Modelling of total soil carbon using readily available soil variables in temperate forest of Eastern Himalaya, Northeast India

Abstract

The study was carried out in temperate forest of Ziro valley, Arunachal Pradesh, Northeast India, during 2015–2016. Stratified random sampling was adopted for soil sample collection from three depths on a monthly basis in each of the permanent plots. Collected samples were analysed using standard methodologies. The XLSTAT (ver.2019) was used for the partial least square regression and modelling. The soil was acidic in nature having moderate soil moisture with very low temperature. Soil texture varies from sandy loam to sandy clay loam in nature. Altogether, 16 readily available soil variables were used for identifying significant variables to be used for modelling the total soil carbon (TSC). Based on the results, five most contributing independent soil variables having high variable importance in prediction were used for modelling the TSC. The average annual TSC was recorded 5.33% for upper, 5.01% for middle, and 4.20% for lower soil surface, respectively. The developed depth-wise equations predict very close TSC with very low root-mean-square error (rmse) to observed values. Hence, the findings of the present study will be very much useful under limited data conditions to predict the TSC and also in inaccessible areas of temperate forest ecosystem.

Keywords

• microbial biomass carbon
• pslr
• soil properties
• vips
• xlstat