On Group Selection

A reader picked up on my parenthesised comment on EO Wilson’s kin selection paper in this post, which led to a substantial e-mail conversation on group selection, which I will distill here to show you that the noise against group selection doesn’t hold much water anymore.

Explaining the existence of altruism has always been one of the central problems of evolution. Altruism is defined as an action that is negative to the actor, and positive for a recipient. Why would an individual organism do something like that? Until the 1960s, group selection – the idea that natural selection can act to benefit groups – was viewed as an ideal candidate, since the positive effect for the group overpowers the negative effect on the individual, leading to that group outcompeting other groups where altruism does not exist. A “group” is nowadays best defined as a population of intercting organisms which affect each other’s fitnesses (Sober & Wilson, 1994).

The staunch opposition to and denial of group selection was popularised in the 1960s with George C. Williams‘s 1966 seminal book, Adaptation and Natural Selection. Among his evidence was that sex ratios in animals are fairly even, whereas group selection would predict biased sex ratios. The entire book consists of him calculating that “group-related adaptations do not, in fact, exist”, because its power would be too weak to influence the power of natural selection on the individual level, since individuals have a much faster turnover rate than groups, thus evolution happens more quickly in them.

Maynard Smith‘s work complemented and expanded on these conlusions (Maynard Smith, 1964, 1976), and he even came up with another alternative to group selection, evolutionary game theory, expounded in his 1982 book Evolution and the Theory of Games. Economists will be aware of game theory; this is the same thing. Behaviour is modelled by seeing the interactions of different individuals. If two altruists interact, the result is positive. If two selfishs interact, the result is negative. If an altruist and a selfish interact, the result is positive for the selfish, negative for the altruist. Therefore, altruism evolves only when altruistic interactions are much more common than selfish interactions. The role of other groups is irrelevant, selection takes place within the group only.

But since then, a lot of research and progress has been made, and group selection has been shown to be possible and even likely, and in my opinion, it is no longer valid to demonise group selection.

I mentioned in my natural selection lecture the classic demonstration of group selection in lab-reared beetles by Wade (1977), which proves that group selection can in fact be a dominant evolutionary force. A valid criticism would be to say that evolution in the lab is too artificial and not reflective of evolutionary forces in the wild.

But in the wild, group selection has also emerged as a favoured explanation for some scenarios. See Heinsohn & Packer (1995) for an example with lions and their territorial defence. The authors set the problem effectively: “If too few females accept the responsibilities of leadership, the territory will be lost. If enough females cooperate to defend the range, their territory is maintained, but their collective effort is vulnerable to abuse by their companions. Leaders do not gain additional benefits from leading, but they do provide an opportunity for laggards to gain a free ride”. The experiments involved changing the conditions slightly, and the diversity and complexity of reactions revealed that just invoking individual selection just wasn’t enough to explain everything; group selection fills the gap.

Even theory and modelling, what initially sounded the death knell for group selection, has recently been turned around to make it a plausible theory again. Models from the 1970s, such as Levin & Kilmer (1974), while adequate for their time, can’t compete with the much more powerful simulations and models computers are capable of today. When a larger multitude of factors is taken into account, group selection does come out as likely. As an example, check out Werfel & Bar-Yam (2004), whose model finds that a reduction in fertility might evolve to avoid resource overconsumption, echoing a classic group selectionist argument (see Borrello’s excellent book, Evolutionary Restraints: The Contentious History of Group Selection).

The opposition to group selection simply does not hold up anymore, since it’s now acknowledged that multilevel selection is the most valid way to look at natural selection, as has been realised since the 1970s. In the 1960s, WD Hamilton developed inclusive fitness theory, now better known as kin selection (Hamilton, 1964a, 1964b), which mathematised the advantage of parental care as an individualistic advantage borne out of the positive effect of making sure relatives have a high fitness, since they carry the same genes. This gave birth to the well-known Hamilton’s Rule: rb > c. Altruism can evolve only when genetic relatedness (r) and the benefit (b) are greater than the cost (c). This would explain why eusocial insect colonies, like ants, are all made up of sisters and daughters. This was initially seen as a futher blow to group selection.

However, in the 1970s, Hamilton combined his inclusive fitness theory with the Price equation and noticed that altruistic traits by themselves are disadvantageous, and only become advantageous when dominant in a population. In other words, an individual altruist is useless, but put many altruists together and the group emerges as a stronger collective than a group of selfish individuals. From such work emerged a new group selection; see Price (1970). Kin selection fits very snuggly within this new group selection. They are not contradictory, but complementary, not the least because the concepts of benefits and costs in both theories differ. Kin selection talks about absolute costs and absolute benefits, whereas group selection talks about relative ones (relative between groups).

This is what should get accepted, either implicitly or explicitly. Group selection is still viewed as a bit of a taboo in some circles, but the thinking behind it as part of a multilevel selection framework is solid and few would have a problem with it, since it has evidence to back it up from all sides. What one needs to do is merely acknowledge that absolutism is wrong: not every social organism evolves exclusively by group selection or by kin selection or by individual selection. Not all adaptations that benefit the group evolved by group selection, and group selection is by itself not sufficient for the evolution of group adaptations. There are multiple causes, and all levels are interlinked. It was failure to recognise the multidimensionality of the problem that resulted in the group selection controversy we all know and hate. In philosophical terms, explanatory pluralism is the way to go, recognising that there can be multiple explanations for the same set of facts depending on the focus of your research (this is also a point I stressed in my natural selection lecture).


Borrello. 2012. Evolutionary Restraints: The Contentious History of Group Selection.

Hamilton WD. 1964a. The genetical evolution of social behaviour. I. Journal of Theoretical Biology 7, 1-16.

Hamilton WD. 1964b. The genetical evolution of social behaviour. II. Journal of Theoretical Biology 7, 17-52.

Heinsohn R & Packer C. 1995. Complex cooperative strategies in group-territorial African lions. Science 269, 1260-1262.

Maynard Smith J. 1964. Group Selection and Kin Selection. Nature 201, 1145-1147.

Maynard Smith J. 1976. Group Selection. The Quarterly Review of Biology 51, 277-283.

Maynard Smith J. 1982. Evolution and the Theory of Games.

Price G. 1970. Selection and Covariance. Nature 227, 520-521.

Sober E & Wilson DS. 1994. A Critical Review of Philosophical Work on the Units of Selection Problem. Philosophy of Science 61, 534-555.

Wade MJ. 1977. An Experimental Study of Group Selection. Evolution 31, 134-153.

Werfel J & Bar-Yam Y. 2004. The evolution of reproductive restraint through social communication. PNAS 101, 11019-11024.

Williams GC. 1966. Adaptation and Natural Selection.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.