Dummy regression analysis for modelling the nutritionally tailored fillet fatty acid composition of turbot and sole using gilthead sea bream as a reference subgroup category

• Published in: Aquaculture Nutrition
• Main Authors: Ballester-Lozano, Gabriel F., Benedito-Palos, Laura, Riaza, Ana, Navarro, Juan Carlos, Rosel, Jesús F., Pérez-Sánchez, Jaume
• Other Authors: Ministerio de Educación y Ciencia (España), Generalitat Valenciana, Diputación de Castellón, European Commission
• Format: Article in Journal/Newspaper
• Language: unknown
• Published: John Wiley & Sons 2014
• Subjects: Turbot
• Online Access: http://hdl.handle.net/10261/144107; https://doi.org/10.1111/anu.12095; https://doi.org/10.13039/501100003359; https://doi.org/10.13039/501100000780

Farmed turbot and sole were sampled at different stages of the production cycle for analysis of fillet lipid content and fatty acid (FA) composition. The entire data set along with our own published data on gilthead sea bream were fitted to dummy regression equations with turbot and sole as dummy variables, gilthead sea bream as a reference subgroup category, and diet FA composition and fillet lipid content as independent variables. The relative contribution of each independent variable to the total variance was found to vary within and among FAs and fish species, but strong correlation coefficients (0.76 < r2 > 0.99) were found for almost all of the FA equations, including saturated FAs, monoenes and long-chain polyunsaturated fatty acids (PUFA) of n-3 and n-6 series. Given the differences in lipogenic activities of the fish species, major interaction effects between fillet lipid content and dummy variables were found for monoenes and saturated FAs. The proposed equations (hosted at www.nutrigroup-iats.org/aquafat) were able to fit different proportions of EPA, DPA and DHA underlying the fish species differences in FA desaturation/elongation pathways. The robustness of the model was proven with extra data from the three fish species, allowing a close linear association near to equality for the scatter plot of observed and predicted values. © 2014 John Wiley & Sons Ltd. This study was funded by Spanish (AQUAFAT, AGL2009-07797, Predictive modelling of flesh fatty acid composition in cultured fish species with different muscle lipid content; AQUAGENOMICS, CSD2007-00002, Improvement of aquaculture production by the use of biotechnological tools) and EU (ARRAINA, KBBE-2011-5-288925, Advanced research initiatives for nutrition and aquaculture) projects. Additional funding was obtained from the ‘Generalitat Valenciana’ (research grant PROMETEO 2010/006). GFB-L was recipient of a Spanish PhD fellowship from the Diputación Provincial de Castellón. Peer Reviewed