Evolution for Everyone

The death of George C. Williams last September coincided with an article that Elliott Sober and I were writing on the significance of his work. The article, titled “Adaptation and Natural Selection Revisited” has now been published in the current issue of the Journal of Evolutionary Biology and is dedicated to the memory of George. In this post I will explain why George made such a lasting contribution to evolutionary thought and why his legacy, properly understood, deserves to continue.

George is widely credited for clarifying the concepts of adaptation and natural selection, which are at the center of evolutionary thought. He said that adaptation is an onerous concept that should only be invoked when certain conditions are met. Consider the example of flying fish that launch themselves out of the water and glide for long distances with their wing-like fins before returning to the sea. Their wing-like fins are an adaptation that evolved by natural selection. Returning to the sea is not an adaptation. That’s what gravity does to all objects and nothing needs to evolve to cause it to happen.

As part of his general examination of adaptation and natural selection, George turned his critical gaze on the idea that traits evolve “for the good of the group”, which was commonly invoked at the time. In one of his most famous examples, he made a distinction between “a herd of fleet deer” and “a fleet herd of deer”. In the first case, predators give a survival advantage to fast deer compared to slow deer in a single herd. This is natural selection at the individual level. The herd as a whole becomes fleet as a result, but this is not relevant to its evolution. If we want to say that natural selection can act for the good of the group, producing “a fleet herd of deer” rather than merely “a herd of fleet deer”, we need to go beyond natural selection operating at the individual level.

For an unambiguous example of a trait that would quality as “for the good of the group”, consider the fabled lemmings that were reputed to commit suicide by running into the sea when their numbers became too high to avoid overexploiting their food supply. This example differs from the previous example because suicidal lemmings have lower survival than their non-suicidal neighbors, no matter how hungry they both are. For suicide to evolve as an adaptation to conserve resources, there must be another layer of natural selection. There must be many groups of lemmings. The groups must vary in their propensity to commit suicide under conditions of high density. If these conditions are met, then the extinction of groups that fail to conserve their resources might cause the suicidal trait to evolve, despite its selective disadvantage within groups. According to George, that would indeed qualify as an example of a group-level adaptation that evolves for the good of the group, in contrast to the previous example.

In our article, Elliott and I encapsulate George’s thinking on this topic and give it a name:

Williams’ Principle: Adaptation at a level requires that there was selection at that level.

Williams’ Principle can be elegantly applied to any level of the biological hierarchy. Consider examples such a meiotic drive and cancer, whereby genes spread at the expense of other genes within the same individual organism, often to the detriment of the whole that is fatal in the case of cancer. Williams’ Principle enables us to identify these traits as adaptive at the gene level (genes bearing the trait are more fit compared to alternative genes within the same organism not bearing the trait), but maladaptive at the individual level (individual organisms bearing the trait are less fit than individual organisms not bearing the trait). The features of this example are identical to the lemming example, frame-shifted down the biological hierarchy to genes in individuals rather than individuals in groups. We can also frame-shift up a multi-tier hierarchy to consider groups within groups within groups.

Or consider the genetic basis of a straightforward individual-level adaptation, such as a gene that causes deer to run faster without any other effects. This gene is not more fit than other genes within the same deer, since all benefit to the same degree. We must move up the biological hierarchy to locate the level at which the adaptation evolves (the differential fitness of individual deer), which in this case is unopposed by selection below the level of the individual. Williams’ Principle reflects the fact that natural selection is fundamentally about differences in fitness, and that we must know where the differences exist in a multi-tier hierarchy to identify what qualifies as an adaptation at any particular level.

In our opinion, this is the most enduring part of George’s legacy, which should be taught to everyone learning about evolution. George earned a permanent place in history not for showing that group-level adaptations never evolve, but for showing how to make the judgment call.

To be continued…


  1. #1 Andrew
    January 15, 2011

    I’m guessing the lemming example is hypothetical… what are some real-world examples of this where such selection is clearly group level? Groups of bacteria in different petri dishes? Strains of decorative plants?

  2. #2 Greg Laden
    January 15, 2011

    OK, let me play devils advocate, as it were.

    1) William’s principle is a tautology. Adaptation is the noun, selection the verb. They are essentially the same thing. Ok, so maybe that’s not a tautology, but it is trivial.

    2) The cellular level is not the same as the group level because cells don’t actually leave the body and mess with the population genetics of other bodies, while individual lemmings can do so with other groups. (This, as you are totally aware, of course, is the classic argument against group selection … groups are genetically leaky).

    3) Damn, I had a third point but this cold medicine is making my brain lose things.

    Andrew: Yes, that’s a hypothetical as far as I know. Lemmings swim. When they are very numerous they may leave one of these little islands some lemming live on in the arctic/subarctic to disperse. A very risky strategy but one that can pay off.

    I’ve always wanted to find a big lake with a swimming-rodent producing island or two and develop an appropriate lure for the pike and muskie that must gorge on them. (Most mouse lures I’ve seen look like house mice, which live in houses and do not disperse across water!).

    OK, back to my cold medicine induced coma … Looking forward to part II.

  3. #3 darwinsdog
    January 15, 2011

    The pendulum of opinion swings.. and will swing back. Unfortunately, George is no longer with us to out-shine the lesser lights. Or, in the dimness of his extinguished light, the roaches come out of the woodwork.

  4. #4 Greg Laden
    January 15, 2011

    Darwinsdog, it’s not THAT bad. He trained students, they trained students, and even where he wasn’t busy training students his influence, in particular, his insistence that adaptation is demonstrated by cost and other key features of his approach, are widely held.

  5. #5 Gerdien
    January 17, 2011

    Adaptation is the noun, selection the verb
    That does not make sense. Selection is a mathemathical relation between variances and covariances in fitness and traits, and adaptation is a statement on how we perceive the function of a trait.

  6. #6 Greg Laden
    January 17, 2011

    Gerdian, let the phrase inform you, assume for a moment that it does make sense, and see how it fits. Because it does.

New comments have been disabled.