For all four species, logistic models provided the best fits to the data. Maximum likelihood parameter estimates for the logistic selectivity curves and goodness of fit measures from the Rep.ttfit models runs are presented in Table 2. The reported selection parameter estimates should be interpreted as relative (Wileman et al., 1996) rather than absolute (i.e., the estimated selection factors should not be used to estimate retention lengths for other mesh sizes) because the control net did not exhibit 100% retention of all size classes.