A brood kept by helpers whose own breeding was settled before they were grown — the workers of the honey bee Apis mellifera, female like their queen and from the same kind of fertilised egg, divided from her not by genes but by what they were fed as larvae, and built with too small an ovary and no working way to mate
R909 closed with a switch held down from outside: take away a dominant marmoset’s scent and sight, and a subordinate’s ovulation returned in days, because nothing structural had been taken away. Its closing turn asked for the opposite — a restraint no removal can release, settled in development itself, with no succession waiting in the helper’s body. This page takes that turn to the hive.
Same egg, two futures. Queens and workers are both females from fertilised eggs. There is no separate worker kind laid in the comb. A very young female larva can still go either way: beekeepers and experiments alike rear queens by taking larvae from worker cells while they are still young enough and feeding them as queens. That openness is brief. In the accounts followed here the larva is still bipotent through roughly the first three days of larval life; by about the third instar the caste is effectively determined, and after about the fourth day it is fixed. Whatever an adult worker later wants, or whatever dominant later dies, she cannot re-enter that window. She is an adult. The window was larval.
Food is the fork. All young larvae receive royal jelly at first. A queen-destined larva goes on receiving it exclusively, in surplus, in a queen cell, for the whole feeding period. A worker-destined larva is switched after about three days to a worker diet to which honey and beebread are added. Later work found one ingredient of that second diet, p-coumaric acid from honey and beebread, by itself shifts caste-associated gene expression and reduces ovary development in larvae reared in vitro. The molecular work points the same way, early: silencing the DNA methyltransferase Dnmt3 in young larvae, in the 2008 experiment, sent most of the treated animals toward queen development with fully developed ovaries, against mostly workers in the controls. Diet, epigenetic marking, and anatomy are not three stories here. They are one decision, taken in the first days and then built into the body.
The body that decision builds. A queen emerges with an ovary built for output — on the order of well over a hundred ovarioles per ovary in the reviews followed here, and a large, working spermatheca in which she stores sperm from a single period of mating flights for years. A worker emerges with only a handful of ovarioles — figures in the literature range from one or a few up to about twenty per ovary, against the queen’s hundred-plus — a much shorter vagina, and a spermatheca that is rudimentary. She cannot mate and cannot store sperm. That last fact is the lock inside the lock: even an egg a worker lays cannot be fertilised by her, so in most Apis mellifera it can become only a drone, a male. She was not given a smaller share of the queen’s job. She was built without the apparatus for the queen’s job at all.
Keeping, on top of a lock. The workers are the keeping: tens of thousands of sisters in a strong colony who nurse and feed the brood, build and repair comb, clean, guard and forage, while the one queen lays — in peak season, by the usual accounts, well over a thousand eggs a day. There is also a switch like R909’s, laid over the developmental one. In a queenright colony, pheromones from queen and brood inhibit most workers’ ovary activation, and other workers remove most worker-laid eggs — worker policing — so successful worker reproduction while a queen is present is small. Name the order correctly, though: remove the queen and that top switch does release, a little. Some workers’ ovaries activate. What it cannot release is the structure underneath. Activation is not succession.
When the queen is lost. A queenless colony can still save itself, but only by going back to the larva, never forward from the adult. If larvae still young enough — roughly nought to three days old for a good emergency queen, with emergency cells attempted on brood up to about five days old at dequeening in one study, at lower quality — are present, workers feed some of them as queens and a new, mated breeder can emerge in about sixteen days from the egg. If no such larva remains, no adult in the hive can take the queen’s place. After a week or more queenless, a share of workers develop eggs and begin laying; thousands of worker-derived drones can be produced, at only a few dozen eggs a day from even an active laying worker, and no new workers at all. The colony does not get a new mother. It spends its remaining workers raising males, and then it ends. The marmoset waiter needed only a signal to stop. The honey-bee worker would need to be a larva again.
One colony, in order: a fertilised egg hatches into a female larva still able to become either caste; feeding divides her future in the first three days — royal jelly throughout for a queen, a switched worker diet for a worker; the worker who emerges keeps the colony’s brood with a handful of ovarioles, a rudimentary spermatheca and no capacity to mate; queen and brood pheromones and worker policing hold her small ovary mostly off on top of that; lose the queen with no young larva left, and no adult can succeed her — laying workers produce drones only, and the colony dwindles out. Schematic, compressed from the published record.
Walk it in order.
Schematic only: the stages compress work across subspecies, studies and settings into one colony. Labelled cautions: the three-day openness and the fixing by about the fourth day are the usual summary of transfer and rearing experiments, not a knife-edge every larva meets at the same hour — intercastes exist, and quality falls off with larval age, so do not state a single instant of decision; ovariole counts vary by study, subspecies and counting method — the contrast, handful against hundred-plus, is the claim, not any one pair of numbers; do not call a worker wholly sterile without the qualification this page makes — she retains ovaries, can activate them queenless, and can lay unfertilised, drone-producing eggs; the famous exception is the Cape honey bee, Apis mellifera capensis, whose workers can lay diploid female eggs by thelytokous parthenogenesis and so can, unusually, produce females and even a queen — this page’s no-succession claim is scoped to the ordinary arrhenotokous majority, not to capensis; do not present royal jelly as a single magic ingredient — quantity, exclusivity, cell, and compounds in the worker diet all contribute, and the mechanism is still worked on; and do not reverse the Dnmt3 result — silencing it in larvae redirected development, which is evidence the decision is early and epigenetic, not evidence any adult worker can be switched back.