In his Jan. 12 statisticsblog.com article, Matt Asher models beehive activity, specifically their propensity to attack. Beehives have sentry bees, whose job it is to sense intrusion and danger. When they discern a threat, they rush off to attack, leaving a scent for the remaining members of the hive to detect so that they, too, can join the attack. As other bees detect the scent and join the attack, they also leave the danger scent, increasing the chances that yet more bees will join the attack.
But the other bees differ in their ability to sense the danger scent, and it turns out that the “counterattack” behavior of the hive is highly variable, and depends on the distribution of “scent sensitivity thresholds” (SST) among the bees.
Asher simulated the hive behavior with random SST’s all drawn from the same uniform distribution, and found that most often the hive counterattacks petered out with just a small fraction of the hive joining. But sometimes, the “bees go nuts,” with the entire hive joining.
So how does this relate to global warming? It is instructive to see that even a very simple model of nature – just one parameter, the SST threshold – can lead to such strikingly different results under repeated simulation trials. Global warming models are extremely complex and infinitely adaptable (some purport to explain recent increased blizzard activity in North America and Europe), so it is difficult to imagine outcomes that cannot be accounted for by some model.
Learn more about the bee simulation, including R code, at www.statisticsblog.com .
