Testing game theory models: fighting ability and decision rules in chameleon contests.

Game theory models of animal contests make many non-mutually exclusive predictions, complicating empirical tests. These predictions regard the relationship between contest parameters and fighting ability, for which body size is usually used as a proxy. However, in many systems, body size may be a limited proxy since multiple traits and contextual factors such as experience influence fighting ability. Using contests between male Cape dwarf chameleons, Bradypodion pumilum, I test alternative game theory models of extended contests. I show how the most likely candidate model can be identified through a process of elimination, based on tests of key predictions. In addition, I present a measure of fighting ability based on multiple traits that allows ability to change as experience changes. In dwarf chameleons, persistence is based on loser thresholds rather than assessment of relative ability, ruling out the sequential assessment model. Winners and losers do not match behaviours in early parts of the contest, arguing against all types of war of attrition models. Although the cumulative assessment model remained as the most likely candidate model, not all specific predictions of this model were upheld.