Increasing confidence in estimates of average weight and recreational harvest ranges

Quantifying recreational fishing harvest by weight is vital for stock assessments and fisheries management. As it is impractical to obtain weights of all fish caught by recreational fisheries, harvest is often calculated from estimated catch in numbers multiplied by an estimate of average weight (usually as an arithmetic mean). The average weight may be derived from measured weights, as well as imputed weights from measured lengths using established length-weight relationships. This study evaluates the impact of uncertainty in lengths and weights from three independent data sources in determining average weights and estimated recreational harvest of four demersal species in Western Australia. Data sources included measured lengths and weights from on-site surveys of boat-based fishers, self-reported lengths from charter-boat logbooks and lengths from biological samples voluntarily donated by recreational fishers for stock assessments. For the latter two data sources, weights were imputed from length measurements. Generalised linear models were used to obtain estimates of standardised average weight and standard errors for each data source, across four years (2011/12, 2013/14, 2015/16 and 2017/18) and three spatial management zones within the West Coast Bioregion of Western Australia (Mid West, Metropolitan and South West). For each species, harvest estimates and 95% confidence intervals were derived from standardised average weights (from aforementioned surveys) and estimated catch in numbers (from off-site surveys). Standardised average weights from GLMs were found to be more precise than arithmetic means and data from the charter-boat logbooks and biological samples generally produced higher harvest estimates. The application of standardised weights from management zones to estimate recreational harvest at the bioregion level reduced the error of estimates. Addressing uncertainty from self-reporting, data sets (charter-boat logbooks and biological samples) and small sample sizes (on-site surveys) can increase confidence in recreational harvest estimates.