The purpose of this forum is to introduce notable papers and books published by you and other persons. The work can be new or old, but it should be of wide interest and high quality. A brief comment on the significance of the work should be attached. The current categories of the subjects are (1) adaptation, (2) behavioral evolution, (3) dosage compensation, (4) evo-devo, (5) gene evolution, (6) genomic evolution, (7) molecular phylogeny, (8) natural selection, (9) phenotypic evolution, (10) sensory receptors, (11) sex chromosomes, (12) sex determination, (13) speciation, (14) symbiosis and evolution, and (15) horizontal gene transfer. However, new categories can be added if necessary. Emphasis will be given on the biological work rather than on the mathematical. Any person may post a paper by sending it to one of the editors listed below. We also welcome your comments on posted work, but we moderate all the comments to control spam. This forum is primarily for scientific discussion and to construct a database for good molecular evolution papers.

Wednesday, March 21, 2012

Honeybee Castes and Altruism

In honeybees the queen and the worker are both females, but the queen has a larger body size than the worker and produces abundant offspring. By contrast, the worker bees are effectively sterile and take care of the queen and her offspring. Whether a female becomes a queen or a worker is determined by the amount of royal jelly provided during embryogenesis. If the amount of royal jelly is high, a queen is produced, but otherwise a worker bee is born. In a recent Nature paper, Kamakura (1) showed that royal jelly contains the protein Royalactin and this protein initiates the development of a queen. Therefore, the presence or absence of this protein generates the castes of the queen and the worker. In a “News and Comments” paper in Nature, Gene Robinson (2) praised this paper stating that Kamakura solved a 100-year old problem. Because the detail of Kamakura’s molecular study is explained by Robinson, I refer the reader to his article as well as Kamakura’s original paper.
In this commentary, I would like to discuss other aspects of evolution of castes and altruism. As is well known, Bill Hamilton (3) published a mathematical paper deriving the condition of evolution of different castes or eusociality. Some time ago, I tried to read the paper, but I was not happy with his formulation. Recently, Nowak, Tarnita, and Wilson (4) re-examined Hamilton’s theoretical work and rejected it. This paper then received hostile responses from the sociobiology community including the five papers simultaneously published in Nature. However, Nowak et al. does not give up their criticism (5). They believe that the caste system in hymenopteran insects has evolved by group selection, as Charles Darwin speculated, and propose one evolutionary scenario of eusociality.
In my view, this problem should be studied by using the molecular approach rather than the mathematical. Kamakura’s study indicates that eusociality can evolve irrespective of haplodiploid or diplodiploid sex determination, because he showed that the ectopic expression of royalactin produces a queen-like female even in Drosophila. This indicates that differentiation of the queen and the worker is initiated by one or a few genes. This finding questions the validity of Hamiliton's principle. The molecular biology of behavioral characters has been studied extensively for more than 40 years after Seymour Benzer’s pioneering work. Of course, the evolution of eusociality is quite complicated compared with the characters studied by molecular biologists. However, it is encouraging that several groups of evolutionists are now working on this subject using various insect species (e.g. 6, 7, 8, 9, 10).
             It is now known that many genes controlling sex determination and eusociality are shared by different insect species, and the expression of several genes is controlled by alternative splicing. For example, the feminizer (fem) gene that initiates the formation of female phenotype in honeybees is orthologous to the transformer (tra) gene in medfly, housefly, and Drosophila, and multiple splicing is necessary for these genes to produce functional proteins.

1. Kamakura M. 2011. Royalactin induces queen differentiation in honeybees. Nature 473:478-483.
2. Robinson G. 2011. Royal aspirations. Nature 473:454-455.
3. Hamilton WD. 1964. The genetical evolution of social behavior, I and II. J Theor Biol 7:1-52.
4. Nowak MA, Tarnita CE, and Wilson EO. 2010. The evolution of eusociality. Nature 466:1057-1062.
5. Nowak MA, Tarnita CE, and Wilson EO. 2011. Nowak et al. reply. Nature 471:E9-E10.
7. Rajakumar R, San Mauro D, Dijkstra MB, Huang MH, Wheeler DE, Hiou-Tim F, Khila A, Cournoyea M, and Abouheif E. 2012. Ancestral developmental potential facilitates parallel evolution in ants. Science 335:79-82.
8. Hasselmann M, Gempe T, Nunes-Silva CG, Otte M, Beye M. 2008. Evidence for the evolutionary nascence of a novel sex determination pathway in honeybees. Nature 454: 519-522.
9. Verhulst EC, Beukeboom LW, van de Zande L. 2010. Maternal control of haplodiploid sex determination in the wasp Nasonia. Science 328:620-3.
10. Foret S, Kucharski R, Pellegrini M, Feng S, Jacobsen SE, Robinson GE, and Maleszka R. 2012. DNA methylation dynamics, metabolic fluxes, gene splicing, and alternative phenotypes in honey bees. PNAS Early Edition.


  1. I am not an expert in this very detailed area. But it is clear that not even the experts agree, even on what the proper theoretical and mathematical techniques should be. This is reflected in a rebuttal of Nowak et al by Rousset and Lion (J Evol. Biol, 24: 1386-1392, 2011). Sociobiology is a food-fight with so much wiggle room that emotive positions can be taken and defended. Maybe the problem is that we are too interested in that, and not interested enough to look beyond these ideas.

    Even the defenders of Hamilton's inclusive fitness and other aspects of evolutionary game theory acknowledge the many ways in which they are problematic or approximate. But 'approximate' may not be good enough for evolutionary understanding.

    Mathematical theories assume either so much intricate and rigid regularity as to be unrealistic or unverifiable, or to allow for alternative explanations. When life works by way of temporally hierarchical, complex interactions, and contingent, opportunistic, local one-way events such as mutation, stochastic fitness, drift including allele loss, it becomes non-mathematical or only probabilistically mathematical. The result can be multiple possible outcomes that are too similarly likely for the theory to be very predictive (or, as in the case of accounting for the origins of present-day insect societies retrodictive).

    But even if mathematical theories are pushed too stridently, I can't really agree with Masatoshi's argument that we can be rescued by taking a molecular point of view. That can be very revealing and give us an understanding of the underlying mechanism, and using the (mathematical!) theory of molecular evolutionary genetics can reveal the plausible history of the mechanism. But something that evolved must have some mechanism at the molecular level.

    Thus, if we look at one gene product like royalactin, and find that it by itself doesn't lead to a constrained picture of how eusociality evolved, the easy and perhaps obviously true response by Hamilton's defenders would be that other aspects of the genome combine (and in the past combined) in different ways to make the mechanisms among existing species, and therefore, essentially by the assumption that evolution is true, must have evolved through population genetics (i.e., mathematically expressable) history. The Drosophila result would seem to show that. Indeed, this kind of argument can be seen in Rousset and Lion's reply to Nowak et al.

    In this sense, it is well known that evolutionary theory becomes irrefutable: once the events are over with, and if hierarchical genome evolution and probabilistic events (duplication, mutation, fixation, loss) are central to the theory, the theory has the ability to fit anything. How else could life happen?

    I think personally this kind of debate shows that the Darwinian method has proven to be an exceptionally powerful way to organize consistent, plausible explanations for life in purely material terms. It shows without serious doubt that evolution is a truth consistent with both observation and experiment.

    But it also shows that we are far from having a definitive theory of life comparable to those of chemistry or physics--unless. Unless the loose, contingent, and probabilistic way evolutionary theory connects the rigors of molecular chemistry to the statistical nature of population history really IS the state of Nature.

    In that case, which seems likely to me, we do understand life, but also have to understand that much of the truth really is ad hoc and can't be addressed definitively with the smoothed-out nature of mathematical population models or evolutionary molecular genetic reconstructions.

  2. I agree with Ken Weiss about molecular versus evolutionary explanations. The royalactin results are really interesting but they provide proximate explanations of mechanism and not ultimate explanations of selective causes. Nothing in them bears on the issue of whether eusociality evolved because of benefits provided to relatives.

    The field of social evolution is not in nearly as much turmoil as it may appear to outsiders. The theoretical arguments are partly a matter of ego and partly a matter of preferences for different modeling strategies. The strategies differ in generality, simplicity, precision, and in how easily we can compare them to the real world. Different modelers weigh these factors differently, often unconsciously. Yet the various models, if we take the trouble to compare them, usually show similar things, viewed from different angles. This leads to priority issues; does this new model using method A show fundamentally new result or is it expressing in a new language something we already knew from method B? The resulting arguments over priority often get confused with disagreements over substance.

    So let me say a few words about what I think the theoreticians agree on. I think they all agree that Hamilton was correct that having fitness effects on genetic relatives is important and that altruism can evolve if sufficient benefits are given to sufficiently related kin. I think they would all agree that Hamilton’s rule is correct when fitness effects are additive. I think they would all agree that non-additivity of fitness effects is another important social phenomenon, for example when it takes two cooperators to generate any benefit. This is the area that game theoretic approaches were devised for.

    Many of the disagreements appear when those two insights – kin selection and non-additivity – are both in play at the same time. Then you can get arguments over when Hamilton’s rule is still correct, how it can be modified to make it correct, or whether it is worth the trouble when other methods are available. But none of these arguments affect Hamilton’s basic insights about kin selection.

    I should mention that at least one person, E. O. Wilson, does not agree. He now seems to believe that social insects evolved for some completely different reason and that genetic relatedness was not important. But Wilson is not a theoretician and there is no theoretical support for his view. Even the models of the evolution of eusociality in the Nowak, Tarnita, and Wilson paper assume high relatedness. That paper is a forced and unproductive marriage between Wilson’s desire to dethrone kin selection to Nowak and Tarnita’s preference for a modeling strategy other than inclusive fitness. People who don’t read the paper (and the math) carefully believe that the two views mutually support each other, but they do not. Masatoshi gave the reference for Nowak et al. and for their response to critics, but not to the criticisms themselves, so those are appended below 1-5, but remember these contain a mix of criticisms of Wilson’s odd view that kin selection is unimportant and issues on the theoretical preferences on modeling strategies.

    I started out trying to be completely dispassionate, and now I mind myself getting a little edgy, so it is time to stop. The main point is that there is really a huge amount of agreement on social evolution theory, and that arguments over preferred modeling strategies should not obscure this.

    1 Ferriere, R. & Michod, R. E. Inclusive fitness in evolution. Nature 471, E6-E8 (2011).
    2 Abbot, P. et al. Inclusive fitness theory and eusociality. Nature 471 (2011).
    3 Strassmann, J. E., Page, R. E., Robinson, G. E. & Seeley, T. D. Kin selection and eusociality. Nature 471, E5-E6 (2011).
    4 Herre, E. A. & Wcislo, W. T. In defence of inclusive fitness theory. Nature 471, E8-E9 (2011).
    5 Boomsma, J. J. et al. Only full-sibling families evolved eusociality. Nature 471, E4-E5 (2011).

  3. It is nice to hear the opinion of an insider (David Queller) of the theoretical sociobiology community with which I have had little contact for the last few decades. Historically, Hamilton’s principle is based on the neo-Darwinian theory of panselectionism with the assumption that abundant genetic variability exists in a population and evolution occurs solely by natural selection. In this sense his principle is similar to R. A. Fisher’s fundamental theorem of natural selection, which states that the rate of increase of mean fitness of a population is equal to the additive genetic variance at that time.

    Although this is conceptually interesting, it does not have much practical utility because genotype fitnesses are dependent on the environmental condition and this condition changes every generation, particularly when geological upheavals such as asteroid hits on earth occur. Furthermore, this theorem cannot be used for predicting the evolutionary change of any specific character controlled by a single locus or a few loci, because the mean fitness is controlled by the entire set of genes in the genome.

    Hamilton’s principle has the same problem when the altruistic character is controlled by one or a few genes. In this case the advantageous mutation may occur only occasionally and even if it occurs it may not be fixed in the population because of genetic drift. Furthermore, the new mutation may be effectively neutral. Recent genomic studies indicate that even some important characters such as sex determination may change by effectively neutral mutations (1). In this case, Hamilton’s rule is not very meaningful. Note that the effective size of a honey bee population has been estimated to be about 1,000 and therefore the effect of genetic drift is quite important (2).

    Kamakura’s study (3) has now shown that the caste of honey bee queens is controlled by one protein, Royalactin, which gives a signal for the development of queen phenotype. Therefore, it would be interesting to study the evolutionary history of the gene encoding this protein by examining related species. The fact that this protein ectopically produced generates a queen-like female even in Drosophila melanogaster suggests that an ancestral species of Diptera and Hymenoptera may have had a developmental pathway for generating a queen-like phenotype. If so, it is possible to explain the evolution of the queen caste in various groups of hymenopteran species by mutations of regulatory genes that trigger the developmental pathway of queen phenotype. In this case, however, the signal protein may not always be Royalactin, but many other proteins or different genetic systems may be used as in the case of sex determination in different groups of insect species (1). See my new blog (Soldier Ants and Caste Evolution).

    Hamilton’s principle is an abstract concept and has no power of explaining the evolution of specific caste systems, in which we are interested in.

    Ken Weiss also made a comment on my commentary, but his position is not very clear to me. He seems to advocate that although the biological world is unwieldy, we should find a way to understand the evolutionary process by using some statistical methods. If this is the case, I would agree with him. When I was young, I was aspired to make the evolutionary biology as a rigorous science and started my career with theoretical population genetics. However, what has happened later is that molecular biology has contributed to evolutionary biology more than population genetics theory. Yet, I am still striding to find some general rules which are biologically meaningful.

    1. Gempe T, and Beye M. 2010. Function and evolution of sex determination mechanisms, genes and pathways in insects. Bioessays 33:52-60.
    2. Yokoyama S, and Nei M. 1979. Population dynamics of sex-determining alleles in honey bees and self-incompatibility alleles in plants. Genetics 91:609-626.
    3. Kamakura M. 2011. Royalactin induces queen differentiation in honeybees. Nature 473:478-483.