Greenhouse-Scale Comparison of 10 Native Pacific Northwest Plants for the Removal of Per- and Polyfluoroalkyl Substances from Stormwater

Phytoremediation has the potential to remove per- and polyfluoroalkyl substances (PFAS) from stormwater. However, there is currently limited knowledge as to how plant type affects PFAS removal from stormwater, particularly in plants commonly used in stormwater bioswales. This greenhouse study evaluated the abilities of ten different Pacific Northwest native plants to remove PFAS from stormwater. The PFAS included C4–C10 perfluoroalkyl carboxylates; C4, C6, and C8 perfluoroalkyl sulfonates; and C6 and C8 perfluoroalkyl sulfonamides (FASAs). Plants were irrigated with the contaminated stormwater once a week for 10 weeks. The presence of plants enhanced PFAS removal over the uncultivated controls, with rushes and dicots having the highest total PFAS removal efficiencies ranging from 75 to 80%. The fate of the PFAS compounds was ultimately controlled by their Log organic carbon/water partition coefficient (K_oc) and molar volume. The PFAS molar volume and Log K_oc were strongly correlated with soil affinity (Pearson’s r = 0.82 to 0.85), resulting in greater removal efficiencies of larger PFAS compounds. Conversely, molar volume and Log K_oc values were strongly negatively correlated with bioconcentration factors (BCFs)(Pearson’s r = −0.72 to −0.82), resulting in the preferential bioaccumulation of smaller PFAS compounds. PFAS molar volume was also strongly negatively correlated with plant translocation factors (TFs)(Pearson’s r = −0.88), resulting in a preferential accumulation of the smaller PFAS compounds in the above-ground biomass. Conversely, root BCFs were positively correlated with the PFAS molar volume and Log K_oc for C4–C7 compounds (Pearson’s r = 0.91 to 0.96) but became negatively correlated for C8–C10 compounds (Pearson’s r = −0.89 to −0.99) as competition for sorption with the surrounding soil increased. The resulting chevron pattern indicated that the plant roots preferentially accumulated those PFAS compounds that were too large and hydrophobic to easily translocate to the above-ground biomass but were also too small and hydrophilic to have a strong affinity for the surrounding soil. Mass recovered for C6 and C8 FASAs was ≤35% in all plants and uncultivated control, indicating microbial transformation. Enrichment of linear perfluorooctane sulfonate (L-PFOS) was observed in all plant components (leaves, stems, and roots) relative to the stormwater influent. Finally, the removal of PFAS compounds from the stormwater could be accurately modeled with a multivariate linear regression containing the PFAS Log K_oc. Additional multivariate modeling revealed that plants with higher evapotranspiration rates had higher PFAS accumulations. However, evapotranspiration rates alone could not accurately model plant PFAS accumulation, indicating that other factors are also responsible for the differences observed. In particular, evapotranspiration rates were not successful in predicting plant PFBA accumulation. Rather, PFBA accumulation increased with increase in root mass, TFs, and leaf BCFs. This suggested that carrier-mediated transport mechanisms were responsible for PFBA accumulation in plants.