Historical Records of Mercury Stable Isotopes in Sediments of Tibetan Lakes

The Tibetan Plateau (TP), known as the “Third Pole”, is a critical zone for atmospheric mercury (Hg) deposition. Increasing anthropogenic activities in the globe leads to environmental changes, which may affect the loading, transport and deposition of Hg in the environment. However, the deposition history and geochemical cycling of Hg in the TP is still uncertain. Our records of Hg and Hg isotopes in sediment profiles of the two largest lakes in the TP, Lake Qinghai and Nam Co, show increased Hg influx since last century, with the maximum Hg influx enrichment ratios of 5.4 and 3.5 in Lake Qinghai and Nam Co, respectively. Shifts in negative δ 202Hg in Lake Qinghai (−4.55 to −3.15‰) and Nam Co (−5.04 to −2.16‰) indicate increased atmospheric Hg deposition through rainfall, vegetation and runoff of soils. Mass independent fractionation of both even-Hg (∆ 200Hg: +0.05 to +0.10‰) and odd-Hg (∆ 199Hg: +0.12 to +0.31‰) isotopes were observed. Positive Δ 200Hg suggest high proportion of precipitation-derived Hg in the TP, whereas the positive Δ 199Hg results from Hg(II) photo-reduction. Both lakes show increasing Δ 199Hg since the 1900 s, and we conclude that with the decrease of ice duration, Hg(II) photo-reduction may have been accelerated in these TP lakes.

Mercury has seven natural stable isotopes (196, 198, 199, 200, 201, 202 and 204), and our understanding of environmental fate of Hg has been enhanced by recent application of Hg isotope geochemistry. In the environment, Hg can undergo both mass dependent fractionation (MDF) and mass independent fractionation (MIF). MDF occurs during a variety of chemical, physical, and biological processes, and has been used to better understand the processes controlling Hg transport, transformation and bioaccumulation 15,16 . MIF signatures can provide a unique fingerprint of specific chemical pathways, such as photochemical reactions 17,18 . Large variations of both Hg-MDF and -MIF signatures have been documented in different environmental compartments 16 , and can provide multi-dimensional information to identify the sources and better understand biogeochemical Hg cycling 15 .
Sediment profiles coupled with high resolution dating (e.g., 210 Pb and 137 Cs) have been broadly used to evaluate historical changes of Hg deposition rate 5 . As the "water tower of Asia", the TP provides an ideal site to reconstruct environmental changes due to its sensitivity to environmental change and the lack of local pollution sources 13,14 . In this study, sediment profiles collected from two of the largest lakes in the TP (Lake Qinghai and Nam Co) were age-dated and analyzed for total Hg concentration (THg) and Hg isotopic composition. The objectives of this study were (1) to elucidate the history of Hg influx and source changes in the TP, and (2) to investigate the influence of global change on the biogeochemical cycle of Hg in this fragile alpine ecosystem.

Experimental section
Study area and sampling. Lake Qinghai (3194 m), the largest lake (4382 km 2 ) in the TP, is located in the northeast of the plateau. Nam Co (4730 m), the second largest lake (1920 km 2 ) in the TP, is centrally located (Fig. 1). Lake Qinghai is fed from a catchment of ~29,660 km 2 , and Nam Co has a catchment area of ~15,000 km 2 . The mean annual precipitations in Lake Qinghai and Nam Co are 357 and 414 mm [19][20][21][22] . The present day climate in both lakes is influenced by the Asian monsoon with dry winters, and precipitation mainly occurring in the summer season. The glaciated area of the catchment of Lake Qinghai and Nam Co is ~10 km 2 and 197 km 2 , respectively, accounting for 0.03% and 1.31% of the catchment. Hence, hydrologic sources to both lakes mainly consist of precipitation, not glacial melt 14 .
Sediment cores were taken from the deepest regions of Lake Qinghai (depth: 25.3 m) in 2006 and Nam Co (depth: ~60 m) in 2009 using HTH gravity corers. The Lake Qinghai core was sectioned in the field using a stainless steel slicer at 0.5 cm intervals from the surface to 5 cm, and then at 1.0 cm intervals to the base of the core. The Nam Co core (21 cm) was sectioned using a stainless steel slicer at 0.5 cm intervals from the surface to the base of the core. Samples were freeze-dried and homogenized prior to 210 Pb dating, total organic contents (TOC), THg and Hg isotope measurements. Sedimentation rate and TOC methodologies have been reported by Lami et al. 20 and Li et al. 22 .
Hg concentration analysis. THg in sediments was analyzed by direct combustion and atomic absorbance detection based on Lepak et al. 23 at the USGS Wisconsin Mercury Research Lab. SRM (IAEA SL 1) recoveries were within 90~110%, and coefficients of variation of triplicate analyses were less than 10%.
Mercury isotopic composition analysis. Based on the measured THg concentration [Table S1 of Hg-MIF is reported in Δ notation (Δ xxx Hg) and describes the deviation from mass dependency in units of permil (‰). MIF is the difference between the measured δ xxx Hg and the theoretically predicted δ xxx Hg value using the following formula 24 .

Results and Discussion
Mercury concentration profiles. Historical sediment profiles from both Lake Qinghai and Nam Co show a general trend of increasing THg over the past century ( Fig. 2A). THg in sediment provides insight into pollution status, however, influx rates of Hg (sedimentation rate × THg) provide the best estimates of inputs of Hg to lake (Fig. 2B). Preindustrial influxes of Hg in Lake Qinghai and Nam Co are about 3.1 and 5.7 ng cm −2 yr −1 , respectively, with the highest Hg influxes in Lake Qinghai and Nam Co at 16.5 and 20.3 ng cm −2 yr −1 . Mercury influxes among remote lakes have shown to be positively correlated to ratios of terrestrial catchment area (A C ) to lake area (A L ) 28 . Nam Co has an A C /A L of 7.8 higher than Lake Qinghai (A C /A L = 6.7). Mercury influx profile shifts are more clearly evident by calculating influx ratios (influx sample /influx background , influxes of each sediment with respect to the geochemical background) (Fig. 2C). Influx sample and influx background were the Hg influx of a given sample and the deepest sediment sample in each core, respectively. The maximum influx ratios of Hg in the 21st century are about 5.4 in Lake Qinghai and 3.5 in Nam Co, consistent with other studies of remote lake sediment cores, where Hg influxes have increased by a factor of 3 to 5 compared to the pre-industrial values 5,29 . Both profiles indicate increased Hg deposition starting from the early 1900 s, with especially intensive increases since the 1960 s ( Fig. 2 A-C). This is in agreement with the increased enhanced global Hg emission (especially China and India) and atmospheric Hg concentrations during the last few decades 5−6,29 .
The increase of Hg influx in Lake Qinghai and Nam Co is also likely synchronous with the rising global temperature, which starts in the early 1900 s, and has accelerated since the 1960 s ( Fig. 2D) 30 . Temperature increase in the TP is twice as high as the global average from 1957 to 2012 (0.036 ± 0.003 °C yr −1 ) (Fig. 2E). This has not only caused increased precipitation at an average annual rate of 10.9 mm per decade from 1961 to 2008, but also resulted in continuous increases of growing season (~1.04 day y −1 ) 31 and vegetation coverage (3 961.9 km −2 yr −1 ) 32 during the past 2 to 3 decades. Precipitation and vegetation (litterfall) are efficiently scavengers of atmospheric Hg 2,4 . Increased precipitation have also caused lake expansion and enhanced soil erosion in the TP 14 . For instance, Nam Co expanded by 20.2% in area between 1976 and 2010, and an average depth increase of 0.11 m −1 yr −1 was observed in Lake Qinghai in recent years 14 . Enhanced soil erosion in Lake Qinghai and Nam Co during the past few decades has been verified by inert tracers (such as Ti, Ni, Al, Fe, etc) [33][34][35] . Precipitation and vegetation (litterfall) are important inputs of Hg to pristine soils. The TP is mostly covered by typical alpine meadow and steppe 31 . Increase of plant production in the TP resulted in the increase of soil organic carbon density (0.1 g C m −2 yr −1 ) during 1981 to 2010 36 . Organic matter has a strong affinity for Hg 37 . Runoff of organic soil particles may effectively capture vegetation and precipitation-derived Hg from soils and the water column, and ultimately sequester it into sediments [37][38] . A recent study observed that organic matter (OM), in sediments of Lake Qinghai is primarily (80%) of terrestrial origin 39 . Significant linear correlations between THg and TOC (P < 0.01, ANOVA test) were observed in Lake Qinghai and Nam Co (Fig. 3). Overall, we suggest that increased anthropogenic Hg emission, enhanced atmospheric Hg deposition (through precipitation and vegetation) and soil erosion, may result in the increased Hg accumulation in the TP lakes.

Mass dependent fractionation of mercury isotopes. Sediments from Lakes Qinghai and Nam Co
showed highly variable δ 202 Hg values, ranging from − 4.55 to − 3.15‰ and from − 5.04 to − 2.16‰, respectively (Fig. 2G), which are much lower than previously reported data for industrial point Hg sources 16,[25][26][27][40][41][42][43][44][45] (δ 202 Hg: − 1.5 to 0‰), consistent with the fact that Lakes Qinghai and Nam Co are less impact by local point sources. Our data are more similar to sediments collected from pristine regions (δ 202 Hg: − 2 to − 3‰) 42,45 , which mainly receive Hg from atmospheric deposition. Previous studies have reported much higher δ 202 Hg (0 to + 1.0‰) in Hg 0 g collected from pristine sites 46 in comparison with that collected from urban-industrial regions (δ 202 Hg: − 3 to − 0‰) [47][48][49] . This indicates that Hg with lower δ 202 Hg values may be preferentially removed during long range transport and deposition through precipitation and litterfall. Indeed, precipitations in Northern America have shown more negative δ 202 Hg (− 4 to 0‰) than that of Hg 0 g (δ 202 Hg: − 0.5 to + 1.0‰) in the same areas 46,49-51 . For instance, precipitations from urban-industrial regions have shown highly negative δ 202 Hg values of down to − 4.27‰ in China 52 and − 4.37‰ in Florida 50 , being similar to our data of TP sediments. Due to close to China and India, it is possible for such highly fractionated rain contributions to the TP. However, our knowledge about the Hg isotope signatures in precipitation of the TP is limited to one single precipitation event, representing large variabilities 53 . Previous studies also reported negative δ 202 Hg of − 4 to − 1‰ for plants, demonstrating that lighter Hg isotopes are preferentially binding within the foliage 46,47,54 . Increased atmospheric Hg deposition through rainfall and litterfall have caused soils in montane regions to have much negative δ 202 Hg values 2 . Our observation of negative δ 202 Hg values is consistent with the fact that atmospheric deposition is the main input of Hg to Lake Qinghai and Nam Co 5 .
In addition to the source effect, post-depositional processes in the water column may also affect δ 202 Hg in sediments. Mercury deposited into lakes can be re-emitted to the atmosphere, while the remaining fraction is adsorbed on and settled by sediment particulate. The product Hg(0) during volatilization, microbial reduction and photoreduction processes could result in more negative δ 202 Hg values and, likewise, the residual Hg in the water column will result in a more positive δ 202 Hg values 17,18,[55][56][57] . Adsorption of aqueous Hg(II) by sediment particles containing thiol groups 58 , goethite 59 and sulfides 60 is likely cause negative shifts of δ 202 Hg (− 0.60‰) in  19 and Ke et al. 21 ) in Lake Qinghai and Nam Co. the solid phase. However, significant shift of δ 202 Hg may only occur when very small fraction of Hg is adsorbed relative to the total Hg in a system. Given the fact that particulate Hg is the dominate form of total Hg in Nam Co (86.7%) and other TP lakes 61,62 , we would not expect a significant negative shift of δ 202 Hg during adsorption of aqueous Hg(II) by sediments.
Like the THg profiles, δ 202 Hg generally increases from the deep part to the surface layer of the two cores (Fig. 2G). This pattern is similar to sediment cores collected near anthropogenic Hg point sources, where increased inputs of anthropogenic Hg with δ 202 Hg ranges from − 1 to 0‰ have shown in upper layer sediments 25,43,44 . It is unclear whether the increase of δ 202 Hg in this study is the result of global anthropogenic Hg input, however, due to the sparse population and industrial activities within the TP, local point sources may not explain the significant δ 202 Hg increase upcore. The shift of δ 202 Hg may be explained by a combined effect of enhanced precipitation, net primary production and soil erosion, all of which could incorporate more isotopic heavier Hg 0 g into waters, soils and sediments. Significant correlations between δ 202 Hg and THg with TOC were observed in Nam Co (P < 0.01, ANOVA test), when compared to that in Lake Qinghai (Fig. 4A,B). This suggests that the shifts of δ 202 Hg in Nam Co are more influenced by input of precipitation and vegetation derived Hg,  and runoff of organic soils, as supported by the smaller lake area (1920 km 2 ) of Nam Co. The lower correlation between δ 202 Hg and THg (P > 0.05) with TOC (P > 0.05) in Lake Qinghai may indicate sediments more influenced by lake processes, with a the much large lake area (4382 km 2 ).

Mass independent fractionation of 200
Hg. In this study, small but detectable MIF of 200 Hg was found in Lake Qinghai (∆ 200 Hg: + 0.07 to + 0.10‰) and Nam Co (∆ 200 Hg: + 0.05 to + 0.08‰) (Fig. 2H). When compared to the analytical uncertainty for ∆ 200 Hg ( ± 0.03‰), these results are considered significant. ∆ 200 Hg values of sediments from both Lake Qinghai and Nam Co were also significantly higher (P < 0.01, T-test) than UM-Almadén. The mechanism for MIF of 200 Hg is still unclear; however, prior studies have suggested that 200 Hg MIF is likely linked to photo-initiated Hg 0 g oxidation 49,51 . Significant 200 Hg MIF has been reported in atmospheric Hg samples 46,[48][49][50][51]  If precipitation ∆ 200 Hg signature did not change over time in the TP, increases of ∆ 200 Hg in sediment profiles result from enhanced precipitation Hg inputs are expected. However, we observed the consistent ∆ 200 Hg pattern in both Lake Qinghai and Nam Co (Fig. 2H). The lack of increased ∆ 200 Hg with elevated precipitation rates in this study may be explained by the isotope dilution of ∆ 200 Hg by other sources. The magnitude of ∆ 200 Hg in precipitation have shown to decrease from pristine to urban-industrial regions [49][50][51] , suggesting the dilution by industrial Hg (∆ 200 Hg: ~0‰) 16,26,27,40 . As mentioned earlier, enhanced input of soil-and vegetation-derived Hg with negative to zero ∆ 200 Hg, may also lessen the increase of ∆ 200 Hg in sediments. However, due to the lack of Hg isotope data in precipitation and soils throughout time, assessment of Hg contributions from precipitation and soil erosion was not performed in our study.

Mass independent fractionation of 199 Hg and 201
Hg. Positive MIF of odd Hg isotopes ( 199 Hg and 201 Hg) was observed in sediment of both lakes (Fig. 2I). The ∆ 199 Hg values in Lake Qinghai and Nam Co range from + 0.19 to + 0.31‰ and + 0.12 to + 0.28‰, respectively. There are two known possible mechanisms for odd-MIF: the nuclear volume effect (NVE) 63 and the magnetic isotope effect (MIE) 17 . Laboratory experiments demonstrated that NVE can be caused during elemental Hg 0 volatilization 64 , equilibrium Hg-thiol complexation 58 and dark Hg(II) reduction 57 with ∆ 199 Hg/∆ 201 Hg of ~1.6. Effects on MIE are mainly due to the photoreactions of aqueous Hg species in the presence of dissolved organic carbon (DOC), resulting in ∆ 199 Hg/∆ 201 Hg of 1.00~1. 30 17,18 . This is comparable with the observed ∆ 199 Hg/∆ 201 Hg ratio (1.07 ± 0.07, 2σ ) in all sediments investigated in this study, suggesting that aqueous Hg(II) photo-reduction is the possible process to cause MIF of Hg isotopes (Fig. 5).
The positive ∆ 199 Hg in TP lake sediments is different from previous data on sediments collected from industrial-urban regions, which mainly have negative to zero ∆ 199 Hg 27,41,[43][44][45] . Industrial Hg sources have shown average ∆ 199 Hg close to zero 16,40,41,43 , and continental soils and vegetation mainly showed negative ∆ 199 Hg values (− 0.5 to 0‰) 2,26,46,47,54 . The positive ∆ 199 Hg of the TP sediments may be explained by the inputs of Hg with positive ∆ 199 Hg or Hg(II) photoreduction in the water column, or both. Positive ∆ 199 Hg values (0 to + 1.0‰) have been reported for precipitation collected from different sites of the world [49][50][51][52][53] . Interestingly, sediment cores in this study reflect a shift of + 0.1‰ in Δ 199 Hg values since the early 1900 s (Fig. 2I), three times higher than our analytical uncertainty for UM-Almadén (Δ 199 Hg: ± 0.03‰, σ ). Increased precipitation tends to cause rise of ∆ 199 Hg in sediments, however, it also results in more input of vegetation-and soil-derived Hg (with negative ∆ 199 Hg) to lakes, likely to lessen the increase of ∆ 199 Hg in sediment profiles. Like the consistent ∆ 200 Hg profiles, we would not expect a significant shift of ∆ 199 Hg due to enhanced inputs of vegetation-and soil-derived Hg.
In this study, increases of Δ 199 Hg are more likely the result from enhanced Hg(II) photoreduction in the lake water column before incorporation into sediments. Photoreduction of Hg is largely controlled by solar irradiation and water conditions 17,18,42,65 . Long-term observation demonstrated no clear patterns on solar irradiation in the TP 66 . The increased Δ 199 Hg patterns in both lakes (Fig. 2I) show similar patterns with temperature rising (Fig. 2D,E). Positive relations between Δ 199 Hg and rising temperatures were observed (Fig. S1 of SI). Temperature rising have caused decreases of ice cover in the TP lakes, which can lead to greater exposure to sunlight for increased photochemical activity 9,10 . Ice cover should play an important role in controlling Hg(II) photoreduction in the TP lakes, considering the long-term of the ice duration in the TP lakes. Due to rising temperature, ice duration in the TP lakes have declined (Fig. 2J). Negative linear correlations between temperature and time of ice duration in 36 lakes in the TP has been also observed 13 . We speculate that the thickness of the lake ice would also decline along with rising temperatures, causing more water to be exposed to sunlight.
Positive relationships between ∆ 199 Hg and δ 202 Hg (Fig. 6A) were observed in Lake Qinghai (P < 0.01) and Nam Co (P < 0.01). Laboratory experiments on Hg(II) photoreduction also revealed positive relations between ∆ 199 Hg and δ 202 Hg with a δ 202 Hg/∆ 199 Hg of 0.83 16 , which is much smaller than that observed for Lake Qinghai (δ 202 Hg/∆ 199 Hg = 8.88) and Nam Co (δ 202 Hg/∆ 199 Hg = 5.75). This suggests that Hg(II) photoreduction may not be the main cause of the positive shifts of δ 202 Hg in the TP lakes. The positive relations between ∆ 199 Hg and δ 202 Hg indicate that enhanced Hg(II) photoreduction and δ 202 Hg shifts are induced by similar reasons, possibly due to the temperature effect. Indeed, warming not only causes decrease of ice duration which leads to higher Hg(II) photoreduction, but also results in higher influxes of atmospheric Hg (with higher δ 202 Hg values) though rainfall and soil erosion into the lakes. Our assumption has been supported by significant positive relations between ∆ 199 Hg and THg (P < 0.01) (Fig. 6B), and TOC (P < 0.01) (Fig. 6C) in Nam Co. Like δ 202 Hg, we also observed less correlation between ∆ 199 Hg and THg (P > 0.05), and TOC (P > 0.05) in Lake Qinghai, which indicates that Lake Qinghai may be more influenced by in-lake processes. Further research on water column Hg processes of Lake Qinghai are needed to better understand the variations of Hg isotopes in this study. Environmental implications. Alpine regions function as important convergence zones for atmospheric Hg, and have a rapid response to environmental change. Environmental changes such as enhanced precipitation, higher terrestrial plant biomass, and erosion of soils, may result in greater atmospheric Hg deposition and transport of historically deposited legacy Hg into the lakes of the in the TP. Dramatic lake ice cover reduction in TP may lead to increased Hg(II) photoreduction and evasion of Hg 0 (g) . The results of this study suggest that environmental change signals can be seen in the Hg isotopic distribution in the TP lake sediments. It should be mentioned that increased precipitation and glacier shrink have resulted in lake expansion and flooding of organic soil horizons 67 , which may affect the food web structures, Hg methylation and demethylation rates, and Hg fluxes on sediment-water-atmosphere interfaces of the TP lakes. Further studies are therefore needed.