Coupled and decoupled carbon-mercury dynamics across forest ecosystem continua

Abstract

Carbon sustains life, whereas mercury is a global toxin, yet their cycling in forests appears to be intimately linked. Here we show, using elemental stoichiometry, carbon and mercury isotopes and a global forest synthesis, that forests simultaneously couple and decouple mercury from carbon along contrasting ecosystem continua. Mercury/carbon ratios remain tightly conserved (0.5–0.9 × 10⁻⁶) along the aqueous-phase continuum, indicating proportional mercury transport with dissolved organic carbon. In comparison, mercury/carbon ratios increase by nearly three orders of magnitude from the atmosphere to soils (0.007–3.6 × 10⁻⁶) along the solid-phase continuum, reflecting progressive mercury enrichment during litter and soil organic matter decomposition. Standing litter acts concurrently as a net carbon source and mercury sink, whereas biomass regulates coupled carbon and mercury storage and litterfall deposition. These contrasting carbon-mercury trajectories reveal how forests both retain and redistribute atmospheric mercury and provide a conceptual framework for understanding terrestrial mercury cycling under environmental change.

Forests regulate both carbon storage and mercury pollution. This study shows that carbon and mercury remain coupled in water but progressively decouple in soils, providing a framework to explain terrestrial mercury cycling under environmental change.

Introduction

Forests are often referred to as the lungs of the Earth, with their canopies essential for photosynthesis and respiration, maintaining the balance between atmospheric oxygen (O₂) and carbon dioxide (CO₂)^1,2. Since the Industrial Revolution and the associated rise in anthropogenic CO₂ emissions, forests have acted as an indispensable buffer against the greenhouse effect and associated climate warming³. Today, global forest gross primary production (GPP) is generally estimated at 50–70 Pg C yr^–1, accounting for ~40–60% of total global GPP^2,4.

Mercury (Hg), a persistent global atmospheric pollutant with an atmospheric lifetime of ~0.5–1.5 years for gaseous elemental form (Hg⁰), can be transported over long distances to regions as remote as the Arctic and Antarctic^5–7. Forest ecosystems efficiently capture atmospheric Hg, particularly through foliage uptake, thereby contributing substantially to global atmospheric Hg deposition^8,9. Globally, vegetation is estimated to assimilate 1300–3000 Mg Hg yr⁻¹ from the atmosphere, accounting for around one-third of total global Hg deposition^10–15.

In nature, Hg strongly complexes with natural organic matter^16,17, thereby linking its occurrence and biogeochemical behaviour with C across aqueous, solid and gaseous phases in the hydrosphere, pedosphere and atmosphere. At the global scale, Hg concentrations in surface soils correlate strongly with soil organic C^18–20. At the catchment scale, Hg export in runoff is commonly associated with dissolved organic carbon (DOC) concentrations^21,22. Vertical Hg migration through soil profiles also occurs predominantly via DOC-Hg complexes, which may subsequently become immobilised following adsorption onto mineral surfaces in deeper soil horizons^23,24. Similarly, Hg in precipitation, cloud water and fog frequently exhibits relatively constrained Hg/DOC ratios^22,25–27. Organic C in atmospheric particulates is also closely linked to particulate-bound Hg concentrations and influences Hg photochemical reduction processes^28,29.

Forests, as major terrestrial reservoirs for both C and Hg, play a central role in regulating the biogeochemical cycling of both elements. Understanding whether the transport and transformation processes of C and Hg are synchronised across forest ecosystems is thus important for improving our understanding of terrestrial C and Hg budgets and their responses to environmental change. Previous studies have reported coupling between canopy uptake of atmospheric CO₂ and Hg^010,11,30, suggesting that the large atmosphere-vegetation C exchange associated with forest productivity may also play an important role in ecosystem Hg uptake. Investigations of forest soils reveal that although soil C loss is frequently accompanied by Hg evasion, C generally turns over more rapidly than Hg²⁰, implying potential decoupling of C and Hg cycling during soil organic matter decomposition. Together, these observations imply that the strength and direction of C-Hg interactions may vary among forest compartments. However, most previous studies have focused on individual ecosystem compartments, such as the atmosphere, soils or aquatic systems, resulting in a fragmented understanding of C-Hg relationships across interconnected forest continua, including the atmosphere-vegetation-litterfall-soil pathway.

Mercury exhibits a unique three-dimensional isotopic fractionation system comprising mass-dependent fractionation (MDF; expressed as δ²⁰²Hg) as well as odd- (odd-MIF; expressed as Δ¹⁹⁹Hg and Δ²⁰¹Hg) and even-mass independent fractionation (even-MIF; expressed as Δ²⁰⁰Hg and Δ²⁰⁴Hg). This characteristic makes Hg isotopes a powerful tool for tracing Hg sources and investigating Hg cycling processes in forest ecosystems^9,31. Carbon possesses two stable isotopes (¹²C and ¹³C), and variations in C isotopic composition arising from MDF during physicochemical and biological processes are expressed as δ¹³C. Such isotopic variations provide sensitive indicators of C sources, turnover, and organic matter transformation in forest environments³². The combined application of Hg and C isotopes offers an integrative approach for investigating coupled and decoupled C-Hg cycling processes across forest ecosystem compartments.

Here, we utilised a Swiss subalpine forest as a model system to examine the elemental and isotopic geochemical behaviour of C and Hg along an integrated atmosphere-phyllosphere-pedosphere-hydrosphere transect. We systematically (1) characterised seasonal variations in atmospheric CO₂ and Hg⁰ concentrations and isotopic compositions; (2) tracked changes in C and Hg contents and isotopic signatures in living canopy biomass, litterfall and soils; (3) determined Hg and C concentrations and Hg isotopic compositions across the forest aqueous-phase continuum; and (4) quantified canopy C and Hg storage in biomass together with litterfall deposition fluxes. In parallel, we compiled and synthesised datasets of C and Hg concentrations, isotopic signatures and surface exchange fluxes from global forest ecosystems (Suppl. Fig. S1) to place the site-based observations into a broader context and evaluate whether similar patterns emerge across forest ecosystems. By integrating elemental stoichiometry, dual-isotope tracing and global literature synthesis, this study aims to identify coupled and decoupled C-Hg cycling processes across forest ecosystem continua, provide mechanistic insights into their controls and derive empirically constrained Hg/C coupling and decoupling coefficients that may help inform future predictions of terrestrial Hg cycling under environmental change.

Results and Discussion

Contrasting C-Hg coupling and decoupling across solid- and aqueous-phase continua in forest ecosystems

The Hg/C ratios and isotopic compositions of C and Hg across major forest compartments, including the atmosphere, living canopy biomass, litterfall, soil and the aqueous-phase continuum, are summarised in Fig. 1. Observations from the Davos-Seehornwald forest reveal two contrasting modes of C-Hg relationships along ecosystem continua: (1) a solid-phase continuum comprising the vegetation-litterfall-soil pathway, and (2) an aqueous-phase continuum spanning precipitation, throughfall, runoff and soil porewater. Here, ‘continuum’ refers to interconnected transport and transformation pathways linking multiple forest ecosystem compartments. These continua exhibit fundamentally different patterns of C-Hg coupling and decoupling across forest compartments.

Fig. 1. Coupled and decoupled C-Hg dynamics across forest ecosystem continua.

Overview of (a) Hg/C ratios and isotopic signatures of (b) C and (c–e) Hg across major forest compartments in Davos-Seehornwald (circles with black outlines) and other forest ecosystems reported in the literature (circles without outlines). Literature data sources are summarised in Suppl. Text S1. Circle colours denote forest compartments: blue, atmosphere; green, living canopy biomass (LV); orange, litterfall (LF); grey, soil; and pink, forest water samples. Numbers beneath the circles without outlines indicate the number of literature-derived data points. Because δ²⁰²Hg, Δ¹⁹⁹Hg, and Δ²⁰⁰Hg values are commonly reported together, the numbers shown in panel (c) also apply to the corresponding data presented in panels (d, e). Error bars represent ± 1 standard deviation of replicate samples. For visual comparison, atmospheric δ¹³C-CO₂ values are plotted at three times their original values. Spr spring, Sum summer, Aut autumn, Win winter, O organic soil, M mineral soil.

In Davos-Seehornwald, progressive decoupling of C and Hg along the solid-phase continuum is reflected by an order-of-magnitude increase in Hg/C ratios from the atmosphere (gaseous phase; Hg/C = 0.007 × 10⁻⁶) to soils (Hg/C = 1.0–3.65 × 10⁻⁶). In contrast, the aqueous-phase continuum is characterised by relatively constrained Hg/C ratios in forest waters (Hg/C = 0.49–1.36 × 10⁻⁶), indicating persistent C-Hg coupling during hydrological transport. Similar Hg/C patterns are also observed in the global literature synthesis (Fig. 1a; Suppl. Text S1), despite uneven spatial and ecosystem representation in currently available datasets.

The compiled isotope datasets further provide multiple lines of evidence for contrasting coupled and decoupled C-Hg cycling processes across forest continua. At the atmosphere-vegetation interface, distinct Hg/C ratios and isotopic signatures between atmospheric gaseous and particulate phases and living canopy biomass reflect differing origins and transfer pathways of C and Hg. Along the solid-phase continuum, the broadly similar MDF directions of C and Hg suggest partially shared fractionation behaviour during organic matter transformation, whereas progressively divergent isotope offsets from canopy biomass to mineral soils indicate increasing decoupling of C and Hg cycling during decomposition and soil accumulation. In comparison, within the aqueous-phase continuum, the MDF characteristics of C broadly correspond with both the MDF and MIF signatures of Hg, mirroring their coupled mobilisation and transport in dissolved organic matter-rich water.

Multiple isotopic and stoichiometric constraints on coupled canopy uptake of CO₂ and Hg⁰

The uptake of Hg⁰ by the forest canopy is coupled to C dynamics, as supported by multiple independent lines of concentration-, isotope- and Hg/C ratio-based evidence presented in this study. This coupling is reflected by (1) the similar annual patterns of atmospheric Hg⁰ and CO₂ concentrations and MDF signatures observed above the canopy; (2) the consistent negative shifts in C and Hg MDF signatures (δ¹³C and δ²⁰²Hg) in canopy biomass relative to the atmosphere; and (3) the close correspondence between Hg/C ratios in canopy uptake fluxes and those in the atmosphere.

Above-canopy Hg⁰ and CO₂ concentrations in Davos-Seehornwald reached 1.27 ± 0.13 ng m⁻³ and 400.7 ± 5.4 ppm during the dormant season, decreasing to 1.17 ± 0.17 ng m⁻³ and 388.1 ± 11.2 ppm during the growing season, respectively (Fig. 2a, b). Moreover, normalised difference vegetation index (NDVI) was significantly negatively correlated with monthly mean atmospheric Hg⁰ concentrations (R² = 0.46, p < 0.05), whereas a weaker negative relationship was observed for CO₂ (R² = 0.39, p = 0.052) (Fig. 2c, d). The seasonal amplitudes of atmospheric Hg⁰ and CO₂ also exhibited synchronised behaviour, becoming negative during the growing season under higher NDVI values (0.70–0.81) and positive during the dormant season under lower NDVI values (0.42–0.63) (Fig. 2e). These coherent seasonal patterns suggest that vegetation activity exerts a common control on atmospheric CO₂ and Hg⁰ exchange in the forest ecosystem. Our observations align with previous studies showing that seasonal Hg⁰ concentration and flux patterns closely track those of CO₂ at the ecosystem scale^10,11,30. The δ²⁰²Hg of atmospheric Hg⁰ exhibited a positive shift during the growing season and reached a maximum in August (Fig. 1c and Suppl. Table S6). This seasonal MDF pattern resembles the seasonal behaviour of atmospheric δ¹³C-CO₂ reported from multiple background forest sites^33,34. We further summarised five long-term forest observations showing significant positive shifts in δ¹³C-CO₂ during the growing season (Fig. 1b). The synchronised seasonal variations in atmospheric Hg⁰ and CO₂ were therefore reflected not only in concentration changes but also in their MDF signatures. Moreover, the δ²⁰²Hg and δ¹³C values of canopy biomass in Davos-Seehornwald and in previously reported forest datasets remain lower than those of atmospheric Hg⁰ and CO₂ (Fig. 1b, c), indicating preferential uptake of isotopically lighter atmospheric Hg and C by vegetation^35,36.

Fig. 2. Coupled seasonal dynamics of atmospheric CO₂ and Hg⁰.

a One-hour-resolution atmospheric Hg⁰ concentrations (red dots) and (b) half-hour-resolution atmospheric CO₂ concentrations (blue dots) measured in Davos-Seehornwald in 2019. Stepwise lines in panels (a) and (b) represent monthly mean Hg⁰ (red) and CO₂ (blue) concentrations, respectively. c Monthly normalised difference vegetation index (NDVI) values. d Linear relationships between monthly mean atmospheric Hg⁰ (red) or CO₂ (blue) concentrations and NDVI. e Seasonal amplitudes of atmospheric Hg⁰ and CO₂ concentrations, defined as the average difference between seasonal and annual mean concentrations (growing season: May-October; dormant season: November-April). Error bars represent ± 1 standard deviation.

Our three- and two-endmember isotopic mixing models based on Hg-MIF (Δ¹⁹⁹Hg-Δ²⁰⁰Hg) (Eqs. 4–8) indicate that atmospheric Hg⁰ contributed 88 ± 6% of total Hg in needles, 87 ± 6% in cones, 78 ± 8% in bark, 79 ± 8% in branches, and 67 ± 13% in lichens (Suppl. Fig. S4; Table S8). These estimates are subject to uncertainties associated with analytical precision and variability in endmember isotopic compositions. Monte Carlo simulations and sensitivity analyses further indicate that, although partial overlap among endmembers may influence source apportionment, atmospheric Hg⁰ remains the dominant Hg source to canopy biomass within the 95% confidence intervals (Suppl. Text S3; Fig. S8). These results indicate that Hg accumulated in canopy biomass in Davos-Seehornwald predominantly originates from atmospheric Hg⁰. Vegetation assimilates atmospheric CO₂ through photosynthesis while simultaneously taking up atmospheric Hg⁰ through leaf gas exchange processes^37,38. Enhanced vegetation activity during the growing season produces coupled seasonal behaviour in atmospheric Hg⁰ and CO₂ concentrations as well as corresponding MDF signatures of C and Hg. From an annual flux perspective, the canopy uptake flux ratio of Hg to C (0.010 × 10⁻⁶) is consistent with the Hg/C concentration ratio of atmospheric gaseous forms (0.007 × 10⁻⁶) in Davos-Seehornwald^39,40. This agreement indicates that canopy assimilation of atmospheric Hg and C occurs at Hg/C ratios comparable to those of their atmospheric gaseous pools, supporting coupled C-Hg uptake at the atmosphere-canopy interface. A comparable uptake relationship is also evident from the global literature synthesis (Suppl. Table S7).

Progressive decoupling of C and Hg along the forest solid-phase continuum

By characterising C and Hg concentrations and isotopic signatures in canopy biomass (living tissues and litterfall) and soils (organic and mineral soils), together with their exchange fluxes at the atmosphere-standing litter and atmosphere-soil interfaces, we provide multiple lines of evidence for progressive decoupling of C and Hg cycling along the vertical solid-phase continuum spanning both the canopy and forest floor. Specifically, whereas C undergoes continuous net loss during decomposition of canopy biomass and soil organic matter, Hg exhibits sustained net retention and accumulation.

Mercury concentrations in litterfall (10.53-426.52 ng g⁻¹, n = 24) were generally higher than those in living canopy biomass (3.69-49.38 ng g⁻¹, n = 12) (Fig. 3a and Suppl. Tables S3 and S4). Significant Hg enrichment was observed in litterfall needles, wood and lichens, with Hg concentrations approximately 28%, 1097% and 355% higher, respectively, than those of their living counterparts (all p < 0.05, t-test). In contrast, C contents in litterfall (41.14–48.59%, n = 24) were slightly lower than those in living canopy biomass (45.99–48.36%, n =12), corresponding to ∼6% loss of C (Suppl. Fig. S5a). The combined effects of Hg enrichment and C loss increased the Hg/C ratio from 0.05 × 10⁻⁶ in living canopy biomass to 0.41 × 10⁻⁶ in litterfall (Fig. 1a). Organic soils further exhibited ∼350% higher Hg concentrations and ∼40% lower C contents than litterfall needles, the dominant litterfall component, resulting in a further increase in Hg/C ratios from 0.41 × 10⁻⁶ to 1.0 × 10⁻⁶. Compared with organic soils, mineral soils exhibited losses of both Hg and C relative to the organic horizon, although C loss (90%) exceeded Hg loss (81%), contributing to a continued increase in Hg/C ratios to 3.65 × 10⁻⁶ (Fig. 1a). Collectively, from living canopy biomass to litterfall and subsequently to organic and mineral soils, the preferential retention of Hg relative to C progressively elevated Hg/C ratios along the solid-phase continuum. These observations indicate that Hg and C undergo increasingly decoupled biogeochemical cycling during vertical transfer through forest solid-phase compartments.

Fig. 3. Progressive decoupling of C and Hg along the forest solid-phase continuum.

a Total Hg concentrations (histograms) and C concentrations (point plots) in living canopy biomass (LV) and litterfall (LF) in Davos-Seehornwald. Litterfall Hg concentration ranges and soil Hg and C concentrations were previously reported in our earlier study^23,62. b Relationship between C/N ratios and Hg/C ratios. c Relationship between Hg concentrations and C contents. d Relationship between Hg concentrations and δ¹³C values. Hg isotopic compositions (e, δ²⁰²Hg versus Δ¹⁹⁹Hg; f, Δ¹⁹⁹Hg versus Δ²⁰⁰Hg) in living canopy biomass, litterfall, precipitation, atmospheric Hg⁰, organic soils, and bedrock in Davos-Seehornwald. The Hg isotopic data for atmospheric Hg⁰, organic soils, and living needles were previously reported in our earlier study³⁹. Inset rectangles in panel (f) represent the Hg isotopic range (mean ± 1 standard deviation) of reactive Hg (RM) reported by Fu et al.⁴⁷. Error bars in panel (a) represent ± 1 standard deviation of replicate samples. Error bars in panel (e, f) represent ± 2 standard deviations of the analytical measurements.

The observed Hg enrichment and C loss likely arise from two concurrent processes: (1) rapid C mineralisation during litter ageing and organic matter decomposition; and (2) continuous atmospheric Hg accumulation by standing litter and forest-floor litter. In this study, ‘standing litter’ refers specifically to marcescent canopy biomass retained on trees prior to detachment, including withered bark, branches, stems and lichens (Suppl. Fig. S3). Accordingly, standing litter is treated here as the transitional phase linking living canopy biomass and forest-floor litterfall. High C/N ratios were observed in fresh litterfall, whereas lower C/N ratios characterised more extensively decomposed litterfall and soils, supporting the use of C/N ratios as indicators of decomposition stage^20,41–43. Thus, C/N ratios were lowest in litterfall wood, fine litter and litterfall lichens (20.0–44.8, n = 15), similar to those of organic soils (20.0–27.4, n = 6), intermediate in litterfall needles and cones (68.1–144.6, n = 9), and highest in living biomass (91.9–268.8, n = 12) (p < 0.05, ANOVA test), reflecting limited decomposition in litterfall needles and cones but substantially more advanced degradation in litterfall wood and lichens (Suppl. Fig. S5c). A pronounced negative correlation (R² = 0.72, p < 0.001) between Hg/C and C/N ratios was further observed along the continuum from living canopy biomass to mineral soils (Fig. 3b), indicating that C is mineralised more rapidly than Hg during standing litter and soil organic matter decomposition. From an annual flux perspective, the Hg/C flux ratio of soil emissions in Davos-Seehornwald (0.021 × 10⁻⁶) was substantially lower than the Hg/C concentration ratio in organic soils (1.0 × 10⁻⁶)^39,44. This discrepancy indicates that C is released from soils to the atmosphere more rapidly than Hg during soil organic matter turnover, resulting in progressive Hg enrichment relative to C in soils. This observation further supports decoupled C-Hg exchange at the atmosphere-soil interface. A similar decoupling pattern is also evident from the global literature synthesis (Suppl. Table S7).

During decomposition of litterfall and soil organic matter, isotopically lighter C is preferentially oxidised to CO₂, resulting in progressively higher δ¹³C values from living canopy biomass (δ¹³C = − 31.04 ± 1.21‰, 1 SD, n = 12) to litterfall (δ¹³C = − 27.40 ± 1.58‰, 1 SD, n = 24), organic soils (δ¹³C = − 26.37 ± 0.50‰, 1 SD, n = 6), and mineral soils (δ¹³C = − 24.96 ± 0.56‰, 1 SD, n = 5) (Suppl. Tables S3 and S4). Although higher δ¹³C values in mineral soils could potentially also reflect geogenic C inputs, contributions from geogenic C at the Davos-Seehornwald site are likely limited because the dominant silicate bedrock lithologies contain negligible carbonate-derived carbon. We further observed a significant negative correlation between C content and Hg concentration (R² = 0.34, p < 0.001), together with a significant positive correlation between δ¹³C and Hg concentration (R² = 0.62, p < 0.001) along the solid-phase continuum in Davos-Seehornwald (Fig. 3c, d). These relationships indicate progressive C loss accompanied by relative Hg enrichment during vertical transfer through the forest solid-phase continuum. Based on the Rayleigh fractionation model (Eqs. 9 and 10), C loss during transitions from living canopy biomass to litterfall and from litterfall to organic soils was estimated to be ∼86% and ∼43%, respectively, under the assumptions of the model. These model-derived estimates agree well with results from the empirical standing-litter decomposition model (Suppl. Text S2). The empirical model further suggests that standing litter acts as a net C source (19.10 ± 0.85 g C m⁻² yr⁻¹) and a net Hg sink (3.22 ± 0.14 µg Hg m⁻² yr⁻¹) during decomposition (Fig. 6). These estimates are derived from an exponential decay framework parameterised using decomposition rates reported for Alpine forests and are therefore subject to uncertainties associated with model assumptions and regional parameterisation. Importantly, the inference of C-Hg decoupling along the solid-phase continuum is primarily supported by observed Hg/C enrichment patterns and isotope evidence, whereas the model-derived fluxes provide quantitative constraints on the magnitude of C loss and Hg accumulation. Despite these uncertainties, the directions of the modelled fluxes well match the observed Hg/C enrichment patterns along the solid-phase continuum. Consequently, Hg enrichment in standing litter and forest-floor material cannot be explained solely by C loss, but also likely reflects continued accumulation of external atmospheric Hg inputs, including Hg⁰ and Hg^II. This interpretation is further supported by litterbag and incubation experiments showing increasing Hg mass concurrent with declining C mass during litterfall decomposition^42,43. The relative importance of standing litter as a coupled C source and Hg sink is nevertheless expected to vary among forest types owing to differences in canopy structure, litter-retention characteristics and decomposition environments. Whether this standing-litter mechanism operates similarly across different forest ecosystems remains to be evaluated.

Fig. 6. Conceptual model of coupled and decoupled C-Hg cycling across forest ecosystem continua in Davos-Seehornwald.

The solid-phase continuum (left panel, green gradient) exhibits progressive decoupling. The widening green bar visualises the order-of-magnitude increase in Hg/C ratios (grey text, 0.007–3.6 × 10⁻⁶) from the atmosphere to mineral soils. In contrast, the aqueous-phase continuum (right panel, orange bar) demonstrates relatively constrained stoichiometric coupling. The consistent width of the orange bar highlights the comparatively conserved Hg/C ratios (0.5–0.9 × 10⁻⁶). Annual exchange fluxes of C and Hg between ecosystem interfaces are indicated by teal and orange arrows, respectively. Positive (+) values denote release fluxes, whereas negative (−) values denote deposition fluxes. Grey labels in parentheses indicate representative coupling or decoupling coefficients (× 10⁻⁶) associated with key C-Hg cycling processes. The coupling and decoupling coefficients shown represent empirically constrained first-order estimates within the proposed framework rather than universally applicable parameters.

Higher Δ²⁰⁰Hg values in litterfall wood, fine litter and litterfall lichens relative to living canopy biomass suggest enhanced capture of atmospheric Hg^II by standing litter, including Hg^II deposited through precipitation and dry deposition processes (Fig. 3f). In addition, litterfall wood, fine litter and litterfall lichens (Δ¹⁹⁹Hg = −0.56 ± 0.07‰, 1 SD, n = 15) exhibited more negative Δ¹⁹⁹Hg values than living canopy biomass (Δ¹⁹⁹Hg = −0.21 ± 0.05‰, 1 SD, n = 12), indicating that although standing litter acts as a net Hg sink, a fraction of S-bound Hg undergoes photoreduction⁴⁵. Mercury reduction and atmospheric Hg^II deposition may both contribute to positive shifts in δ²⁰²Hg values in litterfall^46,47. Nevertheless, no statistically significant difference in δ²⁰²Hg was observed between living canopy biomass and litterfall (p = 0.09, t-test) (Fig. 3e). This isotopic similarity suggests that continued uptake of isotopically lighter atmospheric Hg⁰ by standing litter may partially offset the positive MDF shifts associated with Hg reduction and Hg^II deposition. Collectively, the isotope evidence indicates that standing litter continues to receive newly deposited atmospheric Hg⁰ and Hg^II prior to transfer to the forest floor. Moreover, global observations of net atmospheric Hg⁰ deposition to forest floors together with net CO₂ emissions from forest soils further support continued external Hg uptake concurrent with progressive C loss during litterfall and soil organic matter decomposition^11,48.

Coupled transport of C and Hg in the forest aqueous-phase continuum

Here, we provide multiple lines of evidence that the transport and export of C and Hg during forest hydrological processes are tightly coupled. The dissolution and sorption of biogenic C derived from canopy biomass and soils regulate Hg mobilisation during hydrological transport, as reflected by coherent variations in Hg/DOC ratios and coupled evolution of C and Hg isotopic signatures along the aqueous-phase continuum.

At the Davos-Seehornwald site, both Hg and DOC concentrations exhibited a consistent decreasing trend across hydrological compartments: throughfall (34.4 ± 15.3 ng L⁻¹ for Hg; 29.9 ± 10.4 mg L⁻¹ for DOC, 1 SD, n = 10) > soil porewater (12.3 ± 7.1 ng L⁻¹; 23.8 ± 15.8 mg L⁻¹, 1 SD, n = 5) > runoff (5.5 ± 2.9 ng L⁻¹; 10.9 ± 5.1 mg L⁻¹, 1 SD, n = 6) > bulk precipitation (1.9 ± 1.1 ng L⁻¹; 2.5 ± 1.2 mg L⁻¹, 1 SD, n = 4) (Fig. 4a and Suppl. Table S5). This pattern indicates a coherent transport pathway for Hg and C through the forest hydrological continuum: precipitation becomes enriched in both Hg and DOC during canopy passage, whereas partial retention occurs following interaction with the forest floor. Significant positive correlations between Hg and DOC concentrations were observed within each hydrological compartment (R² = 0.49–0.97, all p < 0.01), indicating coupled mobilities of Hg and dissolved organic matter (Fig. 4a). Notably, the Hg/C ratio in precipitation (Hg/DOC = 1.4 × 10⁻⁶) exceeded those in throughfall (0.9 × 10⁻⁶), soil porewater (0.6 × 10⁻⁶) and runoff (0.6 × 10⁻⁶), indicating that atmospheric Hg and C sources dominate in precipitation, whereas biologically derived C and Hg from canopy biomass and litterfall become increasingly important within internal forest water pathways²⁵. The relatively constrained Hg/C ratios among throughfall, soil porewater and runoff further support persistent coupling of Hg and C during internal forest hydrological transport. Similar Hg/C relationships are also observed in forested regions globally (Fig. 1a). We acknowledge that sample sizes for some compartments, particularly precipitation and soil porewater, remain limited and that observations of the aqueous-phase continuum are primarily based on rainy-season sampling. These limitations introduce uncertainty and warrant cautious interpretation. However, the coherent Hg-DOC relationships observed across hydrological compartments indicate that the overall coupling patterns are robust.

Fig. 4. Coupled mobilisation of Hg and dissolved organic carbon (DOC) in forest waters.

a–d Relationship between DOC and Hg concentration, δ²⁰²Hg, Δ¹⁹⁹Hg, and Δ²⁰⁰Hg in water samples collected from Davos-Seehornwald, respectively. e and f Hg isotopic signatures in water samples, atmospheric Hg⁰, and organic soil in Davos-Seehornwald. The Hg isotopic signatures of atmospheric Hg⁰ and organic soil were reported in our earlier study³⁹. Lines in (a) represent linear regressions for different water samples in Davos-Seehornwald and literatures (the corresponding light-coloured circles without outlines)^22,25,49. Lines in panels (b–d) represent power regressions of water samples from Davos-Seehornwald. The light grey and light orange shading in panels (b–d) represents the Hg isotopic signatures of organic soils and litterfall in Davos-Seehornwald, respectively. Error bars in panel (e, f) represent ± 2 standard deviations of the analytical measurements.

Results from a binary mixing model (Eqs. 7 and 8) based on Δ²⁰⁰Hg indicate that atmospheric Hg⁰ contributes on average 45%, 65% and 22% of total Hg in throughfall, soil porewater and runoff, respectively (Suppl. Fig. S4; Table S8). This pattern suggests that previously deposited atmospheric Hg⁰, likely retained in canopy biomass and soils, is extensively dissolved and remobilised within the forest hydrological cycle. Sensitivity analyses indicate that variability in endmember isotopic compositions exerts only limited influence on estimated source contributions (Suppl. Text S3; Fig. S8). Throughfall exhibited lower δ²⁰²Hg (−0.97 ± 0.53‰), Δ¹⁹⁹Hg (−0.43 ± 0.05‰) and Δ²⁰⁰Hg (0.04 ± 0.02‰) values than precipitation, which showed corresponding values of 0.36 ± 0.77‰, 0.21 ± 0.14‰ and 0.11 ± 0.03‰, respectively (1 SD; throughfall n = 10, precipitation n = 4) (Fig. 4e, f). These isotope shifts cannot be explained solely by reactive Hg pools on canopy surfaces⁴⁷, but instead indicate substantial contributions from canopy-derived biogenic Hg pools. Previous studies similarly reported that 50-80% of Hg in throughfall originates from atmospheric Hg^{0 22,49}, although the specific canopy reservoirs contributing this Hg have remained uncertain. In our study, the similar MIF signatures between standing litter and throughfall suggest that dissolution of Hg associated with standing litter represents an important Hg source to throughfall (Fig. 4f). In comparison, living canopy tissues and litterfall needles and cones exhibit substantially higher odd-MIF signatures than throughfall. We acknowledge that this interpretation is primarily supported by isotope matching and mixing relationships, and that contributions from canopy surface Hg pools and living biomass cannot be fully excluded.

Litterfall contained substantially lower cellulose contents (5.7–17.9 µg g⁻¹) than living canopy tissues (13.7–34.0 µg g⁻¹) (p < 0.01, t-test; Suppl. Fig. S6). During standing litter decomposition, cellulose is rapidly degraded to CO₂ and low-molecular-weight organic compounds⁵⁰, contributing to elevated DOC concentrations in throughfall relative to precipitation¹⁶ (Fig. 4a). Simultaneously, degradation of biopolymers such as lignin and cellulose enhances the mobility of Hg bound to thiol and thioether functional groups^51–53. Consequently, Hg released from standing litter readily binds to dissolved organic matter (DOM), forming Hg-DOM complexes that are subsequently transported by throughfall during rainfall events. The ratio (0.54 × 10⁻⁶) between increased Hg flux and increased C flux in throughfall resembles the Hg/C ratio of standing litter (0.64 × 10⁻⁶), indicating that canopy-derived biogenic Hg and C are incorporated into throughfall through a coupled transport pathway (Fig. 6). The even-MIF signatures of soil porewater mirror those of litterfall and soils, suggesting desorption from litter- and soil-bound Hg as a dominant source. In contrast, runoff even-MIF signatures occur between those of precipitation and soils, indicating mixing between precipitation-derived and soil-derived Hg sources (Fig. 4f). A study from a montane forest in South America showed that precipitation exhibited significantly higher (more positive) δ¹³C-DOC values than throughfall, soil porewater and runoff, whereas the latter compartments displayed more negative δ¹³C values similar to those of global litterfall and organic soils⁵⁴ (Fig. 1b). This pattern implies a predominantly biogenic origin of DOC in throughfall, soil porewater and runoff. Intriguingly, the vertical patterns of Hg isotope compositions (δ²⁰²Hg, Δ¹⁹⁹Hg and Δ²⁰⁰Hg) in water samples from Davos-Seehornwald broadly mirror the δ¹³C patterns reported in the above study, suggesting coupled transport and transformation processes of Hg and C during forest hydrological cycling (Fig. 1b–e).

By examining relationships between DOC concentrations and Hg isotopic compositions across all water samples from Davos-Seehornwald, we further observed that samples with higher DOC concentrations tended to exhibit Hg isotopic signatures closer to those of litterfall and organic soils (Fig. 4b–d). Significant negative correlations were observed between DOC and δ²⁰²Hg (R² = 0.63, p < 0.001), Δ¹⁹⁹Hg (R² = 0.89, p < 0.001) and Δ²⁰⁰Hg (R² = 0.60, p < 0.001), indicating that increasing contributions of biogenic dissolved organic matter are accompanied by progressively stronger incorporation of litterfall- and soil-derived Hg into forest waters. Collectively, δ¹³C, Hg-MDF and Hg-MIF provide complementary process constraints on C transformation, Hg biogeochemical fractionation and atmospheric Hg sources, respectively, thereby strengthening the mechanistic interpretation of coupled and decoupled C-Hg cycling across forest ecosystem continua.

Biomass-mediated coupling of C and Hg storage in canopy biomass and litterfall deposition

Biomass exerts a dominant control on the coupled C and Hg storage in forest canopies, despite differences in their biochemical binding forms and internal transport processes within vegetation. The coupling coefficients of C-Hg deposition fluxes among different litterfall components closely correspond to their respective Hg/C ratios, indicating that biomass regulates the overall pattern of coupled C-Hg deposition via litterfall.

In Davos-Seehornwald, strong positive correlations were observed between biomass and both C and Hg pools in lichens, needles, bark, branches and stems (R² = 0.97, p < 0.05) (Fig. 5a). Canopy C and Hg storage are primarily determined by biomass accumulation rather than by elemental concentrations alone (Suppl. Fig. S7a, S7b). Total Hg pools were also positively correlated with C pools in canopy biomass (R² = 0.97, p < 0.05), a pattern similarly reported in another boreal forest⁵⁵ (Fig. 5b and Suppl. Table S2). Collectively, these results indicate that canopy biomass acts as an integrative reservoir linking C and Hg storage at the ecosystem scale.

Fig. 5. Biomass regulation of coupled C and Hg storage and litterfall deposition.

a Relationships between biomass and C pools (linear regression shown in light green) and between biomass and Hg pools (non-linear regression shown in green). b Relationship between canopy biomass Hg pools and C pools. c Relationships between Hg deposition fluxes and C deposition fluxes for different litterfall fractions based on individual sampling events. d Relationship between Hg deposition fluxes and C deposition fluxes for total litterfall (needles + cones + wood + fine litter + lichens) across different seasons from August 2018 to October 2019. Error bars represent ±1 standard deviation of replicate samples.

Within living canopy biomass, Hg/C ratios followed the pattern: needles (0.090 ×  10⁻⁶) > bark (0.055 × 10⁻⁶) > branches (0.013 × 10⁻⁶) (Fig. 1a). This pattern likely reflects the relatively uniform distribution of C among canopy tissues (Suppl. Fig. S5a), whereas Hg exhibits greater variability among canopy components owing to differences in Hg-binding functional groups and internal translocation processes^53,56. Nonetheless, the relationships between Hg pools and biomass and between Hg and C pools both followed similar logarithmic relationships (Fig. 5a, b). These similar relationships further suggest that biomass exerts a dominant control on coupled C and Hg storage within forest canopies despite tissue-specific differences in Hg accumulation. Beyond the observed statistical correlations, biomass likely contributes mechanistically to C-Hg coupling by simultaneously regulating canopy structural C storage, photosynthetic exchange capacity and atmospheric Hg interception^10,11,38,39. Larger biomass therefore provides both greater C reservoirs and increased surface area and retention capacity for atmospheric Hg accumulation. Nevertheless, the physiological and biochemical controls governing proportional Hg accumulation across different canopy tissues remain insufficiently constrained.

Mercury stored in canopy tissues is transferred to the forest floor through litterfall, resulting in Hg deposition fluxes that are more strongly related to litterfall biomass than to Hg concentrations alone (Suppl. Fig. S7c, S7d). Similarly, litterfall C fluxes closely tracked litterfall biomass (Suppl. Fig. S7e). Consequently, significant positive correlations were observed between litterfall Hg deposition fluxes and C deposition fluxes for individual litterfall components (R² = 0.76-0.91, all p < 0.01) (Fig. 5c), indicating generally coupled deposition of C and Hg during litterfall production. Comparable positive relationships were also observed when all litterfall components were combined, despite seasonal fluctuations in litterfall biomass (R² = 0.79, p < 0.01) (Fig. 5d).

The slopes of Hg versus C deposition fluxes (defined here as C-Hg coupling coefficients) differed among litterfall components and followed the sequence: cone (0.06 × 10⁻⁶) <needle (0.11 × 10⁻⁶) <wood (0.16 × 10⁻⁶) <fine litter (0.40 × 10⁻⁶) (Fig. 5c). Notably, these variations well corresponded to the Hg/C ratios of the respective litterfall components (Fig. 1a and Suppl. Table S4), indicating that Hg/C stoichiometry regulates the coupling coefficients of C-Hg litterfall deposition. By comparing coupling coefficients of individual litterfall components with those of total litterfall, we further observed that the coupling coefficient for total litterfall (0.15 × 10⁻⁶) resembled that of litterfall needles (Fig. 5c, d), although the overall coefficient was partially shifted towards litterfall components with elevated Hg/C ratios, including fine litter and litterfall wood. Because litterfall needles contribute the dominant proportion of total litterfall biomass, biomass ultimately exerts the strongest control on coupled C and Hg deposition at the ecosystem scale.

Implications for terrestrial Hg cycling

This study establishes a process-informed framework distinguishing coupled and decoupled C-Hg pathways across forest ecosystem continua, based on comprehensive observations from an Alpine coniferous forest combined with a global literature synthesis (Fig. 6). Our quantitative reconstruction of Hg/C evolutionary trajectories advances current understanding of terrestrial C-Hg relationships beyond qualitative concentration-based associations towards a more stoichiometrically and process-oriented framework. The synthesis-derived and empirically constrained C-Hg coupling and decoupling coefficients reported here (Suppl. Table S7) provide a process-informed empirical basis for developing process-informed representations of terrestrial Hg cycling in Earth system models (ESMs)⁵⁷.

First, C and Hg exhibit coupled behaviour in the aqueous-phase continuum and at the atmosphere-foliage interface. The relatively constrained aqueous-phase coupling coefficients derived from the literature synthesis (0.3–0.9 × 10⁻⁶) indicate that Hg is transported proportionally with dissolved organic C during forest hydrological transport. This relationship suggests that biologically mediated Hg export may be parameterised in relation to dissolved organic C mobilisation driven by standing litter decomposition and soil leaching processes, rather than represented solely as kinetic Hg desorption. Similarly, foliar uptake of atmospheric C and Hg is coupled, characterised by an Hg/C uptake flux coefficient of (0.009–0.013) × 10⁻⁶ derived from the literature synthesis (Suppl. Table S7). Applying this coefficient together with global vegetation GPP yields a synthesis-based estimate of global vegetation Hg⁰ uptake of approximately 1100–2100 Mg yr⁻¹, broadly consistent with previous independent assessments^12–14. This estimated range reflects uncertainty propagated from variability in both the Hg/C uptake coefficient and literature-derived estimates of global vegetation GPP.

Second, C and Hg progressively decouple along the solid-phase continuum, particularly during canopy litter and soil organic matter decomposition. The enrichment of Hg relative to C from canopy biomass to soils reflects preferential C mineralisation together with relative Hg retention and accumulation, quantified by a standing-litter decoupling coefficient of 0.17 × 10⁻⁶ in Davos-Seehornwald. In addition, the substantially lower Hg/C ratios of atmosphere-soil exchange fluxes (0.01-0.05 × 10⁻⁶ from the literature synthesis) relative to those of soils indicate differing turnover rates of C and Hg during soil organic matter decomposition. These observations suggest that incorporating variable C-Hg decoupling functions linked to litter decomposition and soil respiration processes may improve predictions of legacy Hg retention and re-emission under climate-driven acceleration of terrestrial C turnover.

Third, biomass regulates the coupled C and Hg storage and deposition, resulting in relatively constrained litterfall C-Hg deposition coefficients (0.07–0.11 × 10⁻⁶ from the literature synthesis). This coupled relationship may simplify representation of vegetation Hg reservoirs and litterfall Hg fluxes in terrestrial Hg models. For instance, applying these deposition coupling coefficients together with global forest litterfall C fluxes yields a synthesis-based estimate of global litterfall Hg deposition of ~850-1360 Mg yr⁻¹, broadly consistent with previous independent estimates^58–60. These estimates remain subject to uncertainties associated with forest ecosystem heterogeneity, vegetation-specific variability and uneven geographic representation of currently available datasets. Because atmospheric Hg⁰ concentrations, ecosystem productivity and hydrological transport processes exhibit pronounced seasonal variability, the coupling and decoupling coefficients derived in this study may also vary seasonally. Preliminary comparison of aqueous-phase Hg/C ratios between rainy and dry periods further suggests compartment-dependent seasonal variability, with more pronounced differences observed in throughfall and runoff than in precipitation (Suppl. Text S4 and Fig. S9). In particular, aqueous-phase observations were constrained by relatively limited sample numbers for some compartments and by sampling periods primarily associated with rainfall events. The currently reported coefficients therefore should be interpreted as integrated ecosystem-scale representations constrained by the available observational periods at Davos-Seehornwald rather than temporally invariant parameters.

Overall, our study establishes a foundational coupled-decoupled framework for representing forest ecosystem processes in terrestrial C-Hg ESMs and provides empirically constrained coefficients for several key ecosystem processes. Although the resulting estimates broadly align with previous independent assessments, this agreement alone does not imply universal transferability of the proposed coefficients. The framework presented here is primarily constrained by observations from an Alpine coniferous forest together with literature-derived synthesis datasets and therefore should be interpreted as a process-informed representation rather than a universally applicable parameterisation. Analysis of coefficient applicability (Suppl. Table S7) further highlights the need for additional empirical constraints across especially subtropical and tropical forests, where observational data remain limited. Further evaluation across diverse forest biomes and climatic regions will be essential to assess the broader applicability of these coefficients, which are expected to vary with climate, vegetation composition and decomposition conditions. As climate change accelerates forest C turnover and alters terrestrial Hg deposition and remobilisation, incorporating stoichiometrically constrained C-Hg interactions into terrestrial Hg models may improve predictions of legacy Hg mobilisation and future atmosphere-terrestrial Hg exchange.

Methods

Study site

The Davos-Seehornwald research site (46°48'55.2“N, 9°51'21.3“E) is situated in the middle of the subalpine belt in the eastern Swiss Alps, at an elevation of 1639 m a.s.l. (Suppl. Fig. S2). The site experiences an annual mean temperature of 4.4 °C and annual precipitation of 880 mm. The dominant tree species is the Norway spruce (Picea abies (L.) Karst.), with an average age of 119 years, a height of 17.5 m, and a leaf area index of ~4 m² m⁻² (Swiss FluxNet, https://www.swissfluxnet.ethz.ch). The lichen community is primarily composed of Pseudevernia furfuracea and Usnea subfloridana. Dwarf shrubs and mosses contribute the main components of the patchy ground vegetation, covering about 30% of the forest floor. The soil types are mainly classified as Humic Cambisol and Ferric Podzol, according to the World Reference Base for Soil Resources classification. The Davos-Seehornwald research site is a part of the Integrated Carbon Observation System (ICOS) initiative and the Long-term Forest Ecosystem Research (LWF), which constitutes Switzerland’s contribution to the International Co-operative Programme on Assessment and Monitoring of Air Pollution Effects on Forests (ICP Forests) Level II plot network.

Atmospheric Hg⁰ and CO₂ measurements

From February to December 2019, atmospheric Hg⁰ concentrations at 15 m height were continuously monitored using an automated Hg vapour analyser (Tekran® 2537X, Tekran Inc., Canada) installed in an instrument hut located at the base of the main tower. The sampling line consisted of two Teflon filters, a 1/4-inch Teflon tube and a desiccant tube. The Teflon filters and desiccant were replaced biweekly. The instrument was operated at a flow rate of 1 L min⁻¹. Automatic calibration was performed every 25 h using an internal Hg⁰ permeation source, which was manually calibrated twice annually using an external Hg vapour source (Tekran® 2505, Tekran Inc., Canada). CO₂ concentrations were measured using an enclosed-path infrared gas analyser (IRGA LI-7200; LI-COR, Lincoln, Nebraska, USA) at the top of the tower at a height of 30 m. The CO₂ concentration and flux data, originally collected at 20 Hz, were obtained from the European Fluxes Database Cluster as half-hourly averages (http://www.europe-fluxdata.eu/home). Further details of the CO₂ measurement methods are provided by Zielis, Etzold⁴⁰. The local normalised difference vegetation index (NDVI) for Davos-Seehornwald was obtained from global gridded NDVI datasets with a temporal resolution of 30 days and a spatial resolution of 0.1°. Based on the physiological activity of Davos-Seehornwald vegetation, the growing season extends from May to October, whereas the dormant season spans November to April.

Sampling, pre-concentration, and determination of Hg, C and N

Living canopy biomass (foliage, bark and branch) was collected from several randomly selected spruce trees in Davos-Seehornwald in October 2019. The canopy biomass data were investigated by LWF according to the Swiss National Forest Inventory method⁶¹. The detailed methodology of litterfall, bulk precipitation, throughfall, soil porewater, and soil samples collection has been described elsewhere^23,62. Briefly, twenty circular traps (ICOS-defined design) were installed 1 m above the ground for sampling litterfall from August 2018 to October 2019 in Davos-Seehornwald. Each trap had a collection area of 0.50 m², with a polyvinyl chloride ring attached to it, and a polyester mesh net with a mesh size of 0.25 mm was fixed to the ring. According to the ICP Forests and LWF Manual (Part XIII)⁶³, litterfall samples were sorted into five fractions: needles, cones, wood (branches, twigs and bark), fine litter and lichen. Litterfall biomass data were obtained from our previous study conducted at the Davos-Seehornwald site⁶². Corresponding Hg deposition fluxes were calculated from litterfall biomass and Hg concentrations measured for each litterfall component in this study. Before analysis, the living biomass and litterfall samples were oven-dried at 65 °C for 96 h. Soil samples were air-dried in a cleanroom and then weighed, ground, homogenised and stored in polyethylene zip-type bags. Two stainless steel rain collectors (1 m²) with Teflon coating were installed in the open field and three collectors below the canopy were used to retrieve bulk precipitation and throughfall, respectively. Precipitation samples were taken back to the lab for pre-treatment immediately after each rain event. Soil porewater samples were collected using a tension lysimeter with ceramic cups at four depths (0, 15, 50 and 80 cm). To measure the isotopic composition of atmospheric Hg⁰, ambient air at 18 m above the ground was collected using chlorine-impregnated activated carbon (ClC) traps on the main tower in 2019.

To preconcentrate Hg in precipitation, throughfall and soil porewater, about 5 L water samples were filtered using a 0.45 µm quartz membrane and then completely oxidised with bromine chloride (BrCl) for 24 h⁶⁴. Thereafter, stannous chloride (SnCl₂) was added to generate Hg⁰ and collected Hg⁰ by ClC traps. For Hg stable isotope analysis, living biomass, litterfall and ClC traps were combusted in a double-step oven system and pre-concentrated into 5 mL of 40% mixed acid solution (v/v, 2HNO₃/1HCl)⁶⁵. The total Hg concentration in the living biomass and litterfall samples was quantified using a direct Hg analyser (Milestone DMA-80, Italy). The recoveries of certified reference materials were 101 ± 1% for SRM 1575a (pine needles: 39.0 ± 0.7 ng g⁻¹) and 94 ± 5% for GSB-11 (quince leaves: 150 ± 25 ng g⁻¹), respectively. Total C and N concentrations in living biomass and litterfall samples were determined using an elemental analyser (vario MACRO cube, Elementar, Germany) with standard deviations of 0.06%. The lignin and cellulose contents of living biomass and litterfall were determined using the acid-detergent method. DOC concentrations in bulk precipitation, throughfall and soil porewater were performed with a total organic carbon analyser (Shimadzu, Japan).

Hg and C isotope analysis

The Hg isotopic compositions were characterised at the Institute of Geochemistry, Chinese Academy of Sciences, using a cold vapour multi-collector inductively coupled plasma mass spectrometer (MC-ICP-MS, Neptune Plus, Thermo Scientific, USA). Prior to analysis, the samples were diluted to match with bracketing standards NIST 3133 (1 ng Hg mL⁻¹) and then introduced into a cold vapour phase separator generation system coupled with 3% SnCl₂⁶⁶. The subsequently generated Hg⁰ vapour was carried with Argon into MC-ICP-MS. Internal NIST SRM 997 Tl standard solutions (30 ng mL⁻¹) and standard-sample-standard bracketing with NIST 3133 Hg standard solutions were employed for instrumental mass bias correction. NIST SRM 997 Tl standard was introduced into the plasma as aerosol particles generated by a CETAC Aridus II desolvating nebuliser system. All samples or Hg standards were analysed in one block with 50 cycles to improve internal precision. Between the standard-sample-standard analysis, the system was washed with 10% HNO₃ to reduce the signal intensity of the cold vapour system to the background level. Hg isotopic compositions are reported relative to NIST 3133 using the “δ” notation for MDF, in units of permil (‰), as follows:

$$\delta {}^{xxx}Hg\left(‰\right)=\left[{\left({}^{xxx}Hg/{}^{198}Hg\right)}_{sample}/{\left({}^{xxx}Hg/{}^{198}Hg\right)}_{NIST3133}-1\right]\times 1000$$ 1

where xxx is the mass number of the Hg isotopes (199, 200, 201 and 202). In this study, MIF values are reported by using “Δ” notation and calculated by the following equation⁶⁷:

$$\mathrm{\Delta }{}^{xxx}Hg\left(‰\right)=\delta {}^{xxx}Hg-\left(\beta \times \delta {}^{202}Hg\right)$$ 2

where xxx is the mass of each Hg isotope 199, 200 and 201. β is an equilibrium MDF factor, with the values of 0.2520, 0.5024 and 0.7520 for ¹⁹⁹Hg, ²⁰⁰Hg and ²⁰¹Hg, respectively. NIST 8610 was measured randomly for each analytical session to determine the analytical uncertainty of isotopic analysis. The isotopic compositions of an external Hg standard reference BCR 482 (Lichen, from European Community Bureau of Reference) were measured randomly over different analytical sessions to detect the isotopic bias that may occur during the pre-concentration protocol and the analytical uncertainty of isotopic analysis. The measured Hg isotopic compositions of NIST 8610 (δ²⁰²Hg = −0.54 ± 0.10‰, Δ¹⁹⁹Hg = −0.01 ± 0.06‰, Δ²⁰⁰Hg = 0.01 ± 0.07‰, Δ²⁰¹Hg = −0.01 ± 0.09‰, 2 SD, n = 12) and BCR 482 (δ²⁰²Hg = −1.57 ± 0.13‰, Δ¹⁹⁹Hg = −0.65 ± 0.05‰, Δ²⁰⁰Hg = 0.07 ± 0.07‰, Δ²⁰¹Hg = −0.62 ± 0.08‰, 2 SD, n = 8) were consistent with previously published values^41,68,69.

The Carbon isotope ratios of the samples were determined using an Isotope Ratio Mass Spectrometer (MAT 253, Thermo-Fisher). According to the procedure described by Ehleringer and Dawson⁷⁰, the δ¹³C value was expressed as:

$$\delta {}^{13}C\left(‰\right)=\left[\left({}^{13}C/{}^{12}C_{sample}\right)/\left({}^{13}C/{}^{12}C_{V-PDB}\right)-1\right]\times 1000$$ 3

where “V-PDB” refers to the international standard Vienna Pee Dee Belemnite. In this experiment, IAEA-CH-3 (cellulose) was used as the standard substance and the external standard method was used to determine the value of the parameters. The mean δ¹³C value of the standard substance was −24.72 ± 0.15‰ (2 SD, n = 8).

Hg isotopic mixing model and Rayleigh fractionation model of C

To elucidate Hg sources in canopy biomass and water samples, a Hg isotopic mixing model was developed based on Hg-MIF signatures. Hg in the canopy originates predominately from leaf stomatal uptake of atmospheric Hg⁰, atmospheric Hg^II deposition, and possibly from geological sources of Hg via root uptake and translocation. Hg in lichen and water samples is derived from atmospheric Hg⁰ and Hg^{II 49,68}. Therefore, a triple endmember Hg isotope mixing model (Eqs. 4–6) based on two-dimensional Hg MIF signatures (Δ¹⁹⁹Hg and Δ²⁰⁰Hg) was applied to trace the sources in canopy biomass:

$${\mathrm{\Delta }}^{199}H{g}_{canopybiomass}=f_1{\times \mathrm{\Delta }}^{199}Hg+f_2{\times \mathrm{\Delta }}^{199}Hg+f_3{\times \mathrm{\Delta }}^{199}Hg$$ 4

$$\mathrm{\Delta }{}^{200}H{g}_{canopybiomass}=f_1\times \mathrm{\Delta }{}^{200}Hg+f_2\times \mathrm{\Delta }{}^{200}Hg+f_3\times \mathrm{\Delta }{}^{200}Hg$$ 5

$$f_1+f_2+f_3=1$$ 6

where f_i is the fraction ratio of endmembers (i = 1, 2, 3, representing atmospheric Hg⁰, atmospheric Hg^II and geogenic Hg for canopy biomass). Our earlier study reported the MIF signatures of atmospheric Hg⁰ (Δ¹⁹⁹Hg = −0.22 ± 0.03‰, Δ²⁰⁰Hg = −0.07 ± 0.04‰, 1 SD, n = 11) in the Davos-Seehornwald³⁹. The data were used as endmembers of atmospheric Hg⁰.

A two-endmember Hg isotope mixing model (Eqs. 7 and 8) based on Hg even-MIF signatures (Δ²⁰⁰Hg) was established to estimate the contribution of Hg sources in lichen and forest water samples (throughfall, soil porewater and runoff).

$$\mathrm{\Delta }{}^{200}H{g}_{lichenorwatersamples}=f_1\times \mathrm{\Delta }{}^{200}Hg+f_2\times \mathrm{\Delta }{}^{200}Hg$$ 7

$$f_1+f_2=1$$ 8

where f_i is the fraction ratio of endmembers (i = 1, 2, representing atmospheric Hg⁰ and Hg^II sources, respectively). The even-MIF signatures of atmospheric Hg⁰ and Hg^II sources are the same as above. The Monte Carlo simulation method was used to generate one hundred thousand groups data for MIF signatures and the random range is from mean − SD to mean + SD. Finally, the fraction ratio was estimated as the average of these solutions, which ranges from 0 to 1, during one million times of solving Eqs. 4–6 or Eqs. 7 and 8.

To assess the proportion of C loss in the forest solid-phase continuum, we assumed that the C loss during the decomposition process of canopy biomass and soil follows the Rayleigh fractionation model, and the proportion of C loss can be calculated by the following formula.

$$\delta {}^{13}C_{res}=\delta {}^{13}C_{init}+ϵ\cdot In\left(f\right)$$ 9

$$L=1-f$$ 10

Where $\delta {}^{13}C_{res}$ is the δ¹³C value of the residual substances after decomposition, $\delta {}^{13}C_{init}$ is the initial δ¹³C value, $ϵ$ is the fractionation coefficient⁷¹, f represents the proportion of remaining C and L represents the proportion of C loss during the decomposition.

QA/QC and statistical analysis

To ensure that the samples did not undergo significant isotope fractionation during the pre-concentration, standards with known Hg isotope compositions were enriched into ClC traps using the same procedures as mentioned in Section 2.2. Hg recoveries for spiking 10 ng of the NIST 3133 Hg standard solutions to 5 L ultrapure water were 96 ± 4% (1 SD, n = 3) (Suppl. Table S1). Simultaneously, the pre-concentration process of water samples did not cause significant Hg isotope fractionation (δ²⁰²Hg = 0.07 ± 0.05‰, Δ¹⁹⁹Hg = −0.02 ± 0.05‰, Δ²⁰⁰Hg = 0.07 ± 0.03‰, Δ²⁰¹Hg = 0.01 ± 0.04‰, 1 SD, n = 3). The Hg recovery of the certified reference material BCR 482 was 92 ± 7% (1 SD, n = 5) using the double tube furnace combustion method. The Hg isotopic mixing models, Monte Carlo simulations and sensitivity analysis were performed with R version 4.3.1. One-Way ANOVA or t-test was utilised to determine significant differences in concentrations, isotopes and stoichiometric data with a 95% confidence level on IBM SPSS Statistics v27.

Supplementary information

Suppl. Information ^{(1.9MB, pdf)}

Transparent Peer Review file ^{(1.7MB, pdf)}

Author contributions

X.B.F. and J.H.H. designed research; C.Y.C. and J.H.H. performed research; C.Y.C. and X.W.F. analysed data; N.B., C.A., P.W., S.O. and A.P. provided key resources and technical support; X.W., K.L. and W.Y. provided model output; G.Y.S. and H.Z. provided visualisation; C.Y.C. wrote the manuscript; X.W.F., J.H.H., B.B. and X.B.F. contributed to the interpretation of the results and improvement of the manuscript; X.B.F. and C.A. supervised and funded the study.

Peer review

Peer review information

Nature Communications thanks Jun Zhou and other anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Funding

This work was supported by the National Natural Science Foundation of China (42394093, 42507562, 42350710202), the Postdoctoral Fellowship Program of CPSF under Grant Number GZB20230743 and Swiss National Science Foundation (IZLCZ2_170176, 20F120_198227, ICOS-CH phase 3).

Data availability

The data generated in this study are provided in the Suppl. Information and the accompanying Source Data files.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains Suppl. material available at https://doi.org/10.1038/s41467-026-76140-z.