Bioswales and other forms of green infrastructure can be effective means to reduce environmental stresses in urban ecosystems; however, few studies have evaluated the ecology of these systems, or the role that plant selection and microbial assembly play in their function. For the current study, we examined the relationship between plant transpiration rates for five commonly planted herbaceous species in three bioswales in New York City, as well as bioswale soil microbial composition and soil chemistry. Soils were sampled near individual plants, with distinction made between upper (bioswale inlet) and lower slopes (bioswale outlet). We found high variation in transpiration rates across species, and that Nepeta × faassenii was the highest conductor (13.65 mmol H2O m–2s–1), while Panicum virgatum was the lowest conductor (2.67 mmol H2O m–2s–1) (p < 0.001). There was significant variation in percent N of leaves and soil, which did not relate to the higher water conductance in bioswales. Significantly higher C, N, and water content on the high end of bioswale slopes suggest storm water run-off is mostly absorbed on the inlet side. Bacterial and fungal communities were significantly clustered by bioswale and by plant species within each bioswale implying there are micro-environmental controls on the soil microbial composition, and that plant composition matters for microbial assemblages within bioswales. Plants with higher transpiration rates were associated with greater fungal and bacterial diversity at the level of the bioswale and at scale of the individual plant, suggesting a possible link between plant physiological traits and soil microbial communities. These data suggest that the specific plant palette selected for planting bioswales can have deterministic effects on the surrounding microbial communities which may further influence functions such as transpiration and nutrient cycling. These results may have implications for bioswale management to improve urban water quality and reduce stress on sewage systems after storm events by revising plant species palette selection based on the functional consequences of plant-microbial associations in engineered green infrastructure.