Belowground spectroscopy — Novel spectral approach for estimation of vertical and species-specific distributions of forest soil characteristics and heterotrophic respiration

Abstract

Spatial variations in soil moisture and characteristics (nutrients and organic components) are important for understanding the heterogeneity of soil heterotrophic respiration (Rh) in forest ecosystems. In this study, "belowground spectroscopy", a new method for estimating soil characteristics and Rh in the field, was developed by measuring the spectral reflectance in the short-wave infrared (SWIR) spectral region. In the cool temperate forest of Hokkaido, 184 training samples were collected from litter, organic, and mineral soil layers on the forest floors of 13 tree species. First, a multiple regression model was developed for Rh estimation using the soil characteristics as substrates. Next, the partial least squares regression (PLSR) model was calibrated to estimate soil characteristics using spectral reflectance observed under various moisture conditions, and then spectroscopy-based Rh estimation model was constructed by combining the two models. In the multiple regression model, soil water content and concentrations of total nitrogen and cellulose were selected as explanatory variables of Rh. The PLSR model for these parameters showed estimation accuracies of 23.1–31.8% of relative root mean square error (rRMSE) in the validation analysis. The combined model showed 68.4% of rRMSE in Rh estimation. Furthermore, this method was applied to field observations using a small spectrometer to estimate vertical and species-specific variations in Rh. The estimated vertical profiles of soil characteristics showed specific variations related to vertical changes in the soil layer. The accumulated value of the estimated vertical Rh was significantly correlated with the Rh measured using a gas analyzer on 12 tree species, except Alnus species. The rRMSE of the modeled Rh compared to that of gas analyzer was 29.8%. These results indicate that belowground spectroscopy calibrated using a dataset under varied moisture conditions is effective for developing new, rapid, and non-destructive monitoring of soil characteristics and Rh in the field.