Notes on novelty 5: Evolutionary radiations and individuation

Notes on Novelty series:
1. Introduction
2. Historical considerations – before and after evolution
3: The meaning of evolutionary novelty
4: Examples – the beetle’s horns and the turtle’s shell
5: Evolutionary radiations and individuation
6: Levels of description
7: Surprise!
8: Conclusion – Post evo-devo

Sometime around 1900, Henry Fairfield Osborn (the Ernst Mayr of his day), appropriated the term “adaptive radiation” from botanical writers a few years earlier and published “The Law of Adaptive Radiation”, in which he argued that we could not only identify these radiative events, but the centres of their origin. The notion of an “evolutionary radiation” developed soon after, and by the time of the modern synthesis, the notion of adaptive radiations, adaptive zones, and adaptive niches had become firmly established in the lexicon of evolution. Classifications in terms of these ideas also became popular, leading to the eventual and inevitable (and somewhat absurd) suggestion by Julian Huxley in 1957 that since humans alone occupied the “adaptive zone” of intelligence, they should be classified in an entirely distinct major grade on a par with Metazoa, Psychozoa. Huxley explicitly discussed these in terms of grades.

This gradism is rife among those who adopt evolutionary stances; indeed it has almost become doctrine. And, despite the objections of some cladists, there is nothing wrong with developing and testing hypotheses of grade novelties; the problem arises when clades and grades are forced into hybrid classifications. On the Arthurian definition of novelty, every differentiae that give a clade, no matter how subjectively big or small, is an evolutionary novelty. Novelty is effectively just speciation (or even lower, if haplotype groups and other coalescent trees are acceptable). The evolutionary (in the sense of the modern synthetic account) definition, however, mixes the qualitative criteria of “grade” with the quantitative criteria of apomorphies. The inferential warrant of these hybrid classifications is highly limited.

Effectively, grades are statements that have two places for every predicate: the properties observed, and the criteria salient to the observer. A grade statement is a statement about the observers as well as the organisms. A clade statement, however, is held not to be. It takes the properties that are measurable by any observer, in theory at any rate, and processes the differences more or less algorithmically. It doesn’t matter if the characters being measured were, before the observation, specified as being “significant” or “important” or “novel”; so long as they can reasonably be held to be homologies between the groups. A novelty in this case is a trait (a character state, or roughly the value assigned to the character) that has no homologs in the compared groups.

So individuating the characters/traits is critical to classifying novelties. As the beetle case shows, what looks like it was novel (without homology) at the level of description of the gross phenotype (carapace and mating behaviour), turns out to be a co-option of prior genetic and developmental mechanisms. The homology is at a deeper level of description. Individuation of traits is a fraught topic in biology. Gould and Lewontin accused reductionist evolutionists of “inappropriate atomization” of traits (1979: 585), declaring that organisms are “integrated entities, not collections of discrete objects”. Often the individuation of parts depends on what the assay techniques are: visual inspection, fossilisation, molecular assays, cytological staining, and so forth. Atomisation appears to occur based on what techniques for gathering differentiae are used. Even the notion of individual organisms has been challenged, in part by “extended phenotype” views and in part by endosymbiosis and gastrointestinal florae. As George Williams (1992: 9) once noted, holists and reductionists are largely separated by the size of the glassware they use – test tubes and petri dishes (I would add slides and pipettes)  for reductionists, and terraria, aquaria, and specimen bottles for holists.

To call somebody a reductionist is often thought to be an insult or a dismissal of a simplistic and atomistic philosophy that Everybody Knows has failed. Explanation is supposed to be targeted to the Whole, the organism, or the ecosystem, or even systems in general. And yet, studies like the ones Shubin et al. cite show that the successful approach of reduction (looking for underlying mechanisms) in explanation is not only not dead, but solving problems that have bothered us for some time. However, this doesn’t mean one can leap to the conclusion that reductive explanations will always work, or even that we must be reductive when we can. If an explanation works at a particular level of description, there is no need to reduce it further. I might be able to explain the traffic patterns of Sydney through the use of very general considerations in which the properties of the individual cars are irrelevant to the explanandum of flow rates and snarls. It adds nothing to the understanding to point out that 22% of all sedans are red, or that European cars account for 15% of all traffic, if the relevant properties are indiscernibly different (neither red cars nor European cars go appreciably faster in urban traffic).

So if we can give an explanation at a level of description that satisfies our epistemic and explanatory needs, the homologies of that level of description are enough. However, there will inevitably be other epistemic and explanatory needs not met by a gross description. For example, if I can state with any evidence and authority that of two traits in a given population of organisms in an ecotype, one is fitter and will move to fixation, I have explained the dominance of that trait and made a prediction. However, at that level I am none the wiser about the physiology, development and genetics of that trait. Since these are different explanatory projects to the question of populational distributions, that doesn’t matter in one sense. But it matters to the physiologists, developmental biologists (who really need a better name; might I suggest ontogenists?) and molecular biologists, and so they give their finer grain descriptions in setting up their own research projects.

The question of novelty crosses these grains of description. Take a novel structure like the turtle’s shell. It is novel because we describe the gross morphology of shell, plastron etc in terms we do not apply to near relatives. It is novel because at the level of description of gross adult anatomy, we have an inversion of topological configurations found in relatives. That sets up the problem; the solution is found by appealing to finer grain accounts of development, etc. This leads us to conclude that finer grain accounts should be appealed to in all cases, if we were to be complete about our biology; but everybody knows there is only so many research hours and dollars, and so we leave the details to one side except when there is a problem like this.

So next we need to consider what a level of description is, and how it is chosen.

Gould, Stephen Jay, and Richard C. Lewontin. 1979. The spandrels of San Marco and the Panglossian paradigm: a critique of the adaptationist programme. Proc R Soc Lond B 205:581–598.

Huxley, Julian. 1957. New bottles for new wine: essays. London: Chatto & Windus.

Williams, George C. 1992. Natural selection: domains, levels, and challenges. New York: Oxford University Press.