730 
Fishery Bulletin 95(4), 1997 
tion at the other two regions was due to large fluc- 
tuations in the abundance of zooplankton between 
months. These fluctuations did not appear to influ- 
ence monthly trends in the abundance of fish larvae, 
probably because concentrations of zooplankton typi- 
cally remained high(> 100,000/m 3 ). 
Because in this study we were limited to examin- 
ing the diets of larval fish, a complete assessment of 
dietary relations and the potential for competition 
within the plankton community could not be under- 
taken. However, other zooplankton taxa (e.g. Sag- 
itta minima) sufficiently large to have used the same 
food resources as larval fish were rare in Wilson In- 
let, contributing less than 0.2% of the total numbers 
of zooplankton (Gaughan and Potter, 1995). 
Feeding strategies 
The diets of the fish larvae from Wilson Inlet may be 
viewed as representing a spectrum of feeding 
strategies. The diet of A. suppositus is distinguished 
from those of the other four species by its broader 
composition, the larger size of its prey items, and 
the smaller numbers of prey consumed. At the op- 
posing end of the spectrum, P tasmanianus larvae 
consumed large numbers of small prey items. The 
feeding strategies of the larvae of P. olorum, F. 
lateralis, and U. carinirostris lay between these ex- 
tremes; these species consumed many small and in- 
termediate-size prey which were occasionally supple- 
mented with larger prey items. 
Because the trophic character of a species may be 
influenced by both size and structure as well as be- 
havior (Lavin and McPhail, 1986), the small influ- 
ence of mouth width on the size of prey consumed by 
larval fish in Wilson Inlet indicates that the differ- 
ent feeding patterns among larval species probably 
resulted from behavioral differences (Bremigan and 
Stein, 1994). These patterns, which occurred despite 
the high concentrations of zooplankton, may enhance 
survival, and hence recruitment, if marginally low 
concentrations of zooplankton were present at tem- 
poral or spatial scales beyond those sampled. 
Evaluation of methods 
In this study, we examined a technique for assessing 
DNO that consists of two parts (accounting for zoop- 
lankton abundance in the calculation of BNO and 
objectively assessing significance of DNO with 
bootstrapping). This technique was substantially 
more conservative than that which did not consider 
zooplankton abundance and which did subjectively 
assess significance. Prevalence of significant DNO 
thus doubled (32.6% to 65.1%) when zooplankton 
data were not included in the calculations and an 
arbitrary cutoff point of 0.6 was used to test for sig- 
nificance. The less conservative techniques of mea- 
suring DNO and assessing its significance would 
have therefore overestimated the degree of DNO 
among fish larvae in Wilson Inlet. 
A large overestimation of DNO would likely have 
led to a different interpretation of the data. For ex- 
ample, the higher rate of significant overlap may 
have led to the conclusion that competition for food 
was sufficiently high to influence markedly the sur- 
vival rate of fish larvae in Wilson Inlet. In contrast, 
the lower prevalence of overlap is more consistent 
with our previous hypothesis that food is unlikely to 
be limiting for the open-water assemblage of larval 
fish in Wilson Inlet (Gaughan and Potter, 1995). 
Although concentrations of potential zooplankton 
prey are not necessarily directly related to their avail- 
ability, we suggest that inferences regarding compe- 
tition for food among larval fish may be misleading 
if data on the abundance of the zooplankton are not 
considered when measuring DNO. In studies of other 
taxa, or even of adult fish, where the abundances of 
prey in the environment may be very difficult or im- 
possible to estimate without bias, resource availabil- 
ity can be estimated in a circular manner with pro- 
portional-utilization data (see Winemiller and 
Pianka, 1990). However, because the majority of fish 
larvae and their potential prey are planktonic, small, 
and relatively immobile (thus highly susceptible to 
capture with plankton nets), estimates of prey con- 
centrations in the environment should be used to 
calculate DNO for larval fish. Likewise, because a 
subjective assessment of the significance of DNO is 
inadequate, bootstrapping techniques may prove to 
be useful in making an objective examination of di- 
etary relations, which are typically awkward to ana- 
lyze statistically (Winemiller and Pianka, 1990; 
Baltanas and Rincon, 1992). 
Finally, although the two parts of the technique 
used in this study, i.e. objectively assessing signifi- 
cance and accounting for zooplankton abundance, 
each contributed to the overall result, individually 
the former had a greater influence ( 18.6% and 20.9%) 
on the estimated prevalence of significant DNO than 
the latter ( 11.6% and 13.9%). Even though the direc- 
tion and magnitude of the differences between the 
two parts of this technique may apply only to the 
current study, this finding further suggests that both 
the incorporation of prey abundance data and an 
objective assessment of significance need to be con- 
sidered in an analysis of dietary overlap because ei- 
ther may have more influence on the apparent preva- 
lence of significant DNO. 
