| 328 |
JOURNAL OF PALEONTOLOGY, V. 62,
NO. 3,1988
|
perhaps in the simplified and stereotyped realm of bean-bag models: molecules
in a box, or grains of sand on a beach. But our world is one of massive
constraint mediated by the tree-topology of evolution. Once groups are
lost for whatever reason, they cannot reappear, and the range of disparity
shrinks. More importantly, if evolutionary time has directionality (see
Gould et al., 1987, for a defense of this position), then life can't get
back to its former state following a massive reduction in diversity or
disparity. If genomes coalesce through time, and if the Burgess phyla
had unusual potential for change as a heritage of common ancestry then
not far distant (despite the profound differences in form among phyla),
then loss of the oddballs imposed a permanently reduced disparity upon
the entire clade of Metazoa--for survivors were not free to radiate
this way again. Reduction of variance imparts direction intrinsically.
Let us thank our lucky stars for the survival of Pikaia.
CONCLUSION: HOW DOES THE THEME OF
TRENDS AS CHANGES IN VARIANCE FIT WITH
CURRENT REVISIONIST VIEWS OF
EVOLUTIONARY THEORY?
This paper does not deny the existence of trends properly called anagenetic,
or products of change within unbranched lineages. Conventional anagenetic
change may occur within populations (Sheldon, 1987; Bown and Rose, 1987),
but most reported cases are dubious (Gould and Eldredge, in press), and
I do not believe that this mode accounts for much in the total pattern
of evolution. Valid cases tend to add a rib, a bump, or a millimeter over
millions of years--and such changes simply do not extrapolate to the
evolutionary patterns that historians of life are charged to explain (they
are also too slow to ascribe to conventional directional selection--see
Gould and Eldredge, 1977).
An unconventional form of directional change
may properly be termed anagenetic at the level of species within clades. Trends
may occur within a clade, accompanied by no change in variance, disparity, or
number of species. Suppose, for example, that new species tend to arise in a preferred
morphological direction because allometry and ontogeny channel possibilities for
novelty. In such cases, the number of species in the clade might remain constant
(as extinctions balance originations), and modes, means, medians, and extreme
values at both tails might move, in lockstep fashion, towards new cladal values.
We could not attribute such a trend to changes in variance, nor could anagenesis
within species contribute if punctuated equilibrium prevailed. Such a trend could
be called anagenesis at the clade level, with species assuming the role played
by organisms in traditional anagenesis within populations.This paper
concentrates, instead, upon those "trends" (often falsely so conceived)
that arise as consequences of changes in variance among species within clades.
I try to provide a "bestiary" of examples (to convince colleagues that
this category is large and important), not a logical compilation of all possible
types within this broad domain. In particular, I recognize that at least two rather
different causes underlie the relevant changes in variance: increase or decrease
in number of items, and change in disparity among a constant number of items.
(Many or most real cases may mix both, of course.) Thus, for the two main examples
worked out in detail with original data, foraminifers (p. 323) display an increase
trend based on number of species, while batting averages (p. 326) form a decrease
trend rooted in falling disparity among a constant number of players.I
choose to unite these disparate kinds of trends not only because they have a common
(but usually unrecognized) root in changes of variance, but also because a proper
grasp of this theme might be important for our general understanding of macroevojution.
On the negative side (for debunking is vital), recognition of trends as changes
in variance, not anagenesis of entities, removes the rationale for most justifications
of evolutionary "progress." Progress, after all, has been the culturally-embedded
prod to our main interest in trends all along; we still seek to imbue time and
evolution with some meliorist property, some subtle justification for our own
intrinsic importance as evolutionary latecomers. Why else did we link the notion
of trend to the concept of adaptation in the first place, and why else did we
consider the unproven idea of orthoselection a conceptual victory?Interpreted
anagenetically as the causal movement of an entity, trends seem inevitably linked
with progress. Increase trends advance, and decrease trends hone. The entity is
the essence of the system itself, and must be moving for a reason. But if "entities"
are misleading abstractions, and the real phenomenon is a change in variance among
items, then different interpretations, not linked to progress, must often prevail.
Increase trends may arise automatically from an asymmetry in point of initiation
relative to potential range, and perceived movement in the open direction may
not record progress, but only the marginal position of the starting point. Decrease
trends may arise by random plucking in a directional world with no reaccess to
lost domains; the winnowing of the Burgess fauna may record a random subset of
survivors, not a honing of anatomical excellence.
But I regard a positive theme as more important, for I believe that
the phenomenon of trends as changes in variance has important bearing
on both central themes in current, revisionary work on the basis of evolutionary
theory: 1) concepts of structure and constraint as an antidote or leavening
for overemphasis on adaptation (structural vs. functional thinking); and
2) hierarchical, selection theory to broaden Darwin's reductionist premise
that all causality resides in the struggle among organisms. In trends
produced by changing variance, the primary questions will usually be structural,
rather than immediately adaptational. This shift in emphasis--indeed
(in most cases) this reversal of perspective--may be the most important
reform in thinking that a reinterpretation of trends as changes in variance
could inspire. No longer do we focus on the question: why (for what adaptive
reason) is this entity moving this way? We ask instead: why did the originator
start here, or why is the range of potential realization restricted in
this way? Apparent movement may be an uninteresting consequence.
Questions about starting points and potential ranges are
usually structural; they embody themes of constraint (physiological impossibility
for invasion of land by most marine creatures), or higher-level effects (what
aspects of ecologies and population structures grant small creatures a greater
potential for initiating clades?). For Cope's rule, we must ask "why did
it start here," not "why did it move there?" For batting averages,
we must grasp the change between average and extreme ability, and appreciate the
structural limit upon best performances, not lament a nonexistent loss of former
greatness.
On the theme of hierarchy, we must recognize that many trends formerly
interpreted as anagenetic movement of an entity are actually generated
by the higher-level process of species sorting (Vrba and Gould, 1986)--that
is, by differential success of species leading to expansions or contractions
of variance. For most phenomena of macroevolution, nature works as an
entity making and breaking machine at the level of species, not as a converter
of variation within populations into large-scale trends among clades.
For trends, the key to understanding is often not the perceived excursion
itself (for the "entity" supposedly in motion may be a misleading
abstraction, or a forced and sec-
|