Anita. Behav., 1990, 40, 288-297 Risk-sensitive foraging in bitterlings, Rhodeus sericus: effects of food requirement and breeding site quality R. J. YOUNG,* H. CLAYTON& C. J. BARNARDt Animal Behaviour Research Group, Department of Zoology, University of Nottingham, University Park, Nottingham NG7 2RD, U.K. Abstract. Risk-sensitive foraging preferences in bitterlings were examined in relation to changing food deprivation and intake relative to estimated requirement. When running above estimated requirement and presented with a choice of constant and variable feeding stations, a group of seven fish were all risk-averse (avoiding high variance in reward rate). When running below, six out of the seven fish showed increased risk-aversion and one became risk-prone (choosing high variance). Increasing the degree of food deprivation while running below requirement for one of the risk-averse fish, however, resulted in a switch to risk-proneness, a response that was repeated in a second group of six fish running below requirement. Variance in reward rate also appeared to influence the choice of a potential breeding site by a male bitterling. When breeding sites of apparent equal quality were provided at the constant and variable feeding stations, the fish preferred the constant station; the variable station was preferred only when breeding site quality at the constant station was relatively low. Risk-sensitive foraging, usually characterized as the response of predators to variance in food reward rate in choosing where or what to eat, is now well- established both in theory and empirically (see Stephens & Krebs 1986; Real & Caraco 1986 for recent reviews). While foraging models incorporat- ing risk-sensitivityhave been applied to a number of foraging situations (see Real & Caraco 1986), the commonest scenario involves a predator with a high demand for energy which risks starvation ifthereis a shortfall in its food supply. If such a predator makes decisions so as to minimize the probability of a shortfall, variance in reward rate is likely to be an important criterion on which decisions are based (see e.g. Caraco 1981; Stephens 1981; McNamara & Houston 1982; Stephens & Charnov 1982; Stephens & Paton 1984; Barnard et al. 1985). For any given mean reward rate from feeding sites/prey types, risk-sensitive foraging theory for shortfall minimizers predicts a switch from risk- aversion (avoiding high variance in reward rate) to risk-proneness (choosing high variance) as the ratio of the predator's food requirement to its expectation of reward rate (expected mean) increases. Empirical studies to date have shown that predators such as small passerines (e.g. Caraco et al. 1980; Caraco 1981, 1983) and shrews, Sorex araneus (e.g. Barnard & Brown 1985, 1987; Barnard et al. 1985) whose *Present address: East of Scotland College of Agriculture, West Mains Road, Edinburgh EH9 3JG, U.K. tTo whom correspondence should be addressed. energy demands are high and who are likely to suffer deleterious consequences if there is a shortfall in their food supply respond to variance in reward rate in accordance with these predictions. Other preda- tors, such as rats, Rattus norvegicus (Battalio et al. 1985) and bumble bees, Bombus spp. (e.g. Real 1981; Waddington et al. 1981), which are less likely to be vulnerable to starvation in the event of a shortfall, have tended to be consistently risk-averse (see Stephens & Paton 1984 for a comparative analysis). Although the apparent difference between preda- tors in risk-sensitivity makes sense in terms of pre- sumed differences in the risk of starvation, it may be that normally risk-averse (e.g. low energy demand) predators become risk-prone if their demand is increased sufficiently. Differences in risk-sensitivity between predators may thus be due partly to differ- ences across studies in the degree of food demand imposed. In this paper, we look at the effects of changing estimated food demand on risk-sensitive choice of feeding station in a fish predator faced with two feeding stations and under different conditions of food requirement. While variance in food reward rate may influence foraging decisions, it may also impose a constraint on decisions made in other contexts. Thus, for instance, high variance in food reward rate at a given site may reduce the quality of, and the tend- ency to maintain and defend, a breeding territory at that site. Potentially, therefore, there may be a trade-offin site preference between variance in food 0003-3472/90/080288 + 10 $03.00/0 91990 The Association for the Study of Animal Behaviour 288