How to study adaptation the proper way

One of the recurring discussions in evolutionary biology is the question of how to test adaptive hypotheses. Creationists often claim (wrongly) that the theory of natural selection is tautological, and so did, at some point, prominent philosopher of science Karl Popper (he later acknowledged that his comment was based on a misunderstanding of the theory, something that creationists strangely neglect to mention whenever they quote him in their favor — see this short essay I wrote about it for Skeptical Inquirer). And then, of course, there is the never ending issue of the quasi-scientific status of evolutionary psychology, precisely because adaptive hypotheses are particularly hard (though not impossible) to test in the case of Homo sapiens (more on this below).

It was therefore with delight that I recently happen to stumble on a paper by Mark Olson and Alfonso Arroyo-Santos entitled “How to study adaptation and why to do it that way: adaptation, just so stories, and circularity,” published in the prestigious Quarterly Review of Biology (pdf freely available here).

The basic argument put forth by Olson and Arroyo-Santos is that “circularity” comes in degrees, from vicious to virtuous, which is why they prefer the term “loopiness,” as in feedback loops between hypotheses and empirical evidence. Perhaps the best way to get what they are saying is to examine a few of their figures.

This one is a summary of the standard adaptationist story about why giraffes have long necks:

If you start with the observed pattern (lower portion of the figure), you will agree that, observationally, modern giraffes have long necks and they feed from tall trees. Based on that, you infer (upper portion of the figure) that the following things were true of the ancestors of modern giraffes: (i) they varied in neck length; (ii) this variation was heritable; (iii) short-neck giraffes had less access to food (functional argument); and (iv) short-neck giraffes, as a consequence of their phenotype, had lower fitness (adaptive argument).

The problem is, up to this point you have absolutely no evidence for any of inferences (i)-(iv) above. Which truly does make your “explanation” a just-so story.

But wait!, say — correctly — Olson and Arroyo-Santos, this can be the beginning of a fecund research program, because you can turn (i)-(iv) into hypotheses to be directly or indirectly tested.

For instance, evidence for (i) may be found in the fossil record; evidence for (ii) in extant populations (which would license the inference, though obviously not prove, that this was the case also in the past); (iii) may be tested directly in extant populations, or indirectly via optimization models; and (iv) is a reasonable conclusion from (i)-(iii), and therefore acceptable in direct proportion to the strength of the evidence in favor of (i)-(iii).

If you ponder the above for a minute you will realize why this shift from vicious circularity to virtuous loopiness is particularly hard to come by in the case of our species, and therefore why evolutionary psychology is, in my book, a quasi-science. Most human behaviors of interest to evolutionary psychologists do not leave fossil records (i); we can estimate their heritability (ii) in only what is called the “broad” sense, but the “narrow” one would be better (see here); while it is possible to link human behaviors with fitness in a modern environment (iii), the point is often made that our ancestral environment, both physical and especially social, was radically different from the current one (which is not the case for giraffes and lots of other organisms); therefore to make inferences about adaptation (iv) is to, say the least, problematic. Evopsych has a tendency to get stuck near the vicious circularity end of Olson and Arroyo-Santos’ continuum.

Here, by contrast, is an example of what virtuous loopiness looks like, in the case of the observation that swimming animals tend to have streamlined bodies, regardless of their taxon and phylogenetic relatedness:

You can appreciate the much more clear distinction between the initial observation of a pattern (lower right), the development of a number of assumptions necessary to explain the pattern (upper right), and three types of empirical evidence that feed into the other two components (left side of the figure).

Let us take a closer look at that last bit. Olson and Arroyo-Santos argue, again, correctly, that there are fundamentally three types of evidence that bear upon any adaptive hypothesis: from the comparative method, from population biology, and from optimality considerations.

The comparative method has to do with contrasting large number of species with different degrees of phylogenetic relatedness. The methodology has been fine tuned decades ago, including the development of sophisticated statistical approaches and the use of computer simulations to assess how well different methods of phylogenetic reconstruction work. This is another area where evopsych suffers, however, since there are very few and widely scattered “close” relatives of our species (no other Homo surviving, a couple of species of chimpanzees and the gorillas, separated from us by several million years of independent evolution).

Population biology is a very broad field at the interface among ecology, evolutionary biology and genetics, again characterized by well honed methods that rely on basic population and quantitative genetic theory. This is where our estimates of population variation (and hence also heritability) come from.

Optimality modeling has to do with applying biomechanical considerations to the study of adaptation, developing quantitative assessments of how well certain biological structures — and their variants — perform their function in response to pertinent environmental stimuli. In the case of the streamlined bodies of fish, marine mammals, and some invertebrates like squids, the pertinent theory comes from hydrodynamics.

The third figure, also from the Olson and Arroyo-Santos paper, explains how the loopiness of adaptive explanations works in the recurring case of discussions about the relative roles of natural selection and developmental constraints:

These two classes of explanations of biological structures are too often, even by biologists, considered to be mutually exclusive, and indeed sort of divide evolutionary biologists from developmental biologists, as well as “externalists” (who favor external causes, like natural selection) from “internalists” (also known as structuralists, who favor internal causes, like developmental constraints).

But Olson and Arroyo-Santos rightly point out that this is yet another case of loopiness: in the figure, they show a typical scattergram that hints at a strong linear correlation between two traits, X and Y (say, body size and antler size in the famous “Irish elk,” see figure at top of post). The typical question is: why do the data points line out that way within the phenotypic space defined by the two variables? Why the empty spaces of unrealized phenotypes (say, of large bodied elks with small antlers, or small bodied ones with very large antlers)?

Each of the two alternative explanations, that the empty spaces are explained by the fact that some morphologies are maladaptive, or that they are developmentally inaccessible (i.e., there is no way for the organism in question to build that phenotype, given its genetic-developmental matrix) are testable by stretching the initial circularity into the same sort of loopiness we’ve seen before. Moreover, the two are not mutually exclusive, since developmental constraints can limit, but not nullify the action of natural selection, and selection in turn may favor a genetic-developmental matrix that precludes the accidental invasion of maladaptive areas of phenotypic space. There are, in fact, many examples in the literature of this sort of analysis, and they represent some of the best achievements of modern evolutionary biology.

Here is how Olson and Arroyo-Santos then conclude their excellent paper:

“Studies of adaptation would necessarily seem to require the sort of loopy reasoning [presented in this paper]. Recognizing how adaptationist explanations are structured in actual practice helps give clarity to problems that have plagued biology, such as debates over tautology/circularity, and resolve false conflicts, such as the mutual scorn that often characterizes the adherents of the comparative, population/quantitative genetics, and optimality approaches. Instead, as providers of complementary sources of direct evidence, no single approach has a monopoly on tests of adaptation. An understanding of the real, loopy structure of evolutionary explanation encourages biologists to discuss truly substantial issues awaiting attention, such as how to identify the population of hypotheses from which to select the ‘best’ explanation, how scientists know the best explanation when they see it, or how best to weave disparate sources of evidence into a single explanation. By accepting that studies of adaptation require multiples types of direct evidence, evolutionary biologists can improve current research practice by designing a compelling and long-overdue integration of comparative, populational, and optimality approaches.”