90 
A. C. SANGER 
ted dry, and weighed to the nearest 0.001 g. The 
Gonadosomatic Index (GSI) was calculated by 
expressing ovary weight as a percentage of total 
body weight. Mean egg diameter (to the nearest 
0.1 mm) was determined microscopically with 
the aid of a graduated scale. If mean egg dia¬ 
meter was at least 2.0 mm all eggs were dissected 
from the ovary and counted. Such ovaries were 
arbitrarily considered to be gravid. The stomach 
contents of several samples were examined 
under a stereo microscope. Individual food 
items were assigned to food categories broadly 
matching those used by Jackson (1978b), except 
that all aquatic Coleoptera were combined in a 
single category instead of two, and three new 
categories were used for food items not covered 
by Jackson’s categories. These new categories 
are Decapoda for freshwater crayfish and 
Paratya sp., Isopoda for the semi-aquatic onis- 
coids, and galaxiids for the native fish Galaxias 
olidus. The composition of the diet was analysed 
by the occurrence, number and dominance 
methods as described by Hynes (1950). For com¬ 
parison with the work of Jackson (1978b), the 
rank method of Pollard (1973) was also used. 
Although it is recognised that these methods of 
analysis are inherently biased in one way or 
another, use of several methods concurrently 
should reveal important components of the diet 
of G . bispinosus. Statistical analyses were per¬ 
formed with BMDP statistical software (Dixon 
1983). 
RESULTS 
All samples were collected from a 100 m long 
section of the creek chosen because of its acces¬ 
sibility. The river bed in this section comprised 
stones varying in size from coarse gravel provid¬ 
ing little in the way of cover, to cobbles provid¬ 
ing adequate cover for fish. Additional instream 
cover was provided by several fallen logs and 
branches, and undercut banks in riffle regions 
and on the outer bank of bends. Gadopsis bispi¬ 
nosus was common in all habitats where 
adequate cover was present. The species was 
encountered infrequently in open water or over 
more homogeneous substrates without cover. 
The two exotic species Salmo trutta and Perea 
fluviatilis and the native species Galaxias olidus 
were also collected from this section of King Par¬ 
rot Creek during the study. Of these species, only 
S. trutta was common. 
Mean monthly discharge varied markedly 
over the two year sampling period (Fig. 1); 
however, variation followed a regular pattern of 
high flows (more than 100 ML.day 1 ) in late 
winter and early spring, and low flows (less than 
20 ML.day -1 ) in late summer and autumn. 
Water temperature ranged between 8.5°C in mid 
winter and 24.5°C in mid summer (Fig. 1). 
Length-frequency data for the entire sample 
(n = 1054) and for seasonal samples are given in 
Fig. 2. Mean TL for the entire sample (Fig. 2A)is 
111 mm (range = 20 to 257 mm). The 40 mm 
(30-49 mm) to 180 mm (170-189 mm) size 
classes comprise more than 90% of the total 
sample; the 220, 240 and 260 mm size classes 
comprise less than 2%. Seasonal variation in 
maximal and mean TL is small (Fig 2B). The 
appearance of the smallest size class (20 mm: 
10-29mm) in the summers of 1981 and 1982 
indicates that recruitment of juveniles occurred 
in this period. In the autumn of both these years 
a strong 0T cohort can be seen as the first peak 
on the histograms. 
The relationship between total length and 
total weight for the entire sample, and for males 
and females separately, is indicated in Table 1. 
Females were slightly heavier than males 
(ANCOVA; p < 0.01).' 
Most individuals less than 100 mm TL were 
classified as juveniles (n = 385), with 57 females 
and 19 males comprising the remainder. The 
difference between the number of males and 
females in these small fish is due to the fact that 
females are easier to recognise than males: 
female gonads are larger than those of males, so 
that females were easier to distinguish from 
juveniles than were males. The sex ratio offish 
greater than 100 mm TL (n = 288 males, 306 
females) did not differ significantly from 1:1 
(Chi square = 0.545, p < 0.75). 
Females with gravid ovaries were found 
between September and December in 1980. and 
between October and December in 1981 (Table 
2). The values of AN in Table 2 show how many 
females in each monthly sample were larger than 
the largest gravid female in that sample. These 
values increased during the spawning period in 
both years. Thus, early in the spawning season 
the largest gravid females tended to be the lar¬ 
gest females in the sample. As the spawning sea¬ 
son progressed, the larger females possessed very 
small developing ovaries (GSI < 1,1) and 
smaller gravid females were present (GSI > 9.4). 
In the January samples, all adult females had 
ovaries in the early stages of development (GSI 
values normally < 1). Water temperature rose 
rapidly from about 12°C to 17°C between 
October and November in 1980, and between 
November and December in 1981. Thus, it 
