Résumé
Near infrared reflectance spectroscopy (NIRS) analysis was used to discriminate soils of different agroecosystems in Colombia, with different contents and qualities of organic matter, chemical, and biological properties.
Correlations were sought between absorbance in wavelength classes as determined by NIRS, and a set of variables describing soil quality grouped into three classes: (i) chemical variables (Ca, Mg, K, exchangeable Al, total P, P-Bray II); (ii) organic variables (total C, total N,
N
−
NH
4
+
,
N
−
NO
3
−
, respirometry and carbon content in different fractions separated by the LUDOX physical methods) and (iii) NIRS variables quantifying the absorptions in the near infrared region separated into 101 classes of wavelength).
For each group of variables, a principal component analysis (PCA), associated with discriminant analysis, was run. Each class of variables separated the different soil-use systems (***
P<0.001) similarly. Co-inertia analyses among the different groups of variables verified the sensitivity of the NIRS in detecting significant changes in the soil chemical and organic composition, as well as in microbial activity.
These results show the high potential of the NIRS for evaluating soil quality in large areas, rapidly, reliably and economically, thereby facilitating decision-making with respect to soil management and conservation.