Use of near-infrared reflectance spectroscopy in predicting nitrogen, phosphorus and calcium contents in heterogeneous woody plant species

摘要

Near-infrared reflectance spectroscopy was applied to determine nitrogen (N), phosphorus (P) and calcium (Ca) content in leaf samples of 18 woody species. A total of 183 samples from mountain, riparian and dry areas from the Central-Western Iberian Peninsula were collected for this purpose. The wide intervals of variation observed in nutrient concentrations (6.6-45.0 g kg(-1) for N, 0.24-2.97 g kg(-1) for P, and 1.00-20.06 g kg(-1) for Ca) were due to the great heterogeneity of the samples. To develop calibration equations, multiple linear regression, and partial least-squares regression (PLSR) were used. In both cases, three mathematical transformations of the data were applied: log1/ R and first and second derivatives. The best calibration statistics were obtained using PLSR and derivative transformations ( second derivative for N and first derivative for P and Ca). The following coefficients of multiple determination ( R 2) and standard errors of cross validation were obtained: 0.99 and 0.93 for N, 0.94 and 0.15 for P, and 0.95 and 0.88 for Ca. In the external validation the standard errors of prediction obtained were 0.76 ( N), 0.11 ( P) and 0.60 (Ca).