66 



Current Herpetol. 20(2) 2001 



0.7 



0.6 



o 



(0 



Fig. 2. Proportion in wet mass (mean + SE) of diet composition in large- ( ■ ; n=2 1 ) and small-headed larvae 

 ( □ ; n=23) of Hynobiits retcirdatiis collected from a pond in Tomakomai Experimental Forest. Terres=terres- 

 trial invertebrates; Diptera=Diptera larvae; Epheme=Ephemeroptera nymphs; Tricho=Trichoptera larvae; 

 Cope=Copepoda; Salamander=salamander larvae (H. retardcitus)\ Detritus=fine litter. 



absolute value of HW did not differ 

 between the two groups of larvae. It is 

 therefore likely that the variation in diet 

 was caused not by the difference in gape 

 size between the large- and small-headed 

 larvae, but by other factors. Considering 

 the difference in shape of the snout tip 

 between the two groups (i.e., relatively 

 pointed in the small-headed larvae, and 

 more rounded in the large-headed larvae: 

 Fig. 1), it is probable that the dietary varia- 

 tion actually reflects difference in foraging 

 environments: the small-headed larvae may 

 forage using their snout for digging around 

 the pond bottom where their main food 

 items, detritus and Diptera larvae, chiefly 

 occur, whereas the large-headed larvae may 

 chiefly feed on fallen terrestrial inverte- 

 brates near the water surface. 



For larvae of some salamanders, varia- 

 tion in diet is a cause of variation in 

 growth in body mass (Brunkow and 

 Collins, 1996). The present results suggest 

 that H. retardatus larvae also shift prey 



items under the influence of flexible devel- 

 opment of large head width (Fig. 2). Such 

 an intra- populational variation in diet 

 could influence the population dynamics. 

 For instance, the divergence of diet selec- 

 tion within a population may reduce 

 intraspecific competition, thereby allowing 

 a greater population density and higher 

 averages of individual fitness (Polls, 1984). 

 Because individual variation in head shape 

 depends on the initial density (Kohmatsu et 

 al., 2001), the prediction of population 

 regulation for H. retardatus should include 

 information on such population demogra- 

 phy. 



Acknowledgements 



I am grateful to E. Wada, A. Maruyama, 

 R. Yonekura for invaluable comments. M. 

 Matsui advised me to study this theme. I 

 also thank TOEF for facility of this 

 research. 



