Spatial Distribution of Sequestered Metals in an Arid Southern California Bioswale

Background: Bioswales are a type of permeable green infrastructure designed to slow storm water runoff, and clean runoff by sequestering pollutants such as heavy metals. The influence of bioswale features such as slope, terrain, soil type, depth, and vegetation on the efficacy of bioswales in processing storm water runoff is an active area of research aimed partly at optimizing bioswale design and construction. Methods: A bioswale at Pomona College was used to study several elements of bioswale performance. Concentrations of aluminum, cobalt, chromium, copper, iron, magnesium, manganese, nickel, lead, vanadium and zinc were measured in surface soil, depth soil, and at several plant locations. The distribution of metals in surface, depth, and rhizosphere soil fractions were measured to analyze the roles of location along the water flow axis, bioswale terrain, sampling depth, vegetation types, and soil chemistry on the distribution of metals. Soils were digested according to EPA Method 3051 and for soil fractions, a modified Tessier et. al. (1979) sequential extraction protocol, and metal concentrations were quantified using ICP-OES. Principal Component Analysis (PCA) was used to determine the fractions of measurement variation explained by water flow axis and other variables. Regression models were applied to test whether 2 the surface sample concentrations of metals decreased with distance from bioswale inflow. Empirical Bayesian Kriging in ArcGIS interpolated metal concentrations across the surface of the bioswale, allowing the sources of metal pollution and relative effectiveness of various groundcover at sequestration to be inferred. Results: The PCA found that the location of samples on the axis of water flow explains 62% of the variance between surface soil samples and the linear regressions of these surface samples determined that the anthropogenic metals zinc, lead, cobalt, and manganese all decrease significantly along this axis (p-values from <0.01 to <0.1). These metals are introduced via road runoff from tires and oil, whereas other metals are introduced via aerosols from brake pads. Plant species sampled were local Carex, Juncus patens, Epilobium ciliatum subsp. ciliatum, and introduced Phoenix date palm, none of which exhibited significant abilities to uptake metals. Sequential extraction and ICP-OES quantification revealed that high percentages of all metals are bound to organic matter in soil, and indicated that zinc, lead, cobalt, manganese, and chromium are preferentially bound to non-organic soil fractions whereas aluminum, vanadium, magnesium, and copper are preferentially bound to the organic soil fraction. Conclusions: Orientation of the bioswale to water flow axis, slope and linear length are predictors of runoff metal concentrations. Bioswale vegetation needs to be selected in accordance with bioswale conditions but also metal absorption ability to further clean stormwater runoff. Trees appear superior to rocks and sedge at sequestering metals in soil. The bioswale decreases the concentration of metals in water infiltrating into underground aquifers.