Why Bees Swarm: The Biology of Colony Reproduction

It's a common misconception that swarming is a colony malfunction. In reality, swarming is how a honeybee colony reproduces at the colony level: the old queen departs with a large portion of the workforce to found a brand-new colony elsewhere, while the bees left behind rear a new queen to carry on the original nest. A wild swarm establishing itself successfully in a new cavity is estimated to succeed roughly 67% of the time — a real reminder that swarming, while dramatic, is a genuinely viable reproductive strategy rather than a doomed last resort.

A strong colony doesn't necessarily stop at one swarm. It can produce a sequence of 3 to 5 "afterswarms", each smaller and carrying a virgin queen rather than the original mated queen. Success rates decline noticeably down that sequence — modelled at around 62% for the primary swarm, falling to under 10% for a fourth afterswarm, since each successive swarm is smaller and less resourced.

Once a swarm has left the hive, it doesn't fly straight to a new home — it first gathers in a temporary bivouac cluster (often on a tree branch) while scout bees search for a permanent cavity. Scouts are notably picky, evaluating candidate sites against a range of criteria — cavity volume, entrance size, height, dryness, defensibility — with an ideal cavity volume in the range of roughly 40 to 80 litres and an entrance opening of around 12 to 15 cm², small enough to defend but large enough for traffic.

Nest-site selection is one of the best-documented examples of collective decision-making in the animal kingdom (studied extensively in the real world by biologist Thomas Seeley). Scouts that find a promising site return and perform a waggle dance advertising it; other scouts investigate and, if convinced, dance for the same site themselves. The swarm doesn't commit until a quorum of at least 15 scouts is simultaneously advertising the same location — a genuinely real threshold from published nest-site-selection research, not merely a simulation artifact.

Once the decision is made, the swarm departs together — typically carrying somewhere between 30% and 65% of the parent colony's bees alongside the old queen. In simulation models, the journey to the bivouac and onward to the final site is measured in the range of 200 to 800 simulated ticks, concluding with a final consensus confirmed by piping signals — a real phenomenon in which bees produce a distinctive sound to help synchronise the swarm's collective takeoff toward the chosen cavity.

Meanwhile, the colony left behind has several queen cells developing. The first virgin queen to emerge will pipe — a real, audible signal — to announce her presence and, if rival virgin queens haven't emerged yet, may seek out and destroy their still-sealed cells. If multiple virgins do emerge closely together, they fight, stinging one another, and model estimates put the attacking queen's success rate at around 85% — one of the very few times in a honeybee colony that direct lethal aggression between bees is the norm rather than the exception.

Separately from swarming, colonies can also replace an ageing or failing queen through supersedure — a calmer, non-reproductive queen replacement that happens without the colony dividing. This is typically triggered when egg-laying falls below roughly 400 eggs per day sustained for about a week, an indicator of declining queen quality. Interestingly, model estimates suggest colonies allowed to supersede naturally show a roughly 23% higher two-year survival rate than colonies kept artificially on an ageing queen — a plausible pattern (a genuinely healthy young queen should outperform a failing one) but one presented here as a simulation output rather than a directly cited field statistic.

Swarming and nest-site selection remain one of the richest real-world examples of decentralised, leaderless decision-making in biology — a topic explored in depth in Thomas Seeley's book Honeybee Democracy, drawn from decades of actual field observation. Many of the specific percentages above (afterswarm success rates, queen-fight win rates, supersedure survival benefit) come from an agent-based simulation model rather than that published field research, and should be read as illustrative rather than directly citable. You can watch scouting, quorum-building, and swarm departure play out for yourself in the site's Beehive Colony: Agent-Based Model simulation.