B51B-0425
Building a Probabilistic Denitrification Model for an Oregon Salt Marsh
Friday, 18 December 2015
Poster Hall (Moscone South)
Jessica B Moon1, Hilmar Armin Stecher2, Theodore DeWitt2, Amanda Nahlik3, Rochelle Regutti4, Lauren Michael3, M. Siobhan Fennessy3, Robert Mckane5 and Kusum Naithani6, (1)University of Arkansas, Fayetteville, AR, United States, (2)US EPA, Newport, OR, United States, (3)Kenyon College, Biology, Gambier, OH, United States, (4)Oregon State University, Corvallis, OR, United States, (5)USEPA, Corvallis, OR, United States, (6)Penn State University, University Park, PA, United States
Abstract:
Despite abundant work starting in the 1950s on the drivers of denitrification (DeN), mechanistic complexity and methodological challenges of direct DeN measurements have resulted in a lack of reliable rate estimates across landscapes, and a lack of operationally valid, robust models. Measuring and modeling DeN are particularly challenging in tidal systems, which play a vital role in buffering adjacent coastal waters from nitrogen inputs. These systems are hydrologically and biogeochemically complex, varying on fine temporal and spatial scales. We assessed the spatial and temporal variability of soil nitrate (NO3-) levels and O2 availability, two primary drivers of DeN, in surface soils of Winant salt marsh located in Yaquina estuary, OR during the summers of 2013 and 2014. We found low temporal variability in soil NO3- concentrations across years, tide series, and tide cycles, but high spatial variability linked to elevation gradients (i.e., habitat types); spatial variability within the high marsh habitat (0 - 68 μg N g-1 dry soil) was correlated with distance to major tide creek channels and connectivity to upslope N-fixing red alder. Soil O2 measurements collected at 5 cm below ground across three locations on two spring tide series showed that O2 drawdown rates were also spatially variable. Depending on the marsh location, O2 draw down ranged from sub-optimal for DeN (> 80 % O2 saturation) across an entire tide series (i.e., across days) to optimum (i.e., ~ 0 % O2 saturation) within one overtopping tide event (i.e., within hours). We are using these results, along with empirical relationships created between DeN and soil NO3- concentrations for Winant to improve on a pre-existing tidal DeN model. We will develop the first version of a fully probabilistic hierarchical Bayesian tidal DeN model to quantify parameter and prediction uncertainties, which are as important as determining mean predictions in order to distinguish measurable differences across the marsh.
