Search

Oven dried samples of leaf stem and tuber material taken from a nitrogen field experiment were analysed by Dumas combustion when fresh and by near infrared (NIR) then, and in the next two years, by a number of operators who made estimates of nitrogen concentration, [N]NIR, with differing degrees of error. The errors differed between years in the case of the one operator who made estimates in two years. Leaf, stem and tuber material of high and low nitrogen concentration were treated to produce samples at various moisture contents. These samples were scanned by NIR and the spectral data were examined. Higher moisture was found to decrease the reflectance at all the wavelengths used and would, therefore, introduce error into [N]NIRestimates. The NIR calibration used was found to be applicable to cultivars in a range of maturity classes. Several recommendations are made that will help to minimise the error introduced into [N]NIRestimates from various sources.

Several factors thought to modify the calibration between nitrogen concentration, [N]NIR and [N]Dumas, were examined including particle size using size-classes from the distribution produced by milling through a standard screen. The standard milling procedure produced differing distributions of particle size from different tissues, [N] differed between size classes and the pattern of [N] with particle size differed between tissues. In leaf and stem the smallest particles had the highest [N], in tuber material they had the lowest. Samples for analysis are generally milled in batches and analysed later. These findings show that it is important to ensure that samples are well-mixed and that they are not allowed to stratify. Only well-mixed samples should be used to fill the sample cups on a near infrared (NIR) analyser. A divergence in the relations between [N]NIR and [N]Dumas for stem samples implies that the calibration model used for that material might be particularly sensitive to particle size distribution. Within the range examined, milling speed was not an important variable in the preparation of leaf and tuber material. There is little to be gained from treating small quantities of senescent leaves separately from the remainder of a sample. However, if [N] is to be assessed in large numbers of samples of leaves that are mostly senescent then a separate calibration should be derived.