Bee Biology
Anarchic Behaviour
  David A. Cushman logo  

Worker Policing in Honey Bees

Worker ovary development

In a queenright honey bee colony, the workers have ovaries, but are rarely fertile (only about 1 Apis mellifera worker in 10,000 has fully activated ovaries.[1] and this is similar in other Apidae within the Hymenoptera order.

The queen is normally the only egg producer in the colony and this condition is maintained by a pheromonal 'feed back' system, whereby worker ovary development is inhibited. The queen and the brood both produce pheromones that inhibit worker ovary development, which prevents individual workers from exploiting the system.[2]

However when a honey bee colony becomes queenless, some workers which have intact, but undeveloped, ovaries may develop them and thus become capable of laying male eggs,[3] whose genes reflect the patriline of the worker concerned. After seven or more days of queenlessness around 10% of the workers will have fully formed eggs in their ovaries, and many eggs will actually be laid.[4] This laying worker feature, allows social insects to use different reproductive strategies according to the colony structure at the time. Such strategies favour those patrilines that exhibit larger percentages of laying workers and can result in individual behaviours becoming colony traits. See Thelytoky

From the queen's point of view, individual workers must be prevented from laying eggs and having them raised using resources that could have been used for raising her own offspring.

From nature's point of view there may be diversification benefits in allowing a small number workers to lay a few drones as it allows more speedy changes than pure mutation.

But from the beekeeper's perspective, especially those of us breeders that are attempting to undo some of the gross hybridisation that has gone on for the last century or more, this may be a dangerous source of genetic pollution.

The fringes of the nest

When populations are large and the nest is physically widespread, the distribution of pheromones reduces at the outer edges, simply because of distance from the queen and brood as well as the larger area of the outer periphery of the nest. this gives rise to a condition of reduced suppression of ovary production, but not as severe as in the queenless case. It does give rise to an increase in worker laid eggs, but the numbers of drones arising from them is a very small fraction of those that are laid. 'Worker policing' is the mechanism that causes adult workers to eat 'worker laid' eggs, which are identified by other workers. [5] It is speculated that normal queen laid eggs are marked with a pheromone that is produced by the queen, which is coated on the eggs as they pass over the sting sheath.[6] Worker laid eggs are thought to lack this pheromone and are thus identified as such and eaten by the workers.[7]

It has been suggested that aggression towards workers with activated ovaries is another potential mechanism of worker policing, but I am unsure whether this applies to workers being hoisted out of cells as they attempt to lay an egg or whether the aggression goes further and results it the fertile worker being damaged or stung.


The numbers given here have been lifted wholesale from a paper by Ben Oldroyd, T.C. Wossler and Francis Ratnieks.

Workers are related to sons by 0.5, but to sons of the queen by 0.25.[8] A worker therefore benefits from laying eggs. Nevertheless, few workers in queenright colonies have active ovaries, and only one male in a thousand is the son of a worker.[9]

The evolutionary reason for worker sterility is that although workers are more related to their sons than to those of the queen, they are even less related to sons of their half sister workers. (r=0.125)[10] Because of the high levels of polyandry found in honey-bees [11] the majority of nest mates encountered by workers are 'half sisters' not 'full sisters'.[12] Therefore, on average, workers are much more related to the male offspring of the queen than to the male offspring of other workers (average relatedness to worker's sons=0.15 for a paternity frequency of 10).

I can comment that... As far as Apis mellifera mellifera(Amm) is concerned, the relatedness to worker's sons is even less because the mating frequency is higher than 10. I am not sure how this may be affected if the increased numbers only represents multiple copies of the same alleles from different individual drones in the queen's mating spectrum. In 1958 Steve Taber and J Wendel established that within four different populations in the U.S. and Canada the number of matings was between 7 and 10. Now we know from DNA work that US Bees contain only about 3% Amm genes and 97% of the US bee population contains Am ligustica genes[13]. Some other work (that I cannot track down) gave a figure of 7 for Ligustica so we can infer from this that the mating frequency of Ligustica is lower than that of Amm and as a result is more prone to develop laying workers and that the closer relatedness would give rise to a larger percentage of these worker laid drones reaching maturity. This is further promoted in US bees as generally colonies are larger and contained in hives of large volume so that fringe conditions are more likely to occur.


[1] - - Francis Ratnieks 1993.
[2] - - Shelley Hoover, Christopher Keeling, Mark Winston, and Keith Slessor. (date unknown)
[3] - - Robinson (and others). 1990
[4] - - Velhuis (H.H.W. Velthuis?) 1970.
[5] - - Francis Ratnieks and Kirk Visscher 1989.
[6] - - Francis Ratnieks 1995.
[7] - - Francis Ratnieks and Kirk Visscher 1989.
[8] - - Francis Ratnieks 1988, Kirk Visscher 1998.
[9] - - Kirk Visscher 1989 and 1996.
[10] - - C.K. Starr 1979, M. Woyciechowski and A. Lomnicki 1987, Francis Ratnieks 1988.
[11] - - Kellie Palmer and Ben Oldroyd 2000.
[12] - - Harry Laidlaw Jr. and Robert Page 1984.
[13] - - Nathan Schiff and Walter Sheppard

Dave Cushman.

Page created 19/10/2001

Page updated 01/12/2022

Originated... 19 October , Revised... 14 July 2003, New Domain... 13 May 2004, Upgraded... 29 November 2004,


This page has actually been validated by W3C