Rapid compositional analysis of oysters using visible-near infrared reflectance spectroscopy

Visible-near infrared reflectance spectroscopy (VNIRS) was applied to the quantitative and qualitative analysis of oyster samples. The meat of 183 oysters (Crassostrea gigas and Saccostrea glomerata) were individually homogenised, scanned by VNIRS, subsamples chemically analysed, and calibration models developed to allow VNIRS-prediction. Comparison of predicted to actual (chemically measured) data showed good correlations and low prediction errors: moisture, r2 = 0.92, standard error of prediction (SEP) = 0.53% wet weight (WW); protein r2 = 0.97, SEP = 0.18% WW; fat r2 = 0.97, SEP = 0.11% WW, and glycogen r2 = 0.94, SEP = 0.24% WW. These metrics indicate that the models are sufficiently accurate and reliable for quantitative applications. The key advantage of the methodology is its high throughput, i.e., 250–300 samples can be simultaneously analysed for moisture, fat, protein and glycogen each day. Therefore its use could be directed to applications requiring the rapid analysis of many individuals, e.g., in selective breeding programs where compositional data can provide valuable information on traits associated with animal condition or quality.Three batches of oysters (n = 40–43/batch) were also subjected to discriminant analysis. Based on their respective VNIRS spectra, models were developed that successfully predicted the batch identity of the individual oysters on >95% of occasions. This provided a proof-of-concept that VNIRS may also have application in the qualitative analysis of oysters.