Prediction of the degradability and ash content of wheat straw from different cultivars using near infrared spectroscopy

Abstract Degradability of straw is important in connection with the fermentation process for bioethanol production, while the ash content is important for its suitability for incineration. Therefore, a fast method for assessment of straw quality could be very useful in determining the price and in helping choose between different applications for specific straw batches, such as fermentation for ethanol production, incineration or animal feed. This study investigated the ability of near infrared (NIR) spectroscopy to predict the degradability and ash content of 106 cultivars of wheat straw grown at two different sites. In general, calibrations based on NIR spectra recorded on air-dried samples performed better than those on oven-dried samples. A partial least squares (PLS) calibration based on the spectra of the air-dried samples predicted degradability with r2 = 0.72 and RMSECV = 1.4% with 3 components using samples from the two different sites. The ash content was well predicted with r2 = 0.99 and RMSECV = 0.195% using a complex 15-component PLS model validated using repeated random segmented cross-validation. However, this model proved to be sensitive to site in a validation using the two sites as segments, where the accuracy of ash content prediction decreased to r2 = 0.91 and RMSECV = 0.691% using a 9-component PLS model. NIR spectroscopy proved useful for predicting degradability and ash content of wheat straw from different wheat cultivars. However, when developing predictive models of ash content based on NIR spectra, it should be ensured that the models are transferable to locations other than those used for model calibration.