Abstract

In the Cape honeybee, Apis mellifera capensis, workers lay diploid (female) eggs via thelytoky. In other A. mellifera subspecies, workers lay haploid (male) eggs via arrhenotoky. When thelytokous worker reproduction occurs, worker policing has no relatedness benefit because workers are equally related to their sister workers' clonal offspring and their mother queen's female offspring. We studied worker policing in A. m. capensis and in the arrhenotokous African honeybee A. m. scutellata by quantifying the removal rates of worker-laid and queen-laid eggs. Discriminator colonies of both subspecies policed worker-laid eggs of both their own and the other subspecies. The occurrence of worker policing, despite the lack of relatedness benefit, in A. m. capensis strongly suggests that worker reproduction is costly to the colony and that policing is maintained because it enhances colony efficiency. In addition, because both subspecies policed each others eggs, it is probable that the mechanism used in thelytokous A. m. capensis to discriminate between queen-laid and worker-laid eggs is the same as in arrhenotokous A. m. scutellata.

Inclusive fitness theory (Hamilton, 1964a,b) has been successful in explaining and predicting social behavior (Crozier and Pamilo, 1996; Jarvis, 1981; Ratnieks et al., 2001). Hamilton's rule,

\(Br\ {>}\ C\)
, shows the condition under which a social action is favored in terms of the benefit to the recipient, the cost to the actor, and their genetic relatedness. Genetic relatedness is a key parameter and can now be measured with comparative ease and accuracy (Pamilo et al., 1997; Ross, 2001). Many empirical studies confirm the importance of genetic relatedness in social evolution (e.g., Foster and Ratnieks, 2000; Sundström, 1994). However, it is unlikely that any comparable technological breakthrough will facilitate the measurement of costs and benefits (Ratnieks et al., 2001). One solution to investigating the importance of costs and benefits in social evolution is to chose study systems in which relatedness is held constant but costs and benefits vary. Worker reproduction and worker policing in the Cape honeybee, Apis mellifera capensis, is one such system.

In most eusocial Hymenoptera, workers cannot mate but retain ovaries and can lay eggs (Crozier and Pamilo, 1996). Unfertilized worker-laid eggs are normally haploid (arrhenotoky; Crozier and Pamilo, 1996) and develop into males if reared. However, in a few species (currently six ants and A. m. capensis are known; Wenseleers and Billen, 2000), workers lay unfertilized diploid eggs that develop into females (thelytoky; Crozier and Pamilo, 1996; for rare cases of thelytoky in other subspecies of A. mellifera, see Mackensen, 1943; Tucker, 1958). A. m. capensis is native to the fynbos region (a macchia-like biome) in the western and eastern Cape provinces of South Africa (Hepburn and Radloff, 2002). Thelytokous reproduction by Cape honeybee workers has been long known (Onions, 1912; reviewed in Hepburn and Radloff, 2002), and many of the genetic details are also understood (Greeff, 1996; Moritz and Haberl, 1994; Verma and Ruttner, 1983). Although the thelytoky is via automictic parthenogenesis after meiosis (Verma and Ruttner, 1983), recombination through crossing over is rare (Moritz and Haberl, 1994). Thus, a worker's offspring are almost clonal.

Workers' sons are rarely reared in queenright European honeybee colonies (Visscher, 1989, 1996). This is because few workers have active ovaries (Ratnieks, 1993) and because the eggs they lay are eliminated by worker policing (Ratnieks, 1988). Honeybee workers in queenright colonies eat eggs laid by other workers (Ratnieks, 1993; Ratnieks and Visscher, 1989; Visscher, 1996). Queen-laid and worker-laid eggs are probably discriminated by means of a queen-produced egg-marking pheromone (Ratnieks, 1992, 1995).

Worker policing is selected for in a population of arrhenotokous social Hymenoptera on relatedness grounds alone when each colony has a single queen mated to more than two males (Foster and Ratnieks, 2001; Ratnieks, 1988). Honeybee queens, Apis mellifera, are typically mated to 5–30 males (Estoup et al., 1994; Fuchs and Moritz, 1999; Neumann and Moritz, 2000; Palmer and Oldroyd, 2000). As a result, worker policing is beneficial on relatedness grounds because workers are less related to other workers' sons (nephews:

\(\mathit{r}\ {=}.15\)
, for an effective paternity of 10) than to the queen's sons (brothers:
\(\mathit{r}\ {=}.25\)
). However, in A. m. capensis the situation changes significantly. In particular, workers lay female eggs and are as related to other workers' daughters as to the queen's daughters. But an individual laying worker is still more related to her own offspring (clonal daughter:
\(\mathit{r}\ {=}\ 1\)
) than to the queen's offspring (
\(\mathit{r}\ {=}.3\)
, assuming an effective paternity frequency of 10). Therefore, on relatedness grounds alone, one might expect that A. m. capensis workers either would not police each other or would do so less effectively than in other A. mellifera subspecies (Greeff, 1996). Indeed, brood are frequently observed in hive-boxes located above the queen excluder (a grid too small for the queen, but not for the workers, to pass through, so that the queen cannot lay eggs in the upper boxes) in queenright Cape honeybee colonies (Hepburn and Radloff, 1998; Pettey, 1922; Neumann P, personal observations). Moritz et al. (1999) have shown that this brood is the female offspring of workers.

Hamilton's rule for worker policing in A. m. capensis can be written as follows:  

formula
where rq and rw are the relatedness of police workers to the queen's and other workers' female offspring, and B and C are the relative efficiencies (i.e., total reproduction) of colonies with and without worker policing. Clearly, worker policing is favored if it increases the efficiency of the colony. Colony efficiency could be lowered, e.g., if the laying of additional eggs in a cell reduces the overall efficiency of brood rearing. The existence of worker policing in A. m. capensis would support the hypothesis that worker reproduction is costly.

In the present study, we investigated whether worker policing occurs in A. m. capensis by quantifying the removal rates of worker-laid and queen-laid eggs of both A. m. capensis and of the neighboring arrhenotokous subspecies A. m. scutellata in queenright colonies of both subspecies. Our results show that both A. m. capensis and A. m. scutellata police their own and the other subspecies' worker-laid eggs.

METHODS

Sampling colonies and experimental design

Queenright study colonies of A. m. capensis were obtained near Port Elizabeth, within the native range of A. m. capensis in the Eastern Cape province in southern South Africa. Queenright A. m. scutellata colonies were from the Pretoria area, within their native range. These localities were chosen because morphometrically and physiologically pure A. m. capensis and A. m. scutellata occur there (Hepburn and Radloff, 1998, 2002). Hepburn and Radloff (1998) and Hepburn et al. (1998) review the distribution and biology of these two subspecies.

The colonies were placed in two study apiaries in Grahamstown, South Africa. The A. m. scutellata apiary was distant, greater than 1 km, from any other bee hives to minimize inter-subspecific drifting and/or dispersing (Neumann et al., 2000b, 2001), which may result in social parasitism by A. m. capensis laying workers (Neumann and Hepburn, 2002; Neumann and Moritz, 2002). Both A. m. capensis and A. m. scutellata colonies were studied in order to compare African subspecies with arrhenotokous and thelytokous worker reproduction (Hepburn and Radloff, 2002; Neumann et al., 2000a). The experimental set-up followed standard methods for investigating worker policing via egg removal (Oldroyd and Ratnieks, 2000; Ratnieks, 1995; Ratnieks and Visscher, 1989). We used colonies of both subspecies as discriminator and egg-source colonies. Discriminator and egg-source colonies were different colonies. All discriminator colonies and the source colonies for queen-laid eggs retained their original queens during the time they were used in the study. The source colonies for worker-laid eggs were made queenless 2 weeks before egg-removal trials were started. All colonies were housed in hives composed of two deep or medium Langstroth boxes with a queen excluder between the boxes and the queen in the bottom box.

Quantifying egg-removal rates

Our primary aim was to compare the removal rates of queen-laid and worker-laid eggs within each subspecies. Worker-laid eggs are either haploid or diploid in the two subspecies (male in A. m. scutellata or female in A. m. capensis). Therefore, we compared these to queen-laid eggs of the same sex, and used haploid male eggs laid by A. m. scutellata queens and workers and diploid female eggs laid by A. m. capensis queens and workers. Because A. mellifera queens lay fertilized eggs in worker cells and unfertilized eggs in drone cells (Ratnieks and Keller, 1998), we were able to obtain unfertilized male eggs from drone cells in the queenright A. m. scutellata colonies and fertilized female eggs from worker cells in the queenright A. m. capensis colonies. For both the A. m. capensis and A. m. scutellata discriminator colonies, we used test frames with both drone and worker cells, because A. m. capensis workers naturally lay diploid female eggs mainly in worker cells (Neumann et al., 2000a), and workers of arrhenotokous subspecies lay male eggs mainly in drone cells (Page and Erickson, 1988; Ratnieks, 1993). The drone cells were used for male eggs laid by A. m. scutellata workers and A. m. scutellata queens. The worker cells were used for female eggs laid by A. m. capensis workers and A. m. capensis queens. Following standard procedures (Oldroyd and Ratnieks, 2000; Ratnieks, 1995; Ratnieks and Visscher, 1989), the test frames were placed above the queen excluder in each queenright discriminator colony and sandwiched between two frames containing brood of all ages (eggs, larvae, and pupae). The other frames in this upper box contained a mixture of empty cells, honey, and pollen. Queen-laid eggs were obtained from below the excluder in queenright source colonies (

\(\mathit{N}\ {=}\ 2\)
for A. m. scutellata and
\(\mathit{N}\ {=}\ 2\)
for A. m. capensis). Worker-laid eggs were obtained from the queenless source colonies (
\(\mathit{N}\ {=}\ 2\)
for A. m. scutellata and
\(\mathit{N}\ {=}\ 3\)
for A. m. capensis). For each discriminator colony (
\(\mathit{N}\ {=}\ 2\)
for A. m. scutellata and
\(\mathit{N}\ {=}\ 3\)
for A. m. capensis), we used a single test comb that was initially placed into the hive 2 days before egg removal trials began.

Twenty queen-laid and 20 worker-laid eggs of A. m. capensis were transferred from the source colonies into worker cells of the test frames. Likewise, 20 queen-laid and 20 worker-laid eggs of A. m. scutellata were transferred from the source colonies into the drone cells of the test frames. Eggs were arranged in batches of 20 eggs each on the test combs. The test combs were then reintroduced into their discriminator colonies. After 2 and 4 h, the test combs were briefly removed and inspected to determine which eggs were still present. After 24 h the remaining eggs were counted and removed. A new set of eggs was then transferred. Egg removal trials were made for five consecutive days for each discriminator colony. Moreover, removal rates of queen-laid and worker-laid eggs from A. m. capensis and A. m. scutellata source colonies (

\(\mathit{N}\ {=}\ 2\ queenright\ {+}\ 2\ queenless\)
A. m. capensis and
\(\mathit{N}\ {=}\ 2\ queenright\ {+}\ 2\ queenless\)
A. m. scutellata) in queenright discriminator colonies of the same subspecies (
\(\mathit{N}\ {=}\ 3\)
A. m. capensis and
\(\mathit{N}\ {=}\ 3\)
A. m. scutellata) were also evaluated on three sequential days using the same experimental approach with the exception of the cross tests. In this study a total of five A. m. scutellata and six A. m. capensis were used as discriminator colonies; another 17 colonies supplied the different types of eggs for the trials (four queenright A. m. capensis and A. m. scutellata each for queen-laid eggs, and four queenless A. m. scutellata and five queenless A. m. capensis as source for worker-laid eggs). We compared the removal rates of worker-laid and queen-laid eggs both within and between subspecies by using Kruskal-Wallis ANOVAs and Mann-Whitney U tests (with Bonferroni adjusted levels of significance) using Statistica.

RESULTS

Time to removal was determined for 2720 eggs. The percentages of remaining eggs per egg source in the two subspecies of discriminator colonies are shown in Figures 1 and 2. The results of the Mann-Whitney U tests are shown in Table 1. A. m. capensis and A. m. scutellata discriminator colonies removed worker-laid eggs of their own subspecies significantly faster than did queen-laid eggs of their own subspecies. Likewise, in both subspecies of discriminator colonies, worker-laid eggs of the other subspecies were removed significantly faster than were queen-laid eggs of the other subspecies. Worker-laid eggs of A. m. scutellata were removed significantly faster in A. m. capensis discriminator colonies than in A. m. scutellata colonies. However, A. m. capensis worker-laid eggs were not removed significantly faster in A. m. scutellata than in A. m. capensis discriminator colonies. Finally, there were no significant differences among the individual discriminator colonies of either subspecies in the removal rates of queen-laid and worker-laid eggs of their own subspecies after 24 h (A. m. scutellata: queen eggs,

\(H\ {=}\ 7.56\)
,
\(\mathit{p}\ {=}\ ns\)
; worker eggs,
\(H\ {=}\ 6.99\)
,
\(\mathit{p}\ {=}\ ns\)
; A. m. capensis: queen eggs,
\(H\ {=}\ 5.61\)
,
\(\mathit{p}\ {=}\ ns\)
; worker eggs,
\(H\ {=}\ 5.72\)
,
\(\mathit{p}\ {=}\ ns\)
).

DISCUSSION

The data clearly show that queenright A. m. capensis and A. m. scutellata colonies both police worker-laid eggs of their own and of the other subspecies but accept a much larger proportion of worker-laid eggs than do European honeybees. Although A. m. capensis discriminator colonies remove worker-laid eggs of A. m. scutellata faster than A. m. scutellata colonies, A. m. capensis worker-laid eggs were not removed faster in A. m. scutellata discriminator colonies. Within each subspecies, there were no significant differences among discriminator colonies.

Our data suggest that the two African subspecies are similar in their general pattern to European bees, i.e., worker-laid eggs are less acceptable than queen-laid eggs, but there are differences in the details. In the A. m. scutellata discriminator colonies, many more worker-laid eggs of both A. m. scutellata and A. m. capensis remained after 24 h (17% of A. m. scutellata worker-laid eggs and 14% of A. m. capensis worker-laid eggs) compared with those of colonies of European subspecies (1%; Ratnieks and Visscher, 1989) and even to A. m. capensis (1% of A. m. scutellata worker-laid eggs and 10% of A. m. capensis worker-laid eggs). Thus, it is possible that either policing is lower and/or worker-laid eggs are more acceptable in African than in European subspecies. The similar pattern in both subspecies suggests that the African origin of the tested subspecies may matter but not the genetics of worker reproduction. Moreover, fewer queen-laid eggs remained in the A. m. capensis colonies after 24 h (26%) compared with those in earlier studies on European honeybees (45%; Ratnieks and Visscher, 1989) and on A. m. scutellata (41%). This might be owing to nest-mate recognition for queen-laid eggs, which is well expressed in African honeybee subspecies (Pirk et al. 2001).

Our observations that worker policing in the thelytokous Cape honeybee is at least as effective as in the arrhenotokous A. m. scutellata (and possibly better) contrast to theory, which shows no benefit of worker policing in A. m. capensis (Greeff, 1996). In fact, the observed difference between the two African subspecies (A. m. scutellata with relatedness benefits and A. m. capensis without) is in the opposite direction to prediction (Greeff, 1996). But this prediction is based on relatedness grounds alone (Greeff, 1996). Our observation of worker policing in A. m. capensis is also partly in contrast to the other existing empirical study (Moritz et al., 1999), which showed that laying worker offspring does occur in queenright colonies. However, even if most worker-laid eggs had been removed, some may have remained, and it was these that were detected by using DNA microsatellites (Moritz et al., 1999). Thus, Moritz et al. (1999) showed that some workers' eggs are reared, but they do not show that worker policing is absent. The observed egg-removal rates in this study agree well with our casual observations of the study colonies, which showed that no larvae were being reared above the queen excluders in the test colonies immediately before or during the experimental period. Because the occurrence of brood above the queen excluder seems to be frequent in Cape honeybees (Pettey, 1922; Neumann P, personal observations) and in A. m. scutellata (Wossler T, personal communication), it was quite possible that worker policing would prove to be less effective than we actually observed. Thus, the combined results of Moritz et al. (1999) and this study indicate that worker policing does occur in A. m. capensis but is sometimes not fully effective in preventing worker reproduction.

Why does worker policing still occur in A. m. capensis? Earlier theoretical studies (Greeff, 1996) may have missed a critical piece in the cost-benefit analysis for policing in the Cape honeybee. The occurrence of policing should be dependent on a trade-off between the cost of policing and the cost of worker reproduction to overall colony efficiency and reproduction (Ratnieks, 1988). Given that policing probably costs very little, because eggs are held in open cells that workers are regularly checking anyway, the costs derived from unhindered worker reproduction might easily be higher than the costs of policing. Thus, a large efficiency gain is not needed to favor worker policing. In the A. m. capensis situation, the gain need only be marginal. Even if a queen is single mated, so that policing of worker-laid eggs has a relatedness cost, policing is still favored if colony efficiency increases by 20% (Ratnieks, 1988). A recent theoretical study by Foster and Ratnieks (2001) on the European hornet, Vespa crabro, shows that worker policing can even more easily be selected for at a mating frequency of one as part of a sex allocation biasing strategy of workers. That is, workers want to eliminate males to cause a female-biased sex ratio, and the only eggs they know to be male are workers' sons.

What are the possible costs of worker reproduction in honeybees? There are probably two main potential costs: reduced brood rearing efficiency and a reduced work rate of reproductive dominant workers, when worker reproduction occurs in queenright colonies:

  1. Reduced brood rearing efficiency. Honeybee nests have a limited brood rearing area, which constrains the number of eggs that can be laid by the queen, given that a queen typically will not lay an egg in a cell that already contains an egg (Ratnieks, 1990). Worker egg laying, when common, is characterized by multiple eggs per cell because workers will lay additional eggs in cells that already contain an egg (Gary, 2000). The earlier-laid eggs are often knocked down, squashed, and killed by the abdomen of the laying worker. Only one larva can be reared to adulthood in a single cell, and additional larvae are eaten by workers. This may lead to costs associated with cannibalism (Elgar and Crespi, 1992). Such cannibalism costs are likely to be small because when two larvae occur in one cell, one is removed within a few days of hatching. Moreover, the larvae are cannibalized rather than thrown out of the colony, suggesting that some energy can be recycled. However, when many workers are laying eggs, it may simply take longer for any cell to yield a worker, which is probably the main cost aspect. For example, if it took just one additional day to rear a worker per cell, this would lead to a 5% reduction in the maximal rate of colony build-up, given an egg-adult stage of approximately 19 days in workers of the Cape honeybee (Hepburn and Radloff, 1998).

  2. Work rate of laying workers. Laying Cape honeybee workers in queenless and queenright groups do not participate as much in hive duties such as brood rearing compared with the participation of subordinate workers (Hillesheim et al., 1989; Moritz and Hillesheim, 1985). Thus, a high frequency of laying workers may also reduce colony productivity (Hillesheim et al., 1989). This cost may not be greatly reduced by egg eating, because worker policing via oophagy does not directly stop or penalize egg-laying workers. It may cause a reduction in worker egg-laying over evolutionary time, but for it to be selected for, there has to be an immediate benefit in the colony with policing.

Reproductive dominance seems to be strongly genetically determined (Moritz and Hillesheim, 1985, Moritz et al., 1996). Therefore, almost clonal (Moritz and Haberl, 1994) laying worker offspring are predisposed to develop into laying workers. Thus, worker policing via oophagy may limit the establishment of such laying worker matrilines in queenright colonies, constituting an immediate benefit for policing colonies. Alternatively, but not mutually exclusive, worker-worker aggression in queenright colonies, which is directed toward nest mates with activated ovaries (Visscher and Dukas, 1995), might also restrict the establishment of such laying worker matrilines.

Worker reproduction in the Cape honeybee, if kept to a low level, will probably have almost zero efficiency cost. Therefore, we can expect some worker reproduction, even if there is policing. Indeed, there is considerable variation for egg-laying behavior in Cape honeybee workers (Neumann and Hepburn, 2002). Although in some laying worker colonies, the brood nest is virtually indistinguishable from that of a queen, because there is only one egg per cell (Neumann and Hepburn, 2002), others show the typical pattern of a queenless colony with laying workers such as multiple eggs per cell, as did the colonies in our study (Neumann P et al., personal observations). This suggests that it is possible in Cape honeybees to have a low level of worker reproduction, which does not interfere with brood rearing. Thus, the costs need not be high for a colony if the amount of worker reproduction is low.

The Cape honeybee example is a particularly convincing example. Thelytoky causes a change in kin structure rendering relatedness neutral with respect to worker reproduction (see above). Thus, if worker policing were not beneficial in A. m. capensis, it should be evolutionarily lost. That is, workers should accept eggs laid by other workers. Loss of policing would be a simple adaptation. In fact, loss of policing already occurs in queenless A. mellifera colonies (Miller and Ratnieks, 2001), which have failed to rear an emergency replacement queen. Worker-laid eggs are accepted and reared into a final cohort of males before the colony dwindles in population and dies (Page and Erickson, 1988). The results also show that both races are able to police worker-laid eggs of the other subspecies, indicating that the same underlying mechanism is used for worker policing. Arrhenotoky (Crozier and Pamilo, 1996), multiple paternity (Neumann and Moritz, 2000, Palmer and Oldroyd, 2000), and worker policing (A. mellifera: Ratnieks and Visscher, 1989; A. florea: Halling et al., 2001; A. cerana: Oldroyd et al., 2001) appear to be ancestral in Apis, indicating that thelytoky is a derived condition in A. m. capensis and arose in a clade in which worker policing occurred.

In conclusion, we hypothesize that worker policing still exists in the Cape honeybee owing to colony efficiency grounds. As predicted by Hamiltons' rule, this illustrates that relatedness alone cannot predict the reproductive characteristics of insect societies. The application of the inclusive fitness theory requires knowledge of costs, benefits, and relatedness.

Figure 1

Numbers (

\(mean\ {\pm}\ SD\)
) of queen-laid and worker-laid eggs of Apis mellifera capensis and A. m. scutellata remaining after 0, 2, 4, and 24 h in queenright Apis mellifera capensis discriminator colonies

Figure 1

Numbers (

\(mean\ {\pm}\ SD\)
) of queen-laid and worker-laid eggs of Apis mellifera capensis and A. m. scutellata remaining after 0, 2, 4, and 24 h in queenright Apis mellifera capensis discriminator colonies

Figure 2

Numbers (

\(mean\ {\pm}\ SD\)
) of queen-laid and worker-laid eggs of Apis mellifera capensis and A. m. scutellata remaining after 0, 2, 4, and 24 h in queenright A. m. scutellata discriminator colonies

Figure 2

Numbers (

\(mean\ {\pm}\ SD\)
) of queen-laid and worker-laid eggs of Apis mellifera capensis and A. m. scutellata remaining after 0, 2, 4, and 24 h in queenright A. m. scutellata discriminator colonies

Table 1

Numbers of eggs from different subspecies, colony, and caste sources remaining after 24 h in Apis mellifera capensis and A. m. scutellata discriminator colonies.

Type of comparison Egg source DC (No. of DC, trials) Trend Egg source DC (No. of DC, trials) U p 
Worker-laid versus  queen-laid  C−  C (6,24)  Faster  C+  C (6,24)  107.5  <.0002 
 S− S (5,19) Faster S+ S (5,19) 78 <.003 
 S+ C (3,15) Slower S− C (3,15) 47.5 <.001 
 C+ S (2,10) Slower C− S (2,10) 13.5 .019 
Worker-laid versus  worker-laid  S−  S (2,10)  Slower  S−  C (3,15)  46 .003 
 C− C (3,15) Faster C− S (2,10) 46 ns 
Type of comparison Egg source DC (No. of DC, trials) Trend Egg source DC (No. of DC, trials) U p 
Worker-laid versus  queen-laid  C−  C (6,24)  Faster  C+  C (6,24)  107.5  <.0002 
 S− S (5,19) Faster S+ S (5,19) 78 <.003 
 S+ C (3,15) Slower S− C (3,15) 47.5 <.001 
 C+ S (2,10) Slower C− S (2,10) 13.5 .019 
Worker-laid versus  worker-laid  S−  S (2,10)  Slower  S−  C (3,15)  46 .003 
 C− C (3,15) Faster C− S (2,10) 46 ns 

Results of Mann-Whitney U tests are shown. The Bonferroni adjusted level of significance is

\(\mathit{p}\ {=}.025\)
. DC indicates discriminator colony; C, A. m. capensis; S, A. m. scutellata; +, queenright; and −, queenless.

We wish to thank W.R.E. Hoffmann for technical assistance and S.E. Radloff for statistical advice. We are grateful to H.R. Hepburn for providing laboratory facilities, bee colonies, comments on earlier versions of the manuscript, and an inspiring atmosphere at Rhodes University. Financial support was granted by the Deutsche Forschungsgemeinschaft (P.N.), the Deutschen Akademischen Au-stauschdienst (C.W.W.P.), and the Natural Environment Research Council (F.L.W.R.).

REFERENCES

Crozier RH, Pamilo P,
1996
. Evolution of social insect colonies. Oxford: Oxford University Press.
Elgar MA, Crespi BJ, (eds),
1992
. Cannibalism: ecology and evolution among diverse taxa. Oxford: Oxford University Press.
Estoup A, Solignac M, Cornuet JM,
1994
. Precise assessment of the number of patrilines and genetic relatedness in honeybee colonies.
Proc R Soc Lond B
 
258
:
1
-7.
Foster KR, Ratnieks FLW,
2000
. Facultative policing in a social wasp.
Nature
 
407
:
692
-693.
Foster KR, Ratnieks FLW,
2001
. The effect of sex allocation biasing on the evolution of worker policing.
Am Nat
 
158
:
615
-623.
Fuchs S, Moritz RFA,
1999
. Evolution of extreme polyandry in honeybees (Apis mellifera).
Behav Ecol Sociobiol
 
45
:
269
-277.
Gary NE,
2000
. Activities and behavior of honey bees. In: The hive and the honeybee (Graham JM, ed). Hamilton, Illinois: Dadant & Sons; 289–372.
Greeff MJ,
1996
. Effects of thelytokous worker reproduction on kin-selection and conflict in the Cape honeybee, Apis mellifera capensis.
Phil Trans R Soc Lond B
 
351
:
617
-625.
Hamilton WD,
1964
. The genetical evolution of social behaviour I.
J Theor Biol
 
7
:
1
-16.
Hamilton WD,
1964
. The genetical evolution of social behaviour II.
J Theor Biol
 
7
:
17
-52.
Halling L, Oldroyd BP, Patimus B, Wattanachaiyingcharo W, Wongsiri S,
2001
. Worker policing in Apis florea.
Behav Ecol Sociobiol
 
49
:
509
-513.
Hepburn HR, Radloff SE,
1998
. Honeybees of Africa. Berlin: Springer-Verlag.
Hepburn HR, Radloff SE,
2002
. Apis mellifera capensis: an essay on the subspecific classification of honeybees. Apidologie 33:105–127.
Hepburn HR, Radloff SE, Fuchs S,
1998
. Population and the interface between Apis mellifera capensis and Apis mellifera scutellata.
Apidologie
 
29
:
333
-346.
Hillesheim E, Koeniger N, Moritz RFA,
1989
. Colony performance in honeybees (Apis mellifera capensis Esch.) depends on the proportion of subordinate and dominant workers.
Behav Ecol Sociobiol
 
24
:
291
-296.
Jarvis JUM,
1981
. Eusociality in a mammal: co-operative breeding in naked mole rat colonies.
Science
 
212
:
571
-573.
Mackensen O,
1943
. The occurrence of parthenogenetic females in some strains of honeybees.
J Econ Entomol
 
36
:
465
-467.
Miller DG, Ratnieks FLW,
2001
. The timing of worker reproduction and breakdown of policing behaviour in queenless honey bee (Apis mellifera L.) societies.
Insectes Soc
 
48
:
178
-184.
Moritz RFA, Haberl M,
1994
. Lack of meiotic recombination in thelytokous parthenogenesis of laying workers of Apis mellifera capensis (the Cape honeybee).
Heredity
 
73
:
98
-102.
Moritz RFA, Hillesheim E,
1985
. Inheritance of dominance in honeybees (Apis mellifera capensis Esch.).
Behav Ecol Sociobiol
 
17
:
87
-89.
Moritz RFA, Kryger P, Allsopp MH,
1996
. Competition for royalty in bees.
Nature
 
384
:
31
.
Moritz RFA, Kryger P, Allsopp MH,
1999
. Lack of policing in the Cape honeybee, Apis mellifera capensis.
Behaviour
 
136
:
1079
-1092.
Neumann P, Hepburn HR,
2002
. Behavioral basis for social parasitism of Cape honeybees (Apis mellifera capensis Esch).
Apidologie
 
33
:
165
-192.
Neumann P, Hepburn HR, Radloff SE,
2000
. Modes of worker reproduction, reproductive dominance and brood cell construction in queenless honeybee (Apis mellifera) colonies.
Apidologie
 
31
:
479
-486.
Neumann P, Moritz RFA,
2000
. Testing genetic variance hypotheses for the evolution of polyandry in the honeybee (Apis mellifera L.).
Insects Soc
 
47
:
271
-279.
Neumann P, Moritz RFA,
2002
. The Cape honeybee phenomenon: the evolution of a social parasite in real time? Behav Ecol Sociobiol 52:271–281.
Neumann P, Moritz RFA, Mautz D,
2000
. Colony evaluation is not affected by drifting of drone and worker honeybees (Apis mellifera L.) at a performance testing apiary.
Apidologie
 
31
:
67
-79.
Neumann P, Radloff SE, Hepburn HR, Moritz RFA, Reece SL,
2001
. Social parasitism by honeybee workers (Apis mellifera capensis Esch.): host finding and resistance of hybrid host colonies.
Behav Ecol
 
12
:
419
-428.
Oldroyd BP, Halling LA, Good G, Wattanachaiyingcharo W, Barron AB, Nanork P, Wongsiri S, Ratnieks FLW,
2001
. Worker policing and worker reproduction in Apis cerana.
Behav Ecol Sociobiol
 
50
:
371
-377.
Oldroyd BP, Ratnieks FLW,
2000
. Anarchistic honey bee workers evade worker policing by laying eggs that have low removal rates.
Behav Ecol Sociobiol
 
47
:
268
-273.
Onions GW,
1912
. South African “fertile worker bees.”.
J Dep Agric Union S Afr
 
1
:
720
-728.
Page RE, Erickson EH,
1988
. Reproduction by worker honeybees (Apis mellifera L.).
Behav Ecol Sociobiol
 
23
:
117
-126.
Palmer KA, Oldroyd BP,
2000
. Evolution of multiple mating in the genus Apis.
Apidologie
 
31
:
235
-248.
Pamilo P, Gertsch P, Thoren P, Seppä P,
1997
. Molecular population genetics of social insects.
Annu Rev Ecol Syst
 
28
:
1
-25.
Pettey FW,
1922
. Workers laying in comb of extracting super, Elsenberg apiary.
J Dep Agric Union S Afr
 
4
:
122
-124.
Pirk CWW, Neumann P, Hepburn HR,
2001
. Nestmate recognition for eggs and worker policing in African honeybees (Apis mellifera L.). In: Proceedings of the 2001 Berlin Meeting of the European sections of IUSSI; Sept. 25–29, 2001, Berlin: 106.
Ratnieks FLW,
1988
. Reproductive harmony via mutual policing by workers in eusocial Hymenoptera.
Am Nat
 
132
:
217
-236.
Ratnieks FLW,
1990
. The evolution of polyandry by queens in social Hymenoptera: the significance of the timing of removal of diploid males.
Behav Ecol Sociobiol
 
26
:
343
-348.
Ratnieks FLW,
1992
. Evidence for an egg-marking pheromone in the honey bee.
Am Bee J
 
132
:
813
.
Ratnieks FLW,
1993
. Egg-laying, egg-removal, and ovary development by workers in queenright honey bee colonies.
Behav Ecol Sociobiol
 
32
:
191
-198.
Ratnieks FLW,
1995
. Evidence for a queen-produced egg-marking pheromone and its use in worker policing in the honey bee.
J Apic Res
 
34
:
31
-37.
Ratnieks FLW, Keller L,
1998
. Queen control of egg fertilization in the honey bee.
Behav Ecol Sociobiol
 
44
:
57
-61.
Ratnieks FLW, Monnin T, Foster KR,
2001
. Inclusive fitness theory: novel predictions and tests in eusocial Hymenoptera.
Ann Zoo Fenn
 
38
:
201
-214.
Ratnieks FLW, Visscher PK,
1989
. Worker policing in the honey bee.
Nature
 
342
:
796
-797.
Ross KG,
2001
. Molecular ecology of social behaviour: analyses of breeding systems and genetic structure.
Mol Ecol
 
10
:
265
-284.
Sundström L,
1994
. Sex ratio bias, relatedness asymmetry and queen mating frequency in ants.
Nature
 
367
:
266
-268.
Tucker KW,
1958
. Automictic parthenogenesis in the honeybee.
Genetics
 
43
:
299
-316.
Verma S, Ruttner F,
1983
. Cytological analysis of the thelytokous parthenogenesis in the Cape honeybee (Apis mellifera capensis Escholtz).
Apidologie
 
14
:
41
-57.
Visscher PK,
1989
. A quantitative study of worker reproduction in queenright honey bee colonies.
Behav Ecol Sociobiol
 
25
:
247
-254.
Visscher PK,
1996
. Reproductive conflict in honey bees: a stalemate of worker egg-laying and policing.
Behav Ecol Sociobiol
 
39
:
237
.
Visscher PK, Dukas R,
1995
. Honey bees recognise development of nest mates' ovaries.
Anim Behav
 
49
:
542
-544.
Wenseleers T, Billen J,
2000
. No evidence for Wolbachia-induced parthenogenesis in the social Hymenoptera.
J Evol Biol
 
13
:
277
-280.