The Structure of Evolutionary Theory blogging, chapter 8

Chapters read:1, 2, 3, 4, 5, 6 and 7.

It's been a while since I blogged Stephen Jay Gould's The Structure of Evolutionary Theory. I haven't forgotten it, but once I finished the historical preamble, nearly 600 pages, I was in the mood for a breather. My hunch was that despite Gould's emphasis on contingency in his theory in terms of his narrative there would be only a broad contextual relationship between Part I and Part II, the intellectual history of evolution by Stephen Jay Gould, and the scientific theory of evolution by Stephen Jay Gould. I've finished the 8^th chapter, which is ostensibly the first science chapter, though since Gould promiscuously mixed science & history it is more a notional than substantive distinction. Though a long section, at 150 pages this is nothing compared to the next chapter I'll hit (and probably won't blog for a while), which was spun off into a separate book, Punctuated Equilibrium.

In any case, the general complaints about Gould's prose style holds; despite his shift from narrative history toward a more scientifically inclined exposition the first portion of the chapter is larded with so much florid arcana as to put one to sleep. There are flashes of expository clarity and brilliance embedded within the mush, but there is an interesting caveat to this: my general sense is that Gould is at his best when he is polemically tearing down Richard Dawkin's gene-centered paradigm. I find this ironic, because whatever you might think of Dawkins' science his prose is generally a good balance between sober economy and flourishing engagement at the appropriate points. Gould rises to the challenge specifically when he has to meet Dawkins' prose head-on after a quotation from the latter's body of work. In the The Structure of Evolutionary Theory there are some general arguments about the level at which selection operates, the power of selection vis-a-vis other evolutionary forces, and the scale independence of evolutionary process (i.e., the extrapolation of microevolutionary dynamics to the macroevolutioanry level). The previous chapters have prefigured Gould's arguments copiously, so the introduction of these three points was no surprise. Like a ponderous oil tanker which operates more as a force of nature then an active agent against a physical background the prose rolls on evermore, page after page, example after example. But when confronted by the simplicity of Dawkins' and George Williams' arguments the tenor of the narrative shifts radically as each paragraph is dense with articulation and elucidation, as if Gould was shocked out of a torpor.

Of course, these pools of activity in the deep waters reemerge multiple times and the arguments are only very subtle twists on the general objections. Gould feels that the gene-centered view of Dawkins and the Oxford tradition of evolutionary biology which goes back to R. A. Fisher is exceedingly simple; he objects to the assumption that evolutionary can properly be modeled as the additive substitution of alleles on single loci against a genetic background. The Fisherian tradition assumes that nature works by the "least-squares principle," and Fisher himself aspired to a model of evolutionary genetics which mimicked the deterministic inevitability of thermodynamics. Gould will have none of this, and denies flat out that the assumption of additivity is often not appropriate or valid. He contends that emergence of properties among the network of genetic effects are critical, non-linear responses to genetic variation which are only extracted in particular contexts. These arguments ultimately lead back to the disputes between R. A. Fisher and Sewall Wright as the power of selection to optimize fitness in populations. While Fisher had a rather elegant but simple model in mind of large populations scaling single peaks in a landscape of spare topography, Wright contended that nature was truly characterized by rugged adaptive landscapes where gene-gene interaction and drift operated as essential parameters within and across subpopulations. Gould agrees with Wright, and in fact he suggests that Wright deemphasized the role of stochasticity in his model as the Modern Neo-Darwinian Synthesis crystallized to conform to the orthodoxy as promoted by Ernst Mayr.

But ultimately the aim here is not just to change the weightings of various evolutionary genetic parameters as forces on natural historical temporal scales, rather, Gould also wishes to expand the analysis and applicability of these parameters up to the level of species. Much of the chapter is dominated by the three way conversation between Gould, Dawkins and Williams, with the latter being fellow travelers. Gould argues that Dawkins and Williams improperly privilege the gene, the replicator, as the causal agent of natural selection. In contrast, Gould asserts that interactors, often the vehicles in which genes reside, are the keys to the kingdom. Interactors may be the organism's individual body, or, it may be a deme, or it may be a species. Conversely, it may also be a lower level of organization below the individual, such as the gene! In the domain of selfish genetic elements the gene is both the replicator and the interactor; in other words, here the unit of selection is the gene without any mediating effect of higher level interactors.

Of these interactors Gould's focus as a paleontologist is at the level of the species; he believes that this taxonomical level is critical to understanding the shape of natural history. Here he observes that multi-level selection theorists such as David Sloan Wilson have often put the spotlight on the deme, the population substructure within a species, but extrapolating up to the level of a species the dynamics are qualitatively different and deme-level theories may mislead. For example, one of the major problems with higher-than-individual-level selection models is that the power of selection on each ascending level is generally proportional to parameters which many perceive to be unrealistic; e.g., demes reproduce their community-level traits slowly vs. the level of the individual, and, they tend to eliminate extant genetic variation through interbreeding. On the level of the species Gould notes that interbreeding is not an issue, species are units of selection which don't fall prey to the problem of between population migration which removes the variation selection needs to operate. The problems and strengths of deme-level selection may not apply to higher levels, just as it does not to lower levels.

As to the specific nature of selection, there is a deal deal of attention paid to what Gould term's the "Lloyd-Vrba debate," in reference to Elizabeth Lloyd and Elisabeth Vrba. The latter is a well known collaborator of Gould's, while he published several papers with the former in the 1990s. Overall Gould sides with Lloyd, though he asserts that Vrba's model is simply a subset of Lloyd's, so still valid though superseded in terms of generality. Basically, Vrba focused on the importance of emergent species level functionalities as the targets of selection at that level. Lower level effects which welled up in the aggregate, i.e., the sum of the parts driven selection, did not qualify as it could be reduced to individual level selection. Lloyd adds the twist that what is critical is not some functional characteristic, but the emergent property of species level fitness. In other words, the trait does not need to be an emergent property from below, but its fitness has to be a residual above the level of the individual and deme so that species level effects are a necessary explanation.

There's a lot there, and I'm not really going to repeat everything that is asserted or outlined (e.g., I'm not going to rehash Gould's repeated barbs at the fixation on reductionism which he believes too many biologists have via their physics envy). I have to focus on some stylistic issues here at this point. I've complained ad nauseam about Gould's inability to control his prose excesses. Most people who have purchased or checked out this book simply stop because of its unreadability. I have heard that Gould insisted that there be no editor for this book, his magnum opus. I will offer here that only a trivial amount of effort on the part of an editor would have been necessary to cut the length of the text in half. Gould has a habit of making an assertion, and then supporting it with 10 examples. Then he moves onto another topic, and gives 10 examples. At which point he jumps back to the previous point, and gives another 5 examples. And so on. For lay persons 5 examples vs. 50 makes little difference; the sample space is large enough that a clever scholar can easily load the die. For specialists the strength of the argument is not changed by piling more on; they agree, or they don't. Much of the length of The Structure of Evolutionary Theory can be attributed to the fact that it is basically a core dump. This isn't necessary, and in fact it is probably counterproductive to the communication of science and the elucidation of the ideas which Gould wished to promote. Finally, there is a tendency here in this chapter for Gould to be the final arbiter in scientific disputes. He can do this, this is his book, there's no editor or any peer review. Most perceptive readers will be aware that Gould is speaking as Stephen Jay Gould, not as the consensus of science, but not all will. It is important not to forget that in this arena Gould has total control of his interlocutors, Richard Dawkins and George Williams. They seem silly, wrong-headed, stubborn and plain muddle-headed if you accept Gould's logic. On the other hand, the same has been said of Gould. This is a classic case where the best corrective for speech is more speech. In hindsight I think it would probably be useful to read The Selfish Gene before embarking on Part II of The Structure of Evolutionary Theory; at a little over 300 pages of prose it's a much faster read than one 100 page chapter of Structure.