Reconciling conflicting survey indices of abundance prior to stock assessment

Abstract Indices of relative abundance are one of the most important inputs into a stock assessment model. For many species, we must rely on several indices that routinely conflict with each other and which may result in biased and uncertain outputs. Here, we explored whether reconciled trends obtained from dynamic factor analysis (DFA) applied to conflicting indices can be used as a trend of relative abundance input into a stock assessment model. We simulated an age-structured population of two coastal shark species in the southeast United States to generate multiple disagreeing indices, reconciled the indices using DFA, and then inserted both the multiple conflicting survey indices and the simplified DFA-predicted trend into respective stock assessment models. We compared the results of each stock assessment model to simulated values to evaluate the relative performance of each approach. We found that the DFA-based assessment generally performed similarly to the conflicting index-based assessment and may be a useful assessment tool in situations where conflicting indices with different selectivities, catchabilities, variances, and missing data are present. DFA assessment results were more consistent across simulation scenarios and outperformed many conflicting index assessments when surveys underwent shifts in catchability and the underlying stock abundance exhibited contrast.