Multivariate statistical analysis for the identification of potential seafood spoilage indicators

•Monitoring of seafood quality based on volatile organic compounds (VOCs).•Complexity of the VOC profile calls for multivariate statistics.•HCA, PCA and PLS were applied for seafood data.•PLS was applied for identifying several potential spoilage indicators.•Monitoring multiple instead of single VOCs could enhance quality analysis.

Volatile organic compounds (VOCs) characterize the spoilage of seafood packaged under modified atmospheres (MAs) and could thus be used for quality monitoring. However, the VOC profile typically contains numerous multicollinear compounds and depends on the product and storage conditions. Identification of potential spoilage indicators thus calls for multivariate statistics. The aim of the present study was to define suitable statistical methods for this purpose (exploratory analysis) and to consequently characterize the spoilage of brown shrimp (Crangon crangon) and Atlantic cod (Gadus morhua) stored under different conditions (selective analysis). Hierarchical cluster analysis (HCA), principal components analysis (PCA) and partial least squares regression analysis (PLS) were applied as exploratory techniques (brown shrimp, 4 °C, 50%CO2/50%N2) and PLS was further selected for spoilage marker identification. Evolution of acetic acid, 2,3-butanediol, isobutyl alcohol, 3-methyl-1-butanol, dimethyl sulfide, ethyl acetate and trimethylamine was frequently in correspondence with changes in the microbiological quality or sensory rejection. Analysis of these VOCs could thus enhance the detection of seafood spoilage and the development of intelligent packaging technologies.