Larval marbled salamanders, Ambystoma opacum, eat their kin

The effect of kinship on larval cannibalism was examined in the marbled salamander. In separate behavioural trials, cannibalistic larvae were presented with two smaller conspecifics (a ‘prey group’), matched for size, that were (1) siblings of the cannibal, (2) non-siblings and (3) one sibling and one non-sibling (i.e. a mixture of two sibling groups); larvae were allowed to consume only one conspecific during each trial. This experiment was repeated in 2 consecutive years with larvae from two different populations. In both years, significantly fewer cannibals ate a non-sibling from the mixed sibship group than from groups composed of two non-siblings. In contrast, the number of cannibals that ate a larva from the pure sibling prey group did not significantly differ from the number that ate their sibling from the mixed sibship prey group. Thus, small larvae were significantly more vulnerable to cannibalism by their siblings than by non-siblings. The availability of unrelated individuals as alternative prey did not deter cannibals from eating their siblings, even though cannibals readily consumed prey of either genotype when the alternative one was absent. This kinship-biased cannibalism apparently was not due to size-selective predation or to differences among sibships in their propensity to cannibalize siblings. Moreover, the behaviour of potential prey did not differ towards related versus unrelated cannibals during initial observations of larval interactions. Cannibals required similar amounts of time to capture and process both sibling and non-sibling prey. To our knowledge, our results provide the first evidence of sibling cannibalism when unrelated individuals are available as alternative prey. ? 1995 The Association for the Study of Animal Behaviour Based on Hamilton’s (1964a, b) kinship theory, if all else is equal, individuals should preferentially direct harmful behaviour away from close relatives. However, for inclusive fitness to be maximized, animals must often balance the conflicting consequences of their behaviour to direct and indirect fitness. Such fitness trade-offs may selectively favour harmful behaviour among close relatives in some instances (i.e. when the benefits to direct fitness exceed the costs to indirect fitness), despite the apparent paradox that such cases present to adaptationist models of behaviour. Indeed, numerous examples of harmful (even fatal) aggressive behaviour among close kin have been documented (e.g. Mumme et al. 1983; Mock 1984; Hoogland 1985; Mock et al. 1990; Fraser & Thompson 1991). Thus, depending upon the relative costs and benefits to inclusive fitness, both altruistic and selfish behaviour patterns among close kin may be consistent with Hamilton’s kinship theory (Hamilton 1970; Mumme et al. 1983). Because the gregarious behaviour of many anuran tadpoles has implications for cooperative social behaviour, these amphibians have been model systems for examining the mechanisms of kin discrimination (reviewed in Blaustein et al. 1987a; Blaustein 1988; Waldman 1991; Blaustein & Waldman 1992; Blaustein & Walls, in press). Larval amphibians have also been exemplary subjects for evaluating aggression and cannibalism among kin (Walls & Roudebush 1991; Pfennig & Collins 1993; Pfennig et al. 1993, 1994; see review in Blaustein & Walls, in press). Studies of cannibalism are ideal for inspecting the consequences of behaviour to inclusive fitness, because the indirect costs of cannibalism (the death of kin) and its potential benefits to direct fitness (e.g. nutritional gain, enhanced growth, survival and reproduction; Crump 1992) are feasible to measure. Blaustein & O’Hara (1982) initially suggested that *Present address: Department of Biology, The City College of The City University of New York, Convent Avenue at 138th Street, New York, NY 10031, U.S.A. 0003–3472/95/080537+09 $12.00/0 ? 1995 The Association for the Study of Animal Behaviour