Predicting Fish Age at the Speed of Light

Abstract

Estimating the age structure of fish populations is an important component of sustainable fisheries management. Traditional methods for ageing fish are time-consuming, labor-intensive, and expensive. Previous Australian research has shown that near-infrared spectroscopy (NIRS) has the potential to estimate the age of saddletail snapper (Lutjanus malabricus) (Wedding et al. 2014), deepwater sharks (Rigby et al. 2015, Rigby et al. 2014), barramundi (Lates calcarifer) and pink snapper (Pagrus auratus) (Robins et al. 2015). This current study investigated the effect of seasonal variability for barramundi collected from the Gulf of Carpentaria over four years, 2012-2015. Partial least squares regression (PLS-R) calibration models were developed using samples from successive years with a final model incorporating samples from all four years. The prediction statistics improved as more seasonal variability was introduced into the calibration set. Prediction statistics incorporating seasonal variability from four years with three latent variables, were r2 = 0.81, RMSEP = 7.88 and SDR = 2.3. These results indicate the potential of NIRS to predict the age of barramundi otoliths and the importance of incorporating seasonal variation into a calibration model.