The application of multi-elemental fingerprints and chemometrics for discriminating between cage and free-range table eggs based on atomic absorption spectrometry (AAS) and colorimetry.

Abstract

Mislabeling is a common fraudulent activity in food marketing as producers take advantage of rising demand for ethically produced, high quality animal products such as free-range table eggs. Detection and prevention of this commercial fraud requires robust and widely available tools that can accurately distinguish table eggs from a variety of sources. In this study, the efficacy of multi-elemental fingerprints to discriminate between cage and free-range table whole eggs was assessed using chemometrics. The elemental concentrations of N, P, K, Ca, Mg, Na, Zn, Cu, Fe, and B in cage and free-range table eggs consisting of 99 specimens, with an 80%:20% split between the calibration and verification sets (83 and 16 specimen, respectively) were determined using Flame-Atomic Absorption spectrometry (AAS) and colorimetry. Principal Component Analysis (PCA) for fingerprint determination was applied in combination with Bayesian Machine Learning (PCA-BML), Support Vector Machine (PCA-SVM), and K-Means Clustering (PCA/K-Means). The classification verification set specimens were identified with accuracy and F1-scores ranging from 81.3- 100.0% and 80–100% respectively. PCA/K-Means was the most effective classification model with sensitivity, precision/specificity, accuracy, and FI-score values of 100% while the false positivity rates (FPR) was 0%. The results demonstrated that AAS and colorimetry derived multi-elemental fingerprints and chemometrics were an effective and feasible tool to discriminate between cage and free-range table eggs. Therefore, AAS and colorimetry multi-elemental fingerprints combined with chemometrics can be used to reduce fraudulent marketing practices and improve quality control in the egg industry due to their wide availability, versality, robustness.