Assessing uncertainty in conservation practice performance: effect of sample frequency on bioreactor nitrate removal

One of the most common goals of edge-of-field water quality monitoring in tile-drained landscapes is to assess the performance of conservation practices such as denitrifying bioreactors. However, it is unclear whether sampling frequencies recommended for accurately quantifying nutrient losses in subsurface drainage also apply to monitoring conservation practices that have both an inlet and an outlet. The objective of this study was to quantify uncertainty due to common sampling approaches used at bioreactors treating subsurface drainage. Monte Carlo simulations were used to subsample reference datasets created using high-frequency NO₃-N sensing at seven bioreactors (n = 13 site-years). Four annual metrics (NO₃-N losses at the bioreactor inlet and outlet; removal efficiency; and load removed) were each assessed for three sampling scenarios across a range of sampling frequencies. Results showed that the size of the drainage area as well as variability in discharge and in NO₃-N concentration each influenced optimal sampling frequencies. Sampling every 2.1 ± 2.1 days allowed annual bioreactor performance metrics to be estimated within ±10% uncertainty for the research-oriented monitoring method (scenario 1: linear interpolation of concentration with continuous discharge). The practitioner-oriented instantaneous flux approach (scenario 2) required more frequent sampling to obtain the same level of uncertainty. Event-based sampling (scenario 3) was not recommended due to large overestimation resulting from sampling during high flows. Monitoring conservation practice performance requires more frequent sampling than monitoring subsurface nutrient loss alone. These findings provide benchmark uncertainty levels for monitoring denitrifying bioreactors across a range of edge-of-field objectives.