The defensive response of the honeybee Apis mellifera.
Asian honeybees have been shown to kill hornets by ‘thermoballing’, in which they surround a hornet to form a ball within which the temperature increases to a lethal level. We report here that Cyprian honeybees, Apis mellifera cypria, kill their major enemy, the Oriental hornet, Vespa orientalis, in a different way — by asphyxiaballing, in which the Cyprian honeybees mob the hornet and smother it to death. Cyprian honeybees use a balling defence against the Oriental hornet. But this strategy is not equivalent to the thermo-balling reported for Asian honeybees [1,2], because the temperature inside the ball is only 44 ± 0.5°C (N = 20), while the hornet’s lethal thermal limit is higher (50.6 ± 0.6°C, N = 30) and similar to honeybees (50.5 ± 0.1°C, N = 30). Moreover, in thermo-balling conditions (44°C), hornets die much more quickly (in balls: 57.8 ± 11.4 min, N = 20) when exposed to honeybees than when excluded from them (in incubators: 143.3 ± 33.5 min, N = 30). Furthermore, Cyprian honeybees cannot kill the hornet by stinging, as demonstrated in an experiment where only three hornets were stung in 130 balls. Some observations made on hornets trapped in balls for short periods of time (5–8 min) have also shown that the invader remained motionless for a few minutes before recovering. We there investigated which other factor could explain how Cyprian honeybees kill hornets. As observed in the field, when they form a ball, Cyprian honeybees first target and mob the hornet’s abdomen. A hornet’s respiration depends on abdominal pumping, whereas in other insects, such as large dragonflies, it depends almost exclusively on thoracic pumping . Contraction of the longitudinal segmental abdominal muscles reduces the length of the abdomen and the volume of the abdominal cavity, and causes the outflow of air through the spiracles (expiration). Passive relaxation of these muscles allows the abdomen to recover its initial volume, causing a negative pressure and an influx of air into the tracheas (inspiration) . We there assumed that, in the balling process, honeybees require little effort to maintain the hornet’s abdomen at a minimum length, since inspiration is a passive mechanism. It is important to note that hornets differ anatomically from other insects in having spiracles that are covered by tergites during the expiration phase (Figure 1E,F). As a result, when honeybees maintain the hornet’s abdomen at a minimum length, they also keep the spiracles covered and thereby affect the influx of air through the tracheal system during the inspiration phase. To test if honeybees could kill the hornet by blocking its respiration, we conducted laboratory and field experiments. In the laboratory, we monitored and compared the hornet’s respiration in normal conditions and in conditions simulating balling. Hornets were fixed ventral side down (thorax and abdomen) on a wax platform, and a force displacement transducer was attached on the posterior region of the third tergite (FDT, Figure 1A) in order to monitor the force generated by its movement . Under normal conditions, the respiratory rhythm of four tested hornets was very stable (Figure 1B). During the first and last 5 minutes of a two-hour recording session, the amplitudes (a1 = 5.69 ± 1.62 N and a2 = 6.18 ± 2.51 N, respectively) and frequencies (f 1 = 3.1 ± 0.6 Hz and f 2 = 3.9 ± 1.1 Hz, respectively) of the force generated by the respiratory movements were not different (p > 0.05 for a and f ). In conditions simulating balling, when two and four tergites were successively blocked (locations 1 and 2, Figure 1A), the respiration decreased by 32.8 ± 5.4% (Figure 1C), and by 87.3 ± 1.8% (Figure 1D), respectively. In the field, balling efficiency, as determined by the time required by honeybees to kill hornets, was compared in normal conditions and in conditions where honeybees could not properly block the movement of the tergites. To keep the tergites open, we inserted plastic blocks below them (Figure 2A–C) and recorded the hornet’s respiration (N = 4, a = 94.02 ± 21.27% and f = 98.89 ± 12.14% from the initial stage). These data showed that the presence of blocks did not affect the hornet’s