ORIGINAL ABSTRACT

Glyphosate (GLY) is the most widely used herbicide worldwide, yet its effects on microbially mediated soil carbon-cycling processes remain controversial. Using random-effects meta-analysis, machine-learning interpretation, and path modeling, this study integrates 1099 paired observations from 103 studies worldwide to systematically evaluate how GLY exposure conditions and initial soil physicochemical properties jointly modulate soil carbon cycling responses. Across all observations included in the meta-analysis, GLY increased soil CO2 emission by 16.32% and soil organic carbon (SOC) by 9.98% on average. However, both the direction and magnitude of these responses were strongly context-dependent and could reverse under specific combinations of soil properties and exposure conditions. High GLY concentrations (>100 mg a.i. kg-1) strongly stimulated microbial respiration, increasing the microbial metabolic quotient (qCO2) by 61.22%, and were also associated with higher SOC. In contrast, medium concentrations (10-100 mg a.i. kg-1) significantly suppressed microbial biomass. Exposure times of less than 30 days promoted SOC accumulation, 30-60 days produced the strongest respiratory response, while exposure beyond 60 days weakened the respiration response. Soil pH emerged as a key regulator: acidic conditions generally promoted CO2 emission, whereas alkaline conditions suppressed CO2 emission while enhancing SOC accumulation. Total nitrogen (TN) and soil organic matter (SOM) also influenced the response patterns of soil carbon processes. Overall, the effects of GLY on soil carbon cycling should not be generalized as uniform promotion or inhibition; instead, they represent context-dependent responses jointly shaped by exposure concentration, exposure time, and inherent soil properties. The observed SOC increases mainly reflect short-to medium-term responses in the compiled dataset.