Predictive modeling of European flat oyster (Ostrea edulis L.) fatty acid composition

Abstract

This study investigates polyunsaturated fatty acids (PUFAs) of European flat oyster (Ostrea edulis L.) and their composition shift in relation to the interaction of selected factors in the marine environment of the Adriatic Sea, over an annual farming cycle. We advance the hypothesis that fatty acid composition may be predicted regarding the changes of abiotic environment parameters. Novel approach was developed as a tool for interpretation of PUFA ratio using mathematical models and algorithms, providing association and better prediction of observed variables compared to standard statistical techniques. Data analysis using R language package resulted in models that primarily provide the possibility to predict the PUFA value based on certain fluctuations in the environment. In addition to predictions, results present a useful tool for insight into the oysters’ quality and understanding of their basic biology in different environments. This is the first report of using mathematical modeling to investigate interactions between PUFA compositions of oyster meat relative to interactions of particular environmental variables. We furthermore discuss the potential of environmental variables in affecting the oyster PUFA composition and apply an approach providing a better insight into the health status and quality of oysters throughout the season. Model projection showed a nonlinear, parabolic dependence between the tested variables and dissolved oxygen. Both, the n-3 PUFA and DHA alone, were associated with an open-down parabolic relationship; the maximum levels were reached by 95 % oxygen saturation. The n-3 PUFAs and n-3/n-6 ratio were likely negatively affected by increase in sea water temperature, where a 15 % reduction in n-3 PUFAs content and 50 % reduction in n-3/n-6 proportion were predicted. Also, we detected FA ratios (AA/DHA, AA/EPA) and AA positively affected by increase in sulphate concentration and dissolved oxygen, respectively, and strongly negative prediction of ammonia level and sea water pH with AA, DHA and ARA/EPA ratio, respectively. These models might be applied to other aquatic species, for promotion of sea food quality, improvement of farming conditions and production, as well as for optimizing conditions in controlled trails.