ORIGINAL ABSTRACT

Glyphosate-based herbicides (GBHs) are among the most extensively used agrochemicals worldwide, raising concerns about their unintended ecological hazards. While GBHs are primarily designed to eliminate the weeds, their indirect effects on non-target soil fauna mediated through vegetation change remain poorly understood. We conducted a two-year field experiment in two grassland fallow fields in Korea to evaluate how repeated GBH exposure alters plant communities and, in turn, affects collembolan diversity and community structure. Treatments included mechanical plant removal, a recommended GBH dose (5 L ha⁻¹), and excessive doses (15 L ha⁻¹ in 2016 and 150 L ha⁻¹ in 2017), reflecting realistic agricultural practices and potential overuse scenarios. At Jinju, a site with initially low plant diversity, GBH application further reduced plant richness and Shannon diversity without decreasing total cover, primarily due to the dominance of Xanthium canadense. These vegetation shifts coincided with marked declines in collembolan richness and a reorganization of their community composition. In contrast, Deokso, which had higher initial plant diversity, showed no significant changes in either plant or collembolan communities, suggesting that baseline vegetation diversity can buffer belowground impacts of GBH exposure. Across both sites, Folsomia parvulus consistently increased following GBH exposure, supporting its potential role as a bioindicator of soil disturbance. Our findings provide empirical evidence that GBHs can indirectly restructure belowground biodiversity by altering vegetation, highlighting vegetation-mediated pathways as overlooked routes of herbicide hazard. These results underscore the need to integrate non-target soil fauna into ecological risk assessment frameworks for agrochemicals.