Insect societies

Social insects take cooperation/helping further than any other group of organism - individuals may be completely sterile, spending their entire life raising the young of others. Eusocial insects dominate in terms of biomass: e.g. considering ants alone: On average, ants monopolize 15-20% of the terrestrial animal biomass, and in tropical regions where ants are especially abundant, they monopolize 25% or more (Schultz, 2000). Approximately 12,000 spp. of social insects = more than all known spp. of birds and mammals (E.O Wilson 1975)

Eusociality involves fulfilment of three criteria: 1) Cooperative care of young (not just by the mother), 2) Sterile castes 3) overlap of generations (mother, adult offspring and young offspring alive at the same time) Note that many species show intermediate stages e.g. co-operative nest building but no sterile castes Occurs in three orders - hymenoptera (ants, bees, wasps), isoptera (termites) and homoptera (aphids) Also note that some have argued for this traditional definition being somewhat arbitrary: some are trying to expand it, and others are trying to narrow it (Wcislo 1997) - we should not let this be a blinder to studying natural variation

In hymenoptera, workers are all female (maybe because of haplodiploidy - see Davies et al 2012) Workers usually spend the first few weeks of life inside the colony (handle prey, feed larvae and queen, guard entrance) and spend the later part of their doing jobs outside the colony (foraging/defending). Caste differentiation usually non-genetic. E.g. honeybees - queen suppresses development of new queens through chemical signals that prevent workers from feeding 'royal jelly' to larvae

Marine invertebrates: Pelagic siphonophores (e.g. Portuguese Man'O'War) Social gall thrips - different castes (short winged females act as soldiers). Eusociality first claimed in 1992 (Crespi), but not completely sterile Most recently discovered taxon (Duffy, 1996) = snapping shrimp colonies that live in sponges - at least three origins of 4 eusocial spp. (Duffy et al 2010). Only known eusocial beetle is the Ambrosia beetle - diploid, colonies consist of a mother and a small number of unfertilized daughters eusocial aphids: 50-60 spp have soldiers 17 independent origins of eusociality, either in pemphiginae and hormaphidinae Vertebrates: only in mole rat spp. (defender and disperser sterile castes?)

Primitively eusocial, demonstrate the parasocial route Matriachal family: queens, workers and sexuals Workers are all female, help rear more workers plus sexuals. No sterile castes. Basic life cycle can be drawn as a graph: Biomass versus time - annual cycle slow to increase initially (founding phase) then becomes linear (ergonomic phase) then tails off and rapidly drops to zero (reproductive phase). During the reproductive phase, a subset of the population becomes reproductives (rises then sharply falls).

Advanced, truly eusocial hymenopteran All eusocial Wingless worker caste - first group of predatory esocial insects to colonise thei soil? Could explain their success Only hymenopterans where you get worker subcastes (e.g. soldiers, replets, minors etc) Swarm founding can occur. Typical life cycle (Davies et al 2012): Myrmica rubra = typical of temperate zone ants: single queen fertilized during 'nuptial flight' (lots of reproductive males/females swarming), then loses wings and spends first winter sealed in a nest chamber that she has built. Eggs develop into larvae and then into adult workers (until they mature, queen feeds them off her own reserves). Workers are female, but sterile. After around 9 years, the colony starts to produce a new generation of reproductives who take part in a nuptial flight. Eventually the queen dies and the old colony dies out with her.

All eusocial. Similar to ants, but from radically different phylogenetic base, remarkable superficial convergence of social organization. Different to ants in many ways: exopterygotes, not endopterygotes diploid, not haploid male and female workers, not just females Male reproductive stays with queen, helps make initial nest, and intermittently fertilizes her during colony growth vs no role in the colony for male reproductives in ants colonies potentially immortal in some sp versus colonies not immortal Cellulose diet vs most spp. predatory

Several cycles until the queen dies Independent founding: Same as polistes graph but it continues for several more ergonomic and reproductive phases, before a terminal decline when the queen dies. Swarm founding (e.g. honey bee): no founding phase (reproductive queen and some workers start it off together) - goes straight to ergonomic and reproductive phases [at each reproductive phase, there is fission of the colony, around half the workers leave too]

Independent: risky founding phase but - higher potential rate of reproduction and no local resource competition (no swarming females=don't have to compete for nests) (local resource competition causes swarm founding colonies to produce a very male-biased reproductive sex ratio). Overall, independent founding appears to be the more usual strategy - ancestral? Present worldwide, whilst swarm founding appears to be more common in the tropics Independent founding favoured by most ants, all termits and many wasps/bees Swarm founding favoured by many 'higher' bees e.g. honeybees. stingless bees, many wasps and some ants (e.g. army ants)

Simple societies have 1,000 Low queen-worker dimorphism versus high Highly complex individual worker behaviour, versus less complex (sterile - behaviour does not reflect immediate costs/benefits of eusociality) Simple nests versus elaborate High worker reproductive potential versus zero potential Simple communication versus elaborate Common physical aggression versus rare Little/no chemical signalling of reproductive status versus common Queen replacement present versus absent (not in termites though)

Demographic factors: 1) Larger colony sizes could promote greater division of labour (size-complexity hypothesis) 2) 'Life insurance': 'reproductive head starts' and 'assured fitness returns'. Ecological factors: Fortress construction/maintenance/defence

As colony size increases, both workers and queens will be selected to maximize their inclusive fitness by specializing in non-reproductive and reproductive roles, respectively. (increased division of labour becomes more advantageous when colonies are larger). Ferguson-Gow et al 2014 - phylogenetically controlled comparison = larger colonies of attine ants have greater division of labour, both in terms of worker-queen dimorpism and worker-worker variation. Supports the viewpoint that larger colony sizes drives increased division of labour. Increases in colony size may have thus had a role in evolution of eusociality, by promoting evolution of specialized castes.

Many primitively eusocial insects have relatively short lives compared to the long period for which offspring are potentially dependent on them. This is unlike vertebrate societies (potentially explaining why eusociality is more common in insects). The long period of dependency required for life insurance may have occurred ancestrally due to benefits of reduced impact of parasitism (see later). A lone foundress must survive the whole of this period of offspring dependence, if her offspring are to survive. e.g. in 19 spp. of polistes, 38-100% of lone foundresses were shown to fail before ANY adult offspring emerge. But workers in a nest can reduce these costs: 1) Reproductive head starts As soon as an adult worker emerges, it can raise offspring the instant it becomes active, giving a 'reproductive head start' over solitary breeders. Queller 1989 showed that this can give a selective advantage to worker behaviour in four spp. of polistes wasps, due to factors such as high adult mortality. 2) Assured fitness returns Even if workers die before the offspring reach mortality, other workers can continue to raise the offspring, giving assured fitness returns (Gadagkar 1990). Considering the social wasp Ropalidia marginata, Gadagkar estimated that demographic factors could theoretically increase a female's success by 3.6 fold when part of a group versus when mating alone. The maximum potential benefit from altruism due to haplodiploidy is 1.5 (as can raise sisters for whom r=0.75 instead of offspring, for whom r = 0.5). Thus, the potential importance of assured fitness returns in evolution of altruism may be greater than the importance of haplodiploidy. The theoretical benefits of assured fitness returns do seem to happen in practice too - FIELD 2000 confirmed that, in a tropical hover wasp, part-reared offspring could continued to be raised despite experimental removal of larvae. (large offspring were raised through to pupae, whilst smaller larvae were reared in accordance with the new reduced helper numbers in the nest; smaller offpring may also have been sacrificed to feed the larger brood). Overall, these life insurance advantages promote eusociality by reducing the costs of early mortality for a helper versus a lone breeder.

From Queller and Strassman 1998 1) Polistes - when unrelated females found a nest together - explained by the potential for direct fitness gains (e.g. Leadbeater et al 2011) [2) Unicolonial ants, which are characterized by huge colonies, many queens, and little aggression within a network of interconnected nests. Relatedness may approach zero, so little kin selection is possible (unless individuals can distinguish close kin from random colonymates; see below). Altruism might be maintained because workers in these species are too specialized to revert to a reproductive role. However, although this might account for the maintenance of altruism, a problem remains: With zero relatedness, traits of nonreproductive workers lose all heritability, and worker traits can no longer evolve adaptively (unless there is gene flow from non-unicolonial colonies, as in fire ants; Ross and Shoemaker 1993). Perhaps unicolonial forms are temporary and doomed to failure; this possibility is supported by their scattered, twiggy taxonomic distribution (Holldobler and Wilson 1990).]

Proposed by Hamilton (1964, 1972) as an explanation for the origin of altruism in hymenoptera. Most social insects are haplodiploid, it was argued, because only in haplodiploids does the special 0.75 relatedness among full sisters make it more profitable to raise siblings than offspring (rb > re). This high relatedness applies only among females, explaining why workers are always female in the haplodiploid Hymenoptera, but not in the diploid termites. Finally, because this relatedness advantage applies only when raising sisters, it can explain why workers sometimes still produce sons. However, with time, the importance of the haplodiploid hypothesis has been called into question: 1) Relatedness to females is often much less than 0.75 due to multiple matings or multiple queens, as this means that not all of the sisters are full sisters. e.g. Metcalf (1980): in polistes metricus, workers are related to sisters by an average of 0.65 (less than 0.75 due to multiple queen matings). BUT note - ancestral eusocials may have been singly mated (Hughes et al 2008) -evidence for the monnogamy hypothesis) and even with multiple matings, sister-sister relatedness is still higher in haplodiploids than in diploids with an equivalent number of matings, so the hypothesis could still be relevant. 2)more serious challenge - benefits of haplodiploidy are affected by sex ratio of reproductives: 1:1 sex ratios, favoured by the queen, result in haplodiploidy providing no additional benefits for workers. This is sometimes the case in present-day hymenoptera, where queens can effectively control the sex ratio - e.g. Rhitindoponera ants (Ward 1983). Even if workers can exert some control over the sex ratio such that it is female biased (as is thought to occur in most situations), haplodiploidy still provides no additional advantage unless sex ratios are female biased relative to the population as a whole (Davies et al 2012). Benefits counteracted by reduced mating success of daughters. 3) Thus, haplodiploidy can only have an appreciable affect in very limited contexts - split sex ratios (e.g. due to differing relatedness asymmetries - see below) or partially bivoltine life cycles. e.g. the life cycle of halictines There are also alternative explanations for the three principal phenomena explained by the haplodiploid hypothesis: 1) The fact that most cases of altruism are found in the Hymenoptera might be due to the unusually high frequency in this group of parental care, a useful precursor for evolving care of the young of others. Could favour 'parental manipulation' over 'daughter advantage'. Evidence: within hymenoptera, parental care is particularly elaborate in the aculeate group, to which all eusocial hymenoptera belong (Stubblefield and Charnov 1986). 2) Moreover, the providers of this parental care in the Hymenoptera have historically been female, so workers may be female simply because female-specific adaptations for nest building, homing, capturing and transporting of prey, and stinging have been extended to a helping context. 3) Finally, the fact that workers sometimes produce sons, but not daughters, could simply reflect the fact that daughter production requires the extra effort of mating. NB: recently,there has been a revival of the haplodiploidy hypothesis: Johnstone et al 2012 carried out a modelling study - found that if males disperse further from the natal colony than females, than haplodiploidy IS in fact more favourable than diploidy to the volution of reproductive altruism. [basically the low relatedness to brothers does not 'dilute' the benefits of high relatedness to sisters as much, due to the fact that brothers are more likely to leave the natal patch - thus the requirement for female-biased sex ratio to favour haplodiploidy no longer applies]. Whilst this is a modelling study, they conclude that it is a strong effect that may contribute both to helping among breeding females (e.g. Polistes) and the evolution of sterile female castes, without requiring biasing of the sex ratio or kin discrimination. \"in the case of worker production, the critical ratio of c to b below which invasion is possible is more than 50 per cent greater in haplodiploids than in diploids\" Evidence also suggests that, in nest-building humenoptera, male-biased dispersal is widespread. E.g. can measure sex-biased dispersal by comparing gene flow of mitochondrial (maternally inherited) and nuclear genes - in seven of nine such tests, including four Formica ant species, and in three other ant lineages in which tests have been conducted, gene flow was inferred to be male-biased. So the hypothesis is plausible, but we need more evidence of male-biased dispersal - e.g. investigate the phenomenon in more hymenoptera groups. e.g.2 it seems to be widespread today, but was male-biased dispersal common amongst hymenoptera before eusociality evolved? Without knowing this, we cannot be sure of the true importance of the haplodiploidy hypothesis.

Benefits only occur when sex ratio is female-biased relative to the population as a whole. If not, the extra relatedness to sisters (increase in r1) is counteracted by their decreased mating success relative to males (decrease in r2) (Davies et al 2012). Such circumstances can only occur in specific contexts: 1) Within a single generation - if there are split sex ratios 2) Between different generations - if there is a partial overlap (partially bivoltine life cycles)

[E.g Greenberg (1979) sweat bee workers (primitively eusocial) selectively exclude unrelated individuals from 'intruding' into the colony (probably recognise genetically determined family odours). E.g.2 in honeybees, where the queen may mate up to 20 times, workers can discriminate between sisters who are more or less related, and direct help towards relatives who share the same father (Page et al 1989). They thus derive more benefits from the genetic predisposition due to haplodiploidy.] But, these early results have been challenged: positive results in honeybees being challenged on the basis of statistical bias and artificial conditions (Carlin and Frumhoff 1990, Breed et al. 1994). The bulk of the evidence suggests that social insects do not perform this kind of discrimination (Keller 1997), although it would be premature to assert that they never do.

Instead of daughter advantage, parental manipulation could also be favoured: Subsocial route - mothers make a genetic gain, daughters don't make a genetic loss (Charnov 1978). This creates a genetic predisposition for mothers to 'persuade' daughters to stay at home, whilst daughters are 'willing victims' of this persuasion. This could favour parental manipulation as a driver of eusociality through the subsocial route. This is particularly relevant in aculeate hymenoptera

1) High levels of predation pressure (or ecological pressure from parasites/other conspecifics) can favour kin selected eusociality by increasing the ratio of b:c. According to Queller and Strassmann (1998), this can happen in two ways: i) Thrips, aphids, beetles and termites are 'fortress defenders' - tend to forage inside a nest or in a protected area, hence the main ecological advantage of grouping may be in allowing defence of valuable resources. Good example demonstrating these benefits = eusocial gall-forming aphids - pemphigus spyrothecae - in which soldiers clean the gall of defecated honeydew/dead aphids and defend against insect predators. When soldiers are removed, colony suffers increased predation and accumulation of waste, with reproductive success decreasing as a result (Foster 1990, Benton and Foster 1992). ii) Ants, bees and wasps may instead be referred to as 'life insurers', whose must forage outside of the nest to obtain food, such that the main ecological advantage comes from an overlap in adult lifetimes that provides extensive care to young. Life insurance advantages can relate to 'reproductive head starts' and 'assured fitness returns'. Empirical evidence comes from Queller (1989) and Gadagkar et al 1990. 2) Note that high predation pressure could also favour social behaviour according to two different regimes:

Can occur in primitively eusocial insects that demonstrate the parasocial route e.g. Polistes. Here, each foundress is a 'hopeful reproductive' but some will lose out and end up as subordinate helpers (who may eventually inherit the dominant position themselves). The potential direct fitness benefits from becoming a dominant individual may be evolutionarily beneficial if the chances of reproduction without nest sharing are low enough (as a result of ecological constraints), even if founders are unrelated (West Eberhard 1978). Hence, ecological constraints can drive evolution of eusociality even without genetic predispositions, by favouring reproductive cooperation (not altruism) Evidence for this being the case - in some Polistes spp. (a primitively eusocial wasp) - LEADBEATER ET AL 2011 - study of Polistes dominulus, where nest sharing amongst unrelated individuals is not only present but common. By genotyping offspring produced on more than 200 nests throughout the colony cycle, the study showed that although most subordinates produce few offspring, a minority may become dominants that are extremely productive, so that on average, nest sharing individuals produce more offspring of their own than lone breeders. I.e. the direct fitness benefits alone are enough for individuals to favour nest sharing.

Eusociality in termites (isoptera) may have also been favoured by a unique, third ecological constraint that can favour the subsocial route. Protozoans have to pass from anus to mouth between generations for individuals to effectively digest cellulose. Thus, ancestral termites had to stay at home long enough to become infected with these protozoa. Richard Dawkins (1979) even suggests that termite eusociality may have evolved due to protozoans manipulating them into staying at home to make an ideal environment for protozoans to spread!

Define eusociality. thought to have occurred through both subsocial and parasocial routes. The key question is what can promote reproductive altruism can arise - this can be explained by kin selection, according to hamilton's rule: altruism is favoured if the conditions of Hamilton's Rule are met. 1) Genetic Factors - increase r1 relative to r2 i) Genetic predisposition for altruism as helping is often to close relatives - daughter advantage is possible. ii) Daughter advantage may be particularly high in haplodiploid spp., though the magnitude of these benefits are small, and only occur in specific contexts. iii) Another unique case of high daughter advantage occurs in termites - cycles of inbreeding and outbreeding. v) Parental manipulation could instead provide a genetic predisposition for eusociality in the subsocial route. Particularly relevant in the aculeate hymenoptera. 2) Ecological constraints: increase b relative to c The selective pressures promoting eusociality can be those that make solitary nesting costly (decreasing the c term) as much as any genetic predispositions involving care to related individuals. This can be appreciated considering the example of carpenter bees. Overall, selective pressures that can raise costs of solitary breeding can be considered in several broad areas: i) VARIABLE ENVIRONMENT - e.g. distribution of resources abiotic factors (rainfall) thought to be important in vertebrate eusociality evolution. ii) predation/parasitism/other conspecific risk - fortress defenders versus life insurers. Note that these are not mutually exclusive (just one may have been evolutionarily more important than the other in each group) (e.g. the fact that hymenoptera have a sting could predispose them to fortress defence advantages, as well as the life insurance advantages). The long period of dependency required for life insurance may have occurred ancestrally due to benefits of reduced impact of parasitism (see later) iii) specific factors in termites - protozoa. Dawkins takes this even further - Clearly, ecological constraints and relatedness must work together for reproductive altruism to be favoured, according to Hamilton's Rule, creating context-dependent situations that may favour the origin of eusociality. This can be nicely demonstrated by considering Stark's work on carpenter bees. Non kin selected drivers of evolution: [In some primitively eusocial animals, ecological factors can be so great that they promote eusociality even in the absence of genetic predispositions that promote indirect fitness benefits. Example - polistes dominula (Leadbeater et al 2011). This kind of phenomenon could have been important in the evolution of eusociality through the parasocial route, before sterile castes evolved. [although there is a growing consensus that the parasocial route is relatively unimportant, so this may not be a serious driver]] Final point - there has been recent interest in a socially-imposed constraint for altruism, which create 'enforced altruism'. Possible evolutionary role important in hymenoptera, where coercion is often seen in the present day. Due to coercion due to policing by queen/workers Conclusion - For the most part, we can explain the evolution of eusociality through a mixture of genetic predispositions and ecological constraints that promote kin selection. In limited contexts, non kin-selected drivers of evolution could also play a role, by promoting nest sharing amongst unrelated individuals or creating enforced altruism. The relative benefits of each of the factors discussed vary in different contexts, but in all cases, the evolution of eusociality can be understood as part of inclusive fitness theory.

Queller and Strassman 1998 This is the only way for sterile castes to evolve: For kin selection to produce a sterile caste, genes for sterility must either be expressed conditionally or have low penetrance. A sterility gene that is always expressed never gets reproduced, even indirectly (i.e., through relatives), because any relatives with the gene are also sterile. However, sterility can evolve under kin selection if, for example, a sterility gene expressed only in poorly fed females causes them to help well-fed relatives, which can then transmit their unexpressed sterility genes. Any exceptions are only partial exceptions (genes have an effect within an environmentally determined subset).

A set of individuals specialized to perform one or more roles and limited to those roles. Roles = sets of behaviour that repeatedly occur throughout the species. How do we detect castes? Physical castes - different size = different roles Age = different age, different roles How are castes determined? Usually determined environmentally Physcial castes are a result of different allometric growth Initially they may be overlap in size, which could lead to complete dimorphism. Evidence for physical size determination: 1) - plotting thorax width against head width in army ants - smallest individuals are minors, followed by mediae, submajors and majors. Slopes different for each of the four worker subcastes. 2) Plotting zygomatic arch width against 5th lumbar vertebra in naked mole rats, workers correspond to a linear positive relationship, whilst breeding females are outliers to this relationship.

Early determination is characteristic of complex societies, late determination is characteristic of simple societies. There are often multiple switch points e.g. bombus. Always seems to involve juvenile hormone - see behaviour. Evidence that caste determination is not genetic: Individuals have the necessary genes for producing all possible castes. Example - RAJAKUMAR ET AL 2012 show that in Pheidole ants, individuals retain an ancestral development potential to produce a 'supersoldier caste' - inject a species that does not normally produce a supersoldier with juvenile hormone, and they produce a supersoldier caste.

Parasocial route and subsocial Both demonstrated by modern day primitively eusocial organisms, so both are plausible - sweat bees (halictidae) are subsocial, paper wasps (polistines) are parasocial. Growing consensus that the parasocial route has not been important in the evolution of true eusociality?

Invented by hamilton, named by Dawkins in the selfish gene. Has been discussed as a thought experiment for how a gene for social action might spread in a population. 1) display an obvious trait 2) distinguish between those who do and do not display a trait 3) act selectively altruistically towards those that do display the trait. Potential example in the social insects = the red fire ant - see behaviour. Originally demonstrated by Keller and Ross (1998) Monogynous colonies = BB queens, BB workers who kill all extra queens (ancestral situation) Polygynous colonies = Bb queens (bb die at an early age), Bb workers>>BB ; Bb workers kill BB queens b allele is a green beard gene - whose effects are determined eight nucleotide differences in the coding sequences at the Gp-9 sequence (Bourke 2002), which codes for a pheremone-recognition protein. Furthermore, Gp-9 lies within a large chromosomal inversion (Wang et al 2013), which contains co-adapted sets of alleles that are maintained by suppression of recombination. These other alleles presumably affect other phenotypes of BB versus Bb individuals. Overall, this results in b meeting the requirements of green beard genes: b individuals appear to have a specific surface chemical cue onto the surface = a 'green beard' Differences in gp-9 allow workers with the b allele (Bb) to recognise the lack of this chemical cue on BB queens, acting as a signal for them to be killed, whilst Bb queens are tolerated = selective provision of altruism to individuals with the green beard. Evidence for this comes from the fact that these differences in queen surface chemicals can be experimentally transferred, causing changes in worker behaviour: workers rubbed against BB queens were killed by other workers, but those rubbed against Bb queens were not (Keller and Ross 1998) [they now have surface chemicals of a queen, but lack the cue required for them not to be killed] Overall, this green beard gene maintains social behaviour in polygynous colonies. Monogynous colonies are the ancestral situation. Note that, in polygynous, the presence of Bb workers stops BB workers from killing the queens. But, if BB exceed 15%, they do kill additional queens, creating a transition to a monogamous colony.

Here, c and b can be positive or negative Cooperation - both individuals increase their direct fitness rb+c>0 (relatedness favours the evolution of cooperation) Altruism -actor sacrifices its own direct fitness whilst increasing the direct fitness of the recipient. rb-c>0 (not favoured unless individuals are related) Selfishness - actor increases its own direct fitness whilst decreasing the fitness of the recipient. c-rb>0 (relatedness acts as a brake on selfishness) Spite (only real evidence is in bacteria) harm inflicted on a recipient might become a benefit that potentially outweighs the self-harm experienced by the actor - only occurs when relatedness is negative. -c-rb>0

r1b - r2c =0 r1 = relatedness to beneficiary's offspring r2 = relatedness to own offspring

Work of Robert Stark (analysed by Bourke 1997) Xylocopa sulcatipes - an Arabian carpenter bee sp. [Nest in hollow plant stems Live as a single foundress or a group of two (usually a mother/daughter pair) Females are singly-mated When a mother and daughter pair, mother is the forager and reproductively dominant (eats daughters eggs so all offspring are mother's), whilst the daughter guards the nest - preventing usurpation by conspecifics Behaviour inside nests was observed by attaching small lead shapes to bees and observing them through the nest wall with an x ray machine. Results showed that the guard's presence meant that pairs of bees suffered far fewer usurpations than singletons, as well as boosting the brood size of non-usurped nests. In year 1, these effects were sufficient for helping by daughters to be favoured, according to hamilton's rule, whilst this was not the case in year 2 (where the reduction in frequency of usurpations was smaller). The benefits of sociality here are better defence against intraspecific attack and promotion of effective division of labour. The fact that helping was favoured in year 1 but not year 2 shows that the origin of eusociality depends on comparative costs and benefits - ecological variables can either promote or hinder reproductive altruism in different situations. Selective pressures maintaining sociality could be those that make nesting costly as much as those affecting groups directly (helping was more profitable in the first year because single nesters were usurped more often that year and, if they survived, were relatively unproductive). Maybe this explains why they are facultatively eusocial, and have not evolved true eusociality.

Metcalf and Whitt 1997: paper wasps (Polistes metricus) - sometimes nest as solitary queens and sometimes as two sisters sharing a nest. When two queen shares, one alpha female does most of the egg laying, whilst the 'beta female' passes on her genes mostly through her sister's offspring. By calclulating relatedness between sisters (using detectable enzyme polymorphisms), it is evident that the beta queen receives about as much genetic representation in the next generation as do solitary queens. This is because the benefits of overcoming ecological constraints (discussed above) mean that shared nests tend to produce more young. Hence, relatedness can mean that even the beta queen does not lose out in terms of genetic representation (hamilton's rule). Yagi et al 2012 Study of a halictine bee Measured fitness, maternity, relatedness etc. Showed that the indirect fitness of the sterile workers in their mother's nest was higher than the direct fitness of solitary reproductive females, and that the foundresses achieve a large direct fitness by having helpers. These fitness benefits are attributed to markedly higher larval survival rates in multiple-female nests.

The probability, over and above the average in the population, that a gene in one individual is shared by another. Measured by - pedigrees, genetic methods, DNA techniques esp. microsatellites (sequences of DNA consisting of repeats of brief nucleotide sequences) Microsatellites are useful because they are abundant, very variabl e.g. >30 alleles per locus in some ants Very small amounts of tissue can be typed due to PCR

Queller et al 1993 - colonies are monogynous (single queen) so any deviation from full-sisterhood bust stem from multiple matings. Microsatellite measurements provide evidence for low relatedness, which must imply extreme polyandry: e.g. Apis florea - mean of 8 matings Apis dorsata - mean of 27 matings

Primitively eusocial insects often have high reproductive skew e.g. study of microsatellite sequences in different nests (Sumner et al 2001 - Malaysian hover wasps) - in 11/13 nests, all eggs are laid by a single dominant female. This female obtains a large fitness benefit, whilst that there is evidence that worker's behaviour is influenced by how likely they are to inherit the dominant position. FIELD ET AL 2006 - Stenogastrines hover wasps Working hard increases indirect fitness benefits but decreases a worker's future direct fitness, as they may be less able to take over the dominant position. Lower rank workers work harder (less chance of reaching the dominant position, so makes sense to maximise indirect fitness) This was experimentally shown - if rank 2 females are removed, then current 'rank 3' females now become rank 2. After promotion, they work less hard (as they have a greater chance of inheriting the dominant position) This variation in helper effort can be thought of as due to differences in the 'c' term in Hamilton's rule - after promotion, the likelihood of a large c term increases, so altruism becomes less favourable. (it is not evidence against kin selection).

1) TAYLOR ET AL 2002 - competitive effects modelled how the effect of competition between relatives could affect fitness returns. Modified version of hamilton's rule: r1b/N - r2c>0 Where N is the size of the population of interacting individuals Conclusion = competitive effects cancel out all fitness benefits except those relating to the actor's own production of offspring. (i.e. kin selection does not have an important role). However, this objection is unlikely to be highly applicable in practice: Competition is usually between sterile workers, and sexuals disperse long distances before initiating new colonies - i.e. insect populations very rarely consist of a high density of reproductives competing with their relatives. 2) Biologists such as Wilson, Wilson and Holldobler have recently suggested kin selection should be abandoned altogether, and that individual/group-level selection provide better explanations. However, their arguments seem to stem from an overly-narrow viewpoint of what kin selection predicts: Nowak et al 2010 refer to failures of the haplodiploidy hypothesis as part of the 'rise and fall of inclusive fitness theory' - but the haplodiploidy hypothesis is just one idea arising from kin selection theory. Wilson and Holdobbler 2005 refer to the fact that KS predicts conflicts and thus cannot predict altruism, whilst also stating that KS assumes no colony level effects. In reality, KS includes all of these phenomena. Some authors e.g. Nowak et al 2006,2010 have produced novel models of social behaviour, but Bourke et al 2011 suggest that these are just derivatives of inclusive fitness theory. Nowak et al 2010 also criticised kin selection theory for the fact that relatedness being central in social evolution. However - this is not the case, as comparative fitness costs and benefits are central to kin selection - relatedness and ecological factors work together (the carpenter bees are a good example of this), and group-level effects are incorporated already. Nowak et al 2010 also argue that insect altruism does not need to be explained from the perspective of workers - potential workers (daughters of a colony-founding female or queen) are 'not independent agents' but rather can be seen 'as \"robots\" that are built by the queen' or the 'extrasomatic projection of [the queen's] personal genome'. However, if this was true, how do you explain worker behaviour such as eating of the queen's eggs (Stark et al 1990), egg laying by workers and lethal killing of the queen (Bourke et al 1994). 3) Could be argued that perhaps living in related groups is a consequence, not a cause of eusociality, so kin selected benefits may not have been important in evolution. However, there is robust evidence against this - Hughes et al 2008 found that mating with a single male, which maximizes relatedness, is ancestral for all eight independent eusocial lineages that we investigated. Mating with multiple males is always derived. (i.e. eusocial sp. were ancestrally monogamous, so the existence of closely related family groups preceded evolution of eusociality) (Boomsma belives that the high relatedness to siblings resulting from monogamy promotes obligate eusociality in hymenoptera - the monogamy hypothesis).

Metazoan trematodes - Himasthia sp. Two morphs - reproductives and sterile soldiers Genetics - individuals are genetically identical Ecology - soldier morphs protect the host from attack by flatworms of the same and other species.

If clones are genetically identical, then it is simple - as long as b>c, then altruism will be favoured The genetic problem becomes - what is the level of clonal mixing? As here, individuals won't all be perfect clones Level of mixing is different in different colonies - microsatellite study = avg. 10% mixing in Pemphigus spyrothecae, multiclocus ISSR study - avg. 41% in Pemphigus obesinymphae Evidence for CHEATERS in pemphigus obesinymphae: ABBOT ET AL 2001 Intruders cheat on the host gall by not fighting, growing rapidly and reproducing more (i.e. don't provide altruistic benefits) can be understood by inclusive fitness theory - the fact that they are less related to the host colony members makes altruism less favourable - making pursuit of direct fitness gains more likely.

Galls are important as: source of preadaptations - foundress fighting, intergall migration, gall cleaning Sources of selection could include defendability of the gall (Foster 1990, Foster and Benton 1992), and maintenance of genetic integrity of the clone. Despite the genetic predisposition, and the fact that all aphids live in a gall at some stage of their life, only 1% are social - why? To do with the ecology of galls - if they are slow-growing and long lasting - need defence by soldiers if they can be kept shut until the last moment, then there is no need for soldiers

Pseudoregma sundanica aphids: have 'tending' ants who aid in gall defence by Shingleton and Foster 2000 involved experimentally removing ant predators - the more ants per aphid, the smaller the proportion of aphid soldiers. I.e. division of labour interacts with other forms of defence

subsocial route - 2 independent oriigins, No obvious special genetic factors - naked mole rats are highly inbred (mean = 0.81 - Burland et al 2002), but damaralands are not. Ecology - VARIABLE ENVIRONMENT favours eusociality PATCHILY DISTRIBUTED FOOD - benefit in having multiple individuals returning food to the colony. E.g naked mole rats feed primarily on tubers located by digging, these are patchily distributed and very large (up to 1000 times bodyweight). Thus, it may prove difficult for a solitary indvidual to locate a tuber, but if a member of a colony finds one, it could provide food for a colony for months or even years (Dawkins 2006). SPARSE AND ERRATIC RAINFALL - conditions appropriate for tunnelling occur sporadically for very short periods - during dry periods, they can struggle to expand tunnel systems or even disperse (Jarvis et al 1994). Thus, to ensure location of food, it becomes crucial that many individuals work together to expand tunnel systems when this is possible. Overall, the combination of these factors is referred to as the \"Aridity food-distribution hypothesis\" (Jarvis and Bennet 1991). Predation could also have a role?

GENETIC: 1) high proportion of sex-linked genes All diploid siblings with XX/XY determination are effectively 'haplodiploid' with respect to their sex chromosomes, as all brothers share the same Y chromosome from their fathers. In termites, this can create a genetic predisposition for helping siblings as a significant proportion of the genome is linked to the sex chromosomes (Syren and Luykx 1977). This means that termite siblings may be related to each other by more than the usual 0.5, i.e. they may be more related to siblings than their own offspring. Prevalence of this effect is uncertain though, so it is hard to assess its role in the evolution of altruism. 2) Cycles of inbreeding and outbreeding Hamilton's idea initially, discussed by Bartz (1979) - in termites, reproductives that disperse are often produced by 'secondary reproductives' that are offspring of the original king and queen (i.e. they are the product of brother-sister matings so are more homozygous). After dispersing, they will breed with unrelated individuals - the outbred offspring produced will tend to be related to each other by more than the usual 0.5. This could predispose them to help rather than raise their own offspring (conditions of hamilton's rule more likely to be achieved). Again, there is no direct evidence for this being important. ECOLOGY: Digest cellulose by protozoans living in their intestine - protozoans have to pass from anus to mouth between generations => ancestral termites had to stay at home long enough to become infected with these protozoa = could have favoured the subsocial route Richard Dawkins (1979) even suggests that termite eusociality may have evolved due to protozoans manipulating them into staying at home to make an ideal environment for protozoans to spread!

Bourke et al 2014 refer to 10 studies that have evidenced a negative c but overall fulfillment of Hamilton's rule, thus providing evidence for altruistic behaviour being favoured by kin selection. This ranges from a study of Tiger Salamanders where cannabilism of other eggs is avoided for closely related kin, to classical studies of primitively eusocial insects (e.g. Polistes metricus)

Insurance-based advantages rely on an assumption that long periods of dependency were present ancestrally in solitary breeders. For solitary breeders, this creates a greater risk of death before independent offspring are produced. How then, could long periods of dependency have evolved before eusociality? This apparent paradox was resolved by Field and Brace (2006), whose work can explain why this prerequisite for insurance-based advantages could have occurred ancestrally. For solitary breeders, progressive provisioning of food, resulting in a long period of dependency, may have been more favourable than mass provisioning, which allows for a short period of dependency. This is because extended provisioning reduces the impact of PARASITISM on the mother's fitness: This was shown experimentally in solitary Ammophila wasp spp., as a result of two factors: 1) Progressively provisioned larvae are likely to have hatched and be older when first exposed to parasites, and therefore less vulnerable. (Evidence - larvae were much more resistant to mortality from the parasites than eggs, in both a mass provsioning and progressive provisioning sp.) 2) Under progressive provisioning, the mother was better able terminate investment in the colony if she discovers parasitism, reducing negative impacts on her own fitness (less energy/resources expended on failing broods than if she was a mass provisioner). (~80% less reproductive effort wasted apparently). This is because they are better at recognising fat maggots than small ones, and the longer period of time for maggot growth during progressive feedings means extended care gives a better chance of encountering fat maggots.

In the Hymenoptera (ants, bees, wasps), there have been many transitions in both directions between solitary nesting and sociality. Recent evidence that the transition is plastic, explaining why this is the case: FIELD ET AL 2010 British Isles - sweat bee is eusocial in southern areas and solitary in northern areas - transplantation experiments induced either eusocial or solitary behaviour by females. [Individual variation in social phenotype was linked to time available for offspring production, and to the genetic benefits of sociality, suggesting that helping was not simply misplaced parental care] \"once sociality has evolved, subsequent transitions represent selection at just one or a small number of loci controlling developmental switches between preexisting alternative phenotypes\"

Not very - colony-level effects are incorporated anyway into inclusive fitness via their impact on b term in Hamilton's equation). Holldobler and Wilson created resurgance of the idea of the 'superorganism' in their criticisms of inclusive fitness theory. Do strong colony level effects even occur (group selection) Gardner and Grefen (2009) in a formal analysis suggest that adaptations occur at the superorganism level only under quite strict conditions: when within-group conflict is fully suppressed or there is no conflict anyway because the colony is clonal (as in aphids). Detailed studies show that widespread kin selected conflict occurs in insects, so the role of multi-level selection may appear to be limited.

Fisher (1930) - the rarer sex receives proportionally greater mating success, so stable sex ratios should stabilise around 1:1. However, this is only true for diploid organisms, where relatedness is always equal between male and female siblings. For haplodiploids, we must marry the concept of Fisherian sex ratios with the concept of kin selection, as inclusive fitness is not just the product of mating success, but relatedness x mating success: Thus, the STABLE SEX RATIO should equal the RELATEDNESS ASYMMETRY between sexes, for the controlling party (as here, relatedness x mating success is equal for both sexes, so both sexes provide the same fitness benefit to the controlling party. The queen is equally related to sons and daughters (r=0.5), so she should invest equally in the two sexes Workers are related to their brothers by 0.25, and sisters by 0.75 (assuming a single queen mates once), so stable population sex ratio of 3:1 in favour of females. (here expected reproductive success three times greater for males, whilst relatedness is three times lower - i.e. relatedness x mating success is equal). I.e. workers and queens have different fitness optima. This could create conflict, with each party manipulating the sex ratio to their own advantage. Bulmer and Taylor (1981) suggest that the queen might limit supply of female eggs to the workers, whilst workers may invest more in each queen egg that is available. Mechanisms such as male infanticide and physical distancing from the queen are also likely to play a role (Queller and Strassmann 1998) Who wins? This is a good test of kin selection theory

1) The first empirical study of sex ratio conflicts came from Trivers and Hare (1976),who studied the ratio of investment in males and females (determined by both the number of offspring produced and the weight of offspring, for each sex) in 21 spp of ants. Considerable scatter in the data, but results suggest that the ratio is closer to 3:1 than 1:1 (i.e. workers are in control) Trivers and Hare suggest that workers are in control because they have practical power - they provision the young and are in a position to selectively kill off males and nurture queens. However, this paper was criticised on several grounds: 1) Relatedness asymmetry between males and females may not be 3:1 in practice. E.g. The queen often mates more than once [or there may be multiple queens], decreasing the optimum ratio for workers from the 3:1 that was assumed (as average relatedness to sisters is now less than 0.75). 2) Workers can sometimes lay male eggs (e.g. study of bumblebees => 39% of male eggs laid by workers (Owen and Plowright 1982). This again means that the 3:1 prediction no longer holds, and also provides an alternative explanation for the biased sex ratios that were observed [local competition between brothers competing for mates can mean that Fisher's theory of equal investment no longer holds - the ESS for the queen is not 50:50 but biased in favour of females] Nevertheless, more recent studies still suggest that Trivers and Hare were essentially right - workers do tend to be in control. Monogynous ants: observed sex ratios being close to the expected worker optimum in 5 spp. across 8 different situations (Bourke 1997). However, more generally, these benefits do appear to be context specific - the balance of power can be different in different species: [[[1) Metcalf (1980): in polistes metricus, the queen appears to win. [Workers do not normally lay eggs and males disperse far - can discount local competition effects] Genetic analysis of enzyme polymorphisms => workers are related to sisters by an average of 0.65 (less than 0.75 due to multiple queen matings) Workers should therefore prefer a female-biased ratio, but observed outcomes are closer to the queen's optimum of 1:1. Why does the queen win? Practical considerations - queen produces male offspring early in the season, when few of the workers have emerged, and those that have can be effectively controlled by the queen (so they do not kill male offspring).]]] HELM 1999 - another case where queens have the power to determine sex ratio (bias the caste of eggs)- The phenomenon of different balances in power is supported by Pamilo 1990 - slave-making (where queens can exert control) and polygynous ant colonies (where relatedness asymmetry decreases) have a less female-biased sex ratio than monogynous ones. However, these results have their limitations - on the basis of inaccuracies in measuring allocation, non-random mating and the fact that comparisons were not phylogenetically controlled.

In colonies where relatedness asymmetry is not equal to the population sex ratio, one sex will tend to provide greater per capita fitness returns (as relatedness x mating success will be greater for this sex). Such colonies should invest entirely in one sex, and this can lead to split sex ratios across the population: THEORETICAL EXAMPLE - differences in mating frequency Boomsma and Grafen (1990, 1991) In there is a population of colonies with either singly mated, single queens or doubly mated, single queens, relatedness asymmetry (females:males) will be 3:1 versus 2:1. The stable sex ratio for a colony is where relatedness asymmetry x mating success is equal for both sexes (as here, each sex provides equal fitness returns for the party being considered). If workers have control over sex ratios, this would creates split sex ratios as can be shown in a graph: 1) Outcome 1 (fraction of doubly mated queens 0.33): Under these circumstances, one of the two classes has exceeded a critical frequency that is a function of relatedness asymmetry, allowing it to exert control over the population sex ratio (either when 3:1 colonies have exceeded 75% or 2:1 colonies have exceeded 33%). Thus, the other class will have a relatedness asymmetry that does not equal the population sex ratio, so will be predicted to specialise entirely on males or females (as one sex will provide greater fitness returns than the other). The class that can control the population sex ratio will act as a 'balancing class' - it produces mixed broods that 'balance' the contribution of the single-sex class to achieve a popn. sex ratio equal to its own relatedness asymmetry (so the population sex ratio reflects the stable sex ratio of that class). 2) Outcome 2 (0.25< fraction of doubly mated queens<0.33) Neither class is frequent enough to form a 'balancing class', so the population sex ratio is in between the relatedness asymmetry of each class. Thus, it pays singly mated queen colonies to specialise entirely on females (benefits from greater relatedness exceed costs from reduced mating success), whilst it pays doubly mated queen colonies to specialise entirely on males (benefits from greater mating success exceed costs from reduced relatedness). In general, this leads to the 'class' with higher relatedness asymmetry (singly mated queens) investing more in females than the class with lower relatedness asymmetry (doubly mated queens). Similar theoretical arguments can be applied to other cases where relatedness asymmetry differs, such as due to differences in mating frequency, queen replacement or queen number.

Differences in relatedness asymmetry create split sex ratios as long as workers have some power to bias the sex ratio. 1) EFFECTS OF MATING FREQUENCY Sundstrom 1994 - example of formica truncorum As predicted from the theory of Boomsma and Grafen - in colonies with doubly mated queens, relatedness asymmetry is lower and the colonies invest more in males. In colonies with singly mated queens, relatedness asymmetry is higher and colonies invest more in females. Sundstrom et al 1996 provide direct evidence that this is due to sex ratio manipulation by workers (see below) 2) EFFECTS OF QUEEN REPLACEMENT - Mueller 1991 Augochorella striata (halictine bee) - some colonies are eusocial (daughters stay at home) with high relatedness asymmetry between helpers (usualy 3:1), some are parasocial (founded by sisters) with low relatenness asymmetry between helpers (usually 1:1). Replacement of queens appears to be crucial to maintaining eusocial colonies, and this has effects on the sex ratio. This is because you can convert eusocial colonies to parasocial ones if the queen is removed. This was done in some eusocial colonies, whilst a worker was removed as a control in other eusocial colonies. The investment ratio in females was much lower in experimental versus control colonies, as expected from the now lower relatedness asymmetry created by the switch to parasociality. 3) EFFECTS OF QUEEN NUMBER - Hammond et al 2002 Leptothorax acervorum (ant varying in number of queens). Relatedness asymmetry lower in colonies with multiple queens (raise fewer full sisters and more cousins), and they invest less in males, as you would expect from their lower relatedness asymmetry. Evidence that they achieve this by biasing environmental caste determination (see below) (Overall evidence - Queller and Strassman 1998 - colony differences in MATE NUMBER, QUEEN NUMBER, and QUEEN REPLACEMENT all create relatedness differences among colonies that are predicted to lead to some colonies specialising on females and others on males - these predictions regarding split sex ratios have been found to hold true for at least 17 different spp.)

4) EFFECTS OF QUEEN CONTROL 2) HELMS 1999 - possible in Pheidole ants as they can bias the caste of diploid eggs - i.e. lay eggs pre-determined to be workers or female reproductives. This leads to split sex ratios as a consequence of queens manipulating towards the ESS of a 1:1 population sex ratio ('toss a coin' - either bias towards workers, so reproductives are mainly male, or lay unbiased eggs = workers will be selected to raise these as all-females (as their optimum is a 3:1 ratio). This explains the observed outcomes - strongly split sex ratios (half produce mainly females, half produce mainly males) despite NO differences in relatedness asymmetry. Good example of power shifting to queen - so queen preferred outcome is achieved.

A) By queens 1) Biasing the caste during egg laying (see above) 2) Policing workers (e.g. prevent worker egg laying or male infanticide) - e.g. this is thought to be important in Polistes Metricus, where the queens can effectively police as male eggs are laid early in the season (Metcalf 1980) B) By workers 1) Selective destruction of male eggs/larvae. Evidence = Sundstrom et al 1996: Formica esecta is similar to F truncorum (see above) i.e. some colonies have singly mated queens, others multiply mated. In all of the nests, queens lay a similar fraction of male eggs (60-70%) But, in singly-mated colonies, the outcome is 30% males - i.e. workers are selectively destroying males Experimental evidence that they can detect genetic diversity in brood odour, which may result in this selective killing of males. Also, the manipulation was not observed in the multiply-mated colonies (as would be expected) - showing it is facultative. Thus, this provides direct evidence for active sex ratio manipulation. 2) Biasing environmental caste determination Hammond et al 2002 Leptothorax acervorum - relatedness asymmetry differs due to the number of queens - creating differences in sex bias that appears to reflect the workers selfish interests (see above). How achieved? Sex ratio laid by queens is ~84% female in all colonies, as is the ratio of female (workers and reproductives) to male adults. Thus, workers are not selectively destroying males. Instead - the authors suggest that they may be manipulating the caste of the female eggs - in monogynous colonies, raise a higher fraction as queens, and in polygynous colonies, rear a higher fraction as workers. C) By males For male reproductives, it is severely disadvantageous to mate with a queen where the relatedness asymmetry [f:m] of the offspring produced is less than the population sex ratio (as in this case, workers will be selected to raise entirely male broods to maximise their fitness - resulting in zero fitness for the male that mated in the queen [as the males are haploid and only carry genes from the queen]. Boomsma suggests that male reproductives will therefore be selected to: 1) Avoid mating with already mated queens (as this will result in broods with lower relatedness asymmetry - increasing the risk of male-biased broods) 2) Deliver only a small amount of sperm if they do mate with already mated queens (this can help to ensure that the relatedness asymmetry of the colony is kept fairly high, so fewer female eggs may be killed by workers) But - no rigorous evidence for this yet. Also note, if they male mates with a previously-mated foundress before feritilization has occurred (as may be likely in conditions of swarm founding that occur in ants), then Boomsma's points may not be valid - instead of avoiding mating and delivering small amounts of sperm, it could pay more to actively seek matings and be adapted for sperm competition - could not only result in broods with high relatedness asymmetry but allow them to father more of the female offspring produced. May be likely given the evidence for 'second male advantage' in many other insects. (this is just speculation)

Sex ratio conflicts (Queller and Strassman 1998) 1) Across spp. comparisons - may predict that colonies with singly-mated, monogamous queens have a more female-biased sex ratio than colonies with multiple queens, multiple matings or where there is full maternal control (solitary breeders). There is evidence for this being the case e.g. Bourke 1997 - monogynous ants = tend to be female biased. Pamilo 1990 (although criticised) - polygynous and slave-making ants tend to have lower investment in females, as expected. 2) Within spp. comparisons (split sex ratios) - colony differences in mate number, queen number, and queen replacement all create relatedness differences among colonies that are predicted to lead to some colonies specialising on females and others on males, in order for workers to maximise inclusive fitness. The predicted differences in all three of these categories have been observed in at least 17 spp. of eusocial insects (Queller and Strassman 1998). Would be \"extremely hard to explain\" if queen-worker conflict relating to kin selection was not affecting outcomes. Workers may be exerting control through male infanticide or physical distancing from the queen.

1) Sex ratio conflict (queen-worker) 2) Conflict over male production (queen-worker and potentially worker-worker) 3) Conflict over individual caste determination (undifferentiated individuals versus workers) Note it is important to draw a distinction between potential conflict (as in the Boomsma/Grafen theory) and actual conflict. In reality, the actual conflict depends on the power of workers versus queens (or workers versus workers) and the information held by different parties.

Singly mated, monogynous queens: Workers related to queen's sons by 0.25, but will be more related to their own sons (r=0.5) and those of other workers (r=0.375). Thus, it is in their fitness interests to promote male production by workers. I.e. there is worker-queen conflict - would predict that the queen would police workers. However, as the number of matings by the queen increases, workers become more related to the queen's sons than the sons of a randomly selected worker (the proportion of half sisters- r=0.125 increases). Can show on a graph: relatedness to male produced versus mating frequency of the queen. \"Random sister\"- starts at 0.375 (at 1) then rapidly decreases exponentially, tailing off towards r-0.125. Meanwhile, \"mother\" is a straight line at 0.25. The two lines cross at mating frequency of 2. Would therefore predict that, when there is multiple queen mating, it is in the genetic interests of workers to prevent other workers from laying eggs. This is Francis Ratnieks worker-policing hypothesis.

Seen in singly mated, single queen colonies with 1) Bumble bee matricide - Bourke 1994 In bumble bees (single queen, single mating) - the queen eventually switches to producing haploid eggs (not enough sperm to sustain further female production?). At this point, the workers start to lay their own male eggs, eat the queen's eggs and may kill the queen. Soon after, the colony dies out - kin selected suicide? Why does this happen? because of the queen-worker conflict - they are more related to worker sons (0.375) than the queen's sons (0.25), so once she no longer provides the benefit of producing females, make sense to kill her. Note the way in which kin selected individual conflict can lead to lethal killing of the queen seems to go against Holldobler, Wilson and Wilson's viewpoint of a harmonious super-organism. Group selection less important than individual selection? 2) Stingless bees - Peters et al 1999 Again, single mating and single queen. Thus, as with bumblebees, stingless bee workers are predicted to compete directly with the queen for rights to produce males. Worker egg-laying can be very high and no evidence of worker-worker policing. Worker eggs often eaten by the queen. [Complex laying rituals when the queen lays eggs - stereotyped within a lineage but variable between them. Is this a relic of ancestral dominance struggles for male parentage?]

Arevalo et al 1998 In two spp. of polistes wasps, the workers do not lay male eggs even though the queens are singly mated. Perhaps because the colonies are so small (usually 4-10, but sometimes up to 60) that the queen can police the other workers. (Bit like the polistes metricus example where the queen can police the sex ratio) Has been argued that in larger colonies, queens maintain control using pheremone production. However, this is not a valid argument - mechanistic rather than ultimate explanation. \"pheromonal queen control has never conclusively been demonstrated and is evolutionarily difficult to justify\" (Keller and Nonacs 1993) workers would not maintain responsiveness to a pheremone if it is not within their interest to do so,

Strong reproductive conflicts can occur in queenless ants because the colony is made up of totipotent ants that could all potentially rake up the dominant position. Results in unique mechanisms for conflict resolution 1) over queen determination - Baratte et al 2006 (diacamma australe): Chief reproductive is a mated worker or 'gamergate'. Only those workers with intact pair of appendages ('gemmae') are attractive to mates. The first female to emerge in a colony without a gamergate bites off the gemmae of all other workers so they are not attractive to males. Keeps her own gemmae and bullies the other workers to stop them laying male eggs From then, she stops bullying the workers but they agree to continue not laying eggs. Maybe the unmated gamergate has to bully the female to control them, but once she has mated, they accept her control as she is the only member of the colony that can lay female eggs - thus, it is in their fitness interest to allow her to remain in control, in order to safeguard this production of related females. Experimentally, there is much more aggression to young unmutilated workers (future gamergate with developing ovaries) than to gamergates, supporting this argument. Thus - \"Mutilation seems to maximize colony productivity by creating irreversibly sterile helpers from newly emerged individuals\" 2) Over queen replacement: MONNIN ET AL 2002 (dinoponera quadriceps) If a beta female challenges the gamergate, the gamergate smears a chemical on the beta - low rankers pick up this chemical and immobilise her/bite her for long periods (3-4 days). This causes the beta female to lose rank irrevocably. Can be thought of as honest signalling - once in power, it is in the interest both of the gamergate and the low rankers to prevent the beta from reproducing (Low-rankers probably regulate gamergate number to maintain an efficient balance between egg-layers and workers, given that gamergates work less).

Expected when a queen is multiply mated (as workers will be more related to the queen's sons than the sons of a randomly selected worker). 1) Honeybees mate multiple times (Queller et al showed this using microsatellite measurements - mean is as high as 27 in some spp.) Worker policing evidenced by Ratnieks and Visscher (1989) Dissections show that a minority of workers have developed ovaries (but not sperm receptacle) There is evidence that such workers are policed to prevent them from laying eggs- behavioural observations: they are harassed/recognised by other workers. Experiments - workers can discriminate between male eggs of other workers and those of queens, and destroy worker eggs by eating them. Queens appear to chemically label their own eggs. A stable system of signalling that workers respond to (it is in their interest to preferentially raise queen eggs) Result - genetic evidence that only 1 in 1000 adult males from the colony is worker-produced (not the 54% that would be expected if there was no worker policing - Ratnieks and Welseneers 2006) Therefore strong evidence for worker policing in honey bees. Contrasts with the otherwise similar, but singly mated stingless bees (Peters et al 1999 - high egg laying by males - see above).

OLDROYD AND RATNIEKS 2000 \"Anarchistic bees\" In some colonies (rare) - many (almost 100%) honeybee males come from workers. Why? Demonstrated experimentally by moving workers/eggs from 'anarchistic' colonies to non-anarchistic colonies 1) Anarchistic worker eggs are more acceptable 2) Workers in arachistic hives are worse at discriminating worker-laid eggs. The whole colony is not very productive This is a rare phenomenon, probably not favoured evolutionarily.

FOSTER AND RATNIEKS 2000 Strong evidence for the worker policing hypothesis In this wasp sp. (dolichovespula saxonica), colonies have one queen that may be singly or multiply mated. Would predict worker policing in the multiply mated queen colonies but not singly mated (see previous) This is supported by data collected through microsatellite measurements: singly mated queen colonies- 70% of males produced are from workers, and 70% of the adult males are sons of workers (no worker-worker policing) Multiply mated queen colonies - still some males produced by workers (25-90%), BUT ONLY 7% of adults are workers - i.e. there is worker-worker policing Facultative worker worker policing depending on the queen's mating frequency

1) Low costs - accidentally eating a queen egg is not costly, since so many are produced. Eating an egg is quick - low cost of time spent policing 2) Workers have good information i) male eggs are laid in special drone cells that are intermediate in size between those containing developing workers and queens ii) there may be effective honest signalling between queens and workers to allow discrimination of eggs (note that this breaks down in anarchistic hives) 3) Large colony size facilitates spread of policing (no single worker can dominate male production)

1) Relatedness hypothesis - as described by Ratnieks (see above) and evidenced in bees/facultatively in dolichovespula saxonica [affects r1 term of hamilton] 2) Efficiency hypothesis - workers stop other workers laying because of losses in colony productivity - i.e. the average worker loses if worker egg-laying is allowed [affects b term of hamilton as there may be costs to worker male production]

Hammond and Keller 2004 - 50 species of ants, bees, and wasps in a phylogenetically controlled comparative analysis. Found that that male parentage does not vary with relatedness as predicted by the relatedness hypothesis. Suggested that the efficiency hypothesis is more relevant, as there may be substantial costs to unchecked worker production at the colony level Wenseleers and Ratnieks 2006 disagree: claimed that hammond and keller's data set was too small and analyised ineffectively - carried out their own comparative analysis using 109 spp. of ants, bees and wasps. 1) Behavioral data confirms that worker policing occurs more frequently in species where workers are more related to the queen's sons than to other workers' sons 2) Also, a significantly smaller proportion of adult males are the sons of workers in spp where workers are more closely related to the queen's sons than other workers sons. 1) and 2) support the relatedness hypothesis.

Ratnieks and Wenseelers (2008) - enforced altruism due to coercion (policing by queens and workers) (a socially imposed constraint) could have played a role in the evolution of eusociality in hymenoptera. [[In a large colony, according to hamilton's rule alone, would expect 14% of egg laying workers if sisters are related by 0.75, and 50% if they are related by 0.5. (54% in honeybees) Thus in an absence of policing a significant poportion of workers should be selected to lay eggs. The fact that they often don't (e.g. in honeybees) suggests high levels of policing - i.e. enforced altruism is maintaining colony behaviour today (note that these calculations assumes that egg laying 'workers' do not work)]] Evidence for their hypothesis (in modern ant and bee spp) 1) When policing is effective, fewer workers reproduce (i.e. effective policing may facilitate enforced altruism) 2) i) without a queen, workers reproduce less when they are more related to workers (i.e. relatedness promotes voluntary altruism as under hamilton's rule - suggests they are acting to promote their own inclusive fitness) ii) With a queen, workers reproduce more the more related they are (paradoxical - they are more altruistic when less related to each other - suggests they are not maximising their individual inclusive fitness - hence policing by the queen may be driving enforced altruism). However, it is unclear how important this really was to the evolution of altruism, as we don't know how prevalent policing would have been in the early stages of altruism. I believe that the role in hymenoptera could be limited given that the monogamy hypothesis has been evidenced by Hughes et al 2008 - if monogamy was the ancestral situation, there may have been little incentive for policing by 'helpers' in these ancestral family groups - daughters staying at home would presumably have been more related to the sons of other daughters than the sons of their mother, so there would be little reason for policing between helpers to drive enforced altruism (although policing by the mother could still play a role). Ratnieks and Wenseleers themselves admit there is little reason to believe it was important at the origin of eusociality.

Nanork et al 2005 Dwarf red Asian honeybee: when the queen dies, worker policing breaks down and workers are now able to lay eggs. Even unrelated workers from other colonies fly in and lay male eggs there, parasitising the doomed colony.

Yes - remember that caste determination tends to be environmental. Instead, individuals may have some power in determining their caste, such that their fitness interests diverge from that of workers and/or the queen. Makes sense for undifferentiated individuals to attempt to develop into a queen, due to the large direct fitness benefits that could be obtained (raise offspring rather than neices/nephews). Workers/existing queen indifferent to which larvae become female reproductives, but if too many are produced, it will be costly to the colony (creating conflict). Crucial point is that there needs to be power for self-determination, if this conflict is to arise. 1) For this to occur, it is necessary that developing female has practical control over her own nutrition. This is seen in some cases e.g. bumble bees where larvae feed communally on pollen mass. Stingless bees where larvae can punch through to neighbouring cells and eat food. Some ants, in which larvae are mobile. 2) Self-determination also facilitated by low queen-worker size dimorphism (just a little extra feeding by developing female could allow her to become a queen) 3) Self-determination facilitated when queens and workers are reared simultaneously (ensures appropriate timing for becoming either a worker or queen). Conflict especially likely when there is reproduction by colony fission (an excess of queens will be more costly to the colony [as need fewer reproductives versus swarm founding], as so will likely be culled)

RATNIEKS ET AL 2006 1) Bourke 1997 - Comparison of Melipona stingless bees with Trigonine stingless bees/honeybees (Aphis) All reproduce by colony fission, so the cost of additional queens should be similar. However, in melopina, there is WEAK queen-worker size dimorphism, whilst it is strong in trigonines/apis. Would therefore predict greater conflict in melopina, as individuals have more power for self determination. Data supports this - continual queen production in melonina results in LARGE EXCESS OF NEW QUEENS that are culled on a large scale by workers - suggests conflict occurs In trigonines/apis, there is periodic queen production, with MINIMAL excess of new queens, and no or small scale queen cull - suggests no conflict. Thus, the theory correctly predicts what happens, in this case

Practical considerations can give workers power to coerce females into not becoming reproductives, resolving the conflict. Evidence - Ratnieks et al 2006 Looked at four spp. of swarm-founding bees Inclusive fitness arguments suggest that, in the three sp of stingless bees, which are singly mated, only 20% should be selected to develop as queens. In apis, multiple mating means the queen is less effective in colony resolution, so 54% would be selected to develop as queens based on inclusive fitness calculations alone. In melopina, workers have no practical power to coerce females into becoming workers, so observed % is the 20% predicted from inclusive fitness theory alone. In apis, and nannotrigona, feeding control can effectively coerce females as there are special 'royal cells', so observed % is lower than that predicted by inclusive fitness theory alone. Schwarziana has intermediate power in reducing % below what is predicted by inclusive fitness alone, as some females raised in small cells can still become queens. [take home message - melipona has little power for coercion of developing females due to practical arrangement of cells, apis/nannotrigona have high power due to special 'royal cells', whilst schwarziana has intermediate power as there are special cells but some of the females there still become queens. This means that apis/nannotrigona workers have the greatest ability to resolve the conflict over caste determination, schwarziana has intermediate power, and melipona has little/no power].