Laboratory methods of estimating potentially mineralizable nitrogen in organic potting mixes. II. Development of near infrared reflectance spectroscopy method

Abstract

Near infrared reflectance spectroscopy (NIRS) was tested as a method of predicting potentially mineralizable N (PMN) in organic potting mixes. Initially, calibration models were developed using the near infrared spectra of 100 potting mix formulations and the amount of N recovered by ryegrass herbage after 168 d of growth in the formulations, focusing on which combination of a range of regression, derivatization and scatter correction techniques gave the best calibration model. In the second part of the study, a validation exercise was performed, where the 100 formulations were split into separate calibration and validation sets. In the initial calibration exercise, the most effective combination of regression technique and spectral pretreatments was modified partial least squares in conjunction with a second order derivative and a standard normal variate and detrend scatter correction technique, which had an R2 of 0.99. However, in the validation study the coefficient of determination was much lower (r2=0.73), and results were not in line with expectations. The explanation offered for the poor validation results relates to the constriction of the set of calibration samples. Nevertheless, NIRS appears to have great potential for predicting the performance of organic potting mixes, and despite several practical drawbacks is worthy of investigation by other workers.