Exploring the limits of spatiotemporal and design-based index standardization under reduced survey coverage

• Published in: ICES journal of marine science Vol. 80; no. 9; pp. 2368 - 2379
• Main Authors: Yalcin, Semra, Anderson, Sean C, Regular, Paul M, English, Philina A
• Format: Journal Article
• Language: English
• Published: Oxford University Press 28.11.2023
• Subjects: random fields; geostatistical modelling; abundance index; design-based estimators; model-based estimators; spatiotemporal modelling
• ISSN: 1054-3139; 1095-9289
• DOI: 10.1093/icesjms/fsad155

Abstract Indices of abundance derived from fisheries-independent surveys play a crucial role in sustainable fisheries management. While design-based methods provide unbiased indices in theory, logistical constraints may introduce biases in practice. Spatiotemporal models offer potential for mitigating such bias, but their limitations remain poorly understood. In this study, we compare the performance of model-based and design-based indices using 200 simulated populations representing two virtual species. We simulate stratified-random surveys under various scenarios, including constant effort and coverage, reduced set density, and reduced spatial coverage (e.g. closed areas). We consider three closed-area scenarios: no population recovery, population recovery, and population recovery with spillover. With consistent survey coverage, correctly specified spatiotemporal models demonstrated comparable bias, accuracy, and confidence interval coverage to design-based methods. Spatiotemporal models incorporating appropriate covariates and observation families could mitigate the impact of reduced spatial coverage. However, poorly specified models were sometimes outperformed by design-based methods. Our results, therefore, highlight the potential for spatiotemporal models to mitigate the effects of survey effort reduction on population assessment and the provision of scientific advice. However, they also present a cautionary tale about the critical importance of model evaluation and comparison.