Environmental factors and spatial dependence explain half of the inherent variation in carbon pools of tropical paddy soils

Abstract

The current study examined macro-environmental (climate and topography), micro-environmental (soil chemical properties) drivers, and spatially derived parameters that are significantly associated with soil carbon pools (0–15 cm) across tropical paddy-growing areas in Sri Lanka using the data from 987 sampling sites across the country. Redundancy analysis was performed to identify the relationships between the explanatory variables and the variation in different soil carbon pools i.e., total carbon (TC), Microbial Biomass Carbon (MBC), Permanganate Oxidizable Carbon (POXC), and Dissolved Organic Carbon (DOC). The spatial patterns in soil carbon pools were evaluated using Moran's eigenvector maps. Results indicated that macro, micro-environmental drivers and spatial variables explained 47% of the inherent variation of the TC, MBC, POXC and DOC. Micro-environmental drivers had a larger unique fraction relative to macro-environmental drivers (4% and 1% of the total variation, respectively). Most of the variation explained by macro-environmental drivers was shared by micro-environmental drivers (11% out of 15%). Among macro-environmental drivers, rainfall and enhanced vegetation index were more strongly related to the soil carbon pools compared to the topography-related factors. In terms of micro-environmental drivers, total N, available K, Ca, and soil pH (H2O) were the best explanatory variables of soil carbon pools. Spatial patterns in soil carbon pools were largely induced by the environmental predictors that are spatially structured. Our findings provide insights into improving the reliability of spatial estimation of the soil carbon by incorporating important soil carbon preditors and quantifying the impacts of environmental changes on soil carbon pools.