Escalated fighting despite predictors of conflict outcome:solving the paradox in a South American cichlid fish

Several studies have found behavioural differences between the interactions of opponents, early in a conflict. These differences consistently predict which individual will be the winner. Surprisingly, opponents appear not to use this information: they continue fighting and even proceed to highly damaging escalation. We attempted to resolve this paradox in a series of experiments on interactions between male cichlids. First, the cost of fighting and escalating in terms of energy expenditure (oxygen consumption: gill movements) and injuries was high. Second, previous social experience was important but could not explain the occurrence of predictors of conflict outcome. Third, a less artificial conflict set-up, in which territory ownership was asymmetrical and opponents were allowed to see each other before the conflict, resulted in shorter interactions with reduced escalation phases. This suggests that the occurrence of predictors of conflict outcome and the subsequent lack of response of the opponents is the result of unnatural, highly symmetrical designs, as used in many studies. We propose the following mechanism. Prospective winners postpone escalation as much as possible because of the high cost involved. At the same time, prospective losers continue the fight because they lack alternatives. We tested this 'desperado hypothesis' in a fourth experiment, in which losers were offered alternative options. These conflicts were shorter and both predictors of outcome and escalated fighting were absent. (C) 2001 The Association for the Study of Animal Behaviour.