Zebra finches scrounge more when patches vary in quality: experimental support of the linear operator learning rule

The use of evolutionary game theory and the concept of the evolutionarily stable strategy (ESS) have come under criticism lately because game-theoretic models are often constrained to overly simplistic situations. Furthermore, game-theoretic models commit the behavioural gambit, that is, they assume that individuals have some unspecified decision mechanism that duplicates the outcomes that would be expected from natural selection acting on genetic alternatives. The producer-scrounger game is an ideal illustration of the criticisms aimed at the game-theoretic approach; it has generated a number of game-theoretic models that make specific predictions for highly simplified foraging conditions. Simulation models of the producer-scrounger game that incorporate the linear operator learning rule, however, have been more successful in replicating the empirical results. In these models, groups of animals forage for patches of food using a linear operator learning rule. The linear operator model furthermore provides a new prediction about the effect of variation in patch quality within a producer-scrounger game. Current models based on evolutionarily stable strategies and empirical investigations have always assumed or used patches of uniform quality. Yet patches vary in quality and some environments are potentially more variable than others. The linear operator model predicts an increase in scrounging behaviour when patch quality varies. We tested this prediction using captive foraging flocks of male zebra finches, Taeniopygia guttata, in two treatments with variable and nonvariable patch quality. The results show that as individuals have more experience in the variable environment, they increase their scrounging behaviour. Linear operator models may therefore provide a general and empirically valid means of exploring the outcome of animal decisions in social foraging situations too complex to be addressed by evolutionary game models.