Laserfiche WebLink
recaptures 6-12 months (182-364 days) after initial capture. A minimum of 31 days <br />was used because measurable differences in growth may not be detectable in a shorter <br />interval and to minimize compounding effects of minor measurement errors (Manire and <br />Gruber 1991). <br />Analysis of covariance (ANCOVA) was used to detect statistical differences <br />between the two gear types in growth rate for each species and growing season. <br />Growth rate increment was the dependent variable and logo total lengfh was the <br />covariate. Logo total length was used because residual plots revealed a better fit of the <br />data than with non-transformed total length. The ANCOVA terms for interaction of <br />(log,o total length X gear) and gear were tested for homogeneity of regression-line <br />slopes and intercepts, respectively. Anon-significant interaction term in the first <br />ANCOVA indicated that the regression lines of growth rate as a function of logo total <br />length for each gear type (regression slopes) were parallel. If slopes were parallel, then <br />least-squares means of growth rate (population marginal means; Searle et al. 1980) <br />were calculated for each group of fish captured by the two gear types after accounting <br />for the covariate (log,o total lengfh) as recommended by Trippel and Hubert (1990) and <br />differences in intercepts were tested by the second ANCOVA. A significant difference <br />between intercepts in the second ANCOVA indicated that growth rate differed <br />depending on gear at initial capture. If the first ANCOVA detected non-parallel slopes, <br />then mean growth rates were calculated without adjusting for initial fish length and a <br />t-test was used to detect statistical differences in means. The ANCOVA and t-test were <br />conducted with SAS statistical software (SAS Institute 1988). Biological significance of <br />5 <br />