Age-length keys from RV Kaharoa trawl surveys in the years 1984-1990 and 19
92-1994 were used to estimate mean lengths-at-age of snapper (Pagrus auratu
s) in the Hauraki Gulf of New Zealand. A von Bertalanffy growth curve fitte
d to these lengths-at-age exhibited poor fit, and when the fit was restrict
ed to only cohorts since 1968 (those for which sea surface temperature data
were available) the estimated maximum length parameter dropped 5 cm. A pow
er curve provided a better fit without an increase in the number of paramet
ers and was not sensitive to the restriction to younger cohorts. When water
temperature was added to the power curve model the residual plots indicate
d unexplained year effects and temporal correlation. These features were in
cluded using a mixed-effects model for repeated-measures data, and temperat
ure was found to be statistically significant (p < 0.0001). This conclusion
was corroborated via a simple randomization test. The model predicts growt
h over a year to change by the equivalent of 8 weeks if the average annual
water temperature changes by 1 degrees C. This is sufficient to predict a 4
-year-old snapper from the 1986 cohort as having the equivalent of 6 months
more growth (at the growth rate corresponding to average water temperature
) than a 4-year-old fish from the 1990 cohort.