Comparison between a new application of multivariate regression and current spectroscopy methods for the determination of chlorophylls and their corresponding pheopigments

A multivariate calibration model, partial least squares regression for seve ral independent variables (PLS2), has been developed for the simultaneous d etermination of chlorophyll a, b and c and the corresponding pheophytin a a nd b and pheoporphyrin c. Mixtures with these six components were prepared at four concentration levels with respect to each analyte, in acetone:water . The concentration for chlorophyll ct varied from 1 to 7 mg l(-1). Absorba nce spectra in the visible range, 350-770 nm, were registered. The predicti ve ability, reported as the root mean square error of prediction (RMSEP), w as found to be 0.167 mg l(-1) for chlorophyll a. Similar results were found for the other analytes. The spectral data were also evaluated with the abs orption spectroscopy procedures recommended in Standard Methods for the Exa mination of Water and Wastewater. For chlorophyll a, the proposed multivari ate method improved the predictive ability by a factor 2.4. Data is availab le at: http://www.anchem.su.se (C) 2000 Elsevier Science B.V. All rights re served.