Développement et validation d’un modèle statistique pour la détection de l’odeur de verrat dans la graisse porcine

Boar taint is a major issue in pork production, affecting meat quality and consumer acceptance. It is primarily caused by the accumulation of skatole and androstenone in the fat of intact male pigs. Currently, detection relies on human sensory evaluation, which is rapid but subjective. This study aims to identify key volatile markers and develop a reliable statistical model for boar taint prediction using advanced chemical analysis and statistical modeling.

A non-targeted volatilomic analysis was performed on 129 pork fat samples using HS-SPME-GC×GC-TOFMS. Samples were classified as odorous or non-odorous based on sensory evaluation and UHPLC-MS/MS quantification of skatole and androstenone. A Random Forest classification model was trained on 70 samples, identifying 13 key volatile markers, including androstenone, skatole, indole, and additional compounds such as 2-aminoacetophenone and benzyl alcohol, indicating a more complex odor profile than previously thought. The model was then applied to 59 remaining samples, classifying 7 as odorous.

This approach enhances the objectivity and reliability of boar taint detection, reducing the variability of traditional methods. It also offers potential for slaughterhouse quality control and genetic selection strategies to naturally reduce boar taint, supporting animal welfare by limiting castration.