Development of Southern Hemisphere porbeagle shark stock abundance indicators using Japanese commercial and survey data

Japan holds the longest-term and largest spatial extent of porbeagle shark (Lamna nasus) catch-effort data from longline fishing in the Southern Hemisphere, and also size and sex composition data collected by observers. The various sets of catch and effort data have previously been analysed, but opportunities were
identified for improving the analyses. These include modelling spatial effects at a finer scale, investigating spatial variation in trends, accounting for the effects of targeting behaviour, and accounting for differences
in catchability among vessels. Potential to examine reporting reliability was also identified. We considered that improvements may significantly affect the porbeagle stock assessment, because of the crucial nature of the Japanese data. In a collaborative project, New Zealand and Japanese scientists worked together to reanalyse these data. We explored issues related to reporting reliability, analysed the spatial and temporal distribution pattern of different size and sex classes of porbeagles, and modelled catch rates in several different datasets. In investigating reporting rates, we developed a new method for identifying groups of longline sets that
reliably report sharks, given that their reporting rates are similar to those seen in the observer data. The estimated reliability of shark reporting in logbooks before 2008 appeared to be low, but increased sharply in 2008, with the proportions of nonzero shark catches almost reaching the levels in the observer data. The reasons for the 2008 change are unclear, but it will be important to identify them. Increased reporting rates are likely to affect catch per unit effort (CPUE) indices derived from the logbook data, and a large increase in CPUE in 2008 is indeed apparent in previously estimated porbeagle shark CPUE indices. Identifying the cause of this change will help to determine whether other indices may be affected.
Data preparation included identifying a number of different targeting strategies using cluster analysis on the species composition data for well reported tuna and billfish species. This approach assumes that
different fishing strategies will catch, on average, different mixtures of species. Species compositions were indeed very distinct between the fishing strategies, with effort targeted at southern bluefin tuna
reporting very low catch rates of other tuna and billfish species. Fishing strategies were also spatially and seasonally separated, but the separation was sufficiently complex and variable that categorization based on covariates alone would have resulted in incorrect allocation of sets. There may be other covariates that would reliably differentiate fishing strategies, but if so these have not been recorded, or were not available to the analysts. The strategies that Japanese longline vessels use to target different species have also in
some cases changed through time, suggesting that species composition data may be more useful than, or complementary to, operational covariates as an indicator of the intentions of individual vessels. Grouping the data by fishing strategy will affect abundance indices based on catch rate, because different fishing strategies are likely to have different catch rates for the species of interest (porbeagle).