Gaughan and Potter: Analysis of diet and feeding strategies within an assemblage of estuarine larval fish 
723 
whole, mouth size limits prey size; thus prey width 
is typically the limiting dimension for ingestion (e.g. 
Hunter, 1984, Heath, 1992). It is therefore impor- 
tant to examine mouth size and prey width when 
exploring the trophic relations of larval fish. 
The first aim of this study was to examine the re- 
lation between mouth width, prey width, and dietary 
composition of the larvae of five teleosts in an estu- 
ary. The dietary data were then used to examine the 
extent of DNO between these species with a tech- 
nique that takes into account relative prey concen- 
trations. With this procedure we were able to test 
the hypothesis that divergence in gape size between 
species should be accompanied by a decrease in the 
prevalence of significant DNO between these species. 
Bootstrapping was used to assess whether species- 
pair DNO values were significant. The results of us- 
ing this robust approach were compared with those 
obtained when relative prey concentrations were not 
included in the calculation of DNO and when a sub- 
jective level of >0.6 was considered to be significant 
for the DNO values. 
Materials and methods 
Sampling methods 
This study was carried out in Wilson Inlet (35°00'S, 
117°24’E), an estuary in southwestern Australia that 
comprises a 48 km 2 basin with two main tributaries, 
the Denmark and Hay rivers. Although samples were 
collected monthly between July 1988 and June 1989, 
the data used in this paper are restricted to those 
obtained between October 1988 and April 1989 when 
fish larvae were most abundant. Ichthyoplankton 
and zooplankton were sampled from open waters of 
the upper, middle, and lower basin, and the central 
channel of the lower saline reaches of the Denmark 
River, located 10.7, 8.3, 2.0, and 7.3 km, respectively, 
from the estuary mouth. The water depth in each 
region was between 2 and 3 m. 
Sampling was initiated soon after sunset to reduce 
the likelihood of larvae avoiding the plankton nets. 
Fish larvae were collected with a pair of 500-pm- 
mesh conical nets, each with a mouth diameter of 
0.6 m and a length of 2 m. The nets were attached to 
either side of a powerboat and towed for 10 min just 
below the surface of the water at a speed of 1.5 m/s. 
During each ichthyoplankton tow, three to five zoop- 
lankton samples were taken from the surface with a 
conical, 53-pm-mesh net with a mouth diameter of 
0.35 m. The volumes of water filtered during each 
ichthyoplankton and zooplankton tow were measured 
with flowmeters. The zooplankton tows were 7-10 s 
in duration. The flowmeter in the zooplankton net 
was closely observed during each tow. A tow was im- 
mediately terminated if the propeller speed suddenly 
decreased — a sign that the net was clogging. Samples 
were fixed in a 5% solution of formalin, which was 
replaced with 70% ethanol on the following day. The 
detailed results of the zooplankton sampling are 
given in Gaughan and Potter (1995). 
Laboratory procedures and data analyses 
Zooplankton were identified and counted under a 
dissecting microscope from subsamples of the repli- 
cate samples. Counts were standardized to numbers/ 
m 3 ; thus mean concentrations of taxa at each region 
within each month were able to be calculated. Rela- 
tive proportions of those zooplankton taxa that con- 
tributed to larval diets at any time during the study 
were calculated for each sample. These represented 
relative resource availability (i?.). 
The gobiids Pseudogobius olorum, Afurcagobius 
suppositus, and Favonigobius lateralis , the blenniid 
Parablennius tasmanianus, and the syngnathid 
Urocampus carinirostris were chosen for the present 
study because their larvae are abundant in Wilson 
Inlet from late spring to early autumn (Neira and 
Potter, 1992), collectively contributing 70.8% of the 
total open-water assemblage of larval fish in this 
estuary between September 1987 and April 1989. 
All larvae of each species in a sample were removed 
and counted. Body length (BL) of each larva (i.e. the 
distance from the snout to tip of notochord in 
preflexion and flexion larvae and from the snout to 
the posterior end of the hypural plate in postflexion 
larvae ILeis and Trnski, 1989]) was measured to the 
nearest 0.1 mm. Since a focus of this study was the 
comparison of mouth width with prey width, the di- 
ets of individual size classes of larvae were deter- 
mined. However, because the analyses of DNO were 
undertaken for co-occurring species within individual 
samples, the dietary data for all size classes of each 
species were pooled (see below). 
The smaller larvae of P. olorum, P tasmanianus, 
F. lateralis, and A. suppositus were each grouped into 
1.0-mm length classes. Because P. olorum and P. 
tasmanianus >5 mm BL and F. lateralis and A. sup- 
positus >6 mm BL were rarely caught, larvae of these 
four species longer than these respective lengths were 
each grouped into single length classes. Because the 
length range of U. carinirostris was relatively wide, 
the larvae of this species were grouped into length 
classes with intervals of 3 or 4 mm, depending on 
the numbers caught. 
Items in the gut were identified and counted. Maxi- 
mum widths of intact dietary items, which typically 
