Chuwen et al Changes in the catch rates and length and age at maturity of Cnidoglanis macrocephalus 
255 
Total length (mm) Age (years) 
Figure 4 
Relative selectivities for estuary cobbler ( Cnidoglanis macrocephalus) (A) by total 
length for each of the various gillnet panels and (B) by age for the composite gill 
net in Wilson Inlet. Stretched mesh sizes (mm) are given above the curves in A for 
each gillnet panel. 
(95% CL=514-569 mm TL), respectively. The percentage 
of females that were mature at the end of their third 
year of life was far greater in 2005-08 (62%) than 
during 1987-89 (15%) and the same was true for females 
at the end of their fourth year of life (95% vs 73%) 
(Fig. 5). Consequently, the estimates of the A 50 and 
A 95 for female C. macrocephalus at maturity during 
2005-08, i.e., 2.9 years (95% CL=2. 6-3.1 years) and 
3.9 years (95% CL = 3.7-4.4 years), were significantly 
different (P<0.001) from those during 1987-89, i.e., 
3.5 years (95% CL=3. 4-3.7 years) and 4.7 years (95% 
CL = 4.6-5.2 years). 
The number of large males caught during the spawn- 
ing period was very low because, at this time, they 
tend to occupy burrows within which they brood eggs 
and rear yolk-sac larvae under their pelvic fins (Laur- 
enson et al., 1993a). It was thus not possible to derive 
reliable estimates for the L 50 or A 50 for the males of C. 
macrocephal us . 
Growth 
Although the likelihood-ratio test showed that the von 
Bertalanffy growth curves for females and males of C. 
macrocephalus in Wilson Inlet were significantly differ- 
ent at the a=0.05 level of significance (each P<0.001) 
for both 1987-89 and 2005-08, the differences in the 
estimated lengths at each integer age between 1 and 
4 years, ages that encompassed the majority of the 
data, were less than 3% of the mean of the asymptotic 
lengths of the two growth curves for each period. The 
differences were thus considered not to be of biological 
significance. The length-at-age data for the two sexes in 
each period were therefore pooled. The von Bertalanffy 
growth curve provided a good fit to the lengths at age 
for all but the older C. macrocephalus for both 1987-89 
and 2005-08; the lengths at age of the relatively small 
number of older fish were overestimated by the curve 
(Fig. 6, Table 2). 
The likelihood-ratio test showed that the von Ber- 
talanffy growth curve for C. macrocephalus in Wilson 
Inlet during 2005-08 differed significantly (P<0.001) 
from the corresponding curve derived for this species 
for 1987-89 (Fig. 6). However, because the differences 
in the estimated lengths at each integer age between 
ages 1 and 4 years (ages that encompassed the ma- 
jority of fish) were less than 2% of the mean of the 
two asymptotic lengths, those differences were not con- 
sidered biologically significant. 
Maturation reaction norms 
Logistic regression analyses demonstrated that, for both 
1987-89 and 2005-08, maturity was better described 
as a function of length (log-likelihood [LL\ = -61.2 and 
-21.0, respectively) than of age (LL=- 102.2 and -31.2, 
respectively). Furthermore, the quality of the fit was 
not improved significantly by including age as well as 
length in the model (both P> 0.05, LL=-61.2 and -20.0, 
respectively). Although the coefficient of the length term 
in the logistic curves relating the probability that a fish 
was mature to its length and age, i.e., j3, differed sig- 
nificantly between the two periods (P<0.05), the values 
for the other parameters (a and y) did not differ greatly 
(both P>0.05). 
As expected, given the lack of significance of age 
when fitting the logistic regression equation to the 
combination of both length and age data, each of the 
reaction norms was essentially independent of age 
(Fig. 6). In other words, in each period, fish became 
mature at approximately the same length in that period 
regardless of their age. 
The reaction norm for maturation shifted markedly 
toward lower lengths at age between 1987-89 and 
2005-08, but the slopes of the curves did not change 
(Fig. 6). The von Bertalanffy growth curves did not 
differ in biological terms between the two periods, but 
because the intersection of the growth curves occurred 
