Alternative assessment methods for oceanic whitetip shark

The present study evaluates potential alternative assessment methods for sharks, using oceanic whitetip shark (Carcharhinus longimanus) as a case study, allowing comparisons with the current age-structured integrated stock assessment of this species. The latter was conducted in parallel to the present study, and used Stock Synthesis 3 (SS3) software.

The most recent previous integrated assessment of oceanic whitetip shark concluded that the stock was overfished and that overfishing was continuing. To minimise ongoing fisheries impacts on this species, a non-retention measure (Conservation and Management Measure CMM 2011-04) was imposed by the Western and Central Pacific Fisheries Commission (WCPFC); however, the non-retention of oceanic whitetip shark also introduced additional uncertainty about the value of indicators such as catch-per-unit-effort (CPUE) for the monitoring of population status. In conjunction with limited data about the efficacy of measure CMM 2011-04 for limiting fishing mortality, the current stock status of this species remains uncertain.

Here, we compared three approaches in conjunction with the current integrated stock assessment of oceanic whitetip shark. These approaches were catch-only simulations, a general spatial risk assessment model, and a Bayesian dynamic surplus production model. We also illustrate the impact of different assumptions on estimates of fishing mortality (F) and risk (F/F_crash</sub) to the oceanic whitetip shark stock in the Western and Central Pacific Ocean.

Our findings suggest that catch-only methods are most valuable as a tool to refine Bayesian priors in more sophisticated analyses, as on their own, catch-only methods are dependent on assumptions and provide no relevant management outputs. Nevertheless, we show that by making simple and relatively broad assumptions about the current depletion level, catch alone can constrain initial (unfished and/or starting depletion for the catch time series) population size and productivity and, thereby, serve as a a priori constraint on these parameters.

The application of dynamic surplus production models (DSPMs) showed that these model may provide a reasonable tool to rapidly assess shark stocks, either alongside or instead of fully integrated stock assessments. Dynamic surplus production models can be readily applied to sharks: their implementation in widely-available software packages means that they are a cost-effective assessment tool that requires few assumptions. In addition, these models can provide estimates of management-relevant quantities (e.g., stock status, fishing mortality), which have been shown to be robust for sharks. Furthermore, depletion-based catch-only simulations can be used to construct useful priors for Bayesian implementations of these models. Nevertheless, the reliance of DSPMs on a reliable biomass index (e.g., CPUE time series with contrast) and on complete removal estimates (i.e. the availability of a catch series which accurately reflects total catch) limits their application to species for which these time-series data can be derived. This aspect may exclude the application of DSPMs to species with poor historical identification records such as many shark species.

We also applied a spatial risk assessment (SRA), as this approach only requires recent catch and effort data to estimate fishing mortality, so is less constrained by historical data limitations. Because SRAs generally do not use complete time series of removals, they cannot provide information about stock status. The most commonly employed SRA methods are conceptually similar to fisheries surveys, as they use estimates of gear efficiency to scale observed spatial catch to overall catch via a spatial population density estimate. To derive absolute fishing mortality and risk, however, these methods need to make assumptions about the spatial interaction of the fishing gear with the local population density. This scaling is difficult to establish for longline gear and has a large effect on estimated risk.

For this reason, we suggest that risk assessment methods are employed when 1) no robust time series for catch and CPUE can be derived, and 2) it is possible to make reasonable assumptions about the spatial effect of the fishing gear. Even with these limitations risk assessment methods can be particularly valuable for prioritising assessment and conservation efforts, as they can be readily employed across species in a standardised framework, even for species with limited historical data.

Application of a variety of models to the oceanic whitetip shark stock showed that DSPM, SRAs and SS3 provided similar results, but SRA results were strongly dependent on the assumption of spatial gear effects. All methods suggested that there is a substantial risk that current fishing mortality remains above F_crash</sub, the fishing mortality that would lead to extinction in the long term (and by extension, well above FF_lim</sub and $F_MSM</sub). The SS3 assessment estimated slightly higher overall fishing mortality and lower productivity and stock status, and therefore provides the most pessimistic view of current fishing mortality and sustainable fishing mortality. All methods suggest that reductions of fishing mortality below likely values in the last year of the assessment (2016; about 45% total fishing mortality including haul-back, handling and post-release mortality) would substantially lower existential risks for this stock.

Based on our findings, we suggest that the Scientific Committee considers the following:

• Inferences from different models indicate that oceanic whitetip shark continues to be overfished, and overfishing may still be occurring owing to incidental mortality from fishing, despite non-retention measure CMM 2011-04. Estimated fishing mortality rates for the last year in the assessment (2016) lead to substantial risk that the stock will not persist.
• Spatial risk assessment methods should be employed for species with poor historical records (e.g., poor species identification), but for which recent records are judged reliable. In addition, a standardised methodology based on spatial risk assessment methodology could be employed to prioritise assessment and conservation efforts.
• Surplus production models can provide a robust cost- and time-effective way to assess shark populations, and provide similar outputs to fully integrated stock assessments such as SS3. Therefore, they may be considered as a rapid assessment tool, either alongside or instead of fully integrated stock assessments, which could be employed for species of high priority.
• Depletion-based catch-only simulations should be considered for constructing priors for DSPMs and to understand the amount of additional information provided by fitting the DSPM.