Abstract. Addition of biochar to soils has been shown to increase crop yield and aid in mitigating greenhouse gas emissions by decreasing the extent of soil methane (CH4) flux. Previous studies utilizing metaanalysis to better understand the impact of environmental and management factors on CH4 flux from biochar treated soil systems have provided contrasting results, ranging from significant increase, decrease, to no change in methane flux after amendment. We hypothesized that these discrepancies could be explained by separating studies into two major land use categories, upland and paddy, prior to analysis so that the overall redox conditions are more comparable across studies upon which statistical comparisons are made. Furthermore, past studies did not consider potentially critical soil properties including soil organic carbon, total nitrogen, C/N, and soil texture; a number of biochar properties including biochar pH and C/N; and five additional management and experimental factors. In this study, Hedge's d metric was calculated and Wilcoxon analyses were used in a meta-analysis to determine the impact of these additional factors on methane flux from biochar-amended upland versus paddy soils. We demonstrate that variations in soil characteristics including SOC, C/N, and pH significantly influences the methane flux from biochar treated soils, while biochar characteristics and management practices have less to no effect as determined by the magnitude of the Hedge's d metric. Soils with low SOC, total nitrogen, C/N, acidic or alkaline pH exhibited lowest CH4 emission rates/highest CH4 uptake rates, whereas soils with higher SOC content, C/N, and circumneutral pH exhibited higher CH4 emission with biochar addition. Several possible mechanisms are suggested to explain the role of these variables in CH4 cycling. Results from this study will be used to evaluate the input parameters for building a linear additive model to quantitatively predict soil methane flux in response to biochar additions. Ultimately, implementation of the linear additive model can be extremely valuable for advising agricultural practices toward minimize methane emissions or maximizing methane sink strength. We suggest that additional field and controlled experiments be performed to better define the reaction network that controls methane flux from biochar treated soils, with particular attention to paddy soils where studies are still lacking.
Variable Selection for Partially Linear Additive Model Based on Modal Regression Under High Dimensional Data
