modified by Horn (1966). Q is formulated as 



C, = 



2 ^ Pa  Pjh 



j s 



J. p,ir + 2! pj, 



h = l 



h = l 



where p,;, and Pjj, are the biomass proportions of a 

 prey item h in the diets of species i and j, respec- 

 tively, and s is the total number of major prey taxa 

 in the food spectrum. C^ varies from zero, when 

 there is no overlap between the diets of species i 

 andy, to one, when all prey items are in equal pro- 

 portions. Niche breadth (B) for each species was 

 computed by using the formula developed by Levins 

 (1968): 



B; = 



1 





crustaceans, must have been captured at the sub- 

 stratum. Conversely, the 72 food samples from P. 

 auricilla were composed of 88.2% invertebrates and 

 11.8% fishes by volume; all were predominantly 

 pelagic forms. Major invertebrate prey included the 

 pelagic heteropod Atlanta spp., the pelagic tunicate 

 Pyrosoma spp., and numerous species of cavolinid 

 pteropods. Few fishes, most of them small and 

 unidentifiable, were eaten by this snapper species. 



The dietary overlap value (Q) was calculated to 

 be 0.04. Based on the Langton (1982) convention of 

 0.00-0.29 as the low overlap, 0.30-0.60 as the 

 medium overlap, and X).60 as the high overlap, very 

 little overlap occurred between the diets of the two 

 snappers at Pathfinder Reef. 



The niche breadth values (B) calculated for the 

 two species indicated P. zonatus had a higher food 

 breadth (3.82, .s = 10) than P. auricilla (2.05, s = 

 6), suggesting more food specialization among the 

 latter. Evenness in the proportion of the diet attrib- 

 uted to each prey type, however, was similar for the 

 two species (5„ = 0.38 in P. zonatus; B„ = 0.34 in 

 P. auricilla). 



where again, p^)^ is the biomass proportion of a 

 prey item h in the diet of species i and s is the total 

 number of major prey taxa. The breadth values 

 range from one, when prey items consist only of one 

 category, to s, when all food items are in equal pro- 

 portion in the same diet. These breadth values were 

 then normalized as B„ = Bis, which ranges from a 

 value of zero, representing the most uneven distribu- 

 tion of prey composition in the diet, to one, repre- 

 senting a totally even distribution. For the calcula- 

 tion of both breadth and overlap indexes, only prey 

 items identified at least to the family level were 

 used. 



Results 



The diets of P. zonatus and P. auricilla differed 

 considerably from each other (Table 1). By volume, 

 the 106 food samples from P. zonatus collected for 

 this study were composed of 59.0% invertebrates 

 and 41.0% fishes. The dominant invertebrate prey 

 included the pelagic tunicate Pyrosoma spp. and 

 galatheid crabs, nearly all Munida japonica. The 

 most frequently occurring fishes in the diet were the 

 ophichthid eel, Shultzidia johnstonensis, and the 

 symphysanodontids Symphysanodon maunaloe and 

 iS. typv^. In general, P. zonatus preyed heavily upon 

 benthic organisms. Many of the prey items, such as 

 echinoderms, octopods, and the various benthic 



Discussion 



As previously mentioned, a few fragmentary 

 reports (cf. Parrish 1987) from various localities 

 have addressed the trophic relationships of any 

 tropical snapper inhabiting waters deeper than 100 

 m. He further attributes this lack of information to 

 the normally remote fishing localities and, most of 

 all, to the loss of stomach contents through regur- 

 gitation during capture. 



The problem of regurgitation has plagued trophic 

 studies of demersal fishes, regardless of whether 

 fishing involved trawls (Mauchline and Gordon 1984; 

 Clark 1985; Bowman 1986) or hook and line (Kluegel 

 1921; Forster etal. 1970; Seki 1984). Likewise, most 

 food samples for our study were salvaged, regurgi- 

 tated spe wings retained in the mouth, throat, or gill 

 rakers of the fishes. It is possible that the material 

 occurring as spews may consist predominantly of 

 certain prey items more likely retained because of 

 size or some morphological structure, thereby result- 

 ing in a biased interpretation of the diet. We never- 

 theless employ the assumption that what is recov- 

 ered is representative of the diets at the time of 

 capture. 



Based upon depth of capture information, P. 

 zonatus and P. auricilla occupy nearly the same 

 habitat, and considerable spatial overlap occurs in 

 their foraging zones (Polovina 1986; Ralston and 



808 



