S7 Multi-scale modeling of mercury transport and fate in river ecosystems

The McTier Creek watershed is located in the Sand Hills ecoregion of South Carolina and is a small catchment within the Edisto River basin. Two watershed hydrology models were applied to the McTier Creek watershed as part of a larger scientific investigation to expand the understanding of relations among hydrologic, geochemical, and ecological processes that affect fish-tissue mercury concentrations within the Edisto River basin. The two models are the topography-based hydrological model (TOPMODEL) and the grid-based mercury model (GBMM). TOPMODEL uses the variable-source area concept for simulating streamflow, and GBMM uses a spatially explicit modified curve-number approach for simulating streamflow. The hydrologic output from TOPMODEL can be used explicitly to simulate the transport of mercury in separate applications, whereas the hydrology output from GBMM is used implicitly in the simulation of mercury fate and transport in GBMM. The modeling efforts were a collaboration between the U.S. Geological Survey and the U.S. Environmental Protection Agency, National Exposure Research Laboratory.

We assessed methylmercury (MeHg) concentrations across multiple ecological scales in the Edisto and Upper Hudson Rivers. A general downstream increase in concentrations indicated a continuous supply of MeHg to the stream channel throughout the Edisto basin. In the Upper Hudson basin, higher MeHg concentrations in tributaries than in the main channel indicated that significant sources of MeHg exist in headwater locations but that environmental factors attenuate concentrations as water moves downstream. The results indicated that production in wetland/floodplain areas, hydrologic transport to the stream aquatic environment, and attenuation in open water areas are fundamental controls on dissolved MeHg concentrations in lotic habitats in diverse geographic, climatic, and ecologic settings. These results provide a foundation for several of the Hg modeling efforts being presented in Session S7.

McTier Creek is a small watershed located in Aiken County, South Carolina and forms part of the headwaters for the Edisto River basin. The Edisto River basin is noted for having some of the highest measured fish-tissue mercury concentrations in the United States. In an attempt to improve the understanding of the factors causing these high mercury levels, the National Water-Quality Assessment Program of the U.S. Geological Survey conducted an extensive field investigation of mercury in the McTier Creek ecosystem. This investigation included the collection of hydrologic, biologic, and water-quality data as well as the development of a number of hydrologic and water-quality models. One modeling effort involved the development of a simple water-quality load model that utilized a mass-balance equation in conjunction with hydrologic simulations from the topography-based hydrological model (TOPMODEL). Several variants of this load model were developed including one, called TOPLOAD, which utilized the simulated surface and subsurface flow components taken directly from TOPMODEL. A second variant, TOPLOAD-H, added a groundwater partitioning algorithm (Hornberger and others, 1994) to TOPMODEL, thereby providing for multiple groundwater flow components. A brief description of the development of these simple mercury load models and results of the simulation in the McTier Creek basin will be presented.

Reference:
Hornberger, G.M., Bencala, K.E., and McKnight, D.M., 1994, Hydrological controls on dissolved organic carbon during snowmelt in the Snake River near Montezuma, Colorado: Biogeochemistry, v. 25, p. 147–165.

Multi-scale assessment of mercury and dissolved organic carbon loads (fluxes) was conducted in the Edisto River basin (in the Coastal Plain of South Carolina) and the upper Hudson River Basin (in the Adirondack Mountains of New York). The assessment focused on the role of headwater basins in the transformation of mercury species and their transport to downstream larger river habitats where mercury concentrations in resident fishes have historically been elevated above wildlife and human dietary guidelines. Annual fluxes and yields (area-weighted flux) of dissolved and particulate mercury species (methylmercury and total mercury) were compared between headwater (drainage areas less than 80 square kilometers) and main stem (drainage areas greater than 200 square kilometers) scales in both study areas. Dissolved organic carbon was included in the assessment because of the strong affinity of mercury for organic matter and its potential to control mercury fate and transport.

Fluxes and yields of dissolved and particulate mercury species (methylmercury and total mercury) and dissolved organic carbon were computed from stream discharge and concentrations data using the S-LOADEST software program based on the “rating curve” method. Consistent application of laboratory analytical methods and load estimation procedures were adopted to allow flux comparisons among basins while directly addressing the effect of basin scale on mercury supply. Results of the load estimation and comparison of annual methylmercury yields at different locations within river basins will be presented and discussed in relation to the impact of basin scale on methylmercury supply, the challenges inherent in comparisons between different sized basins, and the appropriate standardization procedures for inter-basin comparisons.

Mercury (Hg) is known to have contaminated food webs of the Sacramento Delta and San Francisco Bay, but the sources and mechanisms of Hg delivery to this ecosystem are still debated. While it is well known that 19^th C. hydraulic gold mining in California produced huge volumes of Hg-contaminated sediments, conventional wisdom favors the idea that hydraulic mining sediment, dosed with Hg used in gold extraction, was largely exhausted from Sierra Nevada foothills as a wave that passed by the mid-1900’s, such that the risk of contamination has declined. However, our work shows that sediment remaining in piedmont storage within valleys used for mining is a persistent, yet episodic source of Hg to the lowland Central Valley of California, posing great risk to downstream ecosystems. We have documented total Hg (HgT) concentrations in >250 sediment samples from a range of deposits along a >100 km longitudinal transect, which highlight consistent patterns of HgT in chemostratigraphy and suggest characteristic redistribution of Hg-laden sediment within this fan system. These findings are supported by historic and modern river longitudinal profiles, spatial mapping of event-based erosion/deposition from aerial photos, and satellite images and washload modeling exhibiting high concentrations of suspended sediment exported by the Yuba during large floods. Considering the HgT within the Yuba hydraulic mining sediment along with the increasing risk of erosion associated with climate change, the Yuba Fan poses a major risk to downstream food webs.

We find HgT varies over two orders of magnitude allowing for discrimination of legacy sediments from those in background (non-mining) sediments. This enables us to map out the spatiotemporal history of toxic fan evolution. Most deposits in active channel banks are composed of several sediment units that decrease in HgT up-section, indicating dilution of mining sediments through time. Sediment budget data for the last century show systematic evacuation of mining sediments from all near channel areas and modest accumulation of sediment in downstream floodplains. This redistribution of sediment occurs during large flood events which erode toxic terraces on decadal timescales, producing high suspended sediment concentrations, such that the toxicity of the fan progrades downstream. These results provide a new context for: interpreting the century-scale evolution of major sediment slugs in fluvial systems; for exploring process links between the history of sedimentation and inundation, and the biogeochemistry of Hg; and for assessing the downstream risk of Hg contamination and bioavailability along sediment transport pathways.

MERGANSER (MERcury Geo-spatial AssesmentS for the New England Region) is an empirical least squares multiple regression model using atmospheric deposition of mercury (Hg) and readily obtainable lake and watershed features to predict fish and common loon Hg (as methyl mercury) in New England lakes. State monitoring programs communicate mercury risks to recreational and subsistence fisherman, however, the majority of waterbodies are unmonitored. The expense of monitoring all potential lakes is prohibitive, therefore we developed a predictive model to address exposure risks in unmonitored waterbodies. We modeled lakes larger than 8 ha and with drainage area completely within the USA (4404 lakes), using 3827 fish (12 species) and loon Hg values from 420 lakes. MERGANSER predictor variables included Hg deposition, watershed alkalinity, percent wetlands, percent forest canopy, percent agriculture, drainage area, population, mean annual air temperature and watershed slope. The model returns fish tissue or loon blood Hg for user-entered species and fish length. MERGANSER explained 63% of the variance in fish fillet and loon Hg concentrations. MERGANSER predicted that 32-cm smallmouth bass had a median Hg concentration of 0.53 µg g-1 (root mean square error 0.27 µg g-1) and exceeded EPA’s recommended fish tissue criterion of 0.3 µg g-1 Hg in 90% of New England lakes. Common loon had a median Hg concentration of 1.07 µg g-1 and was in the moderate or higher risk category of >1 µg g-1 Hg in 58% of New England lakes. MERGANSER can be applied to target fish advisories to specific unmonitored lakes, and for scenario evaluation, such as the effect of changes in Hg deposition, land use, or warmer climate on fish and loon mercury. MERGANSER is intended as a publicly available web-based tool for exposure assessment and risk communication in the New England region.

Issues of scale hamper the integration and modeling of abiotic and biotic mercury (Hg) pools and fluxes across terrestrial and aquatic ecosystems. Mercury monitoring efforts provide some evidence regarding the effects of Hg emissions on water quality and ecosystem health, but not a comprehensive picture because of (1) different spatial and temporal sampling strategies and (2) the complexity of Hg fate and transport dynamics. The role of environmental sampling is to make inferences about the larger population; an objective of this work is to infer spatial and temporal patterns in Hg fate and transport from a diverse collection of sampling data. In order to assess spatiotemporal patterns in Hg concentration across the coastal Maine (USA) landscape, we are integrating legacy data from more than a dozen studies with thousands of environmental Hg samples. Mercury measurements in the coordinated database include atmospheric deposition and several terrestrial and aquatic media types spanning a time period from 1986 – 2010. Some key concepts addressed while integrating disparate data are spatial and temporal support and autocorrelation. Spatial support refers to the physical size and extent over which an observation is made and for any statistical analyses these different spatial and temporal supports need to be reconciled. Spatial autocorrelation refers to the distance over which similarity in Hg concentration values tend to persist. The spatial autocorrelation functions for Hg concentrations are not expected to be the same across different media. This presentation will address the initial characterization of the range in spatial and temporal support across the data set and an assessment of the spatial autocorrelation structures as the first steps in developing an inference framework. The inference framework being developed for this project hopes to elucidate the scales at which the current, diverse monitoring data are applicable and comparable, in order to better predict effects of changes in Hg emissions.

Watershed scale mercury modeling is just beginning to emerge and help further understanding of mercury fate, transport, and transformation within a watershed. The principle processes and equations governing mercury cycling were incorporated into VELMA (Visualizing Ecosystems for Land Management Assessment) to predict daily fluxes and concentrations of total mercury (THg) and methylmercury (MeHg) for a specified catchment. VELMA is a spatially-distributed eco-hydrological model that simulates soil water infiltration and redistribution, evapo-transpiration, surface and subsurface runoff, carbon (C) and nitrogen (N) cycling in plants and soils, and the transport of dissolved organic carbon (DOC), dissolved inorganic nitrogen (DIN), and dissolved organic nitrogen (DON) from the terrestrial landscape to streams. VELMA uses a distributed soil column framework to simulate the vertical movement of water, heat, and nutrients within the soil, between the soil and vegetation, and between the soil surface and vegetation to the atmosphere. The transport of mercury species down the hillslope is accomplished through advection of dissolved mercury (aqueous phase and DOC complexed). Each soil column within the watershed accounts for mercury accumulation through governing processes of deposition, decomposition, inflows, outflows, and transformation reactions. Site and species-specific rate constants, along with complexation constants, have a significant role in mercury transport. The mercury-incorporated VELMA model has been calibrated and simulated for both WS10 of the H.J. Andrews experimental forest and McTier Creek of the Edisto River Basin. WS10 is a 0.102 km² forested catchment located in central Cascade Range of Oregon while McTier Creek is a 79.4 km² watershed in the Sand Hills of South Carolina. The seasonality of in-stream mercury concentration was low during dry summers, while mercury concentration typically peaked during the first fall storm and remained notable throughout the rainy winter season. A large flux in mercury through storm events did not always correspond to an increase of in-stream mercury concentration. At both sites simulated total mercury concentration are within range of observations and help to predict and understand mercury species at the watershed level.