When well-defined paleospecies have been tested for their correspondence with modern biospecies, such status has often been persuasively affirmed. Two recent studies seem particularly convincing. Michaux (1989) studied four living species of the marine gastropod genus Amalda from New Zealand. Fossils of this genus date to the upper Eocene of this region, while all four species extend at least to the Miocene-Pliocene boundary. The four taxa represent good biospecies, based on absence of hybrids in sympatry, and on extensive electrophoretic study (Michaux, 1987) showing distinct separation among species and “no detectable cryptic groupings” (Michaux, 1989, p. 241) within any species. Michaux then used canonical discriminant analysis to achieve clear morphometric distinction among the species based on 10 shell measurements for each of 671 live specimens.
He then made the same measurements on 662 fossil specimens from three of the species (the fourth did not yield enough shells for adequate characterization). Mean values, in multivariate expression based on all 10 variables, fluctuated mildly through time (see Fig. 9-8), but never departed from the range of variation within extant populations — an excellent demonstration of stasis as dynamic maintenance within well-defined biospecies through several million years. Michaux concluded (1989, pp. 246-248): “Fossil members of three biologically distinct species fall within the range of variation that is exhibited by extant members of these species. The phenotypic trajectory of each species is shown to oscillate around the modern mean through the time period under consideration. This pattern demonstrates oscillatory change in phenotype within prescribed limits, that is, phenotypic stasis.”
Jackson and Cheetham's (1990, 1994) extensive studies of cheilostome bryozoan species provide even more gratifying affirmation, especially since these “simple” sessile and colonial forms potentially express all the attributes of extensive ecophenotypic variation (especially in molding of colonies to substrates, and in effects of crowding) and morphological simplicity (lack of enough complex skeletal characters for good definition of taxa) generally regarded as rendering the identification of biospecies hazardous, if not effectively impossible, in fossils. Moreover, Cheetham had begun his paleontological studies (see discussion on pp. 867–870) under the assumption that careful work would reveal predominant gradualism and refute the “new” hypothesis of punctuated equilibrium — so the conclusions eventually reached were not favored by any a priori preference!
In a first study — devoted to determining whether biospecies could be recognized from skeletal characters (of the sort used to define fossil taxa) in several species within three genera of extant Caribbean cheilostomes — Jackson and Cheetham (1990) examined heritability for skeletal characters in seven species. In a “common garden” experiment (under effectively identical conditions at a single experimental site), they grew F1 and F2 generations from embryos derived from known maternal colonies collected in disparate environments [Page 787] and places. Multivariate discriminant analysis assigned all but 9 of 507 offspring into the same morphospecies as their maternal parent. The authors then used electrophoretic methods to study enzyme variation in 402 colonies representing 8 species in the three genera. They found clear and complete correspondence between genetic and morphometric clusterings, and also determined (p. 581) that “genetic distances between morphospecies are consistently much higher than between populations of the same morphospecies”; moreover, they found no evidence for any cryptic division (potential “sibling species”) within skeletally defined morphospecies.
In a concluding and gratifying observation — indicating that paleontologists
9-8. Stasis in three genetically well-defined extant species of the gastropod Amalda from New Zealand based on 662 fossil specimens. Mean values in multivariate expression based on all ten variables fluctuate mildly through time, but never depart from the range of variation within extant populations. From Michaud, 1989.
[Page 788]
need not always humble themselves before the power of neontological genetic analysis of biospecies — Jackson and Cheetham (1990) p. 582) make an empirical observation about the capacity of morphometric data (of the sort generated from fossils):
The identity of quantitatively defined morphospecies of cheilostome bryozoans is both heritable and unambiguously distinct genetically. The importance of rigorous quantitative analysis was underlined by our discovery of three species of Stylopoma previously classified as one, a separation subsequently confirmed genetically. The widely supposed lack of correspondence between morphospecies and biospecies may result as much out of uncritical acceptance of outdated, subjectively defined taxa as from any fundamental biologic differences between the two kinds of species.
Jackson and Cheetham (1994) then followed this study with a more extensive documentaiton, this time using large numbers of fossil species as well as living forms, of phylogenetic patterns in two Caribbean cheilostome genera, Stylopoma (included in the first study as well), and Metrarabdotos (the subject of Cheetham's earlier and elegant affirmations of punctuated equilibrium from morphometric data alone — Cheetham, 1986 and 1987, and extensively discussed on pp. 867–870). Again, and for both genera, they found strict correspondence between genetically defined clusters and taxa established by skeletal characters accessible from fossils.
With increased confidence that the taxa of his classical studies on punctuated equilibrium in Metrarabdotos represent true biospecies, Cheetham (now writing with Jackson) could affirm his earlier work (Jackson and Cheetham, 1994, p. 420): “Morphological stasis over millions of years punctuated by relatively sudden appearances of new morphospecies was demonstrated previously for Metrarabdotos. Our updated results strengthen confidence in that pattern, with 11 morphospecies persisting unchanged for 2-6 m.y., all at p > 0.99, and no evidence that intraspecific rates of morphological change can account for differences between species.”
For Stylopoma, where fossil evidence had not previously been analyzed morphometrically, results also affirmed punctuated equilibrium throughout (1994, p. 420): “The excellent agreement between morphologically and genetically defined species used in this taxonomy suggests that morphological stasis reflects genuine species survival over millions of years, rather than a series of morphologically cryptic species. Moreover, eleven of the 19 species originate fully formed at p > 0.9, with no evidence of morphologically intermediate forms, and all ancestral species but one survived unchanged all with their descendants.”
In a concluding paragraph about both genera, Jackson and Cheetham wrote (p. 407): “Stratigraphically rooted trees suggest that most well-sampled Metrarabdotos and Stylopoma species originated fully differentiated morphologically and persisted unchanged for > 1 to > 16 m.y., typically [Page 789] alongside their putative ancestors. Moreover, the tight correlation between phenetic, cladistic, and genetic distances among living Stylopoma species suggests that changes in all three variables occurred together during speciation. All of these observations support the punctuated equilibrium model of speciation.”
Despite the encouragement provided by these and other cases, problems continue to surround the definition of paleontological species — a subject of central importance to punctuated equilibrium, given our invocation of speciation as the quantum of change for life's macroevolutionary history, and the source of raw material for higher-level selection and sorting. These problems center upon three main issues (in both the inherent logic of the case and by recorded debate in the literature): the first untroubling, the second potentially serious, and the third largely resolved in empirical terms. All three issues raise the possibility that paleospecies systematically misrepresent the nature and number of actual biospecies. (If paleospecies don't correspond with biospecies in all cases — an undeniable proposition of course — but if these discrepancies show no pattern and produce no systematic bias, then we need not be troubled unless the relative frequency of no correspondence becomes overwhelmingly high, an unlikely situation given
the excellent alignments found in the few studies explicitly done to investigate this problem, as discussed just above.) The following three subsections treat these three remaining issues seriatim.
Reasons for a potential systematic underestimation of biospecies
by paleospecies
Might we be missing a high percentage of actual speciation events because paleontologists can only recognize a cladogenetic branch with clear phenotypic consequences (for characters preserved as fossils), whereas many new species arise without substantial morphological divergence from their ancestors? In the clearest case, paleontologists (obviously) cannot detect sibling species, a common phenomenon in evolution (see Mayr, 1963, for the classic statement). Moreover, we may also miss subtle changes in phenotype, or substantial alterations (of color, for example) in features that are often important in recognizing species, but do not achieve expression in the fossil record.
Our harshest critics have urged this point as particularly telling against punctuated equilibrium. Levinton, for example (1988, p. 182), holds that “the vast majority of speciation events probably beget no significant change.” He then views the consequences as effectively fatal for punctuated equilibrium (1988, p. 211): “The punctuated equilibrium model argues that morphological change is associated with speciation and that species are static during their history due to some internal stabilizing mechanism. There is no evidence coming from living species to support this. If anything, recent research has demonstrated that speciation occurs typically with little or no morphological change; hence the large-scale occurrence of sibling species.” [Page 790]
Hoffman (1989, p. 115) invokes this argument to assert the untestability, hence the nonscientific status, of punctuated equilibrium:
Long-term evolutionary stasis of species, however, simply cannot be tested in the fossil record. Paleontological data consist solely of a small sample of phenotypic traits — little more than morphology of the skeletal parts — which does not allow us to make any inference about changes in a species's genetic pool or even about changes of the frequency distribution of phenotypes in a phyletic lineage. The no preserved portion of the phenotype of each fossil species is so extensive that it may always undergo considerable evolutionary changes that remain undetectable by the paleontologist. What appears then to the paleontologist as a species in complete evolutionary stasis may in fact represent a succession of fossil species or perhaps a whole cluster of species, a phylogenetic tree with a sizable number of branching points, or speciation events.
While I freely admit all these arguments for under representation of true species in fossil data, I do not comprehend how punctuated equilibrium could be thus rendered untestable, or even seriously compromised (see further arguments in Gould, 1982c and 1989e; and Gould and Eldredge, 1993). I base my argument on two logical and methodological principles, not on the probable empirical record (where I largely agree with our critics).
The proper study of macroevolution. By consensus, and accepting a criterion of testability, science does not include, within its compass of inquiry, fascinating questions that cannot be answered (even if they address potentially empirical subjects). For example, and for the moment at least, we know no way to ask a scientific question about what happened before the big bang, for compression of universal matter to a single point of origin wipes out all traces of any previous history. (Perhaps we will eventually devise a way to obtain such data, or perhaps the big bang theory will be discarded. The question might then become scientifically tractable.) Similarly, we know that many kinds of evolutionary events leave no empirical record — and that we therefore cannot formulate scientific questions about them. (For example, I doubt that we will be able to resolve the origins of human language, unless written expression occurred far earlier than current belief and evidence now indicates.)
The nature of the fossil record leads us to define macroevolution as the study of phenotypic change (and any inferable correlates or sequelae) in lineages and clades throughout geological time. Punctuated equilibrium proposes that such changes generally occur in discrete units or quanta in geological time, and that these quanta represent events of branching speciation. Thus, we do identify speciation as the source of raw material for macroevolutionary change in lineages. But we do not, and cannot, argue (or attempt to adjudicate at all) the quite different proposition that all speciation events produce measurable quanta of macroevolutionary change. The statement — our proposition — that nearly all macroevolutionary change occurs in increments of speciation carries no implications for the unrelated claim, often imputed to [Page 791] punctuated equilibrium by our critics (but largely irrelevant to our theory), that nearly all events of speciation produce an increment of macroevolutionary change. This conclusion flows from elementary logic, not from empirical science. The argument that all B comes from A does not imply that all A leads to B. All human births (at least before modern interventions of medical technology) derived from acts of sexual intercourse, but all acts of intercourse don't lead to births.
To draw a more relevant analogy: in the strict version of Mayr's peripatric theory of speciation, nearly all new species arise from small populations isolated at the periphery of the parental range. But the vast majority of peripheral isolates never form new species; for they either die out or reamalgamate with the parental population. Similarly, most new species may never be recorded in the fossil record; but, if the theory of punctuated equilibrium holds, when changes do appear in lineages of fossils, speciation provides the source of input in a great majority of cases. Thus, most speciation could be cryptic (and unknowable from fossil evidence), while effectively all macroevolutionary change still arises from the minority of speciation events with phenotypic consequences. Just as peripheral isolates might represent “the only game in town” for forming new species (though few isolates ever speciate), cladogenetic speciation may be “the only game in town” for inputting phenotypic change into macroevolution (though few new species exhibit such change).
The treatment of ineluctable natural bias in science. In an ideal world — the one we try to construct in controlled laboratory experiments — no systematic bias distorts the relative frequency of potential results. But the real world of nature meets us on her own terms, and we must accept any distortions of actual frequencies that directional biases of recording or preservation inflict upon the archives of our evidence. At best, we may be able to correct such biases if we can make a quantitative estimate of their strength. (This general procedure, for example, has been widely followed to correct the systematic under measurement of geological ranges imposed by the evident fact that observed first and last occurrences of a fossil species can only provide a minimal estimate for actual origins and extinctions, for the observed geological range of a species must be shorter (and at least cannot be longer) than the actual duration. Studies of “waiting times” between sequential samples within the observed range, combined with mathematical models for constructing error bars around first and last occurrences, have been widely used to treat this important problem — see Sadler, 1981; Schindel, 1982; Marshall, 1994.)
Often, however, we can specify the direction of a bias, but do not know how to make a quantitative correction. In such cases, the sciences of natural history must follow a cardinal rule: if the direction of bias coincides with the predicted effect of the theory under test, then researchers face a serious, perhaps insurmountable, problem; but if a systematic bias works against a theory, then researchers encounter an acceptable impediment — for if the theory can still be affirmed in the face of unmeasurable biases working against a favored explanation, then the case for the theory gains strength. [Page 792]
For proponents of punctuated equilibrium, speciation represents the primary source for morphological changes that, by summation of increments, build trends in the history of lineages. If a systematic bias in the nature of paleontological evidence leads us to underestimate
the number of speciation events, and if we can still explain trends by this observed number (necessarily less than the actual frequency), then the case for punctuated equilibrium becomes stronger by affirmation in the face of a bias working against full expression of the theory's effect. Thus, although we regret the existence of any bias that we cannot correct, a systematic under representation of speciation events does not subvert punctuated equilibrium because such a natural skewing of evidence makes the hypothesis even more difficult to affirm — and support for punctuated equilibrium therefore emerges in a context even more challenging than the unbiassed world of controlled experimentation.
The Structure of Evolutionary Theory Page 126