Historical catch reconstruction and CPUE standardization for the stock assessment of oceanic whitetip shark in the Western and Central Pacific Ocean

This is a companion paper to the 2019 stock assessment for oceanic whitetip shark (Carcharhinus longimanus)(Tremblay et al. 2019) detailing the historical catch reconstructions and CPUE standardization approaches used to create the data inputs to the assessment. We repeat here the methods from the main assessment report, provide a discussion of key results and present detailed model diagnostics on the observer data-based catch reconstructions and CPUE standardization. Other inputs to the assessment were described in (Tremblay et al. 2019).

Historical catches were reconstructed based on observer catch rates as logbook-reported catches of oceanic whitetip shark were considered unreliable over the assessment period of 1995–2016. We developed a prediction-model from observer catch rates to apply to known longline and purse-seine effort across the Western and Central Pacific Ocean. Catches aggregated across fleets were predicted to peak in 2001 and decline thereafter. The catches for the longline bycatch fleet are predicted to be the highest of all fleets, with catches for the other fleets negligible in comparison. This allocation probably reflects in part the reliability of catch rates recorded by the observer programmes assigned to each fleet, as well as the effort information used to extrapolate catches. The catches predicted were generally higher than those estimated from the same dataset by previous authors but predictions were also made here for CCMs that are typically excluded from these analyses due to a poor characterization of effort. The catch predictions otherwise showed good agreement for key years with an alternative time series that was developed based on global fin trade statistics.

A single standardized index was developed for this assessment based on the longline bycatch fleet following discussions at the 2019 Pre-Assessment Workshop (PAW) (Pilling & Brouwer 2019). This fleet was chosen as the underlying data were deemed the most reliable of those available, and also covered the greatest extent within the assessment region. The standardized year effects resulting from the CPUE model were highly variable at the start of the time series and started steadily declining from 1999 onwards. Influential covariates in the model included sea surface temperature, hooks-between-float as a proxy for set depth, species targeting cluster and observer programme.

The analyses presented here included years from the earlier period of observer programmes in the WCPO. Observer reporting for sharks for this period is known to be unreliable as well as highly variable given the sparse coverage. Any results for this early time period must be interpreted carefully. In addition, we found evidence of observer programmes having an impact on recorded catch rates independent from vessel flag, such that observer training, as would be expected, is a likely influence on the catch rates recorded for sharks across time and programmes. As this is a variable that is challenging to quantify and so cannot be standardized for, the results presented here should be interpreted in light of this caveat.