A question about group selection

I’m doing some work on evolutionary models of game theory and need to understand the debate about group selection. It seems pretty clear that the great majority of evolutionary biologists reject the idea of group selection, but I haven’t found an adequate (to me) explanation of why they do so. A crucial problem for me is that the literature seems, without exception as far as I can see, to conflate group selection with co-operation and altruism. But the problem of group selection arises in non-cooperative settings, provided they are not zero-sum.

To illustrate the problem I’m struggling with, suppose that two previously isolated species meet as a result of some change. In one species (peacocks), competition between males for mates takes the form of elaborate, and energetically costly, displays. In the other species (penguins) males compete by providing food to their mates. In all other respects (diet, predators and so on) the two are similar. It seems obvious to me that the penguins, with their more efficient social arrangements, are going to outbreed the peacocks and eventually drive them to extinction.

It seems to me there are only two possibilities here
(a) My reasoning is wrong, and we can’t judge which species, if either, will dominate; or
(b) Even though it involves one group being selected over another, this isn’t what is meant by group selection

I’d really appreciate some help on this. I’m happy to have thoughts from anyone, but I’d most like to hear from actual experts with contact details.

73 thoughts on “A question about group selection

  1. I haven’t yet seen a clear definition of what “group selection” is supposed to be. I haven’t seen clear definitions of replicators, interactors, “vehicles” and groups. Nor have I seen a clear statement of which of the above is the “real unit of selection”. Until an advocate of group selection can clearly define all of the above (and/or rule some out as meaningless or redundant) then the discussion cannot even begin.

    Herding animals are an interesting case. Is the gene for herding (assuming there is one) sensibly regarded as an individual possession or a group possession? I have no idea how to answer that.

    One researcher has written;

    “Try this new line of thought for a moment: what if we dropped the idea of group selection and instead discussed interaction selection? This would be a study of how the interactions between two or more individuals, between two or more groups, or between two or more species resulted in differential fitnesses of organisms or genes. In discussing interaction selection, we would transcend strict essentialist concepts such as individual selection and group selection, and reach a more pure form of analysis in which the process is emphasized rather than the unit of selection.” – Rebecca Beroukhim

    This contains the germ of a very interesting idea I think. From the point of view of complex systems theory it seems to make a great deal of sense.

  2. @Ikonoclast

    Frustrating, isn’t it. Re your first paragraph, I seem to remember saying very similar things in the thread on the ASIO guy supposedly “ringing around to politicians” re IS and Islam. What is this “normal process” I asked, and how many pollies did he ring, and what did he say to each of them. I never did get an answer, and you’re not going to get one either.

    And you’re absolutely right: without at least some answers, there’s nothing to discuss.

    Never mind, mate, I put it down to bio-quantum entanglement, myself: whatever happens to a dominant individual also somehow affects the group (if only we could know what group is).

  3. @James Wimberley
    ‘ other animals do not share food and other resources like ‘ [humans do] .

    Tho we are certainly very good sharers ,I think it would be hard to define sharing so that there are there are no examples of it in the non-human animal world. The urge to do so is more interesting to me than the definition itself. Rick Dawkins says there is almost no basic kind of human societal structure that cant be found in the ant world. On that a-bit-off-topic line of thought ,although we currently direct massive resources our way, wouldnt the champion groups on long time scales have to be the insects ? There are so many species we cant even count them quicker than they mutate into new species .Our knowledge will never completely envelope them.

    Not having done any supplementary reading, I dont think I have a good handle on the group v’s individual selection debate . As Ronald Brack points out above -it seems that by virtue of the definition of terms ,genetic group change will always be traceable back to genetic change of an individual . However there does seem to be lots of recent evidence showing that expression of genes can be affected by environment and environmental effects on parents can be passed several generations down. There is a continuous complex two way interaction between genes and the world. Is there no way group action could change genes ? Is that what Prof Q is looking for? And wouldnt people then just say ,as Ronald pointed out, ‘well that group action must just be the result of genes anyway’ .As so often happens I suspect the answer may lie in better defining the question .

  4. I can’t understand the debate either. It seems a case of science advancing with the death of each scientist.

    I think the problem might be conceptual tension, as in ‘By discarding inclusive fitness on the basis of its limitations, they [Nowak et al] create a conceptual tension which, we argue, is unnecessary, and potentially dangerous for evolutionary biology.” From Ferriere & Michod Nature 471, E6–E8 (24 March 2011)

    BTW, your example is incorrect. Penguins outcompeting (outcooperating) peacocks is standard evolution. Group selection would be where one group of peacocks have a gene that makes them more penguin-like, grow smaller tails, spend less energy in intra-group competition and collectively prosper relative to the peacock muster across the valley. The problem is working out what would cause the cooperation gene to propagate from an individual mutation into the group if the benefit is applied to all individuals in the group. It may be a smart idea but that isn’t enough. This is a kind of magical thinking which is present in popular thinking about evolution and I think this might be part of the issue with group selection.

    As I understand it, what Nowak did was show that group selection is actually mathematically possible although under a more limited limited set of circumstances than the magical version. In the process he also take a swipe at the inclusive fitness criteria – which has been a bit of a cornerstone of the evolution theory for a few decades – and attempts to replace it with something more like a population model than a simple criteria that applies to the individual.

    There’s some analysis of the arguments here: http://www.christopherxjjensen.com/2010/10/13/robert-trivers-and-colleagues-on-nowak-tarnita-and-wilsons-the-evolution-of-eusociality/

  5. Again, I am no expert, but I’ll have a go at defining some terms:

    According to wikipedia, evolution is: Change in the heritable traits of biological populations over successive generations.

    While the terminology is different, that’s very similar to the one I gave earlier – change in gene allele frequency over time.

    And again, according to wikipedia, group selection is: A proposed mechanism of evolution in which natural selection is imagined to act at the level of the group, instead of at the more conventional level of the individual.

    I would expand on that by saying that for group selection to have occurred, the genetic change selected for would have to benefit the group and disadvantage the individual that has it. If that is not the case and the genetic change benefits the individual, then it is not group selection, but selection at the individual level instead.

    It is easy for people who are not familiar with the subject to think that any selection that benefits the group is group selection, but that is not the case. Individual birds benefit from hanging out with other birds in a flock because more eyes mean predators are more likely to be spotted. So the gene that encourages flocking behaviour is not group selection. I know this may sound odd because if birds didn’t flock there wouldn’t be a group at all, but that’s not what group selection refers to.

    I’ll mention that one behaviour that has been proposed as an example of group selection, the warning cry a bird gives when it spots a predator such as a stalking cat, may not be for the benefit of the other birds. One of the multiple possible reasons, several of which may apply at once, is it is actually a message to the cat telling it, “I’ve spotted you and I’m already escaping. Go for one of the slower birds, not me!” So even for examples that have been put forwards as being group selection alternative, competing explanations always apply. So far we have no clear examples of group selection occurring outside of special examples such as social insects where all the individuals in a colony are closely genetically related.

    Harleymc gave a definition for group above, which I think is the one used for discussing population genetics, but biologists also use the term group more informally to mean, “This bunch of thing I am talking about,” and it might be easier for us to just continue using it that way ourselves.

  6. @Ronald Brak

    And as harleymc explained, in-group (not kin-group) success or failure “doesn’t change the frequency of genes” so there is no selection pressure:

    “A group in evolutionary biology is an association (an in-group) with no more or no less genetic similarity to each other than the out-group. In other words it is not a kin group. So whether the in group or the out group somehow gets a higher number of off spring, doesn’t change the frequency of genes.” – harleymc.

    By that definition of “the group” there can no such thing as biological natural selection stemming from group success.

  7. Ikonoclast, yes, that’s right.

    But I’m not aware of people of people using that definition when discussing group selection. If we strictly define a group as being a bunch of organisms with gene frequencies that are identical to other bunches of organisms, then group selection is impossible. Along with any other form of selection. In other words, if we strictly follow the definition, we freeze evolution.

    ORGANISM 1: Hey! My offspring has a novel mutation!

    ORGANISM 2: No it doesn’t, because if it did, that would make our group’s gene allele frequency different from other groups, and by definition, that is impossible.

    ORGANISM 1: Oh, you’re right. Its DNA just popped back into place via the power of definition.

    That definition of group is a bit of jargon that is useful in the right circumstances, for example discussing population genetics where they also sometimes declare organism populations to be infinite and mating to be completely random to make their equations simpler. But it’s not useful when discussing group selection because it defines away the possibility of group selection or any change in allele frequency at all, and we know from experience that changes in gene allele frequency can and do occur.

    So I would suggest we not use that definition for group, or at least not apply it too strictly, because it contradicts what we have learned from both observation and experiment, which is gene allele frequencies can change over time. (They just don’t appear to change due to group selection.)

  8. Evolution is both adaptive to and exploitative of environments. There are two processes going on here. One is biological, the other cultural.

    “Adaptive” is suitable for looking at passive changes in the biology of large groups or species over very long time spans. “Exploitative” suits the description of behaviour within groups or species reacting more proactively to an environment in the short term. Exploitative evolution is mainly only observable in higher primates – specifically humans. To say a certain species of plant “exploits” an environment is wrong. (This is often said of introduced species – but in fact an introduced species will thrive not through exploitation but simply through being suited to a previously addapted environment.)

    Evolution in most species, then, is a mainly adaptive process taking very long times to work through. Human evolution by contrast also involves an exploitative process leading to relatively rapid changes, mainly cultural. For example the height and longevity of humans has markedy increased over the past 50 of so years, but this is not the result of biological adaptataion, but rather of cultural exploitation, namely the marketing of medical and nutritional advances. It is this mainly exploitative nature of humans that has led to the destruction or farming of other species.

    In the norms of long term evolution, we can’t really predict how the human species will “evolve”. We can only speculate on cultural-type advances (science, politics etc).

  9. @Ronald Brak

    I don’t follow your reasoning at all. The in-group definition given by harleymc does not define away individual selection. It’s just that individual selection under that definition would be precisely that, individual selection.

    What is your suggested definition of group for the purposes of testing a group selection hypothesis?

  10. Suppose biologists studying some particular kind of organism observe an increase over time in the frequency of a particular gene (let’s call it P, which could stand either for ‘peacock’ or for ‘penguin’) in the population’s gene pool.

    They might hypothesise something roughly like this:

    Hypothesis 1: Gene P has effects which result, directly or indirectly, in individuals with gene P producing more offspring that survive to reproductive age (than do other individuals without gene P).

    (If something like Hypothesis 1 were true, the natural expectation would be for the frequency of gene P to increase; and if biologists were thinking along those lines, they could go on to investigate, in more specific detail, what the effects of gene P are and how they might have that kind of result.)

    Something like that could be an ordinary kind of evolutionary explanation that would be unlikely to be referred to as ‘group selection’.

    But suppose there’s another hypothesis:

    Hypothesis 2: Gene P has effects which result, directly or indirectly, in groups that include some individuals with gene P producing more offspring (that survive to reproductive age) than do groups with no individuals with gene P.

    Biologists who favour the idea of ‘group selection’ are interested in finding cases where something like hypothesis 2 is true.

    Biologists who reject the idea of ‘group selection’ tend to argue along the lines that it’s fairly predictable that hypothesis 2 will be true in cases where hypothesis 1 is true (at least often), that in cases where hypothesis 2 is true it’s only as a by-product of hypothesis 1 being true, and that hypothesis 2 gives us no extra information and adds no value.

    I think it would be interesting if there were cases where something like hypothesis 2 is true but something like hypothesis 1 is not true. But are there? Biologists who reject the concept of ‘group selection’ seem to doubt it.

  11. @J-D

    In hypothesis 2, I think you have described a kin-selection variant. Nobody, so far as I can see in this thread, has described any form of group selection which goes beyond kin-selection and which makes any kind of logical sense. The idea of group selection, given what selection is, seems to a non sequitur. I’ve seen nothing so far here or elsewhere to convince me that group selection makes any sense with strict respect to genes as the replicators.

    It seems to me (though as a non-expert I really don’t know) that for any group selection theory to work, the replicator, formally understood, would have to be above the level of the gene. We would have to conflate the replicator with the interactor at some level i.e. with the individual or the group itself. Does this idea work biologically or mathematically? I really don’t know but I doubt it. I think we are all stumbling around here without a clue.

  12. Ikonoclast, let’s say we have two groups of insects with identical genetic similarity. That is gene allele frequencies are the same. One group is on island A and the other group is on island B. On both islands insect numbers are controlled with DDT. On island A a chance mutation results in resistance to DDT and spreads through the population. On island B this does not happen.

    So now we have two populations that differ in gene allele frequency. But if a group has “no more or no less genetic similarity to each other than the out-group”, then we have a problem because we do have a group that differs in genetic similarity. So either the population on island A is no longer a group and we have to call it something else or the situation I have described is imposible and so never happens:

    REALITY: Luke, allele frequencies in group A differ from the out-group.

    DEFINITION: No. No! That’s not true. That’s impossible!

    REALITY: Search your polymerase chain reaction results. You know it to be true.

    DEFINITION: No! No! If that’s true, then you can’t call it a group anymore!

    REALITY: Now you’re just being silly. Join me and together we will document the changes in allele frequency that are occurring before our very eyes.

    DEFINITION: I will never join you. [Jumps down conveniently located chasm.]

    So I don’t find that definition of group very useful. Not when discussing selection since if there is no change in allele frequency then no selection has taken place. So I propose the following definition of the word group: The word group means “the bunch of things that I am talking about.”

    So what do you think of the population on island A that has developed pesticide resistance, Ikonoclast? Do you:

    (a) Think it is impossible for a change in allele frequency to have occurred since the definition of group you quoted states that a group has “no more or no less genetic similarity to each other than the out-group”. That is, no genetic change has actually occurred.

    Or do you:

    (b) Think that a change in allele frequency has occurred and therefore we can’t refer to the population on island A as a group since the definition states a group has “no more or no less genetic similarity to each other than the out-group”. So we have to call it something else now.

    Or do you:

    (c) Think that there has been a change in allele frequency on island A and we can still refer to it as a group. That is, you think the following statement is acceptable, “The group on island A is genetically different from the group on island B.”

  13. @Ronald Brak

    The answer is “c”. You can talk about groups in any way you like (as in dividing by island of residence as you do).

    However, if on island A, a chance mutation happens, in an individual called let us say “Alexander Beetle”, then it can only “spread through the population” by reproduction. All of the later resistant individuals on island A will be descendants of “Alexander Beetle”. It’s a clear case of selection of individuals. Which I assume you realise.

    So you have “proved” the trivial assertion that we can define “groups” any way we like and that this adds nothing to the discussion of group selection as a concept. What is your point?

  14. Ikonoclast, my point is that if we are going to discuss selection, then we have to accept at least the possibility that groups can differ in allele frequency which is something that can’t happen if we define a group as, “…an association (an in-group) with no more or no less genetic similarity to each other than the out-group. In other words it is not a kin group. So whether the in group or the out group somehow gets a higher number of off spring, doesn’t change the frequency of genes.”

    Since you chose (c) you appear to agree with me.

  15. @Ikonoclast

    No …

    … or at least I don’t think that’s the clearest way of putting it.

    As far as I know, biologists generally accept that kin selection happens. A generic description of kin selection using the same kind of wording and structure that I used for the other generic descriptions I gave would be something roughly like this:

    Gene P has effects which result, directly or indirectly, in individuals closely related to the P-bearing individual producing more offspring (that survive to reproductive age).

    In a situation where that is true, both hypothesis 1 and hypothesis 2 would be true; kin selection is one example (but not the only example) of what I described earlier, hypothesis 2 being true only as a by-product of hypothesis 1 being true, so that hypothesis 2 gives no extra information and adds no value (and, in fact, kin selection was an example I had in mind, and not the only one–I left out the examples I was thinking of for simplicity).

    By contrast, the idea of kin selection does add something; it is more specific than hypothesis 1 by itself, although still a fairly general concept.

  16. Kin selection ids favoured because it is micro-founded – it rests on the established mechanism of transmission of traits – the genotype.

    Kin selection: selects for actions which are disadvantageous to individuals but overall advantageous to the genotype. The eusocial insects (and a few others) do this by ensuring a very high degree of genotypic similarity in the group – typically by concentrating reproduction in one or a very few individuals (the queen in bees or ants and the naked mole-rat). So the male mole-rat that flings itself at the invading snake is trading its future sons and daughters for a quantity of nieces and nephews large enough to outweigh the loss.

    The challenge is to find a way that actions disadvantageous to the individual can persist in the absence of genotypical similarity. Humans clearly achieve this (just search the news for “suicide attacks”). How? My guess is the work-around involves neotony, the central role of culture in brain development and the large-ish minimum group size needed to sustain a culture (roughly 1000).

  17. You do know, JQ, that Mary Midgely has described the misapprehension of Darwin in these terms:

    “Thus the picture of natural selection which was shaped around Darwin’s main proposals used only two sombre colours – black for death and white for survival. Nature appeared in it only as an obsessive accountant, spectacles on nose and ledger in hand, testing every action in those terms and destroying the failures. Any trait still appearing in the world was deemed to have passed her audit in some distinct, discoverable way which constituted its evolutionary function.”

    She places the interpretation of Darwin within the distorting ideological landscape of liberalism.

    As to the errors within the group selection dialogue it is not necessary to say more than that the attribution of causality to one component of the whole, the gene, is such a pit of exhausted debate that there are only bones at the bottom. Peering in from the top of the pit are the poets, artists, writers, dancers, musicians, dreamers, philosophers and believers who constitute the rest of the whole of which the gene is but a mere switch.

  18. @J-D

    I was wondering that. I suspect not as we are all amateurs bumbling around with unscientific ideas on this topic. I was idly wondering if this post was a sociological experiment. You know, as in; post something a bit “left-field” and see how silly the replies get. I’ve noticed a strong tendency in myself and in others, for the need to have an opinion on everything, even on things we don’t know anything about.

    Clearly it has to do with discomfit about uncertainty. We humans seem to dislike uncertainty intensely. Uncertainty roughly aligns with insecurity. Indeed, we dislike uncertainty so much we manufacture false certainty by taking positions without evidence. We seem to be wired this way to some extent. It makes humans such fertile ground for dogma, denialisms and fundamentalisms. By contrast, it takes discipline to admit lack of knowledge, go looking for empirical evidence and then form opinions only after that process if such opinions are justified. It takes wisdom and resignation to realise there are many things we can never get certain knowledge about and so we must always live with considerable uncertainty.

  19. I’ve got a couple of benefits from reading the comments here and at Crooked Timber. One is that there have been lots of interesting points raised, in a generally civil fashion. The other is to determine that the idea of group selection is both controversial and not very well defined. So, while I was thinking of referring to an analogy with group selection in some work I’m doing, I’ve decided this would create more confusion than enlightenment.

    So, thanks everyone for talking about it, and feel free to keep going.

  20. Evolution theory is a bit like economics, the actions and impacts are statistical ensembles but everyone wants to distil out coherent actors, intents and narratives.

  21. There was an article about Australian Fairy Penguins taking over the habitat of New Zealand Penguins 400 years ago in The Guardian today.

    A scientist said these two groups of Penguins were difficult to distinguish apart from their behaviour, Australian Fairy Penguins:

    “come ashore in big rafts of 100 or 200…. The Australian ones can breed twice a year, while the native New Zealand one can only breed once, Waters said. Digital audio analysis also revealed they had different calls, Waters said. “It’s a bit like the accent of saying ‘fush and chups’ rather than “feesh and cheeps’,” he said.
    Many of New Zealand’s animal species, birds in particular, have suffered at the hands of people. The really exciting thing about these findings is that they show how quickly nature can respond to human impacts,” said Jonathan Waters from the University of Otago, who led the study.”


  22. I have mentioned why with the example given (b) applies. That is, it is not what is meant by group selection. But I will also make a comment or two on (a) which is whether or not we can judge if one species will dominate. Note that my comments don’t concern group selection as it does not apply in the given example.

    The “penguins” appear to have a better system than the “peacocks” (or “peafowl” if you prefer) because penguin males contribute resources to egg production and chick raising and so it appears that species is able to devote much more resources to reproduction than the peacocks where males contribute nothing but gametes.

    But, a colony of 50 female penguins, 49 contributing males, and one male who spends all his time commenting on the internet, has the same amount of resources to contribute to reproduction as a colony of 99 females and one sufficiently fertile male. So maybe the peahens stumbled onto a good thing by insisting the peacocks have ridiculous tails, as it must result in a lot of males getting killed and so reduce the numbers of the mostly useless male portion of their species.

    However, the ridiculous tails probably don’t kill enough males for an efficient male female ratio to result. But while the surviving male peacocks would be strutting around the place gobbling up resources that could have gone to females and therefore reproduction, they would presumably also be gobbling up resources that could have gone to penguins. So male peacocks could be a pox on both penguins and peahens. While peahens have to waste resources producing males in the first place, once they leave the nest they might be an equal burden to both species. (While not relevant for the example, real peafowl chicks can fly within days of hatching and with help from their mother start finding food for themselves almost immediately, and so take less effort to raise than the chicks of many other bird species.)

    If we assume all else is equal between the “penguins” and the “peacocks” then it looks to me that the “penguins” still seem to have an advantage, but not nearly as much as one might think at first glance. And since all else is rarely equal in real life, other factors would come into play. For example if “peacocks” can reproduce faster than “penguins” under good conditions, the place might become swamped with “peacocks” while the “penguins” only have a foothold.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s