| dc.description |
Simple game theory models predict that individuals should not signal reliable information about he likelihood of attacking their opponents. We investigated the relationship between the likelihood of attacking a conspecific male intruder and the singing behavior of territorial male bobolinks (Dolichonyx oryzivorus). We simulated intrusions by presenting caged males on the territories of male bobolinks. Experiments were performed with 23 territorial males, of which 12 attacked the stimulus during the 15-min trial period and 11 did not. We compared the singing behavior of nonattackers during the 15-min trial periods to the singing of attackers before the attack. Ten song variables plus the grouping variable, attacker and nonattacker, were subjected to discriminant analysis. Only one variable, the number of notes per song initiated, entered the discriminant function, which correctly classified 74% of the territorial males as attackers or nonattackers. The song sung just before an attack also included fewer notes than the nonattacker average. This song showed no differences from the attacker averages for any of the song features we measured. Thus, the data indicate that the singing behavior of territorial male bobolinks contains reliable information about their intentions to attack conspecific male intruders. We believe that the best explanation for this result is that, contrary to the assumptions of the simplest game theory models, there is an added cost to aggressive signaling compared to nonaggressive signaling. [Behav Ecol 1991;2:319–326] |
