Factors which affect the occurrence of cannibalism and the broad-headed "cannibal" morph in larvae of the salamander Hynobius retardatus

Cannibalism in amphibian larvae may be suppressed among siblings in comparison to distant relatives or nonkin, even when a "cannibal" morph that can consume conspecifics shows adaptive advantages. Two experimental studies were undertaken to analyze factors which affect the frequency of cannibalism and the occurrence of the broad-headed "cannibal" morph in larvae of the salamander Hynobius retardatus. The first experiment investigated whether or not the broad-headed "cannibal" morph is only induced after performing cannibalism. Because the broad-headed "cannibal" morph was observed in larval populations that showed no cannibalism, it was concluded that the "cannibal" morph could be induced without actual cannibalism. Second, possible factors affecting the occurrence of the broad-headed "cannibal" morph were analyzed with respect to the density of larvae, level of food supply, and kinship among larvae, alone or in combination. Appearance of the broad-headed "cannibal" morph was affected by interactive effects of density2kinship: although the morph was induced significantly more often at a higher density if the conspecifics were distantly or not related, it was strongly suppressed even at high density if larval kinship was very close or among siblings. In contrast, the frequency of cannibalism was independently affected by larval density, level of food supply, and kinship among larvae: it was significantly larger at high than at low larval densities, at low than at high food supply, and in mixed-sibling groups than in pure sibships. These results suggest that the "cannibal" morphs in H. retardatus are induced even without cannibalism at a high density of conspecifics in mixed-sibling groups, but the occurrence of this morph, which continuously consumes conspecifics, is suppressed among siblings.