MODELING MICROBIAL ACTIVITY AND BIOMASS IN FOREST SOIL WITH SUBSTRATE QUALITY MEASURED USING NEAR-INFRARED REFLECTANCE SPECTROSCOPY

Soil respiration rates before and during glucose decomposition in samples from the A0 horizon of podzolized forest soils were monitored with computerized equipment. Near infrared reflectance (NIR) spectra of the organic matter composition of the same samples were used to model the variation in basal respiration rate, substrate induced respiration (SIR) and lag time. Basal respiration rate was determined for intact sample cores, after sieving the samples, and after adjustment of the water content to 2.5 times the organic matter content. Multivariate regression with the Partial Least Squares algorithm was used in the modelling. Before sieving, 81.8% of the variation in basal respiration could be modelled by 15 NIR wavelengths measured on freeze-dried and milled soil. After sieving, 93.0-97.7% of the variation was explained by the same 15 wavelengths, which also explained 88.4% of the variation in SIR. 93.1% of the variation in lag time could be modelled by 83 NIR wavelengths measured on soil with a water content of 2.5 times the organic matter content.