Context Fourier transform–near-infrared spectroscopy (FT-NIRS) has been proposed as a chemometric method to rapidly age fish. Whether FT-NIRS can be effective in ageing small pelagic species (SPS) has not been fully evaluated. Aims To evaluate the accuracy and precision of FT-NIRS ages estimated for two SPS; and identify factors that influenced SPS FT-NIRS model performance. Methods Spectra acquired from Pacific sardine (N = 2373) and Pacific mackerel (N = 760) otoliths in 2004–2021 were analyzed using partial-least-squares regression, artificial neural network and k-nearest neighbor classification models, and multidimensional scaling analysis. SPS growth patterns were compared with those of slower-growing species with accurate FT-NIRS models. Key results Across a maximum of 10 age classes, SPS FT-NIRS models performed poorly, predicting only Age-0 fish and classifying the youngest or the oldest fish with ≥80% correct classification rates. SPS otolith growth reached a plateau after 3 years, whereas otoliths of the other species grew indeterminately. Conclusions FT-NIRS age prediction performance may depend on SPS ontogenetic growth and habitat characteristics. However, effects of these factors were masked among older age classes because of repeated seasonal feeding and spawning migrations. Implications Population growth and movement dynamics should be evaluated before devoting substantial resources in developing FT-NIRS age prediction models.
Dorval et al. (Thu,) studied this question.