Evaluation of the Near Infrared Spectroscopy (NIRS) to predict chemical composition in Ulva ohnoi

Abstract

Green algae of the genus Ulva have been identified as suitable organisms for biomass production and good candidates for the development of seaweed blue-biotech industries. The fluctuation of chemical composition during the growth of the algae, which depends largely on environmental factors, makes the development of rapid phenotyping protocols necessary. In this work the efficacy of Near Infrared Spectroscopy (NIRS) to predict dry matter, mineral fraction, nitrogen, and carbon contents directly from wet untreated samples and from dried samples was studied. Partial least square (PLS) models from spectra recorded on 80 samples were used to predict dry matter, and 44 samples to predict carbon, nitrogen and mineral fraction on a wet and dry weight basis. NIR models developed from spectra acquired on wet samples had good accuracy (R2>0.9) for the prediction of N (on a ww and dw basis) and C (on a ww basis). Models with lower R2 scores have been obtained for dry matter (R2=0.610) and MF (R2=0.506-0.693). The models developed to predict carbon and nitrogen directly on wet and untreated samples present NIRS as a valuable tool to determine these parameters in a rapid and low-cost way, allowing making decisions about the optimal harvesting time.