No abstract

To be able to monitor in real time the concentrations of DL-lactic acid and acetic acid present in Kefir, a self-carbonated fermented milk product, near infrared (NIR) calibration models were constructed based on NIR spectra of 174 samples. Enzymatic tests and gas liquid chromatography were the reference methods used for DL-lactic acid and acetic acid, respectively. The fit of the models and their prediction power were evaluated using segmented cross-validation and an external validation set. The models obtained for DL-lactic acid were found to be acceptable for both cross-validation ( R2=0.90, SECV=0.110 g 100 mL−1 and RPD=3.16) and external validation ( r2=0.87, SEP=0.156 g 100 mL−1 and RPD=2.57). In contrast, the models for acetic acid were found to be unacceptable. The results obtained for both cross-validation ( R2=0.80, SECV=0.013g 100 mL−1 and RPD=2.21) and external validation ( r2=0.44; SEP=0.017 g 100 mL−1 and RPD=1.17) suggested this model requires further development. The application of principal component analysis (PCA) to the entire sample set showed Kefir prepared with laboratory Kefir grains (LG), traditional Kefir grains (TG) and mass cultured Kefir grains (MG) resulted in similar PCA score values in spite of the Kefir grains not having the same origin. PCA was also able to differentiate between fermented and “milky body-like” samples. The findings of this study could serve the dairy industry in monitoring more efficiently the acidity in terms of DL-lactic acid of fermented dairy products.

Keywords: Kefir, consumer preference, acidity, flavour compounds, diacetyl to acetaldehyde ratio