Addressing Spatiotemporal Data Gaps in Fish Abundance Modelling: Insights From Offshore Wind Impacts in the U.S. Mid‐Atlantic

ABSTRACT Aim Spatiotemporal data gaps in fishery‐independent surveys—arising from offshore wind development, marine protected areas or spatially uneven sampling effort—pose challenges to the consistency and reliability of abundance indices that inform stock assessments. This study evaluates how survey preclusion affects the spatial and temporal behaviour of abundance indices, using offshore wind survey exclusion in key NOAA fisheries surveys in the Mid‐Atlantic as a case study to assess impacts on temporal trends, deviations over space and uncertainty for four iconic stocks. Location Mid‐Atlantic Ocean, US. Methods Using four species as case studies, we compared design‐based abundance indices derived from a Full dataset (without offshore wind) and a wind energy excluded (WEE) dataset that hypothetically removed all survey tows within wind energy areas (WEAs) across the full time series. Model‐based approaches, including statistical models, machine learning methods and vector autoregressive spatial–temporal model (VAST), were then applied to reconstruct tow‐level survey data using the WEE dataset and derive spatial and temporal abundance indices. Model‐based indices were evaluated in terms of deviations in spatial abundance indices, changes in relative precision, and consistency of temporal trends relative to design‐based indices. Results Design‐based indices retained consistent temporal trends despite emerging errors in absolute abundance estimates and increased variance. Model‐based reconstruction partially recovered spatial information within WEAs, but reconstructed spatial abundance indices exhibited greater variability in deviations inside WEAs than outside, whereas changes in relative precision were generally small. Temporal trends in model‐based indices were generally consistent with design‐based indices, though sensitivity varied by species and time series. Differences among modelling approaches were species‐dependent, reflecting contrasts in mobility, spatial aggregation and life history traits. Main Conclusions This offshore wind case study demonstrates that survey preclusion primarily affects the spatial representation and magnitude of abundance indices, whereas temporal trend signals can remain robust under moderate data loss. No single modelling approach was consistently robust to survey preclusion across species and index dimensions, reflecting trade‐offs between spatial reconstruction behaviour, uncertainty propagation and temporal stability. Species life history traits played a role in model performance: mobile species with stronger spatial autocorrelation were generally less sensitive to spatial preclusion than sessile species with patchy distribution. These general patterns highlight the need for dimension‐specific, life‐history‐informed modelling approaches when addressing spatiotemporal data gaps arising from marine development and other forms of conflicting marine use.