Redefining risk in data-poor fisheries
The productivity susceptibility analysis (PSA) is a widely used method to rapidly assess species risk to fishing activities in data-poor fisheries. A step in ecological risk assessments and used in data-poor assessment for sustainable fisheries certification programmes (e.g. MSC) and recommendation lists (e.g. Seafood Watch), the PSA is semi-quantitative, yet little attention has been given to the theoretical basis of this analysis. Current thresholds designating low-, medium- and high-risk categories divide the PSA plot by equal area, assuming area corresponds to likelihood. We show that plot area does not correspond to likelihood, however, and existing thresholds need revision due to the non-uniform distribution of vulnerability scores on the PSA plot. The probability of medium risk assignment increases with the number of attributes used to characterize productivity and susceptibility. Here, we present a novel and statistically robust method to derive vulnerability, where threshold values between the risk categories are adjusted with the number of attributes used in the assessment. Our comprehensive framework accounts for all variations in the method, including logarithmic scaling of axes, weighting of attributes and scoring procedures. Simulated results across a range of conditions and comparative evaluation of 302 species in five studies show that one-third of species may be re-categorized with the new PSA approach. Importantly, the existing PSA approach underestimates risk by up to 35% when compared with the new method. These findings have strong implications for management of data-poor fisheries. We recommend adoption of this approach to the PSA to better resolve species’ risk.