Advancing fish early life-history research: daily age determination of walleye pollock (Gadus chalcogrammus) with Fourier transform–near-infrared spectroscopy

Context Fish daily age information provides insight into growth, phenology and recruitment but is rarely incorporated into the decision-making process for management because of labor-intensive data collection. Aims We tested Fourier transform–near-infrared (FT-NIR) spectroscopy as a rapid method (~1.0 v. ~60 min per otolith for microscopic methods) to determine daily age of walleye pollock (pollock; Gadus chalcogrammus), which supports an important fishery. Methods We reared two annual cohorts of pollock over 3 years to accrue a comprehensive dataset for calibration models. Key results FT-NIR spectroscopy provided an improvement over predicting daily age from otolith weight, fish weight or fish length. A calibration model developed using microscopically aged fish had a root mean square error (RMSE) from cross validation (CV) of 12.2 days. Models developed from the full dataset with ages assigned from cohort hatch dates had a test dataset RMSE of 35.4–40.7 days and a CV RMSE of 29.4–36.9 days. Conclusions Spectra were affected by size-at-age, which could affect model efficacy because growth and size are influenced by environmental variability. This can be addressed with robust calibration datasets, model testing and model updating. Implications FT-NIR spectroscopy age prediction is likely to be sufficient to capture population-scale shifts in hatch dates of pollock.