892 THE STRUCTURE OF EVOLUTIONARY THEORY
discuss why the obvious implication—that a trend towards generalists should
sweep through the clade—fails because ceteris paribus (all other things equal)
does not hold at several levels and strengths of correlation. First of all, "niche
generalists tend to have fewer species per clade than niche specialists" (Kammer et
al., 1997, p. 221), thus illustrating the most pervasive and powerful
macroevolutionary constraint recognized so far at the speciational level (see
general discussion on pages 739-741): the forced and intrinsic negative correlation
between speciation rate and longevity. Secondly, these patterns hold "only during
times of background extinction when Darwinian natural selection prevails" (p.
221). Mass extinctions may then impact species at random, without preferential
regard for their ecological status or prospects for longevity in normal times.
Similar questions may be asked at all scales, including general patterns for
life itself. For example, many robust paleontological data sets show a general
tendency for increased longevity in marine invertebrate species through geological
time. Even a famously iconoclastic thinker like David Raup, who devoted so much
of his career to exploring the power of random systems to render observed patterns
of the fossil record, interpreted this result as our best case for a meaningful concept
of "progress," defined as increasing adaptive excellence of organisms (and leading
to greater resistance to extinction). But several authors (Valentine, 1990; Gilinsky,
1994; Jablonski, Lidgard and Taylor, 1997) have reread this result in terms of
species sorting as a tendency "for high-turnover taxa to be replaced over geologic
time by low-turnover taxa" (Jablonski, et al., 1997, p. 515).
As Gilinsky (1994) notes, such a pattern, thus reformulated, may bear little or
no relationship to general adaptive excellence at the organismal level. If speciation
and extinction rates generally operate in balance (as they do), then some clades
may be designated more "volatile" (in Gilinsky's terminology) as marked by their
high rates of speciation and extinction, and others as more stable for lower rates of
both these defining processes at the species level. In an abstract and general sense,
both "strategies" may be regarded as equal in yielding the same result of steady
cladal persistence, but with volatile clades showing more variation around a stable
mean. However, in our real world of fluctuating environments and, especially,
mass extinctions, volatility may doom clades in the long run because any reduction
to zero (however "temporary" and reversible in abstract modelling) extinguishes all
futures in our actual world of material entities. Since volatile clades, on average,
must cross the zero line more frequently than stable clades, a general trend to
increasing species longevity may only arise as an indirect consequence of the
higher vulnerability of volatile clades and the consequent accumulation of stable
clades through geological time.
Several recent articles on bryozoan evolution illustrate the utility of a
speciational approach to trends—if only as a method for setting base lines and
making distinctions, all the better to document patterns not attributable to
differential speciation. For example, Jablonski, Lidgard and Taylor (1997) found
that "the generation of low-level novelties is effectively driven by