We know less about phosphorus retention in constructed wetlands than we think we do: A quantitative literature synthesis

• Average wetland phosphorus retention was highly variable at 51 ± 164 g/m2/year. • A monitoring period of three years is not sufficient to predict long-term trends. • 70% of studies sampled did not explicitly account for major flow events (storms). • Missing storm events misdiagnosed a wetland as a source of phosphorus. Wetlands are increasingly being constructed to retain phosphorus, but indicators of performance are often unique to individual wetlands. In this quantitative literature synthesis, we highlight two major knowledge gaps and a source of bias in our current understanding of phosphorus retention by constructed and restored wetlands. We performed a literature review to assess differences in wetland characteristics, phosphorus retention, and sampling frequency and duration to better understand how constructed and restored wetlands retain phosphorus. We then examined a series of different sampling frequencies to determine how sampling approaches affect observed trends using a case study at the Old Woman Creek National Estuarine Research Reserve where high-resolution data on phosphorus concentrations was available. We found that, while differences among broad wetland structural groupings exist, measured rates of phosphorus retention are highly variable. Additionally, we observed that the wetlands sampled in the literature are most often newly constructed; 70% of the wetlands were monitored for three or fewer years. Among twelve wetlands that were monitored for over ten years, resampling techniques demonstrated that three-year time scales are unable to predict long-term trends for nutrient retention. Similarly, we found that 70% of wetlands in the literature were sampled weekly or less frequently and did not account for major flow events when estimating nutrient retention. Our case study analysis indicated that excluding storm events lead to a significant underestimate of phosphorus retention. We further demonstrate that when sampling is infrequent and misses storm events, the retention of water and phosphorus by wetlands causes higher underestimates at the inflow than the outflow where phosphorus leaves wetlands more slowly than the inflow. The greater underestimation of loads at outflows compared to inflows can lead to a wetland being mislabeled as a source when it is a sink for phosphorus if storm events are not accounted for. Together this synthesis highlights that our understanding of how constructed wetlands store phosphorus is limited by three major factors: a lack of long-term monitoring, the drivers of high variability in phosphorus retention between different wetland types, and potential bias from the difficulties of capturing storm events. Our analysis indicates that most published works have been conducted at shorter time frames, or lower sampling frequencies than are required for accurate estimates of phosphorus retention. This finding is of significance because substantial investments are made in restoring and constructing wetlands based on inadequately supported assumptions that wetlands are highly effective at phosphorus retention. Ultimately, these results can inform improvements to post-restoration assessment of wetland nutrient function, including decisions on where and when to monitor water quality.