Temporal phosphorus dynamics affecting retention estimates in agricultural constructed wetlands

Abstract

Data from seven constructed wetlands (CWs) in the south of Sweden were analyzed to investigate the effects of water flow and season on inflow phosphorus (P) concentrations and temporal P retention variations in CWs receiving runoff from arable land. The form of P (dissolved or particulate) during different water flows (high and low) and seasons (warm and cold) was investigated using the results of total P (TP) and phosphate analyzed in grab samples that had been collected regularly or occasionally during two to nine years, along with continuous water flow measurements.The form of inflow and outflow P (particulate or dissolved P) differed between CWs, and also varied with season and flow. For instance, in three of the CWs, particulate P (PP) dominated the inflow during the cold period with high flow, while during the other periods the proportion of PP was approximately 50%. In one CW situated in a catchment with high clay content, PP dominated both inflow and outflow at all times. The average clay content in catchment top soils was positively correlated to the flow-weighted inflow TP concentrations.In three CWs receiving runoff through drainage pipes, the relationship between TP concentrations (TPin) and water flow was positive, both during high and low flow, and during warm and cold period. However, in four CWs that received surface water runoff, the relationship between TPin and water flow was positive during high flow periods (i.e. the 25% sampling occasions with the highest flow), and during low flow and warm period, the relationship was negative in these four wetlands, indicating either anoxic stagnant water upstream or influence from rural wastewater.The temporal dynamics of P concentrations mean that in some of the CWs, the main part of the annual P retention may occur during a few days with high water flows. The correlation between concentration and water flow suggests that the water sampling strategy may have a considerable impact on retention estimates, as exemplified by some calculation examples.