be not casts but sediment accumulations glued by mucus. In order for these originally
positive-relief structures to be preserved on the underside of the overyling bed, they must
somehow have become transferred from one bed to another. Even if this is possible, they
still apparently show the same composition of the bed they now reside in; and they do not
show any sign of being compressed.
A more complex taphonomic history for these structures is hinted at in a recent
discussion of their nature (Conway Morris et al. 2002). Here, the original authors suggest
that the original mucus-bound strings consisted of mud, and have been later replaced/cast
by sand. Such a scenario would require the mucus strings to be partly covered by more
mud, and then during compaction of the upper layers, to collapse more than this mud
(because of decay of the mucus within). Even if this scenario had any evidential support, it
seems highly implausible that the end result would be the sharply defined ridges as now
preserved. All of these difficulties render the likelihood of these structures representing
true trace fossils extremely low. The simple fact is that whatever the true status of these
and other candidate early trace fossils, none of them can be straightforwardly accepted as
trace fossils; they all require a certain or high amount of special pleading: in itself a
significant point.
The above points, we think, make it clear that whatever the resolution of the misfit
between the fossil record and molecular evidence for the origin of animals, it does not
come about through a misunderstanding of the known fossil record. The other options are
to investigate what our reasonable expectations of the fossil record should be, and to re-
examine the molecular evidence. Perhaps we should not be too surprised about the
mismatch.
Telling the true time from the fossil recordThe most persistent and telling critics of the fossil record have, perhaps surprisingly, been
palaeontologists. Hence, Smith (1999) and Smith and Peterson (2002) have pointed out
important features of the fossil record that, at least potentially, greatly weaken its
reliability. These include the facies dependence of fossils; the reliance on outcrops of the
right age and type; the variability of preservation rate through time; and the problems of
identifying basal members of clades. More importantly, a considerable amount of effort is
being expended in quantifying what this weakening of fossil reliability means (e.g. Strauss
and Sadler 1989; Marshall 1990, 1997; Foote et al. 1999; Tavaré et al. 2002). The
simplest, and most unrealistic, approach is to consider the actual fossil finds during an
interval to be Poisson distributed, and use the calculated mean to place confidence limits
on the exponentially distributed ‘gaps’ that extend above and below the known range.
Bayesian inference (Strauss and Sadler 1989) and likelihood estimates (Huelsenbeck and
Rannala 1997) of true range can also be made (see Huelsenbeck and Rannala 1998 for
discussion). Other ways of relaxing the Poisson distribution criterion would presumably
include modelling fossil finds with a gamma distribution, although as Huelsenbeck and
Rannala (1998) point out, this sort of refinement can hardly be done justice to by the
general imperfection of the stratigraphic record.
An important critique of these distribution-based analyses is that they fail to take into
account important facts that particularly apply to the beginnings (although not always to
174 DATING THE ORIGIN OF BILATERIA