Trophic polyphenism in certain species of salamanders seems to have developed specifically for cannibalism (Powers, 1907; Collins and Holomuzki, 1984; Crump, 1992). The morph of Ambystoma tigrinum larvae is characterized by its larger body size, broader head, and enlarged vomerine tooth patch in comparison with the larval morph (Powers, 1907; Rose and Armentrout, 1976). These traits provide cannibals with the ability to consume conspecifics as well as large heterospecific prey (Loeb et al., 1994; Maret and Collins, 1996). The range of A. tigrinum extends from coast to coast in North America (Rose and Armentrout, 1976; Stebbins, 1985); however, the cannibal morph of this species has been documented only in midwestern and western populations: Nebraska (Powers, 1907), Oklahoma (Glass, 1951), Arizona (Gehlbach, 1967), Colorado (Reese, 1969), Texas (Rose and Armentrout, 1976), Iowa (Lannoo and Bachmann, 1984), and Indiana (D. Pfennig and H. Whiteman, unpubl. data). The nonuniform distribution of this trophic polymorphism suggests the potential for behavioral and morphological variation among populations of the A. tigrinum complex in North America, as well as variation in the selection pressures influencing cannibalism (Rose and Armentrout, 1976; Collins et al., 1980; Collins et al., 1993). There also appears to be variation across cannibalistic populations with respect to morphological characters, particularly those involved in the relationship between head shape and body size [e.g., populations from Texas (Pierce et al., 1983), Iowa (Lannoo and Bachmann, 1984), Arizona (Collins et al., 1993), and Colorado (this study)]. Pierce et al. (1983) envisioned three processes that may contribute to the development of the large head characteristic of the cannibal morphology (Fig. 1). Each hypothesis depends on the relationship between head size and body size, estimated using regression analysis. In the first hypothesized process, positive allometry results in a proportionately larger head in the largest individuals, which are cannibalistic; thus, the regression lines of both morphs should have similar slopes that are greater than one (Fig. 1A). The difference observed in the head size of the two morphs would simply be a result of the cannibal morph's larger body size. A second possibility is that cannibals exhibit a faster rate of head growth than that experienced by typicals, and thus the slope of the cannibal regression line should be greater than that of the typical morphs (Fig. 1B). In the third scenario, cannibals experience a rate of head growth that is high early in ontogeny but then reduced during later stages of development to a level similar to or less than that of typicals. This process would result in a greater y-intercept for the cannibal morph regression (Fig. 1C). Pierce et al. (1983) also suggested that some combination of these hypotheses might operate under some conditions (i.e., the processes are not mutually exclusive). In this study, we quantify the relationship between head size and body size in cannibal and typical A. tigrinum larvae from south-central Colorado. Although the relationship between head and body size can be explored using measurements of various head dimensions, gape width (width of head atjawjoint) holds particular relevance. Gape width is an important determinant of foraging success in A. tigrinum because it directly influences the type and sizes of prey that can be ingested (Loeb et al., 1994; Maret and Collins, 1996). We compare the head/body size relationship of our Colorado population to data published for other populations of this species [i.e., Texas (Pierce et al., 1983), Iowa (Lannoo and Bachmann, 1984), and Arizona (Collins et al., 1993)] with respect to the hypotheses outlined above. We also suggest possible explanations for apparent geographic differences in this relationship across populations.
Read full abstract