Detection of Illicit Conservation Treatments in Sea Bass (Dicentrarchus labrax): Application and Data Integration of NIR Spectrometers

Abstract

Global fish and seafood consumption is increasing annually, frequently leading to the emergence of food fraud, mainly related to mislabeling and adulteration like, for example, the use of illicit/unauthorized food additives to mask or delay fish spoilage. Among the available diagnostic tools for control purposes, spectroscopic techniques have often been proposed to identify these kinds of illicit practices in fish and seafood products. The presented study aims to test two cheap and portable near infrared (NIR) spectrometers, a handheld MicroNIR and a pocket-sized SCiO, to uncover use of the illicit food additive Cafodos, a mixture of sodium citrate and hydrogen peroxide used to preserve some fish characteristics (like smell, color, na dtexture). The NIR spectroscopy in combination with chemometric approaches, allowed the successfully classification of (81-100%) samples of sea bass (Dicentrarchus labrax) treated with Cafodos. The study highlights the potential of this technique that, by not requiring pre-treatment of samples with further reagents, is cheaper and safer for the environment. In conclusion, the study confirmed the potential of portable devices for rapid NIR spectroscopy analysis to identify food fraud and ensure consumer safety.

Keywords: Cafodos; PLS-DA; chemometrics; food fraud; near infrared spectroscopy; sea bass.

Figure 1

Confusion matrix obtained with model normalised, pre-treatments: average scans and SNV (standard normal variate). (a) Gill (G), muscle (M), skin (S). NT0G; NT0M; NT0S: not treated; T3G; T3M; T3S: treated and acquired post 3 h; (b) NT24G; NT24M; NT24S: not treated; T24G; T24M; T24S: treated and acquired post 24 h. The percentage of samples correctly classified by category is represented in green, the percentage of samples classified in an incorrect category is represented in pink.

Figure 2

Average spectra of a control fish and its corresponding treated fish at both short- and long-term for each matrix separately: eye (top left); gills (top right); muscle (bottom left); skin (bottom right). Blue lines represent the short-term measurements, while red lines correspond to long-term treatments. Solid lines represent controls, while dotted lines represent the treated samples.

Figure 3

Score plot of the first two (eye and gills) or three LVs (muscle and skin) calculated for the best models obtained for each sample matrix for the short-term effect: eye (top left); gills (top right); muscle (bottom left); skin (bottom right).

Figure 4

Score plot of the first two (eye) or three LVs (gills, muscle, and skin) calculated for the best models obtained for each sample matrix for the long-term effect: eye (top left); gills (top right); muscle (bottom left); skin (bottom right).

Figure 5

Plots of the coefficients for the best models calculated for each sample matrix in both short- and long-term: eye (a), gills (b), muscle (c), skin (d). The coefficients for short-term (3 h) are represented in blue; those for long-term (24 h) are represented in red. In all the cases the variables are reported on the x-axis and the coefficients are reported on the y-axis.