Як бджолиний розплід обирає нову оселю без лідера

A homeless cluster with a deadline

When a honeybee colony outgrows its hive, roughly two-thirds of the workers leave with the old queen and cluster on a nearby branch — a swarm with no walls, no stored food to speak of, and days, not weeks, before exposure or a rainstorm kills it. It has to choose a new home fast, and it has to choose well: a cavity that is too small, too damp, or too exposed dooms the colony months later. There is no queen bee directing the search — she barely flies during it — and no single bee ever visits more than a handful of the dozens of candidate sites under consideration at once. The decision is made collectively, by a few hundred scout bees out of a swarm that can number ten thousand.

The waggle dance as an honest vote

A scout who has inspected a tree cavity — crawling around inside to gauge its volume, checking the entrance size and how high it sits off the ground, sensing draftiness and dampness — flies back to the cluster and performs a waggle dance: a repeated figure-eight in which the angle of the straight run encodes the site's direction relative to the sun and the run’s duration encodes distance. What decades of work by Thomas Seeley and colleagues added to that picture is that the dance’s vigor — how many times it repeats, how long each round lasts — is not fixed. It scales with the scout’s own assessment of quality. A cramped, leaky knot-hole gets a handful of half-hearted waggles that most nestmates ignore; a spacious, south-facing cavity forty litres in volume with a small, defensible entrance gets a dance repeated a hundred times or more, and it recruits heavily.

Recruitment is positive feedback, not persuasion

Bees watching a dance don't take the dancer's word for it. Most follow the dance a few rounds, then fly out and inspect the advertised site themselves, forming their own independent opinion of its quality. If they agree it’s good, they dance for them too, adding their own recruitment to the pool — a snowball that grows fastest around the genuinely best sites, because better sites produce longer, more repeated dances that recruit more inspectors, who independently confirm the quality and dance again. Poor sites simply fail to sustain their own advertising: dancing for them decays faster than new recruits arrive, and they drop out of contention within a few hours. No bee ever compares two sites directly or does arithmetic; the comparison happens implicitly, in which site’s recruitment curve wins the race.

for each scout at a candidate site:
  score      = f(volume, entranceArea, entranceHeight, dryness)   // 0..1
  danceReps  = round(score * MAX_REPS)                            // longer dance = stronger vote
  danceReps decays a little every trip — must be refreshed by revisiting

recruit = bee that follows a dance, flies out, inspects independently
if recruit's own score agrees → recruit also dances for that site (positive feedback)

if scoutsPresentAtSite(site) >= QUORUM:      // ~15-20 bees simultaneously
    trigger "piping" + "buzz-run" → whole cluster warms up and launches

Quorum sensing: the moment of commitment

The swarm doesn’t wait for every scout to agree, and it doesn’t wait for one site’s dancing to simply outlast all rivals — that could take too long. Instead it uses a quorum rule: once roughly fifteen to twenty scouts are physically present at the same site at the same time, those scouts sense the crowd (through direct contact, not counting) and switch behavior. They return to the cluster and produce a distinctive high-pitched piping signal, then perform buzz-runs — running through the resting cluster while vibrating their flight muscles — which wakes up thousands of non-scouting bees and gets the whole mass warm enough and coordinated enough to lift off together, usually within an hour of quorum being reached.

Stop-signals: how a split vote is avoided

Two good sites discovered around the same time could, in principle, both accumulate dancers and pull the swarm apart. Bees have a second signal for exactly this failure mode: a scout returning from one site can deliver a brief stop-signal — a headbutt paired with a buzz — to a bee dancing for a different site, which measurably suppresses that dancer's further advertising.

The more competition a site faces, the more stop-signals its dancers absorb, which brakes the weaker alternative and helps the swarm converge on a single quorum rather than fracturing. Researchers have described this cross-inhibition as functionally similar to competing populations of neurons suppressing each other during a perceptual decision — the "bee brain" analogy that made this system famous well outside entomology, as a clean natural example of decentralized, error-correcting collective decision-making with no chief executive anywhere in the loop.

Frequently asked questions

Спробуйте наживо

Усе, що вище, працює прямо у вашому браузері — відкрийте the simulation і змінюйте параметри під час роботи. Нічого не встановлюється, нічого не завантажується на сервер, уся модель живе в одній вкладці.