Relating Fish Hg to Variations in Sediment Hg, Climate and Atmospheric Deposition

This article addresses total fish Hg concentrations (THg) by variations in lake Sediment THg, atmospheric Hg deposition (atmHgdep), and climate, i.e., mean annual precipitation (ppt) and air temperature. The Fish THg data were taken from the 1967-to-2010 Fish Mercury Datalayer (FIMDAC). This compilation was standardized for 12-cm long Yellow Perch in accordance with the USGS National Descriptive Model for Mercury in Fish (NDMMF [1]), and documents Fish THg across 1936 non-contaminated lakes in Canada. About 40% of the standardized Fish THg variations related positively to increasing ppt and Sediment THg, but negatively to increasing mean annual July temperature (TJuly). Only 20% of the Fish THg variations related positively to atmHgdep alone. Increasing TJuly likely influences Fish Hg through increased lake and upslope Hg volatilization, in-fish growth dilution, and temperature-induced demethylization. FIMDAC Fish THg effectively did not change over time while atmHgdep decreased. Similarly, the above Fish Hg trends would likely not change much based on projecting the above observations into the future using current 2070 climate-change projections across Canada and the continental US. Regionally, the projected changes in Fish Hg would mostly increase with increasing ppt. Additional not-yet mapped increases are expected to occur in subarctic regions subject to increasing permafrost decline. Locally, Fish THg would continue to be affected by upwind and upslope pollution sources, and by lake-by-lake changes in water aeration and rates of lake-water inversions.


Introduction
There is much uncertainty about how climate and atmospheric Hg deposition affect Fish THg [2] [3] [4] [5]. In part, this uncertainty is due to large and yet to be quantified land-to-lake and in-lake variations by which Hg uptake by fish increases or decreases. The primary concern is that the bioaccumulation of toxic methyl Hg (MeHg) in fish and other aquatic organisms may increase with increasing atmospheric Hg deposition and climate warming. In principle, the presence and accumulation of Hg and MeHg in fish and sediments starts with the sequestration of atmospheric Hg deposition on land and water, and with the release of surface-exposed geogenic Hg minerals. A part of land-retained Hg is gradually released into streams and lakes through 1) direct Hg-containing litter inputs (detritus), 2) upslope soil and stream bank erosion, and 3) transfer of Hg bound by water-carried particulate and dissolved Hg-containing matter [6] [7] [8] [9]. With regard to detrital Hg inputs, tree foliage, twigs, branches, bark, and wood generally have lower Hg concentrations than mosses, fungi, and lichens [10] [11]. In soils, THg decreases from the organic litter layers on the surface to the subsoil layers below [10]. In downstream, the total amount of Hg generally increases with increasing organic matter transfer [12], which in turn leads to increasing Hg accumulations in stream and lake sediments [13] [14] [15] [16]. Due to particulate Hg retention in wet areas and wetlands, first-order stream and lake sediments downslope from forested and wetland dominated watersheds have higher THg and organic matter contents than elsewhere [17]- [22].
This article focuses on analyzing the extent to which standardized data for total Hg concentrations (THg) in fish-as compiled within the Fish Mercury Datalayer (FIMDAC [1])-co-vary with lake Sediment THg, atmospheric Hg deposition (atmHg dep ), mean annual precipitation (ppt), and mean annual air temperatures for January (winter, T Jan ) and July (summer, T July ). This analysis was enabled by cross-referencing the Fish THg data to the modelled and mapped Sediment THg, atmHg dep , ppt, T Jan and T July variations across Canada, with special reference to potential climate-induced changes up to 2070. The hypothesis is that at least some of the Fish THg variations can be quantified in this way for tentative trend mapping across Canada, the continental USA, and over time.
Data pertaining to lakes downwind and downslope from major anthropogenic American Journal of Climate Change Hg release locations (e.g. [34]) are not part of this analysis.

Methods
The 1967 to 2010 FIMDAC Fish THg data (reported in ng•g −1 , wet-weight) were compiled for 1936 pristine lakes across Canada from 1967 to 2010. These data were standardized for one fish species (Yellow Perch) and one fork length (12 cm [37]). Fish sampling per lake varied from one time to multiple times. Total sample size used for the analysis below was n = 3179. Fish THg data from water bodies other than lakes were not included; also excluded: The data for lake Sediment THg (in ng•g −1 ) were obtained from the open geochemical survey files of the Geological Survey of Canada [38] [39], and were compiled for the provinces of Quebec and Nova Scotia as well [40]. This compi- With ArcMap, cross-referencing was done by data-layer extracting the lake Sediment THg, atmHg dep , ppt, T Jan and T July values for each FIMDAC Fish THg location, row-by-row. This was followed by multivariate regression and factor analysis, with Fish THg, Sediment THg, atmHg dep , ppt, T July and T Jan as variables. Table 1 provides a basic statistical summary of the variables used in this article. Most notably, these variables differ across Canada in range, such that the lake-by-lake entries for Fish THg have a wider maximum to minimum ratio (i.e. 38) than the corresponding model-derived entries for Sediment THg (i.e. 15) and for atmHg dep (i.e. 12) For mean annual precipitation, the maximum/minimum ratio is 20. Mean annual T Jan is about three times more variable than mean annual T July . The longitudinal range is 4 times larger than the latitudinal range.

Results
Using Sediment THg as dependent variable, and atmHg dep , ppt, T Jan and T July as independent variables produced: M. Nasr, P. A. Arp Figure 1. Mean annual January and July air temperature (T Jan , T July ) and mean annual precipitation (ppt) according to the Coupled Model Intercomparison Project (CMIP5-rcp8.5) at http://climate-scenarios.canada.ca/?page=download-intro with contours. Also shown: mean annual net atmospheric Hg deposition (atmHg dep ), based on the Global/Regional Atmospheric Heavy Metals Model (GRAHM2005). Extent: across Canada (provinces and northern territories outlined except most northern parts) and the continental USA.
By climate variables alone, the following was obtained: thereby accounting for about 90% of the variations captured by Equations (2) and (3). Using atmHg dep , T July and T Jan as predictor variables produced .303 R Hence, using atmHg dep together with T July and T Jan improved the results, but using Sediment log 10 THg or ppt with and without T July as additional predictor variables was more effective in capturing more of the Fish log 10 THg variations (Equations (2), (3), and (6), (7)). In noting that 80% of the GRAHM2005-modelled atmHg dep variations can also be attributed to climatic and geographic variations via Equation (8) The T July effect on atmHg dep is most likely a surrogate for the geographic Hg American Journal of Climate Change emission and subsequent Hg deposition pattern, which is highest along the southeastern USA-Canada border, and least for the snow-and ice-covered alpine and arctic areas. T Jan likely accounts for increased Hg volatilities from snow-and ice-covered surfaces.
The location-specific adjustments for the Pacific Rim and High-Arctic coastal areas in Equation (8) can be used to compensate 1) for the ppt-induced atmHg dep dilution, and 2) for the T Jan -induced underestimation for atmHg dep along the arctic coastlines, where oceanic Hg upwelling contributes to land-based and aquatic Hg sequestration and bioaccumulation [44] [45]. With respect to Fish THg, increases in ppt and Sediment THg affect Fish log 10 THg in direct proportions.
Checking the correlations among the Equation (1) to (8) variables reveals that T July only has a weak direct effect on Fish log 10 THg (Table 2). Nevertheless, factor analyzing this matrix revealed that T July is an important negative co-variant of Fish log 10 THg as per factor 2 in Table 3.
Projecting the Equation (2)      Equations (8). This was also the case for the earlier North American Hg deposition study by [50] and [51].

Discussion
The above results account for about 40% of the standardized Hg concentrations in fish, with Sediment THg, atmTHg dep , ppt, T July , and T Jan as significant co-variants in varying combinations. The co-varying influences on Fish THg are particularly strong for Sediment THg and ppt, likely due to direct and indirect precipitation-induced Hg inputs into lake water from the atmosphere and surrounding land. While these inputs would have the most direct and causative Hg lake inputs. In addition, while Equations (2) and (5)  In part, this is due to increasing and decreasing trends by regions and lakes. In some areas, gradual declines in Fish THg have occurred [46]. Increases in Sediment THg with increasing atmTHg dep -as implied by Equations (4) and (5)-have been reported for the upper layer of non-bulked mid-to high-latitude lake sediments by [52] and [57].  Figure 4-is likely due to a cancelling of positive and negative lake-by-lake trends. For example, [59] and [46] reported declining to non-declining levels of Fish Hg with respect to climate warming. In contrast, and 2) extrapolated these into the future, up to 2050. Changing to non-changing levels in Fish THg are likely due to complex interactions involving changes in trophic Hg transferences, fish predation, lake-water conditions (temperature, aeration, chemical composition) and reductions in atmospheric and terrestrial Hg inputs into lakes. The latter, as implied by Equations (3) and (4), would especially be the case in areas subject to drought (i.e., low ppt, high rates of Hg volatilization, low rates of Hg immobilization). On land, Hg retention and land-to-lake transfers are affected by upslope land use, extent and type of vegetation type including wetland coverage, and Hg-containing mineral exposures [22] [61]. Increased surface temperatures on land stimulate 1) Hg re-volatilization from open areas [62], 2) Hg losses due to forest fires [63], and 3) Hg immobilization in vegetation biomass and in soil organic matter due to increased warmer and longer summer growth, especially where soils remain moist to wet [8] [64].

Equation (9) and in
Increasing upstream lake-to-lake wetland coverage tends to increase THg in water and fish through brown-water transfer of organically complexed Hg [7] [65] [66]. In lakes, fish uptake of Hg varies by, e.g., extent and type of lake input from land and atmospheric sources, and by lake size and morphology. Daily, seasonal and annual water-intake and related aeration and de-aeration dynamics affect in-lake trophic bioaccumulation of methyl Hg directly [67]. For example, increased Hg uptake by fish would occur through increased net Hg methylization as hypolimnetic water temperatures and biological oxygen demands rise from cool to warm [68]. In this regard, brown-water discharge renders small lakes warmer than clear-water lakes, thereby reduxing epi-and hypolimnetic aeration which would stimulate Hg methylization. Hence, Fish THg (small brown-water lakes) > Fish THg (clear-water lakes). In addition, Fish Hg increases with decreasing lake-water pH, likely due to higher levels of Hg solubility and increased Hg methylization rates at lower pH [68] [69]. However, Fish THg can also increase with increasing pH, particular where in-lake de-aeration trumps the pH effect on Hg methylization [70] [71].
Detailed studies have shown that, while increasing water temperatures increase the production and uptake of methyl Hg from sediments and epilimnetic and anoxic hypolimnetic waters [72], there is also a concurrent temperature increase in Hg demethylation [53] [73]. [71] reported that deepening the thermocline and oxycline of one lake relative to a reference lake (done by way of pumping at about 15,000 m 3 •day −1 ) increased epilimnetic temperature and oxygen saturation leading to continuing reductions in epilimnetic MeHg and Fish THg, with post-experiment effects lasting for at least one year. Hence, future changes in Fish Hg are also related to climate-induced thermocline inversions, lake-by-lake.
Across northern Canada, the influx of Hg and MeHg into lakes and rivers peaks during the snowmelt season, and is lowest during the winter season [74].
Post-2000 Sediment THg fluxes decreased towards northern latitudes after showing a steady increase since about 1900. This was likely in response to the steadily falling of net atmospheric Hg deposition rates over the last 25 years [75] [76] [77]. Mercury levels in fish, birds, and mammals, however, varied, with inter-annual trends remaining uncertain across arctic inland and marine waters [5] [67] [78] [79]. In contrast, the subarctic region is currently subject to increasing permafrost decline. This decline produces water-filled collapse scars across peat plateaus [80], with each scar supporting algal growth, which in turn leads to peat-and mineral-released Hg accumulations in lakes and sediments [81] [82] [83].
Given that total Hg concentrations in retail fish should not exceed 500 ng•g −1 [84] (which corresponds to log 10 THg = 2.70), most of the standardized Fish THg values fell below this limit. The percentage extent to this value to be exceeded in 2070 would mainly be limited to southern Quebec where the maximum projected change in log 10 THg is ≈0.1, and more strongly so along the Pacific coast, where the maximum projected change in log 10 THg is ≈0.4.
Altogether, 60% of the Fish THg variations remain unexplained by way of the above trend analysis. This situation may improve by adding lake-by-lake variables such as lake size, morphology, upland watershed, topography and upslope wetland coverage to the analysis.

Concluding Remarks
As shown, climate affects in-lake Hg accumulations due to variations in weather, season, atmospheric Hg deposition, surrounding vegetation, and geological Hg exposure patterns. This would also include lake-by-lake variations in thermocline inversion and related recovery times. As quantified above, cross-regional changes among the co-varying patterns of atmospheric Hg deposition, Sediment THg and Fish THg will likely remain small, with potential changes more likely related to changes in precipitation amounts, frequencies and intensities than to co-varying changes in atmospheric Hg deposition and temperature. A major exception to this would be occurring across the subarctic, where sustained permafrost losses lead to a widening incidence of peat-plateau scars followed by mineral-and peat-accumulated Hg release into lakes and rivers. In total, lake-by-lake variations in Fish THg will remain large as documented, and will require continued monitoring due to changing Hg inputs and changing in-lake dynamics.
Whether or not future levels in sediment and Fish THg will trend as projected in Figure 4 remains unknown. The assumption made is that the co-dependencies of the variables in Table 1 are functionally linked to changing climate variations not only across space, but also across time, in the same way.