S2 Canadian Clean Air Regulatory Agenda (CARA) Mercury Science Program

In 2007, the Government of Canada (GOC) announced its intention to develop and implement regulations and other measures to reduce emissions of greenhouse gases, smog precursors and hazardous air pollutants including mercury under the Clean Air Regulatory Agenda (CARA). The CARA is a comprehensive approach to addressing these atmospheric issues and, as such, includes a scientific program to support regulatory activities and accountability. The CARA Mercury Science Program was developed to provide scientific information to support regulatory activities and accountability pertaining to atmospheric emissions of mercury. The first phase of the program, entitled “Setting-the-Baseline”, identified key indicators of the state-of-the-Canadian environment with respect to the transport, fate and effects of mercury and set out to define these indicators and understand the processes that underpinned the relationship between the indicators and anthropogenic emissions of mercury. This presentation describes the process that was used to develop the CARA Mercury Science Program and sets the stage for the presentations to follow in this special session.

Increased rates of atmospheric inorganic mercury (Hg(II)) deposition have been shown to increase methylmercury (MeHg) concentrations in fish; however the relative importance of regional versus local Hg sources are generally not known. In Canada, estimates from the global/regional atmospheric heavy metals model (GRAHM) indicate that while in most regions only ~2% of Hg deposition originates within Canada, there are “hotspots” where ~60% of deposition originates from local point sources. As part of the Clean Air Regulatory Agenda (CARA) Hg science program, we are quantifying Hg deposition to aquatic ecosystems from local sources using analysis of dated sediment cores from lakes located downwind of major point sources. In 2009-2010, sediment cores, water samples, and catchment soils were collected from: 18 lakes located varying distances from a metal smelter located at Flin Flon, Manitoba and coal-fired power plants in central Alberta; 5 lakes in Kejimkujik National Park, Nova Scotia where Hg concentrations in biota are known to be high despite the absence of local point sources; and 5 lakes at the Experimental Lakes Area (ELA) in Northwestern Ontario, where Hg deposition rates should represent regional background. In Flin Flon lakes, total Hg (THg) fluxes (HgF) began increasing in the ~1930s, when metal smelting began in this region and peaked in the late 1980s-early 2000s, reaching up to 6529 µg/m²/year in lakes near the smelter. However, THg fluxes (HgF) dramatically decreased with distance from the smelter and flux ratios (FR; HgF_{post 1990}/HgF_{pre 1900}) were 27-47 in lakes within 5 km of the smelter, 10-17 within 40 km, and only 3-5 70 km away. Lake water THg concentrations also decreased with increasing distance from the smelter (r²=0.51, p=0.03), but were low (1.08±0.40 ng/L) indicating that most of the Hg emitted from the smelter is particulate-bound and rapidly deposits to the lake sediments. In contrast, in Alberta lakes, Hg fluxes showed no trend with distance from the power plants and FRs were only 2-6. Lake water THg concentrations also did not vary with proximity to the power plants and averaged only 0.46±0.34 ng/L. These results suggest that the coal fired power plants are emitting fine particulate-bound or gaseous Hg(0) species which travel long distances after emission. Catchment soils are currently being analyzed for THg and lithogenic elements so that catchment and atmospheric Hg inputs to each lake can be distinguished. Results from the Kejimkujik and ELA lakes will also be presented.

An important knowledge gap in understanding how Hg moves from the atmosphere to the biosphere involves the role of terrestrial plants in the deposition of Hg. Plants act as conduits of atmospheric Hg into the biosphere, and Hg loading beneath forest canopies (in water that passes through the canopy [throughfall] and senescent vegetation [litterfall]) is much higher than in wet deposition to adjacent open areas. We have been examining the biogeochemical cycling of mercury at the Experimental Lakes Area (ELA) field station situated in the boreal ecoregion of NW Ontario, Canada since 1990. The goal of our ongoing research and monitoring program at this site is to quantify dry deposition of Hg to boreal forest landscapes using both direct (subtracting wet deposition of Hg from deposition of Hg in throughfall + litterfall) and inferential methods (using atmospheric Hg concentrations and meteorological data to model Hg dry deposition velocities). Towards this goal, since 2005 we have been quantifying atmospheric concentrations of gaseous elemental Hg (GEM), reactive gaseous Hg (RGM) and particulate-bound Hg (Hg-PM2.5) using Tekran^® Hg air speciation units and the Research Data Management and Quality Control System-RDMQ (for quality control). RGM and Hg-PM2.5 concentrations at the ELA are on average very low. Concentrations of GEM are ~1.2-1.4 ng m^-3 at the ELA, but can fluctuate quite dramatically daily, seasonally and during periods when forest fires are burning nearby. Using measured concentrations of GEM, RGM and Hg-PM2.5, modeled deposition velocities, and in forested areas, the foliar surface area (leaf area index) available to scavenge these atmospheric Hg species, we are modelling Hg dry deposition rates to this region. These modeled estimates will also be compared to estimates of dry deposition calculated from open area wet deposition, throughfall and litterfall fluxes of Hg from this same site.

Under the Clean Air Regulatory Agenda (CARA), a regulatory framework for air emissions, Environment Canada is studying lakes impacted by major atmospheric mercury (Hg) emitters. The goal of this study is to assess the localized impact of Hg pollution from major industry on lakes in two areas, the Precambrian shield of northern Saskatchewan and Manitoba and the boreal plain of central Alberta, and to investigate differences in food web biomagnification between the two study locations. Shield lakes are relatively isolated, surrounded by boreal forest and have received emissions from the Hudson Bay Mining and Smelting complex in Flin Flon, MB since 1930; Hg levels are highly elevated in nearby lake sediments. Central Alberta lakes are highly productive and receive inputs from the surrounding cottage and agricultural landscape in addition to emissions from a complex system of four large coal-fired power plants, in operation since the 1950’s; Hg levels are only slightly elevated in lake sediments. In summer 2009, we collected northern pike (Esox lucius), forage fish, benthic invertebrates and zooplankton from five lakes in each area. We hypothesized that the high productivity of the lakes in central Alberta would result in lower (as compared to Flin Flon lakes) Hg levels in northern pike due to faster growth rates; however pike are growing at similar rates in both areas i.e. length-age and weight-age regressions are similar. In the immediate Flin Flon area, Hg levels in northern pike do not reflect those of the high concentrations in the sediments; they, like Alberta pike, are below consumption guidelines of 0.5 µg/g. The notable exception is McLurg Lake, approximately 70 km northwest of the Flin Flon smelter, where mean THg concentrations in fish is 1.22±0.57 µg/g. In the summer of 2010, we sampled two additional lakes in this area; pike from these lakes also had high Hg concentrations (mean THg 0.88±0.37 µg/g). Similar observations were made in a study conducted in the 1980’s. Also being discussed is the characterization of the food webs of the Flin Flon and Alberta lakes using d¹⁵N and d¹³C stable isotopes and the use of d¹⁵N and Hg regressions to determine biomagnification rates within each lake. These findings will highlight the importance of implementing emissions controls specific to ecological and environmental factors and will have key implications for future industry.

Most studies aiming at predicting mercury (Hg) levels in fishes of boreal lakes require extensive, time consuming and expensive sampling campaigns of key biogeochemical variables. In a first step of our study within Environment Canada’s Clean Air Regulatory Agenda (CARA) program, we managed to model Hg levels in predatory fish frequently consumed by Québec fishers of the Abitibi-Temiscamingue administrative region simply using readily available GIS data for 38 lakes. To build our models, we namely used watershed morphometrics (lake order, drainage density, weak slopes in the drainage basin, drainage area on lake area ratio and mining activities) and land cover characteristics (fraction of watershed comprised of forested, unforested and wetland areas). Our northern pike (Esox lucius) model explained 71% of Hg level variations with three variables, the fraction of weak slope in the watershed being the strongest predictor followed by the percentage of catchment as forest and the lake order (32.1%, 22,6% and 16,3% of total value predicted respectively). Our northern pike model failed to explain Hg levels in walleye (Sander vitreus), a more pelagic fish species. To model Hg levels in walleye, we had to split lakes into three categories to obtain significant results. The number of mining sites in the watershed explained 77% of walleye Hg level variations. In the absence of mining sites, the geographic positioning of the lakes over the Ojibway-Barlow clay plain or not gave different responses both regulated by the fraction of the catchment as wetlands (86% of the variability explained for lakes located over the clay plain, 57% outside the clay plane). As our first walleye model was not of general use, we upgraded it by adding 30 new lakes in the Northern Québec administrative region and by including fish growth rates as a key variable that by itself explained 74% of Hg variability in walleye flesh. After having validated our northern pike and walleye models with new data collected in lakes not sampled so far, we tested the long-term influence of intensive logging in watersheds on fish Hg levels. To do so, we also validated our models with fish Hg data gathered over the past three decades in a series of 10 lakes sampled twice or three times since the 1980’s, and with watershed that have been altered by intensive logging during that period.

We investigated Hg exposure in fish-eating wildlife (common loons) breeding on selecting lakes in Nova Scotia, Quebec, Manitoba, Saskatchewan, and Alberta, Canada. We collected samples of surface sediment, water, small pelagic fish (8-20 cm fork length; mainly yellow perch), and blood from adult loons and their chicks from a number of sites, representing >50 lakes in total, including areas in the vicinity of coal burning power plants, and a large base metal smelter, as well as areas impacted by acid deposition, and a series of suitable reference lakes. Our study sites exhibited a wide range of lake size, surface sediment Hg concentration (<0.005 – 2.5 ug/g dry wt.), water chemistry (e.g.- pH range ~4.5 – 8.5), fish-Hg concentrations, and other variables. Fish-Hg and loon-Hg levels were not well correlated with proximity to power plants or metal smelters, nor with Hg concentrations in surface sediments, but were highly correlated with certain water parameters such as lake pH. Lake pH explained most (>70%) of the among-lake variability in fish-Hg and loon chick-Hg concentrations across our 4 study areas. Biological factors such as size and trophic level; and environmental factors such as water chemistry (especially pH) were the most important determinants of fish and loon Hg concentrations in this study.

It is well documented that methylmercury (MeHg) biomagnifies within aquatic food webs and that the consumption of Hg contaminated fish can be toxic to both humans and wildlife. However, it is much less clear whether the current environmental levels of MeHg in fish pose a risk to non-human consumers (i.e. wild fish and piscivorous vertebrates), and how widespread such risk might be. As part of Environment Canada’s Clean Air Regulatory Agenda Mercury Program (CARA), with the co-operation of various government, academic, industry and non-profit partners, we have compiled a database of over 415,000 records of Hg measurements from 158 species of fish. These data span nearly 40 years from all 12 provinces and territories in Canada, making it the largest of its kind in the world. In order to assess potential risk to piscivorous fish and wildlife, we used a multiple species modeling approach developed by the USGS (Wente 2004) to partition variation in fish Hg burden due to space, time, species and sample type in order to derive estimates of the Hg concentration in a theoretical prey fish (12 cm Yellow Perch) at all possible sampled locations. In this presentation, we compare the theoretical prey fish Hg concentrations to dietary MeHg thresholds developed for fish and common loons, and examine spatio-temporal patterns to determine areas in Canada where wild fish and common loons may be at risk of adverse effects due to MeHg contaminated prey. Results from this work build on previous efforts to develop large scale synoptic assessments of environmental MeHg risk, and results will be discussed in relation to complimentary work relating to atmospheric Hg deposition and MeHg bioavailability that will be presented in this special session.

A process-based watershed scale ecosystem mercury modeling framework is developed under the leadership of Environment Canada. The primary objective is to develop a predictive capability to assess the benefits of mercury emission reductions slated under Canada’s Clean Air Regulatory Agenda (CARA) and elsewhere globally. Other objectives are to identify the extent to which the mercury benefits of CARA may vary on a broad national scale and to estimate the impact of a changing climate and chemical environment on mercury cycling in Canadian ecosystems. Despite current knowledge gaps in mercury biogeochemical processes, deterministic mercury models can provide valuable information relating mercury emissions to mercury concentrations in biota for decision making and enhance our understanding of mercury cycling in the environment.

The model framework consists of sequentially linked dynamic models of atmospheric, terrestrial, aquatic and bioaccumulation cycling of mercury. Global/Regional Heavy Metal’s Model (GRAHM) simulates meteorological processes and atmospheric transport, transformations and deposition of elemental and oxidized mercury in gas and aerosol phases. Anthropogenic, natural and re-emission of deposited Hg from land and oceans constitute the mercury emissions in the model. Global anthropogenic emissions from 1990-2005 and detailed Canadian anthropogenic emissions for 2006 (CARA base year) are used to simulate the atmospheric mercury distribution and deposition in Canada. Mercury deposition from GRAHM is introduced into Integrated Catchment (INCA) - Mercury model. INCA-Hg simulates the transport of gaseous, dissolved and solid forms of Hg and transformations between elemental (Hg⁰), ionic (Hg²⁺) and methyl (MeHg) mercury in components of the watershed including vegetation, snow, litter and soil and surface waters. Dissolved and sediment-bound Hg²⁺ and MeHg are transported from the catchment to the lakes by stream flow. Atmospheric deposition from GRAHM and terrestrial mercury inputs from INCA-Hg feed into aquatic models: Hg Environmental Ratios Multimedia Ecosystem Sources (HERMES), and an aquatic and bioaccumulation model, Dynamic-Mercury Cycling Model (D-MCM). HERMES simulates concentrations of Hg⁰, Hg²⁺ and MeHg in the water column and sediments of lakes. D-MCM simulates Hg⁰, Hg²⁺ and MeHg in water column, sediments and food webs including six trophic levels (phytoplankton, zooplankton, benthos, piscivore fish, omnivore fish and non-piscivore fish).

Seven lakes in Canada covering a wide range of watershed to lake area ratios (3:1 - 38:1) and a variety of physical, geographical and chemical characteristics were selected for the ecosystem model application. These include North Cranberry and Big Dam West, Nova Scotia, Lake 240, Dickie and Harp, Ontario, Phantom, Manitoba and Wabamun, Alberta representing contaminated to remote sites. These lakes are in regions of CARA measurement super-sites that provide a comprehensive and geospatially explicit datasets for model input and evaluation. Model evaluation and sensitivity of the model to various watershed characteristics will be presented.