Prediction of the quality of forage maize by near-infrared reflectance spectroscopy

The aim of this work was to assess the ability of near-infrared reflectance spectroscopy (NIRS) to be a rapid and reliable method for estimating quality variables for forage maize with an extremely wide range of maturity and growth stage, as it occurs, e.g. in experiments dealing with growth analysis. For maximum variability samples for analysis were collected over 4 years from field trials which comprised a wide range of nitrogen (N) fertiliser (0–150 kg N ha−1) and slurry application (0, 20 and 40 m3 ha−1) treatments, but one maize variety named Naxos. Phenological variability was obtained by sampling in intervals of 14 days during growth of maize. A total of 6296 samples was collected and 398 samples were selected to calibrate estimated net energy and crude protein contents and validate the equations derived. To increase accuracy separate calibrations were developed for reproductive and vegetative fractions of maize. For both groups NIRS equations were scrutinized by a set of validation samples which were selected from the same sample pool before the samples for calibration were chosen. Predictions of crude protein and net energy contents over a wide range of plant maturity showed satisfactory accuracy. Standard error of prediction was below 0.3 MJ NEL kg DM−1 for energy value and below 123 g kg−1 of dry matter (DM) for crude protein. In the course of 4 years the number of necessary samples to be included in the calibration decreased to between 2.6 and 15% of sample number in the first year and the validity of the equation increased.