A new paper showing the usefulness of the kin-selection model

February 4, 2016 • 9:45 am

There’s a new paper in the Proceedings of the National Academy of Sciences USA by David A. Galbraith et al. (free link and reference at bottom) that has a very cool result: one predicted by kin-selection theory. Kin selection, as you may know, is the idea that the adaptive value of a gene (and hence its evolutionary fate) must include information about how that gene affects its copies in relatives (e.g., a gene in parents for taking care of offspring can promote the replication of the copies that also occur in those offspring). Wikipedia describes this idea pretty succinctly.

Kin selection has been a very useful concept in understanding things like behaviors directed at offspring and relatives, and particularly in understanding the evolution of altruism and of one of its forms: eusociality—the behavior in which a colony of individuals is divided up into castes, some of which reproduce and some of which are nonreproductive but tend the “queen’s” brood (honeybees and naked mole rats are examples).

There are a few people, though, most notably Martin Nowak and E. O. Wilson at Harvard, who have questioned the usefulness of kin selection, arguing that group selection theory (or “multilevel” selection theory) is the only way to study the evolution of eusociality. I’ve written a lot on this site questioning their ideas (see some links below) as well as their claim that kin selection is not a useful way to study evolution in nature. The paper below, I think, shows the usefulness of the kin-selection paradigm, which seems to make predictions—ones that are verified—that don’t flow in any obvious way from a perspective of group or multilevel selection.

Because the paper is complex, I’ve asked my friend Phil Ward, a professor of entomology at the University of California at Davis (and a student of insect evolution) to explain its predictions and results. His explanation may be a bit difficult for non-biologists, but there is no simpler way to explain the study. Give it a go!


by Phil Ward

There has been a vociferous debate over the relative merits of group selection theory and inclusive fitness theory (or kin selection theory) as explanations for the evolution of altruistic behavior, especially following a contentious paper by Nowak et al. (2010) which claimed the superiority of the group selection approach. This was met with a resounding rebuff by a large group of evolutionary biologists who argued for the much greater explanatory power and heuristic value of inclusive-fitness thinking (e.g., Abbot et al. 2011). Some previous postings on WEIT about this topic have appeared here, here, here and here.

One fruitful area of inquiry in which kin selection theory makes explicit and testable predictions is in the study of genomic imprinting, a form of intragenomic conflict in which there is differential expression of genes inherited from the mother versus the father. In a theory paper published more than a decade ago, David Queller pointed out that this form of intragenomic conflict can be expected to be particularly widespread in colonies of social insects, and he employed kinship theory to predict the outcome of such conflict under different social contexts.

Now a recent empirical paper by Galbraith et al. (2016) provides convincing evidence that intragenomic conflict in honey bees indeed reveals itself in a way predicted by kin selection theory.

The authors first point out that genes inherited from mothers (matrigenes) and those inherited from fathers (patrigenes) are expected to be in conflict in honey bee workers that have an opportunity to reproduce. Why? Because a honey bee queen mates with multiple males, and the resulting workers are mostly half-siblings. These half-sibling individuals share half of their matrigenes but none of their patrigenes (see Figure 1 of the paper). So, consider a colony in which the queen has died, and half-sibling workers begin to compete over egg-laying (this behavior is inhibited by the queen while she is still alive). A worker’s matrigenes can be passed on when either she or her siblings reproduce, but her patrigenes are present only in her own offspring. Hence, as the authors put it, “compared with matrigenes, patrigenes will favor worker reproduction and exhibit enhanced activity on worker reproductive traits”.

This prediction was tested by quantifying the extent of genomic imprinting, i.e., the differential expression of genes of paternal origin.

The authors’ predictions were upheld. Using a series of genetic crosses that allowed them to distinguish matrigenes from patrigenes, they found that workers in queenless honey bee colonies showed greater expression of paternal than maternal genes, and this patrigene-biased expression was even higher in those workers that actually reproduced. In addition, when comparing parent-of-origin effects on reproductive traits such as ovary size and ovarian activity, patrigenes were shown to exert a much greater influence than matrigenes.

It should be emphasized that the worker reproduction occurring in queenless honey bee colonies produces only one sex: males.The workers lay unfertilized eggs and, as a consequence of the peculiar genetic system (haplodiploidy) found in bees, wasp and ants, these haploid eggs develop into males (which thus carry only one set of chromosomes). With no further production of workers, the colony will soon decline.

So, this last gasp of haploid reproductive effort that occurs when a queen dies (and is not replaced) will have selective significance only if the males that are produced have an opportunity to mate with queens from other colonies, something that takes place in population-wide mating swarms. Presumably this process of rearing and releasing drones (male bees) in a timely manner works best if some workers reproduce while the remainder continue to forage for food and feed the developing drone brood. Thus, colonies in which all reproductively capable workers give in to their patrigenic impulses might produce fewer reproductively successful drones than those in which there is some degree of reproductive restraint by the workers. One could argue that this is a kind of “colony-level” selection that weeds out disruptively high levels of patrigene expression, but inclusive fitness theory would explain this as a consequence of cost-benefit ratios that moderate the expression of both matrigenes and patrigenes.

Finally, for the small fraction of workers in a honey bee colony that are full siblings, the genetic interests of matrigenes and patrigenes are quite different: patrigenes can be equally well propagated through a worker’s own reproduction or that of a full sibling. Most competition for reproduction in honey bees is among half-siblings, however, so this should have little effect in honey bee colonies. Nevertheless, among other social insects in which the queen mates only once (such as bumble bees and many species of ants) all workers are full siblings and, as the authors note, the prediction is reversed: matrigenes should favor worker reproduction and show enhanced gene expression relative to patrigenes. Apparently this has not yet been studied, but it would constitute an elegant complementary test to the ground-breaking results of Galbraith et al.

Honeybee workers surrounding their queen, who’s been marked with a dot


Galbraith, D. A., S. D. Kocher, T. Glenn, I. Albert, G. J. Hunt, J. E. Strassmann, D. C. Queller, and C. M. Grozinger. 2016. Testing the kinship theory of intragenomic conflict in honey bees (Apis mellifera). Proc Nat. Acad Sci. USA 113:1020-1025. doi:10.1073/pnas.1516636113




30 thoughts on “A new paper showing the usefulness of the kin-selection model

    1. I’m assuming that’s a compliment and not intended to mean you had to read it twice to figure out what it was saying! 😉

  1. Very interesting.

    I have always thought it a bit of human psychological bias to think of the ‘workers’ serving a ‘queen.’ We think of the queen as in a superior position merely because that reflects our social forms – if you’re sedentary and taken care of, you must be on the top of the social ladder. But we could just as easily think of the social structure from the reverse psychological angle: the queen is the worker’s breeding-slave, a uterus stuck in a box for the purpose of making more workers, while the workers get to go out doors, have regular bee lives, etc…

    Biologically it doesn’t make a difference whether we think of the colony as one queen with a bunch of worker servants or a worker society with a breeding servant. The bees themselves could care less. 🙂 Its just interesting that we gravitate towards the former thinking rather than the latter. If your life choices were between queen (sit in one place, eat honey, make kids, that’s all you do) or worker (go collect pollen and tend the kids; probably don’t eat as well and die sooner), which sort of bee would you want to be? Do you really think the “queen” position is the most desirable one in the hive?

    1. The bees themselves could care less.

      As Dawkins pointed out in (I think) The Selfish Gene, the bees do care, and there is an actual fact of the matter about who’s in charge.

      The queen’s reproductive interests are favored by a 1:1 ratio of female to male offspring, but (because of haplodiploidy) the workers’ interests are favored by a 3:1 sex ratio. The queen decides what sort of eggs to lay (fertilized or unfertilized) but the workers decide which of those eggs to feed and nurture. And at the end of the day, it’s the 3:1 ratio that prevails.

    1. Just one clarification that needs to be made: the model proposed by Nowak et al (2010), as an alternative to kin selection theory, is not a group selection model. They refer to it as “standard natural selection theory in the context of precise models of population structure”.

      1. Yes, I was going to try to make that point. The Nowak, Tarnita, Wilson paper was anti inclusive fitness (kin selection), not pro group selection. As long as kin selection is used as a heuristic explanation and not an exact science, there is much less to argue about.

  2. Because of the “haplodiploid” nature of inheritance in this species, half-sibling individuals share half of their matrigenes but none of their patrigenes

    Unless I’m very confused, this is not unique to haplodiploid species. Even in humans, half-sibs from the same mother share half of their matrigenes and none of their patrigenes.

    What makes haplodiploidy unique is that full sibs share all of their patrigenes, not just half. The fact that worker bees are predominantly half-sibs rather than full sibs is a consequence of reproductive specialization combined with multiple matings by the queen, not of haplodiploidy. (Reproductive specialization may itself be selectively favored by haplodiploidy, but that’s another story.)

    The workers lay unfertilized eggs and, as a consequence of the peculiar genetic system (haplodiploidy) found in bees, wasp and ants, these eggs develop into males, which are therefore haploid (that is, they carry only one set of chromosomes).

    This seems to be putting the cart before the horse. The workers lay unfertilized eggs, which are therefore haploid, and (as a consequence of bee genetics) develop into males. It’s the haploidy that determines the maleness, not the other way round.

    1. Hi Gregory,

      You are right. That phrase “because of the haplodiploid nature of inheritance” was inserted by Jerry as he tried to clarify my writing, and I didn’t see it before it was posted! As you point out, diploid half-siblings with the same mother and different fathers would also share matrigenes but not patrigenes.

      I also agree that the statement about unfertilized eggs would read more clearly as: “The workers lay unfertilized eggs and, as a consequence of the peculiar genetic system (haplodiploidy) found in bees, wasp and ants, these haploid eggs develop into males (which thus carry only one set of chromosomes).”

      Thanks for the comments!

  3. It took me two readings, but I think I got it. Workers in a queenless colony can start laying eggs, and this is associated with expression of patrigenes. Kin selection is supported b/c we can link patrigene expression to favoring ones own offspring that carry these genes over the good of the colony members that do not carry those genes.
    What would the group selectionists say? Perhaps they would counter by pointing out that many workers defer reproduction, saying that is consistent with group selection. Or they could say that the worker egg laying is a kind of pathological behavior induced under admittedly unusual circumstances.

    1. Some of the workers that refrain from reproducing upon loss of the queen may do so because they are older individuals for whom activation of ovaries is more difficult.

      The idea that this behavior might be pathological and induced by unusual circumstances is one that I have thought about too. But upon reflection I suspect that orphaning of social insect colonies (honey bees and many other species as well) is a common occurrence in nature, and these last gasps of reproduction (of haploid males) are indeed adaptive–as long as the males are produced at the right time of year and have an opportunity to mate.

      1. I think it quite reasonable that worker behavior is influenced by their age or other conditions effecting their physiology. I know from a bee-keeping course I took many years ago that the workers normally change their behavior with age. I am not sure if I remember this right, but I think the younger ones perform duties in the nest. Then they take up guard duty. Later, they become field bees.

  4. Questions

    1. In the second last paragraph, Professor Ward suggests that one could see reproductive restraint as colony-level selection. I admit that that was what I was thinking while reading his account there. He counters that possible argument by explaining that inclusive fitness explains reproductive restraint as “a consequence of cost-benefit ratios that moderate the expression of both matrigenes and patrigenes”. I feel as though this is a key point, and I don’t know if I quite get it. Can this be spelled out more explicitly? Is it that the workers who continue to forage while other workers in the same colony mate with drones are actually acting according to a mathematically demonstrable strategy of moderation being better than giving in to “patrigenic impulses”? If I’ve got that right, why do some behave moderately and others not? Shouldn’t inclusive fitness guide all the bees to the same strategy? Or is it that workers who do give in to their patrigenic urges are equally acting on the better strategy for them? (Or am I misunderstanding the theory in that it explains behaviour on average, not in every individual case?)

    2. In the last paragraph, Professor Ward notes that queens of bumblebees and some ants only mate once. What happens after? Do they die?

    1. To your first question, the worker bees may all be performing the same cost/benefit calculation, but with different inputs depending on their individual age and fitness.

      To your second question, the queen’s mating is basically a one-night stand, with one partner or with many partners. Either way, she stores the sperm internally and dispenses it gradually over the rest of her life as needed to fertilize the eggs she lays. The males don’t participate in that part; they’re just donors to her internal sperm bank and are out of the picture as soon as the nuptial flight is over and the queen is safely ensconced in her hive.

      1. Thanks for both answers.

        I didn’t think of the different inputs. So, the different inputs create different calculations, and that results in different responses. And the professor’s point is that it could look like colony-level selection (or group selection) in some cases but that it is really still kin selection. Is that basically right?

  5. Even with such a consideration of different inputs which modify the calculation of the benefit/cost ratio, one cannot assume that every individual acts optimally (or to put it more precisely, the genes effecting action in that individual optimize their propagation). There are limits to what natural selection can achieve. There might be a general rule of thumb: “after foraging for x days, forget about trying to activate the ovaries.” Mind you, such a giving-up rule could also be differentially moderated by the action of matrigenes and patrigenes (the patrigenes are expected to raise the threshhold for foregoing reproduction).

    I am not a theoretician but I suspect you could probably model such reproductive restraint under both a colony-level selection framework, and a kin selection framework. But the latter approach arguably generates more interesting questions.

    1. Why would there be a giving-up rule like the one that you suggest? Just too costly to keep trying? Better to tend to what you have rather than to add to what’s there? Better just to survive?

      As well, I understood colony-level selection to be group selection and thought that your point at the end of your second-last paragraph had to do with showing that kin selection offers better insight into what happens. Did I have that right or not?

      1. Too costly; or perhaps the use of a simple rule-of-thumb, moderated by a few additional variables in some situations, is the best that can be done under the neurological/physiological constraints of a honey bee brain.

        Colony-level selection is a special kind of group selection where the group under selection is the family unit. To be honest, I am not sure how something like genomic imprinting could be incorporated into a group selection model or the kind of population genetics model espoused by Nowak et al. (2010). Kin selection provides a more useful way to think about the problem.

Leave a Reply