evolution but also to our understanding of the sequence of character changes that shaped all
subsequent events in chordate phylogeny.
Neoproterozoic refugia and the origin of vertebratesOne inevitable development of molecular clock estimates is that attempts are being made
to link intrinsic evolutionary change to extrinsic environmental factors. For instance, van
Tuinen et al. (1998) proposed that the origin of ratites is not just coincident with, but
inextricably linked to, the separation of Africa and South America during the Early to mid-
Cretaceous (see Cracraft 2001 for an excellent analysis). More recently, Hedges (2001;
Chapter 2) proposed that the proximity of the molecular clock estimate for the
divergence of crown-vertebrates to radiometric dates for the first major Neoproterozoic
glaciation (Sturtian; 750–700 Ma) may not be coincidental. It is argued that both the
Sturtian and Varanger glaciations (610–570 Ma) would have led to contraction in the
topological range of species and, through long-term genetic isolation in small refugia, to
considerable speciation. As worthy as this approach may be in demonstrating an integrated
approach to the questions of when, where, how, and why vertebrates first evolved, there
are two significant problems with regard to this linkage of intrinsic evolutionary and
extrinsic environmental factors. First, the nature, timing, and tempo of the Cryogenian
period of the Neoproterozoic is utterly unresolved, in terms of the timing, duration, and
number of glaciation episodes (Knoll 2000). Second, and more intractably, the standard
errors on molecular estimates are currently so vast (and unrealistrically conservative) that
they render worthless any attempt to match biotic events to radiometrically dated
environmental events.
Evolutionary scenarios based upon palaeontological datingAlthough molecular estimates fail to provide the necessary temporal constraint to under-pin
attempts to uncover any possible link between intrinsic evolutionary events and extrinsic
environmental events, palaeontological data provide no panacea either, at least with
regard to the origin and early evolution of vertebrates. It has been recognized for many years
that evolutionary history cannot be read directly from the rocks, but many scenarios for
the origin of major clades remain current, even though the supporting data have not
expanded from those on which they were originally contrived. For instance, Romer’s
celebrated ‘eurypterid influence on vertebrate history’ (Romer 1933) is based upon the
co-occurrence and vaguely comparable diversity trends of eurypterids and the then
earliest skeletonizing vertebrates in the Silurian. Thus, the origin of the skeleton has been
attributed to the selection-based effect of predating eurypterids upon early vertebrates.
But not only are the earliest known skeletonizing vertebrates now Cambrian in age, and
the earliest undisputed ‘armoured’ vertebrates Ordovician in age, but our phylogenetic
tests and internal assessments of the consistency of stratigraphic data both reveal that these
lineages probably existed even earlier. Thus, the co-occurrence and evolutionary history
of vertebrates and eurypterids is no longer apparent and Romer’s evocativer theory must
finally be laid to rest.
PHILIP C.J.DONOGHUE ET AL. 217