Designing and evaluating length-frequency surveys for trap fisheries with application to the southern rock lobster

A survey design for estimating the length distribution of harvested souther n rock lobsters (Jasus edwardsii) was developed for the South Australian fi shery. Experimental sampling was carried out by volunteer fishers in spring 1996 and autumn 1997 to test three proposed survey designs. A variance com ponents analysis indicated that it would be more efficient to sample one po t per trip from all trips rather than the previous design of sampling multi ple pots from a few trips. The variation among licenses (fishers) accounts for most of the remaining sample variance. Onboard research sampling by sci entists, who in the past measured from all pots on selected trips, was show n to be the least efficient design option in comparison with volunteer samp ling by fishers. A sampling protocol where fishers measure one to three pot s per trip has been adopted by the South Australian rock lobster fishers. E stimators, based on a three-level sampling hierarchy of pot, day, and licen se, are presented for estimating the mean and sample variance of the number s harvested overall and within each length category.