Inputs to the stock assessment of Southwest Pacific shortfin mako shark

Shortfin mako (Isurus oxyrinchus) are large pelagic predators that are caught as bycatch in tuna and billfish fisheries worldwide. This report details data inputs for the southwestern Pacific stock assessment for shortfin mako, including length frequency information from regional observer programmes, reconstructed catch histories, and a number of alternative catch per unit effort (CPUE) series.

Observed biological data held by the Pacific Comminity (SPC) were relatively sparse for shortfin mako, leading to high spatial variablty of samples, with few clear spatial trends. Despite the paucity of records, higher latitude fisheries appeared to capture juvenile (1–2 m) fish that do not appear as frequently in other fisheries. Based on the occurence of age zero and juvenile sharks in these latitudes, we hypothesised that fleets in these latitudes form a separate fishery from those in lower latitudes. In addition, observer data suggests very low relative abundance between the equator and 15^∘ South, therefore, we excluded these equatorial data. Fleets were therefore structured latitudinally, as combining data within latitudinal bands helped to overcome these data deficienies.

Catch was reconstructed from observer data using Bayesian spatial GLMMs to extrapolate observed CPUE to unobserved effort. Due to the importance of extrapolation, we used blocked cross-validation by vessel-flag to weight models of varying complexity in terms of their prediction skills for data from different fleets. Model weighting using stacking of Bayesian posterior distributions, showed that no single model performed best across all fleets, and the most complex model was not usually the best one. We combined models by averaging weights across all holdout sets, producing a weighted ensemble prediction of catches. The model produced relatively high catches between the mid-1990s and early 2000s, with relatively strong reductions in catch since about 2010. Additionally, analysis of discarding rates suggested that across many fleets with consistent recent obersever coverage, discarding rates have been high, with a large proportion of mako sharks being cut free and released alive in the most recent years.

Logsheet based CPUE series were attempted, using delta-lognormal and negative binomial GLM models, for a number of areas and fleets, including New Zealand, Australian, Japanese and topical south Pacific fleets. We found that there was little consistency in CPUE trends for Southwest Pacific mako, especially in latter years. Trends in the 1990s are relatively uncertain, due to poor observer coverage, and poor reporting of sharks in logsheet data. While early CPUE in the 1990s often showed a decline, recent CPUE in some fleets has been increasing, i.e., New Zealand, while CPUE in other areas has been relatively flat or even declining in recent years. We suggest that this discrepancy may be due to these indices measuring different components of the stock, as evidenced by latitudinal length frequencies.

We found that recent rates of cutting sharks free from lines may have resulted in lower recent CPUE in many fleets if cut-free sharks are not recorded in log-sheets and the possibility that they may not all be seen by observers. To adjust for this, we produced alternative CPUE time series that included the rate of cutting free to provide a more realistic measure of encounter rate of mako sharks. Despite these adjustments we found relatively little consistency in fleet specific CPUE trends, which may hint at either regional abundance patterns, or problems with using logsheet CPUE to index shortfin mako abundance.

The following recommendations are made:

• Future assessments should spend increased effort to reconstruct spatio-temporal abundance patterns for shortfin mako, and develop a better understanding of how these patterns drive regional abundance indices.

• Providing more time, either as inter-sessional projects, or by extending time-frames for shark analyses will allow more thorough investigation of input data quality and trends, which shape assessment choices. In addition, this approach would allow input analyses to be completed in time to be presented to the March pre-assessment workshop prior to the stock assessment commencing. Moreover, this will provide more time for the assessments themselves allowing a more thorough investigation of alternative model structures or assessment approaches.

• Increased effort should be made to re-construct catch histories for sharks (and other bycatch species) from a range of sources. Our catch reconstruction models showed that model assumptions and formulation can have important implications for reconstructed catch. Additional data sources, such as log-sheet reported captures from reliably reporting vessels, may be incorporated into integrated catch-reconstruction models to fill gaps in observer coverage.