Detection of lard in fish feed and optimizing the processing conditions for pelleted fish feed production using chemometric and response surface methodology

In order to save the cost of fish production, some small-scale farmers prefer to produce fish feed using inexpensive pelleting methods and ingredients. However, due to a lack of knowledge and expertise, the processing conditions of fish feed production are often neglected, and the basic ingredient such as oil is deliberately replaced with cheaper material such as lard. From a religious perspective, lard in any product is strictly prohibited, especially for the Muslim community. This research thus intends to optimize the processing conditions of the pelleting method using response surface methodology (RSM) and also to detect lard as an adulterant in fish feed using fatty acid (FA) and triacylglycerol (TAG)-based analyses separately, which are then integrated with chemometrics. The optimum processing conditions were determined based on the desired properties of bulk density, floatability, water absorption and water solubility indices of the fish feed produced. For lard detection, sn-2 FA composition was coupled with total FA composition to enhance the FA-based analysis by gas chromatographymass spectrometry (GC-MS). Meanwhile, to improve the TAG-based analysis, thermal analysis by differential scanning calorimetry (DSC) was combined with TAG composition analysis by high-performance liquid chromatography (HPLC). These two types of analysis (FA and TAG-based) were then incorporated with chemometric techniques, namely principal component analysis (PCA), orthogonal partial least squarediscriminant analysis (OPLS-DA), and orthogonal partial least square-regression (OPLS-R). Results of RSM revealed that the combination of water temperature at 70 ºC, mixing speed at 75 rpm, and mixing time of 5 min were the optimal processing conditions in producing pelleted fish feed with the best quality. For lard detection, the enhanced- PCA model exhibited a preferable distribution pattern of fish feed by differentiating them into separate clusters without overlap. The OPLS-based prediction models constructed from the most discriminating variables pre-defined by the enhanced-PCA also showed good performances. The OPLS-DA model provided significant class discrimination between lard adulterated and non-adulterated fish feed; meanwhile, the OPLS-R model revealed a stronger relationship between the observed and predicted value of lard percentage with the increasing value of the coefficient of determination (R2). The optimized processing conditions determined by RSM showed it was possible to produce relatively good quality pelleted fish feed comparable to extruded fish feed. In addition, the involvement of complementary analysis with chemometrics indicated that lard in fish feed could be detected and quantified more effectively, and its relevant biomarkers could also be discovered.

Text MOHAMED HANIFF HANAFY - IR.pdf Download (5MB)

Actions (login required)

View Item