Analytical strategies based on near infrared spectroscopy and multivariate calibration for rapid quantification of florfenicol at low-concentrations in medicated-feed pellets

Abstract

Florfenicol (FF) is an antibiotic that is widely used in the salmon industry to decrease mortality caused by several diseases. Aquiculture salmon cages require rapid techniques to quantify this antibiotic in the pellets before liberation in water. In this study, multivariate calibration models using near-infrared spectroscopy (NIRS) and partial least squares (PLS) were developed to monitor the dose of FF in supplied medicated feed pellets, where the concentration of FF is much lower compared to other components of the complex matrix (˂1% w/w). Different data pre-treatments were performed to minimize the errors of FF prediction in commercial pellets obtained by surface coating method. The increasing of the number of samples and the FF concentration range in calibration set, improved significantly the prediction of FF in the commercial pellets, even if only physical mixtures of FF and the pellet matrix were used in the calibration set. The results show that NIRS is an excellent alternative to more laborious analytical methodologies used to quantify FF at low concentrations in the pellets. Good predictions were obtained even in the presence of additional interferences such as the complexes used in some pellets to increase FF solubility.