Abstract

Herbicides considered selective to rice are generally evaluated based on their direct crop safety and weed suppression effects, yet it remains unclear whether they may also trigger indirect or context-dependent effects on rice under rice–barnyardgrass co-cultivation. To address this question, we compared rice performance and associated microbial dynamics under six conditions: rice–barnyardgrass co-cultivation and rice monoculture, each treated with a water spray control or sublethal doses of propanil (Pro, 66.7 mg a.i. L−1) or cyhalofop-butyl (Cyh, 5.86 mg a.i. L−1). Barnyardgrass exhibited visible injury and stronger leaf-level oxidative stress responses, whereas rice displayed no discernible phytotoxic symptoms. Nevertheless, under co-cultivation, herbicide treatment significantly suppressed rice growth, with up to 17.8% lower root lengths and 24.8% lower shoot fresh weights, with reductions varying by herbicide and trait. By contrast, comparable suppression was not observed under herbicide exposure or co-cultivation alone, identifying this response as an emergent, context-dependent negative effect. Microbiota reassembly emerged as an early and stage-specific component of the herbicide-associated response under co-cultivation, with the most pronounced changes detected on day 5 and occurring primarily in bacterial communities. Moreover, bacterial community variation was negatively correlated with root length (ρ = −0.664), and urease activity declined under herbicide treatment. Together, these findings indicate that in paddy fields, herbicides act not only on individual plants but also as an external disturbance to the coupled rice–barnyardgrass system, for which microbiota reorganization represents a key component of the ecological response. Our results suggest that herbicide selectivity should be interpreted within a crop–weed–microbiome context, rather than being inferred solely from their direct crop safety and weed suppression effects.