A novel analytical approach in Amazon fish chain: Using infrared-spectroscopy with chemometric tools to identify non-typhoid Salmonella

The tambaqui (Colossoma macropomum) is one of the main native fish species in the Amazon region of Brazil, with economic, nutritional, cultural and environmental importance for the country. However, the presence of pathogens such as Salmonella in this matrix poses a threat to the safety and potential of this food chain. In response, this study aimed to develop a rapid, non-destructive approach to detecting Salmonella enterica serovar Schwarzengrund in tambaqui samples, using Fourier Transform Mid-Infrared Spectroscopy (FT-MIR) spectral data combined with the Data-Driven - Soft Independent Modeling of Class Analogy (DD-SIMCA) method. For this purpose, seventy-two samples of tambaqui ribs were purchased from supermarkets. Two groups of samples were used to build the model: control and contaminated with Salmonella. FT-MIR spectra were checked and four relevant regions were analyzed: all spectrum (4000–550 cm-¹), region 1 (1490–500 cm-¹), region 2 (1500–1730 cm-¹), and region 3 (2835–4000 cm-¹). The results revealed that region 1 proved to be the best for classifying contaminated samples from those not contaminated with Salmonella, with the best predictive performance with an accuracy of 94.2%. Our model exhibited the potential to be applied to the identification of Salmonella in tambaqui and to be a valuable tool for guaranteeing the safety and authenticity of fish products in the Brazilian Amazon region and, potentially, beyond. However, the use of FT-MIR combined with DD-SIMCA could be further explored in the future with a larger sample database, in order to verify the model's performance when the entire spectrum, regions 2 and 3 are used.