Schweber (1980, 1985, 1988) noted Darwin's trouble with discordance between [Page 238] levels, though he does not provide the technical arguments detailed below. In Schweber's view, Darwin was driven to formulate an argument that “does not cohere” (Schweber, personal communication) because his century's ignorance of hereditary mechanisms drove him to describe variation within species and varieties treated as units, while the causal structure of natural selection rested upon individual organisms. Arguments about organisms and species are not comfortably intertwined or mutually supporting within Darwin's conceptual structure: rather, the two levels remain discordant and inadequately (if not illogically) bridged. “This difference in the 'units' used is important. It accounts for the fact that at times levels of description were interchanged and some confusion necessarily crept in” (Schweber, 1980, p. 240). “There was no link between adaptation and speciation, except whatever could be supplied by a quasi-historical developmental idea of optimizing the amount of life” (ibid., pp. 287-288). “The problem of the different levels of descriptions was confined to how the properties of variations in individuals ... were responsible for the assumed variability characteristic of varieties and species. This problem Darwin never solved” (ibid., p. 288).
We can exemplify Schweber's perceptions about Darwin's incoherence of argument by dissecting the logic of Darwin's attempt to use ordinary natural selection as the basis of divergence. For three basic reasons, his attempt to invoke selection among organisms as an explanation for patterns in speciation and extinction — the heart of the “principle of divergence,” and the primum desideratum for a complete theory of natural selection — fails because the level of species must be addressed both directly and causally, while Darwin's rationale for explanation from below includes gaps and fatal weaknesses.
The calculus of individual success
Darwin treats the principle of divergence in two extensive discussions — the long and even labored account of chapter 4 in the Origin of Species (1859, pp. 111-126), and the even more detailed exposition intended for the “big species book” that Wallace interrupted and Darwin never published because he rushed to compose the Origin instead. The manuscript for most of this larger project survives, including the full discussion of divergence intended for chapter 6 “On Natural Selection.” This text, published under R. C. Stauffer's editorship in 1975, treats the principle of divergence on pages 227-251.
When, in the 1970's, I first read the Origin with the notion of hierarchical selection in mind, I was fascinated by Darwin's struggle to bridge the levels, and his ultimate lack of success. Schweber speaks of “incoherence”; I would rather describe Darwin's “moves” of argument as an oscillation between one mode and the other. In some passages (including those cited above), Darwin speaks of ordinary natural selection and the advantages enjoyed by extreme variants. In others, he judges the success of a parental form not by the vigor or competitive prowess of offspring, but by the number of descendant species emanating from a rootstock.
These themes could, of course, be complementary. One perspective might [Page 239] imply and grade into the other; the two levels could represent alternate solutions of the Necker cube (see Dawkins, 1982) — that is, views of the same configuration from different vantage points. The individual success of extreme organisms may simply imply, ipso facto and necessarily, the ultimate multiplication of species, and success measured by number of descendant taxa might therefore act as a surrogate for, or simple extension of, natural selection. (For this reason, I assume, Darwin includes the principle of divergence as the largest section of chapter 4, entitled natural selection.)
But this argument does not cohere, and I think that any careful reader must be struck by Darwin's discomfort as he mixes and juggles the argument for success of extreme organismal variants with the calculus of advantage mapped by number of descendant taxa. Darwin's argument falters because the use of a lower level (success of extreme variants, in this case) to explain a phenomenon at a higher level (multiplication of species) can only work if “perfect transfer” can be defended — that is, if the lower level entails the higher as a direct consequence without any intervention (even a synergistic boost, not to mention a contrary force) from causes at the higher level itself. Darwin understood this principle perfectly well. Indeed, he was probably the only man who, in this infancy of evolutionary science, had carefully and consistently thought the logic of selection through to this correct interpretation. Thus, Darwin tried to construct an argument for perfect transfer — but he failed. Darwin advanced these claims in Natural Selection, but ultimately dropped them from the Origin, because (I suspect) he recognized their weaknesses. (I criticize the arguments for perfect transfer in points two and three below; in this subsection, I document Darwin's bold claim for a higher level calculus of individual success.)
Consider Darwin's first hypothetical example of the principle of divergence in the Origin (1859, p. 113). He speaks of a “carnivorous quadruped” that, by ordinary natural selection, has expanded in population to the limits of local environments. To do even better in the struggle for life, this form must now diversify into several descendant taxa. But how can the canonical argument for natural selection be cashed out in terms of multiplicity of descendant species? The logic of individual struggle carries no implications about the splitting of populations (especially for an in situ sympatric splitting that implies general predictability rather than the simple good fortune of geographic isolation).
Take the case of a carnivorous quadruped, of which the number that can be supported in any country has long ago arrived at its full average. If its natural powers of increase be allowed to act, it can succeed in increasing (the country not undergoing any change in its conditions) only by its varying descendants seizing on places at present occupied by other animals: some of them, for instance, being enabled to feed on new kinds of prey, either dead or alive; some inhabiting new stations, climbing trees, frequenting water, and some perhaps becoming less carnivorous. The [Page 240] more diversified in habits and structure the descendants of our carnivorous animals became, the more places they would be enabled to occupy (Darwin, 1859, p. 113).
Darwin continually invokes the calculus of individual success by number of descendant taxa (see also 1859, p. 116, and Stauffer, ed., 1975, p. 228): “As a general rule, the more diversified in structure the descendants from any one species can be rendered, the more places they will be enabled to seize on, and the more their modified progeny will be increased” (Darwin, 1859, p. 119).
Often, he mixes both criteria — the adaptive success of extreme variants in struggle, and the calculus of descendant taxa — in a single statement.
Here in one way comes in the importance of our so-called principle of divergence: as in the long run, more descendants from a common parent will survive, the more widely they become diversified in habits, constitution and structure so as to fill as many places as possible in the polity of nature [organismic level], the extreme varieties and the extreme species will have a better chance of surviving or escaping extinction, than the intermediate and less modified varieties or species [taxon level]. But if in a large genus we destroy all the intermediate species, the remaining forms will constitute sub-genera or distinct genera, according to the almost arbitrary value put on these terms (in Stauffer, ed., 1975, p. 238).
But are these two statements really equivalent? Does the lower level claim for organismic advantage imply the higher level phenomenon of species proliferation without reference to any higher level causes?
In his most striking passage, indicating that he did grasp the need for higher level sorting based upon such group properties as the range of variation, Darwin attributes the success of introduced placentals in Australia not, as we might anticipate from ordinary natural selection, to the adaptive biomechanical superiority of placental design (honed in the refiner's fire of more severe competition in Eurasia and America), but to a greater range of
placental variation across taxa, produced by their later stage in the historical process of divergence.
A set of animals, with their organization but little diversified, could hardly compete with a set more perfectly diversified in structure. It may be doubted, for instance, whether the Australian marsupials, which are divided into groups differing but little from each other, and feebly representing, as Mr. Waterhouse and others have remarked, our carnivorous, ruminant, and rodent mammals, could successfully compete with these well-pronounced orders. In the Australian mammals, we see the process of diversification in an early and incomplete stage of development (Darwin, 1859, p. 116).
The causes of trends
Trends represent the primary phenomenon of evolution at higher levels and longer time scales. Trends therefore pose the key challenge, the ultimate making [Page 241] or breaking point, for extrapolationist theories that seek the causes of macroevolution in microevolutionary processes centered upon organismic selection. Darwin understood and accepted this challenge; his principle of divergence marks his attempt to depict trends as extrapolated results of natural selection. (The principle of divergence attempts to explain morphological trends by specialization and progressive departure from ancestral form, and also to account for numerical trends by multiplication of some taxa at the expense of others within a clade.)
Advocates of species selection hold that trends must be described as the differential birth and death of species (not the simple anagenetic extrapolation of change within a population), and that the causes for such differentials must be sought, at least in part, in irreducible species-level fitness (see Chapter 8). The standard extrapolationist rejoinder invokes two arguments: (1) Differential death and survival rather than differential birth (of species) usually fuels trends. The death and persistence of groups can be reduced more easily to organismic competition, while differential production of species more often demands irreducible causes, for an organism cannot speciate by itself, while the death of a population may represent no more than the accumulated demise of all organisms. (2) The cause of differential survival or death must be reducible to ordinary natural selection.
Darwin did not offer his principle of divergence as a rejoinder to any explicitly developed theory of species selection, for no such formulation existed when he wrote. But he understood the logical requirements of his theory so well that he provided the necessary rationale without the spur of a formally stated alternative. He also, and uniquely, reinforced his argument with an illustration of the need for differential survival of certain kinds of variants within random arrays. In Natural Selection, Darwin presents his case as a second figure (reproduced here as Fig. 3-6) that he did not include in the abridged Origin of Species in 1859. (Virtually no one knew about the existence of this figure or argument until Stauffer published the manuscript of Natural Selection in 1975.)
The basic figure of both Natural Selection and the Origin illustrates Darwin's claim that only a few vigorous species will produce the variants leading to the “recruitment” of new species. (These vigorous species are the extreme forms favored by ordinary natural selection — A and M in Natural Selection, A and I in the Origin, see Figs. 3-5 and 3-6.) The variants of these vigorous species radiate in an even fan, or random array, about the modal form of their ancestor. A trend then arises because ordinary natural selection favors extremes within the fan. Darwin recognizes that a trend to specialization and diversity cannot be generated only by the greater vigor of extreme species in the initial array; he must also defend a second proposition about differential survival among offspring of these favored extremes.
The second figure of Natural Selection now comes into play (Fig. 3-6). Darwin shows us what happens under a regime of random survival within the fan of variants generated by the favored vigorous species at ecological extremes, as opposed to a regime of selection positively directed towards extreme members
[Page 242]
3-5. The famous and only diagram published by Darwin in the Origin of Species, 1859. Darwin did not construct this much misunderstood diagram as a simple and general model of branching phylogeny, but quite specificaiiy as an illustration of his Principle of Divergence. Note that each species tends to produce a fan of variants and that the most successful forms tend to emerge from extreme positions of each fan. See text for details.
[Page 243]
3-6. The expanded version of Darwin's figure, drawn in preparation for the long version, Natural Selection, that was never published because Darwin hurried to complete the shorter Origin of Species after receiving Wallace's manuscript. This version was not published until 1975 (see Stauffer, 1975). Here, Darwin shows us how the expectation of enhanced success for extreme variants within each fan, as predicted by his Principle of Divergence, will lead to trends (upper part of diagram); whereas random survival of variants (lower part of diagram) yields no trend and no ecological expansion. See text for details.
of each fan. The trend to diversification halts in the random regime. The final products need not become any more distinct than the initial parents (f10, h10, and l10 lie right under ancestor A, while m10 does not differ from parental M). On the other hand, differential survival of extremes within the fans will produce trends (also seen in Fig. 3-6). Darwin writes of the second diagram [Page 244] (the lower half of Fig. 3-6), where variation to the left and right of A represents greater or lesser adaptation to drought in plants: “Everything is the same as in diagram I... except that it is left to mere chance in each stage of descent, whether the more or less moisture loving varieties are preserved; and the result is, as graphically shown, that a10 and l10 [sic, he has no a10 in the drawing, but represents the leftward extreme as f10] differ in this respect; and so in other respects, hardly more than did the first varieties (a1l1) which were produced” (in Stauffer, ed., 1975, p. 244).
The argument for a trend that can be reduced to natural selection therefore hinges upon reasons for differential survival of extremes within the fans of varying species; for the trend cannot emerge simply from the greater evolutionary vigor of the ancestral extremes themselves. (Interestingly, Darwin never considers the alternative, more congenial to species selection, of greater production of variants at the extremes, with random survival within fans.) Darwin now makes his crucial move for ordinary natural selection, using his principle of divergence: extremes enjoy differential survival within the fans of variants, because natural selection favors adaptation to peripheral, over adaptation to central, “stations” in any region. (We must remember that all members of the fan are well adapted to their own local bits of the environment). Now, at the crux of his development, Darwin tries to defend his position on the differential value of extreme stations, and his argument falls apart — to be rescued only with a forced and self-contradictory ad hoc hypothesis (explicitly stated in Natural Selection, but wisely omitted from the Origin).
Darwin provides two potential reasons for differential success of organisms adapted to extreme environmental stations. The first remains perfectly acceptable, and would pass muster today as a standard ecological argument featured in all textbooks — reduced competition in less “crowded” extreme environments: “From our principle of divergence, the extreme varieties of any of the species, and more especially of those species which are now extreme in some characters, will have the best chance, after a vast lapse of time, of surviving; for they will tend to occupy new places in the economy of our imaginary country” (in Stauffer, ed., 1975, p. 239).
If Darwin had stopped here, his argument would have remained consistent, if dangerously weak. But his relentless probing would not permit such a course — for he knew that a key problem remained unsolved:* extreme variants may be favored in their own extreme environments, but why should they [Page 245] prevail over other variants in more central environments? After all, the more central ancestor continues to survive after the extreme variant buds off. The ancestor should therefore be favored in its own central environme
nt, and the descendant in its new peripheral station. Why, then, should the descendant ever replace the ancestor — so long as central environments persist along with marginal places? In all his writing on divergence, Darwin recognized that trends to specialization could not occur unless extreme descendants tended to wipe out more central ancestors in competition: trends, in other words, required a pattern of differential extinction as well, for the number of species in a region cannot increase indefinitely.
And here, after so much effort and careful development, Darwin bogged down. For this most resolutely higher level phenomenon of the supposed differential success of extreme vs. central species, Darwin could not provide a tenable argument based upon natural selection. With evident discomfort, Darwin resorted to an ad hoc assumption: he argued that while extreme variants adapt to their marginal stations, they also retain all the adaptations of their parents for the original central habitats. Thus, the descendant extremes remain as good as their parents in the ancestral environment, while adding a capacity for survival in marginal habitats.
But how can such a proposition be defended? Why should a species that has left one environment, and explicitly adapted to another, still retain all its prowess in an environment no longer inhabited (and from which it has actively diverged)? Not only does this proposition make no sense prima facie; such a claim also contradicts the canonical argument (often embraced by Darwin and his contemporaries) that specialization leads to “locking in” and decreased flexibility. In short, Darwin knows that he has run into a severe logical problem in trying to justify a central implication of his general argument: the differential survival of extreme taxa with a consequently preferential extinction of central species. How can such a pattern be explained — for central and marginal species should not, after all, be in overt competition, and central environments cannot be regarded as generally more evanescent than extremes? Darwin therefore invokes his ad hoc argument (described just above) for an expanded range of adaptation in extreme species, in order to place the organisms from these extremes into competition with their parents, thus generating a hypothetical explanation for differential parental death in terms of natural selection.
The Structure of Evolutionary Theory Page 40