Effects of a herbicide on paddy predatory insects depend on their microhabitat use and an insecticide application.

Indirect effects of agrochemicals on organisms via biotic interactions are less studied than direct chemical toxicity despite their potential relevance in agricultural landscapes. In particular, the role of species traits in characterizing indirect effects of pesticides has been largely overlooked. Moreover, it is still unclear whether such indirect effects on organisms are prevalent even when the organisms are exposed to direct toxicity. We conducted a mesocosm experiment to examine indirect effects of a herbicide (pentoxazone) on aquatic predatory insects of rice paddies. Because the herbicide selectively controls photosynthetic organisms, we assumed that the effects of the herbicide on predatory insects would be indirect. We hypothesized that phytophilous predators such as some Odonata larvae, which cling to aquatic macrophytes, would be more subject to negative indirect effects of the herbicide through a decrease in abundance of aquatic macrophytes than benthic, nektonic, and neustonic predators. Also, we crossed-applied an insecticide (fipronil) with herbicide application to examine whether the indirect effects of the herbicide on the assembling predators act additively with direct adverse effects of the insecticide. The herbicide application did not decrease the abundance of phytoplankton constitutively, and there were no clear negative impacts of the herbicide on zooplankton and prey insects (detritivores and herbivores). However, the abundance of aquatic macrophytes was significantly decreased by the herbicide application. Although indirect effects of the herbicide were not so strong on most predators, their magnitude and sign differed markedly among predator species. In particular, the abundance of phytophilous predators was more likely to decrease than that of benthic, nektonic, and neustonic predators when the herbicide was applied. However, these indirect effects of the herbicide could not be detected when the insecticide was also applied, seemingly due to fipronil's high lethal toxicity. Our study highlights the importance of species traits such as microhabitat use, which characterize biotic interactions, for predicting indirect effects of agrochemicals. Given that indirect effects of the chemicals vary in response to species traits and direct toxicity of other chemicals, efforts to explain this variation are needed to predict the realistic risks of indirect effects of agrochemicals in nature.