. None the less, studies utilizing NPRS do
appear to be reducing the discrepancy between fossil- and molecular clock-based
estimates (e.g. angiosperms; Wikström et al. 2001).
Research into the molecular clock is still in its infancy, and it would be extremely
valuable if it could be accurately utilized. There are a number of potential problems, chief
among them rate variation, and these are particularly problematic when considering
divergences that occurred deep in geological time. Hopefully as more research is
undertaken, more reliable methods to compensate for rate variation will appear, and we
can have more confidence in the results. Until such time I consider it premature to rely solely
on data from the molecular clock.
ConclusionsIt is concluded that the most reliable date for the origin of embryophytes is the minimum
age of Llanvirn (Ordovician) provided by the earliest occurrence of dispersed microfossils
(specifically spores) that are widely accepted as deriving from land plants. Other lines of
evidence, namely geochemistry and the molecular clock, are considered to be unreliable at
DATING THE ORIGIN OF LAND PLANTS 139