Prediction of Fatty Acid Profile of Fish Skin from the Fish Flesh Fatty Acid Composition: A Regression Modelling Approach

This study intends to investigate whether the fatty acids of fish flesh could be used to predict the fatty acids of skin using regression analysis. Since it was evident from literature that though the fat content of the skin is larger than the flesh, there has been little effort to measure the same. Fatty acids from flesh and skin of Labeo rohita, Catla catla, Cirhinus mrigala and Lates calcarifer were analyzed. Fatty acids in the flesh were correlated (α = 0.05) with those of the skin, and significant correlation was found for total fatty acid of the flesh (r(8) = 0.919, p = 0.001), C16:0 (r(8) = − 0.919, p = 0.001), saturated fatty acids (r(8) = 0.707, p = 0.05), C16:1 (r(8) = 0.718, p = 0.045), monounsaturated fatty acids (r(8) = 0.761, p = 0.028), C18:2c (r(8) = 0.081, p = 0.017) and polyunsaturated fatty acids (r(8) = 0.844, p = 0.008). Bivariate correlation test between fatty acids of fish flesh showed no correlation greater than 0.7 (r > 0.7). Twelve linear regression models were built and validated. Two among them (models 3 and 8) best explained the experimental and literature data. This study shows that a model can be built to predict fatty acid content of the skin using data from the flesh of fish. However, due to significant difference in the lipid profiles among different fish species, constructing a single model applicable to all fish species can be difficult and needs further exploration. © 2023, Zoological Society, Kolkata, India.