Sources of soil dissolved organic carbon in a mixed agricultural and forested watershed in Rwanda

Abstract

Dissolved organic carbon (DOC) plays a key role in linking terrestrial and aquatic carbon cycles. Most of the work on soil and water DOC has been conducted in temperate watersheds. There is still a gap in knowledge on DOC dynamics within the tropics. This study assesses water extractable organic carbon (WEOC) in topsoils and describes the relationship between WEOC and land use/land cover (LULC), slope position, curvature and soil properties using linear regression in the Rukarara River Watershed (RRW) in Rwanda. The study analyzes DOC concentration in soil percolation water (pDOC) and describes its relationship with antecedent precipitation index (API) and mean antecedent temperature (MAT) within the watershed using quadratic regression. Generalized linear model (GLM) and linear mixed effect model (LME) with site and/or LULC random effects are used to predict WEOC within the watershed. WEOC concentrations range from 124 to 855 mgC/L in the study area. The highest WEOC concentrations were observed in natural forest, followed by tree plantations, tea plantations and croplands. t-test results did not reveal a significant difference between concentrations of WEOC in valleys, upper slopes, ridges, flat, concave and convex areas in the Rukarara River Watershed (RRW). Considering the relationship between WEOC and soil properties, significant positive correlation coefficients were 0.60, 0.53, 0.50, and 0.36 respectively for the total organic carbon (TOC), the total nitrogen (TN), the cation exchange capacity (CEC), and the aluminum (Al). The best predictor WEOC as a function of soil properties was the generalized linear model (GLM) and indicated soil TOC as the overarching soil factor of WEOC in the RRW by 71%. The pDOC concentration ranges between 0.34 and 10.03 mgC/L and its relationship with both API and MAT was concave upward. APIs explained 12 to 17% of the pDOC variation in the RRW whereas MATs explained 8 to 25%, the natural forest site showing the highest values and the cropland site the lowest values. This result means that a conversion from forest to cropland within the RRW could decrease DOC in both soils and in percolation water. Also, an increase of precipitation and temperature up to respective optima in the RRW, could increase DOC in percolation water and consequently in streams. This increase of instream DOC can impact the water quality of the Rukarara River and its streams, with implications for their ecological function. Strategies of land management and water resources should be enhanced to preserve soil and water quality in the RRW.