Global health effects of future atmospheric mercury emissions

Mercury (Hg) is a global pollutant whose organic form, methylmercury (MeHg), causes neurocognitive deficits in fetuses and cardiovascular effects in adults. Human MeHg exposure occurs mainly via food consumption (seafood, freshwater fish, and rice). Substantial health and economic burdens have been documented, motivating international control via the Minamata Convention (effective 2017). Future primary anthropogenic Hg emissions are uncertain and influenced by socioeconomic and technological trajectories, while legacy re-emissions from soils and oceans are large. Exposure is governed by a chain of processes from emissions, atmospheric transport and deposition, air–sea/land exchange, methylation, food web transfer, to human intake, all modulated by climate, land-use, ocean circulation, and ecosystem dynamics. Earlier studies often did not connect emissions to exposure; more recent efforts considered subsets of processes or used simplified representations. This study develops a comprehensive coupled atmosphere–ocean–land–ecosystem and exposure-risk framework to project how future Hg emissions translate into environmental MeHg levels, dietary exposure across countries, and health risks, enabling policy-relevant assessments under multiple emission scenarios.

Previous work projected wide ranges of future Hg emissions based on economic and technological change (e.g., Streets et al. 2009; Pacyna et al. 2016) and highlighted the importance of legacy emissions from soil and ocean re-emissions. Many studies used atmospheric deposition as a proxy for MeHg in seafood or considered limited components (e.g., deposition-only indicators; effects of climate/land-use change on deposition; box models for land/ocean re-emission responses). Regional assessments estimated IQ decrements and cardiovascular risks tied to MeHg exposure, with valuation via VSL and earnings loss. However, fully integrated linkages from emissions to environmental levels to actual exposure across seafood, freshwater fish, and rice and then to health endpoints were limited. This study advances prior work by coupling 3D atmosphere and ocean Hg models with a 2D terrestrial Hg model and an ecosystem model to simulate environmental Hg/MeHg dynamics and by combining these with country-specific intake inventories and epidemiology-based dose-response functions for global-scale risk and valuation.

The authors constructed an integrated modeling framework coupling: (1) GEOS-Chem (atmospheric Hg chemistry and transport) driven by GISS ModelE2 climate fields; (2) MITgcm ocean Hg model (inorganic and methylated Hg species, 14 tracers) driven by IGSM ocean physics and the Darwin marine ecosystem model (plankton functional groups and biogeochemistry); and (3) GTMM terrestrial soil Hg model (four soil carbon-tied Hg pools with turnover times from years to millennia). Models are two-way coupled via NJUCPL, exchanging atmospheric deposition and concentrations with soil and ocean evasion/re-emissions at hourly frequency. Emission scenarios for 2010–2050 include: CP (current policy; near-constant emissions), A1B (business as usual; increasing), A2 (divided world; increasing), NP-Delayed (new policy delayed; slower reductions to 2050), and MFR (maximum feasible reduction; aggressive controls), with spatial distributions based on the 2010 WHET inventory and fixed speciation/spatial patterns over time. Environmental outputs include atmospheric deposition, soil Hg concentration, and marine planktonic MeHg. Exposure modeling: Country-level per-capita consumption of seafood, freshwater fish (and other aquatic animals), and rice is taken from FAO statistics. A literature-based database of MeHg concentrations is compiled (≈210,000 data points from 395 publications), excluding point-source contaminated sites. Fish/aquatic species are grouped by trophic level bins (2–2.5, 2.5–3.5, 3.5–4.5, 4.5–5) and by wild-caught vs farmed; geometric mean MeHg concentrations per trophic bin are used with consumption shares estimated via the marine trophic index where detailed data are lacking. Present-day exposure is calculated by summing intake times concentration across categories. Future concentrations are scaled by modeled environmental proxies: freshwater fish by atmospheric Hg deposition; rice by soil Hg concentration; seafood by planktonic MeHg weighted by fish harvest distribution. Food consumption patterns are assumed constant over 2010–2050. Health impact assessment: Two endpoints are included: fetal IQ decrement and fatal heart attacks (FHA) in adults. A linear no-threshold dose-response from MeHg intake to IQ loss is applied via standard pharmacokinetics linking intake to blood Hg, blood to hair Hg, and hair Hg to IQ decrement. FHA risk is derived from epidemiological dose-response relationships relating Hg biomarkers to cardiovascular outcomes. Country-level valuation monetizes IQ losses via lifetime earnings decrements per IQ point and FHA via value of a statistical life (VSL), scaling values by PPP-adjusted per-capita GDP and discounting (3%) to 2050 under SSP2 socioeconomic trajectories. Population, births, and baseline FHA incidence are taken from UN and WHO sources. Uncertainty is assessed via sensitivity analysis and Monte Carlo, varying food consumption, MeHg concentrations, dose-response parameters, pharmacokinetics, and economic valuation (notably VSL).

• Baseline (present-day) global health impact from MeHg exposure is estimated at $117 billion per year (2020 USD, PPP), comprised of 1.2 × 10^7 IQ points lost (0.086 point per fetus on average) and 29,000 FHA deaths annually.
• Country patterns: Highest per-capita seafood MeHg exposure in Maldives (33 µg/d), Greenland (16 µg/d), Iceland (15 µg/d), Kiribati (13 µg/d); lowest in inland countries with minimal seafood, e.g., Ethiopia (0.0018 µg/d), Uganda (0.0093 µg/d), Chad (0.014 µg/d). Rice exposure is highest in Southeast Asia (e.g., Indonesia 1.7 µg/d; Laos 0.90 µg/d; Cambodia 0.77 µg/d) and can dominate total exposure in some inland, rice-consuming countries (Nepal 58%, Afghanistan 50%, Bhutan 45%). Freshwater fish exposure is highest in Cambodia (6.3 µg/d), Myanmar (3.5 µg/d), Japan (2.9 µg/d), with high values also in Russia (3.5 µg/d), Finland (3.2 µg/d).
• Economic losses by country: IQ loss—USA $12B/y, China $7.3B, Japan $6.2B, Russia $2.9B. FHA VSL loss—Russia $9.1B/y, USA $9.0B, China $7.7B, Japan $3.2B. Total loss—USA $21B/y, China $15B, Russia $12B, Japan $9.3B. Regional totals: Asia $48B, Europe $34B, North America $23B (≈90% combined), Africa $6.4B, South America $5.4B, Oceania $1.5B.
• Model–biomarker agreement: Estimated blood Hg across 40 countries (2.5 ± 1.8 µg/L) matches observations (2.2 ± 2.1 µg/L), r = 0.71. Hair Hg correlation r = 0.53; estimates within ~factor 2 of observations.
• Environmental response to emissions (2050 vs CP): MFR and NP-Delayed reduce atmospheric deposition by 48% and 28%, respectively; increases under A1B and A2 are 87% and 59%. Changes in planktonic MeHg mirror deposition changes; soil Hg changes are small (−3% to +4%). Primary anthropogenic emissions changes are damped at the deposition level due to large legacy contributions.
• Health trajectory to 2050: Under CP, global IQ loss rises slightly from 11.1 to 11.6 million points/y; MFR and NP-Delayed reduce 2050 IQ loss by 24% and 15% relative to CP, while A1B and A2 increase by 51% and 34%. FHA deaths: CP increases to 40,000/y by 2050 (cumulative 1.6 million, 2010–2050). MFR and NP-Delayed trajectories are flat (cumulative 1.4 and 1.5 million); A1B and A2 rise by 120% and 94% vs 2010, cumulative 2.0 and 1.9 million.
• Economic valuation: CP cumulative loss 2010–2050 is $19 trillion (2020 USD, discounted at 3% to 2050). Benefits vs CP: MFR $2.4T, NP-Delayed $1.5T. Additional losses vs CP: A1B $4.9T, A2 $3.3T. By 2050, FHA VSL contributes ~60% of annual losses due to faster population growth relative to births.
• Uncertainty: CP cumulative loss ranges $10–27T based on component-wise sensitivity; broader Monte Carlo range $4.7–54T. Largest contributors to uncertainty are dose-response parameters (IQ and FHA risk coefficients) and economic valuation (notably VSL, up to factor ~10), with smaller contributions from pharmacokinetics (10–20%) and food intake/MeHg concentration data (each causing factor ~2 variability).

The integrated framework directly links emissions to environmental Hg/MeHg levels, dietary exposure, and health risks, addressing the research question of how future atmospheric Hg emissions translate into global human health impacts. Results show that despite large projected changes in primary anthropogenic emissions, health outcomes respond more modestly due to legacy emissions and slow dynamics in soils and oceans, underscoring the importance of early and sustained emission reductions. The findings identify dietary structure—especially reliance on seafood, freshwater fish, and rice—as a key determinant of risk, implying heterogeneous benefits of controls across countries and the potential importance of dietary guidance focusing on lower-trophic-level species selection. The model’s ability to simulate ocean planktonic MeHg and terrestrial soil Hg offers improved proxies for seafood and rice MeHg dynamics compared to deposition-only approaches. Policy-relevant insights include substantial cumulative benefits of aggressive controls (MFR) and significant foregone benefits when reductions are delayed (NP-Delayed), with even larger additional burdens under high-emission scenarios (A1B/A2). The framework can support evaluation of Minamata Convention effectiveness and prioritization of sectoral measures (e.g., ASGM controls), recognizing the damped and delayed system response and compounding economic impacts.

This study develops a comprehensive, coupled climate–atmosphere–land–ocean–ecosystem and exposure-risk modeling framework to quantify global health impacts from MeHg under future Hg emission scenarios. Present-day annual losses are ~$117B with 1.2 × 10^7 IQ points lost and 29,000 FHA deaths. Under current policies, cumulative losses by 2050 are ~$19T (3% discount), while aggressive mitigation (MFR) yields ~$2.4T in benefits versus CP; delayed mitigation erodes benefits, and high-emission pathways impose $3.3–4.9T additional costs. Responses are damped by legacy reservoirs, highlighting the necessity of early action. The analysis emphasizes the role of national dietary patterns and fish trophic levels in exposure and suggests that while seafood often dominates exposure, rice contributions are persistent and relatively insensitive to emission controls. Future research should: expand endpoints (beyond IQ and FHA) and include high-exposure subpopulations; improve spatial resolution and fishery dynamics; refine dose-response and valuation parameters; and better constrain land/ocean legacy processes (including potential permafrost sources) and future non-atmospheric Hg releases. The framework can inform detailed scenario analyses to assist parties in implementing and assessing Minamata Convention measures.

• Exposure assessment excludes high-exposure groups (e.g., ASGM workers, Arctic and certain riverine/coastal communities) due to limited global data, potentially underestimating total risk.
• Food consumption patterns are held constant through 2050, not capturing future dietary shifts or fishery changes; fishery harvest structure and trophic dynamics are simplified and may change under climate and management.
• Freshwater fish and rice MeHg concentrations are scaled using proxies (atmospheric deposition and soil Hg); river/lake and paddy biogeochemistry are not explicitly simulated, and higher-trophic-level food web dynamics are simplified.
• Dose-response relationships for IQ and FHA and valuation parameters (especially VSL) contribute large uncertainty; pharmacokinetic variability and genetic/intrinsic factors affecting biomarkers are not fully captured.
• Potential future changes in Hg releases to water/soil and emerging sources (e.g., thawing permafrost) are not explicitly modeled.
• Limited interannual variability is simulated at global scale; regional variability due to dietary shifts and high-trophic-level dynamics may be larger.
• Validation data (biomarkers, food MeHg) are sparse in many regions and species; some food categories beyond seafood/freshwater fish/rice are not included.