Thursday, August 26, 2010

Down with Kin Selection?

Oecophylla weaver ants in Gombe. Photo by Axel Rouvin via Wikimedia Commons.

The concepts of kin selection and inclusive fitness are centerpieces of modern evolutionary biology. Almost every biologist has some familiarity with at least the outline of the issue, and they are staples in textbooks and in the classroom. Briefly, these ideas were introduced in the mid-20th century to explain the apparent paradox of how altruism and cooperation might evolve in a biological system. If natural selection favors the spread and fixation of genes that cause an individual to have more offspring, how can we explain the evolution of traits which involve one individual aiding another at some cost to itself?

Cooperation might seem to be intuitively beneficial, but consider a population of animals which all sacrifice a bit of their potential individual fitness in order to cooperate for some greater good, e.g. defending the nest from predators. They will get good returns on their investment if they all cooperated honestly. However, such a group is prone to invasion by so-called 'cheaters', which take but do not give back. A group of cooperators without defences against cheating will rapidly be overtaken by cheaters, and hence any cooperative endeavour is doomed to crumble.

It was to resolve theoretical problems like these that the idea of inclusive fitness was introduced. Individual fitness is simply the number of offspring that one produces. Inclusive fitness, on the other hand, counts not just one's own offspring, but the offspring of relatives, multiplied by the factor of relatedness. The process by which it functions is kin selection - if one helps one's relatives and contributes to their individual fitness, then one is contributing to one's inclusive fitness. A gene for cooperation could hence spread and be resistant to cheating, because it is more likely to be present in relatives of an individual that already bears that gene, than in unrelated individuals. The essence of the idea is found in a prescient quip by the geneticist JBS Haldane, who said that he "would [lay down his life] to save two brothers or eight cousins" (Wikiquote) - because the relatedness (more rigorously, the probability that a gene is identical by descent) of siblings is 1/2 and that of cousins is 1/8.

A new look at the math (the derivations are in the supplementary material, but the main paper is unfortunately hidden behind a paywall) behind inclusive fitness theory is now claiming that it is no better than standard natural selection theory in explaining one of the key problems of biology, the evolution of eusociality (edit 28/8/2010: Harvard Gazette press release). The group, comprising mathematical biologists Martin Nowak and Corina Tarnita, and the father of sociobiology EO Wilson, explain that inclusive fitness is merely a different way of 'doing the accounting' which is somewhat more complicated than it needs to be. That is, it is too specific, relying on certain assumptions (additive fitness effects, pairwise interactions only, weak selection) which are fulfilled in only a few exceptional cases. It is not an adequately general theory, in the mathematical sense of the word. They also challenge the validity of Hamilton's rule, the simple-looking equation that states that a gene for a cooperative behavior will spread if:

Relatedness > cost/benefit

Population biologists trying to test Hamilton's rule in actual organisms have found it difficult to measure these quantities. The authors of this paper, using a more explicit derivation, found that cooperation does spread when something is greater than the cost to benefit ratio, but this 'something' (as they put it) is not relatedness. In place of inclusive fitness theory, they propose that the more general game-theoretic theory of natural selection, used in conjunction with precise models of population structure, is adequate to explain the evolution of cooperation, without invoking kin selection.

What then, of the crown jewel of kin selection, the explanation of eusociality in hymenopterans? Hymenopterans are the insects which include bees, wasps, and ants. They have an unusual genetic system called haplodiploidy: females lay either fertilized (diploid) or unfertilized (haploid) eggs. The former hatch into females, and the latter into males. The ants are perhaps the most successful and best known eusocial animals. Kin selection theory explains the cooperation between female, sterile worker ants (who might easily defect and start laying eggs of their own) in terms of their relatedness - sisters are more closely related to each other (relatedness of 3/4) than daughters are to mothers (1/2), and so cooperation among sisters can spread. Not all eusocial animals have haplodiploid sex determination - termites, for example, do not. As more eusocial animals have been found, it turns out that the number of haplodiploid lineages is in the minority. An alternative explanation has to be found for eusociality, because the kin-selection based explanation as given above was now most probably an exception to the rule.

Naked mole rats are eusocial mammals. Photo via Wikimedia Commons.

Ed Wilson has proposed for some time now that eusociality is to be explained by theories other than the traditional haplodiploid hypothesis (in Quarterly Rev. Biol., Bioscience (pdf), PNAS) . Briefly summarizing, his model, which is adopted and refined in this paper, involves the formation of groups subject to selection, the predisposition to group-formation by certain 'preadaptations' (they cite the example of solitary bees being behaviorally programmed to complete tasks in sequence - hence they naturally divide labor when forced to cohabit), and selection acting at multiple levels. Hence organismal traits and population structure are sufficient, without having to invoke the concept of inclusive fitness.

This analysis is quite satisfying because it represents the convergence of two different approaches to population biology: that rooted in mathematical theory and that rooted in natural history. The strong claims that it makes will certainly trigger robust debate. However it turns out, it will be interesting to see how the study of eusociality will respond. As the authors note (quite provocatively, given the highly charged reception to the original publication of Sociobiology in the 1970s), "[w]e have not addressed the evolution of human social behavior here, but parallels with the scenarios of animal eusocial evolution exist, and they are, we believe, well worth examining."

No comments: