Retrospective investigation of assessment uncertainty for fish stocks off southeast Australia

Abstract There is a need to provide quantitative measures of uncertainty to support fisheries management decision making. A retrospective analysis of historical assessments for fish stocks off southeast Australia is conducted to quantify the extent of uncertainty associated with estimates of spawning stock biomass in absolute terms and when expressed relative to spawning stock biomass over a sequence of reference years. This approach to quantifying uncertainty captures more sources of uncertainty than alternative approaches, such as the estimate of the variance of terminal year spawning stock biomass from asymptotic methods, the extent to which estimates of spawning stock biomass vary among the sensitivity tests that form part of most assessments, and conventional retrospective analyses. By all measures, estimates of spawning stock biomass in absolute terms are much less certain than estimates of relative stock size (i.e. spawning stock biomass relative to a reference level), although application of most current harvest control rules rely on estimates of biomass in absolute terms. Overall, uncertainty in estimates of spawning biomass in absolute terms can be represented as a log-scale standard error of 0.37, while this standard error is 0.18 for estimates of spawning biomass in relative terms. There is considerable variation in among-assessment uncertainty in stock assessment outputs across species groups, with, for example, higher variation for assessments of chondrichthyans compared to other species.