In urban areas, untreated stormwater runoff can pollute downstream surface waters. To intercept and treat runoff, low-impact or "green infrastructure" approaches such as using biofilters are adopted. Yet, actual biofilter pollutant removal is poorly understood; removal is often studied in laboratory columns, with variable removal of viable and culturable microbial cell numbers including pathogens. Here, to assess bacterial pollutant removal in full-scale planted biofilters, stormwater was applied, unspiked or spiked with untreated sewage, in simulated storm events under transient flow conditions, during which biofilter influents versus effluents were compared. Based on microbial biomass, sequences of bacterial community genes encoding 16S rRNA, and gene copies of the human fecal marker HF183 and of the Enterococcus spp. marker Entero1A, removal of bacterial pollutants in biofilters was limited. Dominant bacterial taxa were similar for influent versus effluent aqueous samples within each inflow treatment of either spiked or unspiked stormwater. Bacterial pollutants in soil were gradually washed out, albeit incompletely, during simulated storm flushing events. In post-storm biofilter soil cores, retained influent bacteria were concentrated in the top layers (0-10 cm), indicating that the removal of bacterial pollutants was spatially limited to surface soils. To the extent that plant-associated processes are responsible for this spatial pattern, treatment performance might be enhanced by biofilter designs that maximize influent contact with the rhizosphere.
Keywords: biofilters; microbial community; source tracking.; stormwater; transient flow.