Improving design-based estimates of biological data collected from a restricted spatio-temporal access point survey of recreational fishers

Abstract

Accurate and precise estimates of recreational harvest are required where there are formal catch allocations between commercial and recreational sectors. Yet they are challenging to obtain without appropriately designed surveys, especially in the recreational sector, which often require complex survey designs. Estimates of mean body mass are required for recreational fisheries when converting catch by numbers to catch by weight and should account for pseudoreplication to truly represent the characteristics of the data. However, methods for determining mean mass from recreational fishing surveys are rarely defined, leading to assumptions that an arithmetic mean has been applied, which does not account for pseudoreplication or hierarchical, clustered sampling. This study uses western rock lobster (Panulirus cygnus), commonly targeted by recreational fishers in Western Australia, as a case study to determine a mean mass for retained lobsters using several survey design-based methods. The six-year, restricted spatio-temporal access point survey of boat-based recreational fishers, who target lobsters by trapping or diving, addresses pseudoreplication by aggregating survey data and calculating a weighted mean and uncertainty for non-normally distributed data using nonparametric hierarchical bootstrapping. While we present the mean and confidence intervals for all levels of the survey design, our preferred level is the ultimate sampling unit (lobster) because there are insufficient replicates to conduct nonparametric hierarchical bootstrapping at the primary (day) or secondary (interview) sampling units. For this species, we recommend using a fishing method-based mean rather than an overall mean to capture annual variability in catch and participation and provide an index for determining harvest when catches by number are known for each fishing method. This approach for analyzing recreational fishing data is relevant to other fisheries estimating total recreational harvest from restricted spatio-temporal surveys.