Origin traceability of bottled mineral water imported into China using chemical and stable isotope fingerprints

Abstract

A comprehensive study of 96 bottled mineral water (BMW) samples consisting of 21 of the most popular origins and brands sourced from China (n = 38), France (n = 20), Italy (n = 10), New Zealand (n = 17) and Fiji (n = 12) has been undertaken. The δ18O and δ13C of dissolved inorganic carbon (DIC), δ18O and δ2H of water, mineral elements (Li, Na, Mg, K, Ca, Sr) and ions (Cl−, SO42−, F−, CO32−, HCO3−) were investigated and used for origin traceability. The origin of BMW was confirmed by applying Fisher discriminant analysis (Fisher - DA) and artificial neural network (ANN) models to these variables. δ18ODIC, δ13CDIC, δ18O, δ2H values of BMW showed significant deviations with origin. Mineral elements and anions varied widely due to different regional geology and hydrology characteristics. A discriminant model based on Fisher-DA showed that the identification accuracy of mineral water from different origins was in the range of 81 % to 100 %, whereas the discriminant accuracy was 100 % for the training samples and 96.0 % for the testing samples calculated by ANN. HCO3−, NO3−, δ2H, δ18O and F− were the most important variables that characterized BMWs from different origins. Using a combination of stable isotopes and different water chemical components, BMWs from China can be accurately distinguished from other origins using multivariate statistical models.