SEARCH

SEARCH BY CITATION

Keywords:

  • caste conflict;
  • kin conflict;
  • Melipona;
  • stingless bees

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. The model
  5. Individual caste allocation interests
  6. Worker interests
  7. Empirical support for caste conflict theory
  8. Discussion
  9. Acknowledgments
  10. References
  11. Appendix. Caste conflict from a levels-of-selection perspective

A caste system in which females develop into morphologically distinct queens or workers has evolved independently in ants, wasps and bees. Although such reproductive division of labour may benefit the colony it is also a source of conflict because individual immature females can benefit from developing into a queen in order to gain greater direct reproduction. Here we present a formal inclusive fitness analysis of caste fate conflict appropriate for swarm-founding social Hymenoptera. Three major conclusions are reached: (1) when caste is self-determined, many females should selfishly choose to become queens and the resulting depletion of the workforce can substantially reduce colony productivity; (2) greater relatedness among colony members reduces this excess queen production; (3) if workers can prevent excess queen production at low cost by controlled feeding, a transition to nutritional caste determination should occur. These predictions generalize results derived earlier using an allele-frequency model [Behav. Ecol. Sociobiol. (2001) 50: 467] and are supported by observed levels of queen production in various taxa, especially stingless bees, where caste can be either individually or nutritionally controlled.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. The model
  5. Individual caste allocation interests
  6. Worker interests
  7. Empirical support for caste conflict theory
  8. Discussion
  9. Acknowledgments
  10. References
  11. Appendix. Caste conflict from a levels-of-selection perspective

Societies of eusocial Hymenoptera (ants, and many bees and wasps), are characterized by a high degree of cooperation (Wilson, 1971; Hölldobler & Wilson, 1990; Seeley, 1995). However, the nonclonal structure of their societies also causes conflict. Traditionally, four major conflicts have been studied (Bourke & Franks, 1995; Crozier & Pamilo, 1996): (1) queen-worker conflict over the sex ratio, (2) conflicts over male parentage, (3) conflicts over female parentage (‘nepotism’; Ratnieks & Reeve, 1992) and (4) conflict over reproductive dominance (e.g. in foundress groups). Recently, however, an additional conflict – that over female caste fate – has attracted considerable attention (Ratnieks, 1989, 2001; Bourke & Ratnieks, 1999; Reuter & Keller, 2001; Strassmann et al., 2002). In social Hymenoptera, immature females are generally genetically totipotent, and in species with morphological castes each can be reared irreversibly into either a queen or a worker (Wilson, 1971). However, according to caste conflict theory, females would often benefit if they could develop as queens rather than workers, in order to gain greater direct reproduction (Bourke & Ratnieks, 1999; Ratnieks, 2001). In social Hymenoptera without morphological castes, where all females are totipotent, conflicts over who takes on the breeding role have long been recognized (e.g. West-Eberhard, 1967, 1978, 1981; Strassmann, 1989) and their study continues to be an important area of research (e.g. Reeve & Nonacs, 1992; Monnin & Ratnieks, 1999, 2001; Monnin et al., 2002). However, that the same conflict can also occur in species with morphological castes is a relatively new idea (Ratnieks, 1989, 2001; Bourke & Ratnieks, 1999). Yet, the consequences of caste conflict on the functioning of the colony can be profound. When females can control or influence their own caste fate, caste conflict is predicted to lead to more females selfishly choosing to develop as queens than is optimal for colony function (Ratnieks, 2001; Reuter & Keller, 2001; Strassmann et al., 2002). With more females developing into queens the workforce is diminished leading to a reduction in total lifetime colony reproduction (i.e. a smaller colony reproduces less, survives less or both).

Although caste conflict in social Hymenoptera is expected under a wide range of conditions (Bourke & Ratnieks, 1999; Ratnieks, 2001; Reuter & Keller, 2001; Strassmann et al., 2002), in societies with morphological castes, it is usually present only in latent or ‘potential’ form (Bourke & Ratnieks, 1999). This is because in such species, the passive female brood rarely have the ability to control their own caste fate, as caste is usually determined by the amount or quality of the food given to larvae by workers (‘nutritional caste determination’; Wilson, 1971; Michener, 1974; Wheeler, 1986). By food control, the adult workers can coerce immature females to develop as workers. For example, in honeybees, female larvae develop into queens only when fed large amounts of special food, royal jelly, provisioned in special queen cells (Winston, 1987).

However, in one group of highly eusocial insects, Melipona stingless bees, the caste fate of female brood can be controlled by the immature female individual herself (Bourke & Ratnieks, 1999, reviewed by Wenseleers & Ratnieks, 2003). This is because Melipona queens and workers are of the same size and are reared from identical, mass-provisioned sealed cells (Wilson, 1971; Michener, 1974). These conditions prevent the adult workers from coercing the immature females and allow caste to be self-determined – prerequisites for actual caste conflict to be observed. In agreement with this prediction, Melipona produce a great excess of queens. Stingless bees reproduce by swarming (Engels & Imperatriz-Fonseca, 1990). That is, daughter colonies are founded by queens accompanied by a swarm of workers. Colonies, therefore, require only few queens to be reared, just enough to permit seasonal swarming or supersedure of a failing mother queen. In Melipona, however, up to 25% of all females are reared as queens, and most of these are executed by the workers within hours or days of emergence (Kerr, 1950; Engels & Imperatriz-Fonseca, 1990; Imperatriz-Fonseca & Zucchi, 1995). Although such queen overproduction has long been considered paradoxical (Engels & Imperatriz-Fonseca, 1990), it can now be explained on the basis of caste conflict theory (Ratnieks et al., 2001). In other stingless bees (Trigonini), queens are normally larger than workers, and caste is determined by nutrition, as in the honeybee (Engels & Imperatriz-Fonseca, 1990; Wenseleers & Ratnieks, 2003). In trigonine stingless bees few queens are reared, in keeping with colony needs (e.g. in Tetragonisca angustula, only approximately four queens are reared over an entire season; van Veen & Sommeijer, 2000).

Although the above observations provide good general support for caste conflict theory, additional, more detailed, tests would be desirable. However, such tests require a solid theoretical basis in order to make predictions about the expected level of excess queen production. Ratnieks (2001) showed that for swarm-founding social Hymenoptera, the optimum or ‘evolutionarily stable’ proportion of females that should develop as queens ranges from 14 to 20% depending on whether the colony's males were produced by the workers or the queen (workers in many species of eusocial Hymenoptera, including Melipona, retain ovaries and can lay unfertilized, male, eggs). These predictions were derived using an allele-frequency model and although rigorous, this method has several disadvantages. In particular, it requires evolutionary stable strategy (ESS) ratios to be derived separately for each colony kin structure (e.g. single or double mating) which is impractical for all but the simplest colony structures (e.g. one queen mated to one or two males).

The aim of this paper is to develop caste conflict theory using inclusive fitness methods (Hamilton, 1964; Frank, 1998). We develop our model specifically for swarm-founding social Hymenoptera, as Melipona, the prime example where overt caste conflict occurs, is swarm-founding and because it is only in swarm-founding species that predictions can easily be tested (Bourke & Ratnieks, 1999). This is because colonies only need a few queens to swarm, so that excess queen production is easily quantified (i.e. all queens are basically excess). By comparison, in species with queens that found nests independently the colony optimum would also be to produce many queens and it would be hard to determine what fraction of them represents an excess (i.e. reared in excess of the optimum of the queen and workers collectively). A major conclusion of the model is that when caste fate is under larval control, an excess of queens should occur that is inversely proportional to intracolonial relatedness. For simple relatedness structures, we rederive the earlier results (Ratnieks, 2001). But the new model also makes more complex social structures, such as extreme polyandry (multiple mating by queens) or polygyny (multiple queens occurring in a single colony) open to prediction. These additional predictions will allow critical tests of caste conflict theory once more date are available, and will also suggest key species for data collection. We also analyse how the colony-level cost of excess queen production influences the expected ESS ratios. Finally, we consider when a transition from individual to social control over caste fate is expected (Bourke & Ratnieks, 1999), that is, when workers should attempt to prevent immature females from developing as queens.

Individual caste allocation interests

  1. Top of page
  2. Abstract
  3. Introduction
  4. The model
  5. Individual caste allocation interests
  6. Worker interests
  7. Empirical support for caste conflict theory
  8. Discussion
  9. Acknowledgments
  10. References
  11. Appendix. Caste conflict from a levels-of-selection perspective

First we derive the optimal probabilities with which individual females should develop as queens vs. workers in swarm-founding social Hymenoptera. We use the direct-fitness method of Frank (1998), whereby the direct fitness of various classes of relatives is written as a function of the individual and average behaviour of individuals in the group. In our case, this behaviour is the probability with which females develop as queens. We use y, z and Z to denote the focal individual, group-average and population-average probabilities with which females develop as a queen (Table 1 summarizes the notation used in this paper). Note that the model analyses optimal caste allocation (worker vs. queen production), rather than reproductive (worker vs. sexual, male + queen) allocation (Pamilo, 1991a; Bourke & Chan, 1999; Herbers et al., 2001).

Table 1.  Parameters and terminology used in this paper.
WfFitness through queen (swarm) production
WmFitness through male production
GRelative group success (lifetime colony productivity)
IIndividual success of one female relative to another
yProbability with which a focal female larva develops into a queen
zAverage probability with which female larvae develop into queens in the colony
ZPopulation average probability with which female larvae develop into queens
z*ESS probability of developing as a queen
kParameter describing how excess queen production affects total colony reproduction (swarm and male production) (k = 1/2: square law, k = 1: linear, k = 2: quadratic, etc.)
ksParameter describing how excess queen production affects swarm production (ks = 1/2: square law, ks = 1: linear, ks = 2: quadratic, etc.)
kmParameter describing how excess queen production affects male production (km = 1/2: square law, km = 1: linear, km = 2: quadratic, etc.)
MEffective maternity (number of laying queens)
PEffective paternity
cCost of preventing excess queen production
RfLife-for-life relatedness to sister queens reared in the colony
RmLife-for-life relatedness to males reared in the colony

It is clear that total colony reproduction, the production of males and swarms, is a decreasing function of z. The more females that develop as queens (higher z), the fewer that develop as workers, leading to less work and a lowered ability of the colony to rear males or produce swarms. For simplicity, and as in the model of Ratnieks (2001), we initially use a linearly decreasing function, 1 − z, for the absolute productivity of the focal colony, but we consider other cost functions later. Relative colony productivity, i.e. the productivity of the focal colony relative to a random colony in the population is then given by G = (1 − z)/(1 − Z). We also assume that queen overproduction reduces male production to the same extent as swarm production. This assumption is also relaxed later. Clearly, at the level of the colony, it is undesirable to have many females develop as queens (high z). However, at the individual level, there are distinct benefits associated with becoming a queen rather than a worker: if females develop as queens with higher than average probability, it allows them to increase their chances of inheriting a daughter swarm. Formally, over the course of a single queen replacement or swarming event, the relative probability for a female to inherit a daughter colony is y/z, the ratio of the individual probability with which it becomes a queen and the probability with which other females in the colony become queens.

We can now write the fitness through male (Wm) and swarm (Wf) production as a function of the individual (y) and average (z) probabilities with which female larvae become queens within the colony:

  • image(1)

In these equations, (1 − z)/(1 − Z) is the relative productivity of the colony in terms of male and swarm production (relative group success G) and (y/z) is the success of any female larva relative to another (relative individual success I). In these equations we also assume that the population is in sex ratio equilibrium, so that the relative valuation of the sexes is not affected by it. The above model is closely related to that of Frank (1995b), which was developed to determine the optimal level of exploitation within social groups. The only difference is that we specifically address the hymenopteran situation where only females can vary their level of exploitation (i.e. become queens rather than workers) but costs are to both sexes.

Whether an increased probability of developing as a queen is favoured can be calculated from a simple Hamilton's rule condition (Hamilton, 1964) with three affected parties: (1) self, (2) sisters and (3) males. The personal benefit of developing as a queen is given by ∂Wf/∂y = (1 − z)/[(1 − Z)z] (for a full account of the method used, see Frank, 1998). In other words, the benefit of becoming a queen is proportional to the relative productivity of the colony, (1 − z)/(1 − Z), because this sets the ability for the colony to produce workers and swarms, but is inversely proportional to the number of competing queens (z), because a female that develops as a queen will only inherit the colony or swarm with a probability of 1/z. Similarly, the personal cost that occurs as a result of other females developing into queens is given by −∂Wf/∂z = y/[z2(1 − Z)]. As the role of any female in the colony is symmetrical, this is also the cost of developing as a queen to a sister. Lastly, the cost to male production is given by ∂Wm/∂z = −1/(1 − Z), that is, relative to the population average productivity of a colony (1 − Z), a unit change in the proportion of females that develop into queens causes a corresponding reduction in the net production of males. Following kin selection theory (Hamilton, 1964), we need to weigh these costs and benefits with the relatedness to the affected parties, that is self (r = 1), sisters (life-for-life relatedness Rf), and males (life-for-life relatedness Rm), to calculate the change in frequency of the caste determining gene. Summing the three relatedness-weighted costs and benefits, the net inclusive fitness benefit of developing as a queen with probability y is given by

  • image(2)

Near the ESS, all female larvae will evolve to become queens with the same probability, so we can set y = z = Z. Setting the resulting equation to zero and solving for Z yields the ESS probability with which female larvae should develop into queens:

  • image(3)

In equivalent form, females should develop into queens or workers in a ratio of

  • image(4)

The intuitive explanation is that a female gains from becoming a queen in proportion to its own value (1) relative to that of the sisters it is competing with (value Rf), whereas becoming a worker has indirect benefits in terms of increased swarm (value Rf) and male production (value Rm). A similar derivation from a levels-of-selection perspective is also possible (see Appendix). Note that the fact that most Melipona queens are killed upon eclosion (Kerr, 1950; Engels & Imperatriz-Fonseca, 1990; Imperatriz-Fonseca & Zucchi, 1995) does not make it less profitable to develop as a queen. This is because if the queens were allowed to survive this would only delay the time at which one was ‘chosen’. Or if the queens could share reproduction this would divide each queens share of colony reproduction by the number of queens. Therefore, queen killing does not alter the expected payoff. Rather, it just converts it from a small equal payoff to a large jackpot won by one individual.

Extending the result for nonlinear colony cost functions is straightforward. Instead of using a linear colony-level cost function, we use a power function (1 − z)k (or any other monotonically decreasing function) for colony productivity, instead of the linear (1 − z)1. This gives a family of convex (k < 1) and concave (k > 1) curves. Relative colony productivity G is now equal to (1 − z)k/(1 − Z)k and analysis by the same methods as before yields the ESS

  • image(5)

This optimum is plotted for different values of k in Fig. 1a. It is also possible that the costs of excess queen production to male and swarm production are different, because they are very different modes of reproduction. We can generalize the model by describing costs to male and swarm production with separate power functions with coefficients km and ks. Fitnesses through male and swarm production then become

  • image(6)
image

Figure 1. ESS probability z* for a female larva to become a queen in a colony headed by a single once-mated queen, assuming different colony-level cost functions. The optimum is plotted for cases when all males are either queen (full line) or worker (dashed line) produced. (a) Queen overproduction reduces male and swarm production equally, and the cost function follows a power law with coefficient k (plotted from eqn 5). (b) Queen overproduction reduces male and swarm production unequally with costs to swarm and male production following power laws with coefficients ks and km.

Download figure to PowerPoint

Analysis as before yields the ESS

  • image(7)

This optimum is plotted for arbitrary values of ks and km (Fig. 1b).

From eqns (3)–(5), the prediction is that larvae should selfishly develop as queens with a higher probability when the relatedness to sister larvae or males or both is lower (Fig. 2), or when the costs to colony function are less severe (k < 1, i.e. with a convex colony productivity function, Fig. 1a).

image

Figure 2. ESS probability z* for a female larva as a function of the colony kin structure. It is assumed that queen overproduction linearly reduces total colony reproduction. (a) Optimal levels of queen production when the colony is headed by multiple, once-mated queens. The figure is plotted for effective maternities (M) ranging from 1 to 4, assuming that the males are either the queens’ (unbroken line) or the workers’ sons (dashed line). The bold and fine curves are for when the mother queens are unrelated or related by 0.5. (b) Optimal levels of queen production when the colony is headed by a single multiple-mated queen. The optimum is plotted for effective paternities (P) ranging from 1 to 15, with all males either queen (full line) or worker produced (dashed line). The optima are calculated from eqn (3) with relatedness values Rf = (1 + 2/P + 0.5(M − 1))/4M and Rm = (1 + 0.5(M − 1))/4M (all males queen produced) or Rf = Rm = (1 + 2/P + 0.5(M − 1))/4M (all males worker produced) (Pamilo, 1991b).

Download figure to PowerPoint

Table 2 shows some numerical examples for typical colony kin structures as they occur in Melipona and honeybees. Most Melipona are headed by a single once-mated mother queen (Kerr et al., 1962; da Silva et al., 1972; Kerr, 1975; Contel & Kerr, 1976; Peters et al., 1999; Paxton et al., 2001), so that Rf = 0.75. In one species, M. bicolor, however, the colony can be headed by one or several (2–7) related queens, and the average Rf has been estimated at 0.6 (Peters et al., 1999; Table 2). The relatedness to the males (Rm) reared in a single-queen colony can range from 0.25 to 0.75 as a result of variable degrees of worker reproduction (Tóth et al., 2002). For example, in M. favosa, nearly all males are workers’ sons, so that Rm = 0.75, whereas in M. beecheii, all males are the queens’ sons, so that Rm = 0.25 (Table 2). Assuming a linear cost function (k = 1), the prediction for Melipona is that larvae should develop as queens with a probability of (1 − 0.75)/(1 + 0.25) = 20% when all males are the queens’ sons (Rm = 0.25), or with a probability of (1 − 0.75)/(1 + 0.75) = 14% when all males are workers’ sons (Rm = 0.75, Table 2). When all males are workers’ sons the optimum is lower because the cost of individual selfishness then falls on more closely related males (nephews, Rm = 0.75 rather than brothers, Rm = 0.25). From a colony-level perspective both of these values represent great queen overproduction as only few queens are needed for swarming. These values agree with those obtained from the exact genetic model (Ratnieks, 2001), but the current method is more general as it makes any colony kin structure open to prediction. For example, in M. bicolor, as a result of polygyny, females should develop as a queen with even higher probability, 23–32% (Table 2). Similarly, in the honeybee, polyandry causes very low relatedness among sisters (Rf = 0.3; Estoup et al., 1994), and the individual optimum is to develop as a queen with a probability of 56% (Table 2). Figure 2 shows the optimum for different levels of polygyny and polyandry.

Table 2.  Kin structure of various stingless bees and the honeybee Apis mellifera and the individual-optimum probabilities for females to develop as queens. Lower sister–sister relatedness causes more females to develop as queens while worker production of males has the opposite effect as the cost of queen overproduction is then borne by closer relatives – workers’ sons (nephews) rather than queens’ sons (brothers). Life-for-life relatedness to females (Rf) and males (Rm) is the product of regression relatedness (rf, rm) and the sex-specific reproductive value (the contribution of a given sex to the future gene pool). The relative reproductive value of males is 1 + p where p is the population proportion of males that are queens’ sons (Pamilo, 1991b).
SpeciesKin structurerf = RfrmRelative reproductive value of males, vmRm = rmvmOptimal probability for females to develop as queens* =  (1 − Rf)/(1 + Rm), %References
  1. *Assuming that queen overproduction linearly reduces total colony reproduction (eqn 3). The optimum is higher or lower with a convex or concave productivity function (eqn 5).

  2. †Double mating by queens was erroneously reported for M. beecheii due to the accidental inclusion of workers that drifted between colonies (Paxton et al., 1999).

Melipona favosaSingle once-mated queen, nearly all males workers’ sons3/43/413/40.14Sommeijer et al. (1999), Chinh et al. (2002)
Melipona beecheiiSingle once-mated queen, nearly all males queens’ sons3/41/21/21/40.20Paxton et al. (2001)
Melipona bicolorMultiple-related queens, most males probably workers’ sons0.61/2–3/41/2–11/4–3/40.23–0.32Nogueira-Neto (1970), da Silva et al. (1972), da Silva (1977), Bego (1989), Peters et al. (1999), D. Koedam (personal communication)
Apis melliferaSingle multiple-mated queen0.31/21/21/40.56Estoup et al. (1994)

The numerical predictions above are for the case where queen overproduction linearly reduces total colony reproduction (production of males and swarms). The predictions for other colony-level cost functions are plotted in Fig. 1. Figure 1a plots the optima when both male and swarm production are affected in a nonlinear way. As can be seen, the optimum probability to develop as a queen is higher when the colony-level cost is less severe (with a convex colony productivity function, k < 1, Fig. 1a), or lower when it is more severe (with a concave colony productivity function, k > 1 (Fig. 1a). Figure 1b plots the optima when effects of queen overproduction on male and swarm production differ. As can be seen, the relative influence of relatedness is the same in all cases, but the predicted optima are different, both in terms of magnitude and range. In other words, predictions are affected quantitatively but not qualitatively [Maynard Smith (2002) comments that biological systems are generally too complex for precise quantitative predictions, but strong qualitative predictions can be made]. However, if the cost functions were different for the different species, then that might obscure the predicted effect of relatedness. Currently, no information is available on the shapes of the cost functions. It would be hard to determine them empirically as a colony normally only swarms once every few years (Engels & Imperatriz-Fonseca, 1990) but can produce males over a long period, even the whole year (Moo-Valle et al., 2001).

In deriving the optimal level of queen production (eqn 3) we make no assumptions as to how it may be caused at the proximate level. For Melipona, Kerr (1950) hypothesized a two-locus two-allele genetic system that would give 25% queens, on average, in each colony. But such a genetic system, which still lacks strong empirical support (Velthuis & Sommeijer, 1991), is not strictly required. If it did occur, however, it might constrain what ratios can be attained (Ratnieks, 2001). In the model above, we assumed there are no constraints. Instead, we assumed a very large number of alleles, each of which could slightly alter the probability with which females develop as queens, and with the allele that codes for the optimum going to fixation.

Worker interests

  1. Top of page
  2. Abstract
  3. Introduction
  4. The model
  5. Individual caste allocation interests
  6. Worker interests
  7. Empirical support for caste conflict theory
  8. Discussion
  9. Acknowledgments
  10. References
  11. Appendix. Caste conflict from a levels-of-selection perspective

Immature females benefit from developing as a queen with a high probability because they value themselves more highly than their sisters. The adult workers, however, should value all developing females equally as they are equally related to all of them (this is neglecting the possibility of strong nepotism which could potentially occur under polyandry or polygyny). Formally, from the perspective of a worker, the relatedness to a potential beneficiary in eqn (3) is Rf rather than 1, which yields the optimum z* ≅ 0. The caste fate optimum of the adult workers is thus the same as the colony optimum of near-zero allocation to queens. By the same argument, the mother queen also favours the production of few queens.

A direct consequence of this is that workers (or the mother queen) should be willing to police larvae that try to selfishly become queens in much the same way as parents should try to prevent young from being overly greedy (Clutton-Brock & Parker, 1995; Mock & Parker, 1998). As larval manipulation of caste fate brings colony productivity down from 1 to 1 − z* (cf. eqn 1) with z* = (1 − Rf)/(1 + Rm) (eqn 4), workers are expected to prevent larvae from becoming queens if this can be done at a lesser cost (c), i.e. when

  • image(8)

(The cost of policing is measured as that which would occur in a colony where all females would become workers; for a more formal derivation see Wenseleers, 2001.)

From eqn (6), the prediction is that workers should try to gain social control over larval caste determination whenever it can be achieved at low cost. This is more probable when queens are larger than workers, as workers can then withhold food from larvae to hinder them from developing into a queen (Bourke & Ratnieks, 1999). However, costs, such as accidental starving of larvae, may be larger in societies of low relatedness, as policing of larval selfishness can then cause greater potential increases in colony productivity. This parallels the conclusion of Frank (1995b) that low relatedness is conducive to the spread of policing although his result, due to the simpler kin structure modelled, was simply that policing is favoured when c < 1 − r. Even in societies of high relatedness, however, policing may be expected if it is not too costly. For example, in societies with a once-mated queen, workers could prevent larvae from becoming queens if social control had a cost of <20% (eqn 6).

Empirical support for caste conflict theory

  1. Top of page
  2. Abstract
  3. Introduction
  4. The model
  5. Individual caste allocation interests
  6. Worker interests
  7. Empirical support for caste conflict theory
  8. Discussion
  9. Acknowledgments
  10. References
  11. Appendix. Caste conflict from a levels-of-selection perspective

Strong general support for the model predictions comes from the clear association between queen overproduction and conditions allowing individual control of caste fate. In Melipona, where queens and workers are reared from the same cells, there is great queen overproduction: on average between 5 and 15% of all females usually develop as queens across different species (e.g. Kerr, 1950; Darchen & Delage-Darchen, 1975; Koedam et al., 1999; Moo-Valle et al., 2001; Gioli & Bego, 2002; Sommeijer et al., 2003). Similarly, excess queens develop in some trigonine stingless bees where females have the ability to develop as dwarf queens from worker cells – a situation that has been specifically likened to Melipona (Engels & Imperatriz-Fonseca, 1990; Schwarziana quadripunctata: production of 25–30 gynes in 10 days, Ribeiro & Alves, 2001; Plebeia julianii: Juliani, 1962, 1967; Plebeia remota: production of up to six gynes per month, Ribeiro et al., 2003). In most species, however, excess queen production can be effectively prevented by nutritional control. When queens are larger than workers, developing females can be forced into a worker role by giving them insufficient food to allow them to develop as queens. This is a type of social policing (Ratnieks, 1988), and could be termed ‘caste fate policing’. Nutritional caste control can be extremely effective at preventing excess queen production. Taxa where feeding control is possible, such as army ants, honeybees and most trigonine stingless bees (with the exception of the genera that produce dwarf queens), all produce few queens, in keeping with colony needs. Apis mellifera, for example, produces about 15–25 queens vs. 150 000 workers over one season when swarming occurs (Seeley, 1985; Winston, 1987). This implies that the chance for a female to be reared as a queen is only 0.01–0.02%. Similar ratios have been observed in the trigonine stingless bee Tetragonisca angustula (van Veen & Sommeijer, 2000), where approximately four queens develop over a single season, amounting to 0.01–0.02% of the female brood. In army ants, finally, where queens are 50 times the size of workers, few queens are generally reared (Aenictus gracilis: <6, Macevicz, 1979; Dorylus wilverthi: 1–2, Raignier, 1959; Eciton hamatum: 1–14, Schneirla & Brown, 1952; Schneirla, 1971; Franks & Hölldobler, 1987; the concurrent worker brood ranges in size from 30 000 to 300 000; Schneirla, 1971). That caste conflict is resolved so efficiently is surprising given the huge potential for conflict that is present. As mentioned above, in the honeybee relatedness is low (Rf = 0.3; Estoup et al., 1994) because of polyandry and the inclusive fitness interest of an individual female is to become a queen with a probability of 56% (assuming a linear cost function).

In future, more detailed predictions of the theory could also be tested. For example, it is predicted that fewer females should selfishly develop as queens when all males are workers’ sons (Table 2). Unfortunately, there are currently good data on male parentage and caste ratios for only few species. Therefore, a meaningful test will be possible only once more data have accumulated. One species that would be particularly valuable for a more detailed test is M. bicolor. In this species, the predicted queen production is higher than in other Melipona as a result of polygyny, which reduces relatedness (Table 2). Although no detailed caste ratio data are available for M. bicolor, da Silva et al. (1972) noted that bicolor is exceptional within Melipona in that it ‘produces an unusual [high] quantity of virgin queens’. If M. bicolor larvae could detect how many laying queens there are in the colony, they would also benefit by facultatively adjusting their probability of becoming a queen (Ratnieks, 2001). Given that there are other examples of facultative adjustment of reproduction in social Hymenoptera, a facultative response seems quite possible (e.g. Sundström et al., 1996; Foster & Ratnieks, 2000). Future studies could attempt to test this.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. The model
  5. Individual caste allocation interests
  6. Worker interests
  7. Empirical support for caste conflict theory
  8. Discussion
  9. Acknowledgments
  10. References
  11. Appendix. Caste conflict from a levels-of-selection perspective

In this paper, we have investigated a key topic in insect sociobiology – the conflicts associated with the rearing of individuals of different reproductive and working ability. More than 20 years ago, Oster & Wilson (1978) noted that ‘the interaction of individual-level and colony-level selection complicates the evolution of caste in ways that are just beginning to be explored’. Indeed, this is exactly what our current analysis shows. The decision to become a queen or a worker depends upon an interaction of individual benefits and costs to the whole colony. Feeding control of larvae by adults can maximize the success of the colony by preventing excess numbers of individuals from selfishly developing as queens. For example, in the honeybee, only 1 in 10 000 females are reared as queens (Seeley, 1985; Winston, 1987), although the individual optimum is to become a queen with a probability of 56%. Clearly, nutritional caste determination is a very powerful form of social suppression, which makes individuals work for the benefit of society even when this goes against their inclusive fitness interests. In general terms, our results reinforce the view that parental or sibling manipulation, in addition to kin selection, is of major importance in resolving social conflicts (Alexander, 1974, 1987; Michener & Brothers, 1974; Hamilton, 1975; Leigh, 1977; Frank, 1995b; Monnin & Ratnieks, 2001; Strassmann et al., 2002).

If the honeybee illustrates a situation where caste conflict is efficiently resolved by social control then Melipona illustrates the reverse. Weak size differences between queen and worker and rearing in sealed cells away from any interference by adults allow every female larva to control her own fate. In line with predictions, queens in Melipona are produced in great excess of colony needs. Excess queen production also occurs in some trigonine stingless bees where females evade social control by developing as dwarf queens in worker cells (Engels & Imperatriz-Fonseca, 1990; Ribeiro et al., 2003). Significantly, excess queen production only occurs in those trigonines where self-determination is possible. This represents good and phylogenetically independent evidence for caste conflict theory.

Our inclusive fitness model also makes more detailed predictions on the expected level of excess queen production as a function of the colony kin structure. For single mating, the model recovers the predictions of Ratnieks (2001) that between 14 and 20% of all females should develop as a queen depending on male parentage. But the model also makes several new predictions, for example, deriving the optima under polygyny (Table 2), or for nonlinear colony-level cost functions. These additional predictions will be useful for future tests of caste conflict theory.

Recently, Strassmann et al. (2002) gave evidence for caste conflict in the swarm-founding epiponine wasp Parachartergus colobopterus. They suggest that in epiponines, caste conflict may lead to occasional episodes of queen overproduction, leading to the colony becoming polygynous. Parachartergus colobopterus is a species without morphological castes, and who becomes a queen depends purely on whether females mate and activate their ovaries (Strassmann et al., 2002). Caste conflict in P. colobopterus is, therefore, over an individual female's behavioural rather than morphological caste fate. Consistent with the interpretation of caste conflict, queens in polygynous societies tend to be aggressively suppressed by the workers (Herman et al., 2000; Strassmann et al., 2002). Aggression can force queens to work rather than reproduce, which has the advantage that neither they nor the other workers have to pay the cost of queen killing, as in Melipona (Herman et al., 2000; Strassmann et al., 2002). It is possible, however, that aggressive suppression does eventually cause a reduction in queen number, and that it helps to drive the queen number cycles typical of epiponine wasps (Jeanne, 1991). An additional observation consistent with occasional queen overproduction is that polygyny does not increase colony productivity (Forsyth, 1980). Strassmann et al. (2002) also developed a model to show why P. colobopterus only produces excess queens when there is just a single mother queen left. The conclusion of their model was that this could be the result of collective worker interests, which favour the adoption of new queens especially at low queen number (because they would then replace full siblings). An alternative that was not discussed, however, is that females may benefit most from becoming a queen at low queen number, as they will then have fewer competitors.

Although our discussion has been primarily concerned with caste conflict in the social Hymenoptera it is worth considering whether similar conflicts occur in the other major group of social insects, the termites. In particular, in the ‘lower’ termite families Kalotermitidae and Termopsidae, overt caste conflict might be expected, as all individuals (except soldiers) are totipotent and can moult into either a queen or king (Shellman-Reeve, 1997; Thorne, 1997; Myles, 1999). In fact, replacement reproduction, the process whereby some individuals moult into a queen or king in case of accidental loss of the royal pair, seems formally identical to caste conflict. Any individual would benefit from becoming a replacement reproductive, but future colony productivity would be reduced if an excess do so (for a review, see Myles, 1999). Indeed, removing the royal pair in some species results in a vast overproduction of replacement reproductives (up to 50%), and the excess are subsequently killed (Termopsidae: Porotermes adamsoni, Mensa-Bonsu, 1976; Kalotermitidae: Cryptotermes brevis, Lenz et al., 1985; Kalotermes flavicollis, Lüscher, 1952, 1964; Ruppli, 1969; Neotermes connexus, Myles & Chang, 1984; Neotermes jouteli, Nagin, 1972). This closely resembles the queen cull in Melipona. Similar conflicts also occur at other evolutionary levels. For example, in slime moulds, the decision of a cell to become either a reproductive spore or a sterile stalk cell parallels the decision of a social insect larva to develop as a queen or a worker (Bourke & Ratnieks, 1999). As in caste fate conflict, conflict over cell-fate is expected especially at low relatedness, which can arise as a result of the mixing of different clones during fruiting body formation (Hudson et al., 2002). Indeed, chimaerism appears common in nature (Strassmann et al., 2000), and there is some evidence that actual conflict does occur (Atzmony et al., 1997; Csaba & Balázs, 2000). This shows how the concept of caste conflict will probably help towards the understanding of a wide variety of social conflicts.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. The model
  5. Individual caste allocation interests
  6. Worker interests
  7. Empirical support for caste conflict theory
  8. Discussion
  9. Acknowledgments
  10. References
  11. Appendix. Caste conflict from a levels-of-selection perspective

We thank Andrew Bourke, Kevin Foster, Steve Frank, Adam Hart, Judith Korb, Dave Queller, Thomas Platt, Max Reuter, Joan Strassmann, Lotta Sundström and Eva Tóth for discussion or comments. This work was supported by the ‘Social Evolution’ and ‘INSECTS’ EU-TMR networks, by a grant from the Flemish Fund for Scientific Research (FWO-Vlaanderen, grant no. 2.0014.94) and by a EU Marie Curie fellowship to TW.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. The model
  5. Individual caste allocation interests
  6. Worker interests
  7. Empirical support for caste conflict theory
  8. Discussion
  9. Acknowledgments
  10. References
  11. Appendix. Caste conflict from a levels-of-selection perspective
  • Alexander, R.D. 1974. The evolution of social behavior. Annu. Rev. Ecol. Syst. 5: 325383.
  • Alexander, R.D. 1987. The Biology of Moral Systems. Aldine de Gruyter, New York.
  • Atzmony, D., Zahavi, A. & Nanjundiah, V. 1997. Altruistic behaviour in Dictyostelium discoideum explained on the basis of individual selection. Curr. Sci. 72: 142145.
  • Bego, L.R. 1989. Behavioral interactions among queens of the polygynic stingless bee, Melipona bicolor bicolor Lepeletier (Hymenoptera, Apidae). Braz. J. Med. Biol. Res. 22: 587596.
  • Bourke, A.F.G. & Chan, G.L. 1999. Queen-worker conflict over sexual production and colony maintenance in perennial social insects. Am. Nat. 154: 417426.
  • Bourke, A.F.G. & Franks, N.R. 1995. Social Evolution in Ants. Princeton University Press, Princeton, NJ.
  • Bourke, A.F.G. & Ratnieks, F.L.W. 1999. Kin conflict over caste determination in social Hymenoptera. Behav. Ecol. Sociobiol. 46: 287297.
  • Brandon, R.N. & Burian, R.M. 1984. Genes, Organisms, Populations: Controversies Over the Units of Selection. MIT Press, Cambridge, MA.
  • Chinh, T.X., Gijs, G.B.J., Meeuwsen, F.J.A.J. & Sommeijer, M.J. 2003. Patterns of male production in the stingless bee Melipona favosa (Apidae, Meliponini). Apidologie 34: 161170.
  • Clutton-Brock, T.H. & Parker, G.A. 1995. Punishment in animal societies. Nature 373: 209216.
  • Contel, E.P.B. & Kerr, W.E. 1976. Origin of males in Melipona subnitida estimated from data of an isozymic polymorphic system. Genetica 46: 271279.
  • Crozier, R.H. & Pamilo, P. 1996. Evolution of Social Insect Colonies. Sex Allocation and Kin Selection. Oxford University Press, Oxford.
  • Csaba, P. & Balázs, P. 2000. Selfish cells threaten multicellular life. Trends Ecol. Evol. 15: 351352.
  • Darchen, R. & Delage-Darchen, B. 1975. Contribution à l’étude d'une abeille du Mexique Melipona beecheii B. (Hymenoptère: Apide). Ap i dologie 6: 295339.
  • Dugatkin, L.A. & Reeve, H.K. 1994. Behavioral ecology and levels of selection – dissolving the group selection controversy. Adv. Stu. Behav. 23: 101133.
  • Engels, W. & Imperatriz-Fonseca, V.L. 1990. Caste development, reproductive strategies, and control of fertility in honey bees and stingless bees. In: Social Insects. An Evolutionary Approach to Castes and Reproduction (W.Engels ed.), pp. 167230. Springer-Verlag, Berlin, Heidelberg.
  • Estoup, A., Solignac, M. & Cornuet, J.-M. 1994. Precise assessment of the number of patrilines and of genetic relatedness in honey bee colonies. Proc. R. Soc. Lond. B Biol. Sci. 258: 17.
  • Forsyth, A. 1980. Worker control of queen density in Hymenopteran societies. Am. Nat. 116: 895898.
  • Foster, K.R. & Ratnieks, F.L.W. 2000. Facultative worker policing in a wasp. Nature 407: 692693.
  • Frank, S.A. 1995a. George Price's contributions to evolutionary genetics. J. Theor. Biol. 175: 373388.
  • Frank, S.A. 1995b. Mutual policing and repression of competition in the evolution of cooperative groups. Nature 377: 520522.
  • Frank, S.A. 1998. The Foundations of Social Evolution. Princeton University Press, Princeton, NJ.
  • Franks, N.R. & Hölldobler, B. 1987. Sexual competition during colony reproduction in army ants. Biol. J. Linn. Soc. 30: 229243.
  • Gioli, L.D. & Bego, L.R. 2002. O estoque de alimento e sua influência no processo reprodutivo de Melipona asilvae (Moure, 1971), com especial ênfase à produção de machos (Hymenoptera: Apidae: Meliponinae). In: V Encontro sobre abelhas (C.A. Garófalo & G. Freita, eds), p. 323. Ribeirão Preto, Brazil.
  • Gliddon, C.J. & Gouyon, P.H. 1989. The units of selection. Trends Ecol. Evol. 4: 204208.
  • Hamilton, W.D. 1964. The genetical evolution of social behaviour. I & II. J. Theor. Biol. 7: 152.
  • Hamilton, W.D. 1975. Innate social aptitudes in man: an approach from evolutionary genetics. In: Biosocial Anthropology (R.Fox, ed.), pp. 133155. Wiley, New York.
  • Herbers, J.M., DeHeer, C.J. & Foitzik, S. 2001. Conflict over sex allocation drives conflict over reproductive allocation in perennial social insect colonies. Am. Nat. 158: 178192.
  • Herman, R.A., Queller, D.C. & Strassmann, J.E. 2000. The role of queens in colonies of the swarm-founding wasp Parachartergus colobopterus. Anim. Behav. 59: 841848.
  • Hudson, R.E., Aukema, J.E., Rispe, C. & Roze, D. 2002. Altruism, cheating, and anticheater adaptations in cellular slime molds. Am. Nat. 160: 3143.
  • Hölldobler, B. & Wilson, E.O. 1990. The Ants. Springer-Verlag, Berlin, Heidelberg.
  • Imperatriz-Fonseca, V.L. & Zucchi, R. 1995. Virgin queens in stingless bee (Apidae, Meliponinae) colonies: a review. Apidologie 26: 231244.
  • Jeanne, R.L. 1991. Chapter 6. The swarm-founding Polistinae. In: The Social Biology of Wasps (K. G.Ross & R. W.Matthews, eds), pp. 191231. Cornell University Press, Ithaca, NY.
  • Juliani, L. 1962. O aprisionamento de rainhas virgens em Trigonini. Bol. Univ. Paraná Zool. 20: 111.
  • Juliani, L. 1967. A descriçao do ninho e alguns dados biológicos sobre a abelha Plebeia juliani Moure. Rev. Bras. Entomol. 12: 3158.
  • Keller, L. 1999. Levels of Selection. Princeton University Press, New Jersey.
  • Kerr, W.E. 1950. Genetic determination of castes in the genus Melipona. Genetics 35: 143152.
  • Kerr, W.E. 1975. Evolution of population structure in bees. Genetics 79: 7384.
  • Kerr, W.E., Zucchi, R., Nakadaira, J.T. & Butolo, J.E. 1962. Reproduction in the social bees (Hymenoptera: Apidae). J. NY Entomol. Soc. 70: 265276.
  • Koedam, D., Contrera, F.A.L. & Imperatriz-Fonseca, V.L. 1999. Clustered male production by workers in the stingless bee Melipona subnitida Ducke (Apidae, Meliponinae). Insectes Soc. 46: 387391.
  • Leigh, E.G.J. 1977. How does selection reconcile individual advantage with the good of the group? Proc. Natl. Acad. Sci. USA 74: 45424546.
  • Lenz, M., Barrett, R.A. & Williams, E.R. 1985. Reproductive strategies in Cryptotermes: neotenic production in indigenous and ‘tramp’ species in Australia (Isoptera: Kalotermitidae). In: Caste Differentiation in Social Insects (J. A. L.Watson, B. M.Okot-Kotber & C.Noirot, eds), pp. 147163. Pergamon Press, Oxford.
  • Lüscher, M. 1952. Die Produktion und Elimination von Ersatzgeschlechtstieren bei der Termite Kalotermes flavicollis Fabr. Z. vergl. Physiol. 34: 123141.
  • Lüscher, M. 1964. Die spezifische Wirkung männlicher und weiblicher Ersatzgeschlechtstiere auf die Enstehung von Ersatzgeslechtstieren bei der Termite Kalotermes flavicollis (Fab.). Insectes Soc. 11: 7990.
  • Macevicz, S. 1979. Some consequences of Fisher's sex ratio principle for social Hymenoptera that reproduce by colony fission. Am. Nat. 113: 363371.
  • Maynard Smith, J. 1998. The units of selection. CIBA Found. Symp. 213: 203211.
  • Maynard Smith, J. 2002. Equations of life – the mathematics of evolution. In: It Must be Beautiful – Great Equations of Modern Science (G.Farmelo, ed.), pp. 193211. Granta Books, London.
  • Mensa-Bonsu, A. 1976. The production and elimination of supplementary reproductives in Porotermes adamsoni (Froggatt) (Isoptera: Hodotermitidae). Insectes Soc. 23: 133154.
  • Michener, C.D. 1974. The Social Behavior of the Bees. Harvard University Press, Cambridge, MA.
  • Michener, C.D. & Brothers, D.J. 1974. Were workers of eusocial Hymenoptera initially altruistic or oppressed? Proc. Natl. Acad. Sci. USA 71: 671674.
  • Mock, D.W. & Parker, G. 1998. The Evolution of Sibling Rivalry. Oxford University Press, Oxford.
  • Monnin, T. & Ratnieks, F.L.W. 1999. Reproduction versus work in queenless ants: when to join a hierarchy of hopeful reproductives? Behav. Ecol. Sociobiol. 46: 413422.
  • Monnin, T. & Ratnieks, F.L.W. 2001. Policing in queenless ponerine ants. Behav. Ecol. Sociobiol. 50: 97108.
  • Monnin, T., Ratnieks, F.L.W., Jones, G.R. & Beard, R. 2002. Pretender punishment induced by chemical signalling in a queenless ant. Nature 419: 6165.
  • Moo-Valle, H., Quezada-Euan, J.J.G. & Wenseleers, T. 2001. The effect of food reserves on the production of sexual offspring in the stingless bee Melipona beecheii (Apidae, Meliponini). Insectes Soc. 48: 398403.
  • Myles, T.G. 1999. Review of secondary reproduction in termites (Insecta: Isoptera) with comments on its role in termite ecology and social evolution. Sociobiology 33: 191.
  • Myles, T.G. & Chang, F. 1984. The caste system and caste mechanisms of Neotermes connexus (Isoptera, Kalotermitidae). Sociobiology 9: 163321.
  • Nagin, R. 1972. Caste determination in Neotermes jouteli (Banks). Insectes Soc. 19: 3961.
  • Nogueira-Neto, P. 1970. A criação de abelhas indígenas sem ferrão (Meliponinae). Chácaras e Quintais, São Paulo.
  • Oster, G.F. & Wilson, E.O. 1978. Caste and Ecology in the Social Insects. Princeton University Press, Princeton, NJ.
  • Pamilo, P. 1991a. Evolution of colony characteristics in social insects 1. Sex allocation. Am. Nat. 137: 83107.
  • Pamilo, P. 1991b. Evolution of colony characteristics in social insects 2. Number of reproductive individuals. Am. Nat. 138: 412433.
  • Paxton, R.J., Weissschuh, N., Engels, W., Hartfelder, K. & Quezada-Euan, J.J.G. 1999. Not only single mating in stingless bees. Naturwissenschaften 86: 143146.
  • Paxton, R.J., Ruhnke, H., Shah, M., Bego, L.R., Quezada-Euan, J.J.G. & Ratnieks, F.L.W. 2001. Social evolution in stingless bees: are the workers or is the queen in control of male production? In: II Seminario Mexicano sobre abejas sin aguijón, pp. 104107. Universidad Autónoma de Yucatán, Mérida, Yucatán, México.
  • Peters, J.M., Queller, D.C., Imperatriz-Fonseca, V.L., Roubik, D.W. & Strassmann, J.E. 1999. Mate number, kin selection and social conflicts in stingless bees and honeybees. Proc. R. Soc. Lond. B Biol. Sci. 266: 379384.
  • Price, G.R. 1972. Extension of covariance selection mathematics. Ann. Hum. Genet. 35: 455458.
  • Queller, D.C. 1992. Quantitative genetics, inclusive fitness, and group selection. Am. Nat. 139: 540558.
  • Raignier, A. 1959. Het ontstaan van kolonies en koninginnen bij de Afrikaanse trekmieren. Meded. Koninkl. Vlaamse Acad. Wetensch., Letteren en Schone Kunsten België 21: 124.
  • Ratnieks, F.L.W. 1988. Reproductive harmony via mutual policing by workers in eusocial Hymenoptera. Am. Nat. 132: 217236.
  • Ratnieks, F.L.W. 1989. Conflict and Cooperation in Insect Societies, 242 pp. PhD Thesis, Faculty of the Graduate School of Cornell University, Cornell University, Cornell.
  • Ratnieks, F.L.W. 2001. Heirs and spares: caste conflict and excess queen production in Melipona bees. Behav. Ecol. Sociobiol. 50: 467473.
  • Ratnieks, F.L.W. & Reeve, H.K. 1992. Conflict in single-queen hymenopteran societies – the structure of conflict and processes that reduce conflict in advanced eusocial species. J. Theor. Biol. 158: 3365.
  • Ratnieks, F.L.W., Monnin, T. & Foster, K.R. 2001. Inclusive fitness theory: novel predictions and tests in eusocial Hymenoptera. Ann. Zool. Fenn. 38: 201214.
  • Reeve, H.K. & Nonacs, P. 1992. Social contracts in wasp societies. Nature 359: 823825.
  • Reuter, M. & Keller, L. 2001. Sex ratio conflict and worker production in eusocial Hymenoptera. Am. Nat. 158: 166177.
  • Ribeiro, M.F. & Alves, D.A. 2001. Size variation in Schwarziana quadripunctata queens (Hymenoptera, Apidae, Meliponini). Rev. Etol. 3: 5965.
  • Ribeiro, M.F., Imperatriz-Fonseca, V.L. & Santos Filho, P.S., 2003. Exceptional high queen production in the Brazilian stingless bee Plebeia remota. Stud. Neotrop. Fauna Environ, 37: in press.
  • Ruppli, E. 1969. Die Elimination überzähliger Ersatzgeschlechtstiere bei der Termite Kalotermes flavicollis (Fab.). Insectes Soc. 16: 235248.
  • Schneirla, T.C. 1971. Army Ants: A Study in Social Organisation. W.H. Freeman, San Francisco, CA.
  • Schneirla, T.C. & Brown, R.Z. 1952. Sexual broods and the production of young queens in two species of army ants. Zool. NY 37: 532.
  • Seeley, T.D. 1985. Honeybee Ecology. A Study of Adaptation in Social Life. Princeton University Press, Princeton, NJ.
  • Seeley, T.D. 1995. The Wisdom of the Hive. Harvard University Press, London, England.
  • Shellman-Reeve, J.S. 1997. The spectrum of eusociality in termites. In: The Evolution of Social Behavior in Insects and Arachnids (J.Choe & B. J.Crespi, eds), pp. 5293. Cambridge University Press, Cambridge.
  • Da Silva, D.L.N. 1977. Estudos bionomicos em colônias mistas de Meliponinae. Bol. Zool. São Paulo 2: 7106.
  • Da Silva, D.L.N., Zucchi, R. & Kerr, W.E. 1972. Biological and behavioural aspects of the reproduction in some species of Melipona (Hymenoptera, Apidae, Meliponinae). Anim. Behav. 20: 123132.
  • Sommeijer, M.J., Chinh, T.X. & Meeuwsen F.J.A.J. 1999. Behavioural data on the production of males by workers in the stingless bee Melipona favosa (Apidae, Meliponinae). Insectes Soc. 46: 9293.
  • Sommeijer, M.J., De Bruijn, L.L.M., , Meeuwsen, F.J.A.J. & Martens, E.P. 2003. Natural patterns of caste and sex allocation in the stingless bees Melipona favosa and M. trinitatis related to worker behaviour. Insectes Soc. 50: 3844.
  • Strassmann, J.E. 1989. Early termination of brood rearing in the social wasp, Polistes annularis (Hymenoptera, Vespidae). Journal of the Kansas Entomological Society 62: 353362.
  • Strassmann, J.E., Zhu, Y. & Queller, D.C. 2000. Altruism and social cheating in the social amoeba Dictyostelium discoideum. Nature 408: 965967.
  • Strassmann, J.E., Sullender, B.W. & Queller, D.C. 2002. Caste totipotency and conflict in a large-colony social insect. Proc. R. Soc. Lond. B Biol. Sci. 269: 263270.
  • Sundström, L., Chapuisat, M. & Keller, L. 1996. Conditional manipulation of sex ratios by ant workers: a test of kin selection theory. Science 274: 993995.
  • Thorne, B.L. 1997. Evolution of eusociality in termites. Annu. Rev. Ecol. Syst. 28: 2754.
  • Tóth, E., Strassmann, J.E., Nogueira-Neto, P., Imperatriz-Fonseca, V.L. & Queller, D.C. 2002. Male production in stingless bees: variable outcomes of queen-worker conflict. Mol. Ecol. 11: 26612667.
  • van Veen, J.W. & Sommeijer, M.J. 2000. Colony reproduction in Tetragonisca angustula (Apidae, Meliponini). Insectes Soc. 47: 7075.
  • Velthuis, H.H.W. & Sommeijer, M.J. 1991. Roles of morphogenetic hormones in caste polymorphism in stingless bees. In: Morphogenetic Hormones of Arthropods (A. P.Gupta, ed.), pp. 346383. Rutgers University Press, New Brunswick.
  • Wade, M.J. 1980. Kin selection: its components. Science 210: 665667.
  • Wenseleers, T. 2001. An introduction to the evolutionary modelling of conflict. In: Proceedings of the EU ‘INSECTS’ Training Workshop ‘The Mathematical Modelling of Social Behaviour’, pp. 2153. Great Hucklow, UK, 1418 September.
  • Wenseleers, T. & Ratnieks, F.L.W. 2003. Caste determination in stingless bees. In: Mexican Stingless Bees: Their Biology and Management (J. J. G.Quezada-Euán, ed.), in press.
  • West-Eberhard, M.J. 1967. Foundress associations in polistine wasps: dominance hierarchies and the evolution of social behavior. Science 157: 15841585.
  • West-Eberhard, M.J. 1978. Temporary queens in Metapolybia wasps: nonreproductive helpers without altruism? Science 200: 441443.
  • West-Eberhard, M.J. 1981. Intragroup selection and the evolution of insect societies. In: Natural selection and social behavior (R. D.Alexander & D. W.Tinkle, eds), pp. 317. Chiron, Comcord, MA.
  • Wheeler, D.E. 1986. Developmental and physiological determinants of caste in social Hymenoptera – evolutionary implications. Am. Nat. 128: 1334.
  • Wilson, E.O. 1971. The Insect Societies. Harvard University Press, Cambridge, MA.
  • Wilson, D.S. 1983. The group selection controversy – history and current status. Annu. Rev. Ecol. Syst. 14: 159187.
  • Winston, M.L. 1987. The Biology of the Honey Bee. Harvard University Press, Cambridge, MA.

Appendix. Caste conflict from a levels-of-selection perspective

  1. Top of page
  2. Abstract
  3. Introduction
  4. The model
  5. Individual caste allocation interests
  6. Worker interests
  7. Empirical support for caste conflict theory
  8. Discussion
  9. Acknowledgments
  10. References
  11. Appendix. Caste conflict from a levels-of-selection perspective

Caste conflict is expected because females are more highly related to own offspring than to a sister's offspring. Alternatively, one could say that females benefit from developing as queens because this increases their reproductive success relative to other females within the colony, although this might reduce the total reproduction of the group. This leads to an equivalent interpretation of caste conflict as a levels-of-selection problem, which can be analysed using the multilevel Price equation (Price, 1972; Frank, 1995a). For structured populations (e.g. individuals structured into groups), Price (1972) showed that selection may be partitioned into two components, one due to the differential success of individuals within groups, and the other due to the differential success of groups. Price's (1972) equation for gene frequency change, in simple form, ignoring mutation and drift, is βGzVB + βwyVW, where βGz is the effect of the average level of selfishness on group productivity (i.e. the group cost), βwy is the effect of the individual level of selfishness on individual fitness (i.e. the individual benefit) and the VB and VW are between- and within-group genetic variances, respectively, which give appropriate weightings to the group and the individual as units-of-selection. Hamilton (1975) showed that when groups are large, the between-group genetic variance is proportional to relatedness, and the within-group genetic variance is proportional to 1-relatedness. The Price equation, therefore, both resolves the kin vs. group selection controversy and offers a precise answer to the dilemma of what really is the unit of selection (Brandon & Burian, 1984; Gliddon & Gouyon, 1989; Maynard Smith, 1998): groups can be considered units of selection in proportion to intragroup relatedness while individuals are units of selection in proportion to 1-relatedness. When groups are clones, natural selection simply strives for maximal group success and for maximal intragroup altruism. On the other hand, when groups are composed of unrelated individuals, natural selection favours high within-group selfishness, leading to lowered group success. In our case, the within-group variance is proportional to 1 − rf, but the between-group variance has two components, one for the genetic variation present in females, rf, and one for the genetic variation present in males, rm (rm does not feature in the within-group selection component as males do not actually express the caste determination phenotype. Therefore, no within-colony selection can act on them. However, between-colony selection does occur, as males contain copies of the caste determination gene, and fewer of these will be passed on if the colony productivity is reduced.) The individual benefit βwy is ∂Wf/party with Wf = [(1 − z)/(1 − Z)](y/z), i.e. (1 − z)/[(1 − Zz]; the group cost βGz is −1/(1 − Z) as G = (1 − z)/(1 − Z) and ∂G/∂z = −1/(1 − Z). Equilibrium occurs when components of selection, weighted by reproductive value, balance each other:

  • image(9)

with VB = Rf + Rm and VW = 1 − Rf. This equilibrium condition is exactly the same as that derived using the kin selection approach (eqn 2). The only difference is that the logic and calculation to arrive at the result were different. This reinforces the view that kin and group selection are merely two different ways of analysing gene frequency change (e.g. Hamilton, 1975; Wade, 1980; Wilson, 1983; Queller, 1992; Dugatkin & Reeve, 1994; Keller, 1999).