International trade shapes global mercury–related health impacts — TEI-tagged view

Zhencheng XingpersName, Ruirong ChangpersName, Zhengcheng SongpersName, Yanxu ZhangpersName, Marilena MunteanpersName, Kuishuang FengpersName, Yifan LiupersName, Zongwei MapersName, Jigan WangpersName, Jie ZhangpersName, Haikun WangpersName (2023date). “International trade shapes global mercury–related health impacts.” PNAS Nexustitle·j.

Object of study — relocation of mercuryterm·dom emissions, pollution, exposure and health impactsterm·key through international tradeterm·dom

Research question — how trade relocates the mercuryterm·dom cycle along the whole chain from economic production to human health

Contribution — four linked global models and a counterfactual no-trade scenario quantifying trade's net effect on mercuryterm·dom health impactsterm·key

Full text — tagged

The reference list is not part of the text we hold for this paper. Our copy was extracted from the publisher's page and the bibliography did not survive that extraction, so the article below ends before its references. Nothing has been removed by us.

International tradeterm·dom shapes global mercuryterm·dom–related health impactsterm·key Abstract Mercuryterm·dom (Hgterm·dom) is a strong neurotoxin with substantial dangers to human health. Hgterm·dom undergoes active global cycles, and the emission sources there of can also be geographically relocated through economic trade. Through investigation of a longer chain of the global biogeochemical Hgterm·dom cycle from economic production to human health, international cooperation on Hgterm·dom control strategies in Minamata Conventionterm·key can be facilitated. In the present study, four global models are combined to investigate the effect of international tradeterm·dom on the relocation of Hgterm·dom emissions, pollution, exposure, and related human health impactsterm·key across the world. The results show that 47% of global Hgterm·dom emissions are related to commodities consumed outside of the countries where the emissions are produced, which has largely influenced the environmental Hgterm·dom levels and human exposure thereto across the world. Consequently, international tradeterm·dom is found to enable the whole world to avoid 5.7 × 10 5 points for intelligence quotient (IQ) decline and 1,197 deaths from fatal heart attacksterm·dom, saving a total of $12.5 billion (2020 USD) in economic lossterm·key. Regionally, international tradeterm·dom exacerbates Hgterm·dom challenges in less developed countries, while resulting in an alleviation in developed countries. The change in economic lossterm·key therefore varies from the United States (−$4.0 billion) and Japan (−$2.4 billion) to China (+$2.7 billion). The present results reveal that international tradeterm·dom is a critical factor but might be largely overlooked in global Hgterm·dom pollution mitigation.

Effective international mercuryterm·dom mitigation requires a good knowledge of the longer global biogeochemical mercuryterm·dom cycle from economic activities to human health. Here, we combine four global models to investigate the effect of international tradeterm·dom on the relocation of mercuryterm·dom pollution and related human health impactsterm·key across the world. We find that international tradeterm·dom enables the whole world to avoid considerable health impactsterm·key and economic lossterm·key, but this comes at the cost of environmental degradation in less developed countries. Our results highlight the role of international tradeterm·dom in international cooperative mercuryterm·dom control.

Introduction Mercuryterm·dom (Hgterm·dom) is a potent neurotoxic substance, since the organic form thereof, methylmercuryterm·dom (MeHgterm·dom), can result in neurocognitive abnormalities in kids and cardiovascular issues in adults ( 1 , 2 ). In the annals of history, MeHgterm·dom exposure has resulted in substantial health concerns to human beings. For example, 1,784 people died of the Minamata disease in Japan ( 3 ). More than 10,000 deaths from fatal heart attacksterm·dom (FHAterm·dom) are thought to occur each year in China as a result of exposure to MeHgterm·dom ( 4 ). It is calculated that MeHgterm·dom exposure costs the United States and Europe $16 billion in lost productivity due to the intelligence quotient (IQ) decline of developing brains ( 5 , 6 ). About 60% of the Hgterm·dom emitted to the atmosphere each year derives from anthropogenic Hgterm·dom that was previously deposited in soil and water ( 7 ). Just 10% of the current emissions come from natural sources, and the remaining 30% come from manmade sources ( 8 ). Human-related Hgterm·dom emissions play a sizable part in the global Hgterm·dom budget ( 9 ). Among the background of such context, the Minamata Conventionterm·key on Mercuryterm·dom was signed by 128 countries to establish a mandate for measures to combat global Hgterm·dom contamination ( 10 ), including reducing artisanal and small-scale gold miningterm·dom (ASGMterm·dom) as well as the purposeful use of Hgterm·dom in goods and industrial processes.

Through international tradeterm·dom, locations where commodities are consumed are separated from those where emissions and the resulting pollution and health effects occur ( 11 , 12 ). Although items from downstream industries don't contain Hgterm·dom, using these products can cause upstream Hgterm·dom emissions from the manufacture of raw and intermediate materials ( 13 , 14 ). Moreover, developed economies frequently transfer Hgterm·dom emissions to emerging countries in order to boost financial returns by utilizing cheap labor and energy as well as lax environmental regulations, which can be attributed to differences in division of labor in the global value chains ( 15 , 16 ). As such, there is debate among academic circles concerning whether existing production-side controls (for example, putting up Hgterm·dom removal facilities and adopting cleaner production techniques) can effectively reduce emissions, since the drivers (that is, consumption) behind such emissions are neglected ( 17 , 18 ). Thus, the role of international tradeterm·dom should be identified in the relocation of global Hgterm·dom emissions, pollution, exposure, and related health burden.

A series of processes, such as Hgterm·dom emission, atmospheric transport and deposition, environmental exchange, Hgterm·dom methylation, food web transfers, and dietary intake, affect MeHgterm·dom exposure to humans ( 19 ). The various phases of the Hgterm·dom cycle are interrelated and may have an impact on one another. Several specific segments of the chain have been evaluated in previous studies. For example, Hgterm·dom emissions are generated from production activities and induced by consumers through international tradeterm·dom ( 20 , 21 ). The physical transport of Hgterm·dom emissions from other countries can significantly affect domestic Hgterm·dom pollution ( 22 , 23 ). Hgterm·dom levels in environmental medias can influence the MeHgterm·dom levels in various foods including seafood, freshwater fish, and rice ( 24 , 25 ). MeHgterm·dom intake in one place may result from a variety of meals from distinct worldwide regions ( 26–28 ). Despite the aforementioned findings, the global biogeochemical Hgterm·dom cycle has not been identified along a longer path from economic production to human health.

Notably, the effects of subnational trade on Hgterm·dom emissions, transportation, deposition, food exposure, and related health impactsterm·key have been investigated in China ( 4 ). However, in such research, the atmospheric Hg depositionterm·dom was taken as the sole proxy to scale the change of food MeHgterm·dom levels without inclusion of marine plankton and soil Hgterm·dom levels which are more suitable proxy indicators for seafood and rice ( 29 ). At the same time, studies focusing on a sole country result in a failure to provide effective policy instruction for the implementation of the Minamata Conventionterm·key from a multinational perspective. The limitations in previous studies make it difficult to identify how international tradeterm·dom affects the Hg-related health risks and to propose practical policy solutions to mitigate those risks through cooperation on a global scale. In this context, a comprehensive investigation of the Hgterm·dom health impactsterm·key associated with global trade is urgently needed for promoting international cooperation on the Minamata Conventionterm·key ( 30 ).

In the present study, the extent to which international tradeterm·dom results in relocation of Hgterm·dom emissions, pollution, exposure, and related health impactsterm·key across the world is evaluated, and the net effect thereof on the global Hgterm·dom challenges is further investigated. Such efforts are achieved by linking a Hgterm·dom emission inventory ( 31 ), a global multiregional input–outputterm·mth (MRIOterm·mth) model ( 32 , 33 ), a coupled atmosphere–land–ocean–ecosystem model ( 29 ), and an exposure–risk–valuation model ( 34 ). Notably, it is assumed that a certain country's MeHgterm·dom exposure via seafood is proportional to the global marine plankton MeHgterm·dom levels weighted by its spatial distribution of fish catch rather than a simple global average concentration adopted by ZhangpersName et al. ( 29 ). This enhancement may help improve the accuracy of food MeHgterm·dom exposure estimates for various countries. The present method allows for a more comprehensive assessment and could be used to look into a longer chain of the global biogeochemical Hgterm·dom cycle from economic production to human health. The results may facilitate the implementation of an ecological compensation mechanism to advance international cooperation on Hgterm·dom control strategies in the Minamata Conventionterm·key and formulate targeted emission reduction measures for different countries from both production and consumption perspectives (further details are provided in the policy implications in a later section).

Results and discussion Relocation of Hgterm·dom emissions of countries Globally, human activities release a total of 1,833.3 Mg of Hgterm·dom in 2011 ( 31 ), 47.1% (864.2 Mg) of which is embodied in international tradeterm·dom. In other words, nearly half of the global Hgterm·dom emissions can be linked to exports to other countries (see Fig. S1 for emissions embodied in bilateral trade). Figure 1 displays the net impact of international tradeterm·dom on Hgterm·dom emissions in 13 world regions (see Fig. S2 for regional definitions). The top net Hgterm·dom receptors are all less developed economies (red shading, Fig. 1 ), such as sub-Saharan Africa (+199.4 Mg), Latin America (+166.3 Mg), and China (+80.9 Mg). Such economies are found to be major sources of global Hgterm·dom emissions (Fig. S3 ), of which China is the world's largest emitter (546.1 Mg), followed by sub-Saharan Africa (291.8 Mg) and Latin America (291.3 Mg). From the view of sector, the smelting and pressing of nonferrous metals are the primary direct contributors to the Hgterm·dom emissions of the three regions, even accounting for over 90% in sub-Saharan Africa (Fig. S3 ). The aforementioned findings can primarily be attributed to the heavy use of Hgterm·dom by the ASGMterm·dom in such regions ( 35 ). The top net Hgterm·dom outsourcers are mainly developed economies (blue shading, Fig. 1 ), such as the United States (−145.2 Mg), Western Europe (−140.9 Mg), and Japan (−36.9 Mg). However, several developing economies such as the Middle East and North Africa (−112.7 Mg), India (−20.6 Mg), and Eastern Europe (−5.9 Mg) are also found with net negative emissions, which could be ascribed to their lower emissions thereof from ASGMterm·dom, thereby contributing to reducing production-based emissions (Fig. S3 ).

Spatial distributions of net Hgterm·dom emissions embodied in the international tradeterm·dom and prominent net flows of Hgterm·dom emissions between the 13 world regions. Bilateral flows of Hgterm·dom emissions embodied in the international tradeterm·dom between 13 world regions are plotted in Fig. S3 . The shading of regions represents the difference between where Hgterm·dom emissions are physically produced and where the related goods and services are consumed, thereby highlighting the emissions embodied in the net international tradeterm·dom. Arrows show the 10 largest net flows of embodied Hgterm·dom emissions, and the sizes of arrows correspond to the amount of net emission flows.

Developing economies, the main suppliers of primary and semimanufactured products (for example, metals and nonmetallic mineral products), are located upstream of the global supply chains. In contrast, industrialized economies have higher consumption-based emissions since they are further away from global supply chains (Fig. S3 ). The specialization of global value chains results in a vast difference between the Hgterm·dom emission intensities of industrialized and emerging economies (Fig. S4 ). As such, flows of Hgterm·dom emissions induced by international tradeterm·dom are found to mainly go from developed to developing economies (Fig. 1 ). For example, Middle East and North Africa outsources 84.1 Mg of net emissions to sub-Saharan Africa, followed by the United States → Latin America (73.2 Mg), Western Europe → Latin America (40.3 Mg), Western Europe → sub-Saharan Africa (39.9 Mg), and the United States → China (34.7 Mg). These results demonstrate that developed economies can avoid significant amounts of Hgterm·dom emissions by transferring Hg-intensive production activities to emerging economies through international tradeterm·dom ( 18 ), thereby highlighting a necessity for international responsibility sharing on Hgterm·dom control from an environmental equity perspective.

Redistribution of environmental Hgterm·dom levels One central question is whether the international tradeterm·dom has inadvertently increased or decreased global Hgterm·dom pollution relative to a world with no trade. In the present study, to assess the effects of international tradeterm·dom on global Hgterm·dom pollution, a counterfactual scenario with the absence of international tradeterm·dom (“no trade” scenario) is set up for comparison with the existence of international tradeterm·dom (“with trade” scenario). There is an assumption that trade partners can manufacture the same commodities that are originally imported through international tradeterm·dom at the same emission level ( 4 , 12 ). As such, the global total emissions remain unchanged, but emissions of various countries are counted according to consumption rather than production. See Table S1 for more detailed descriptions. The gridded emission inventory under the “with trade” scenario and production- and consumption-based emissions of countries are combined to scale the corresponding gridded emissions for each country under the “no trade” scenario. The Hgterm·dom emission inventory under “with trade” and “no trade” scenarios and the differences therebetween are provided in Fig. S5 . Once released into the environment, Hgterm·dom can diffuse and move through various environmental medias (for example, the atmosphere, water, and soil). Figure 2 shows the environmental Hgterm·dom levels under “with trade” and “no trade” scenarios and the differences therebetween, indicating the net impact of international tradeterm·dom on the distribution of Hgterm·dom pollution. Findings are made that the relocation of emissions induced by international tradeterm·dom has, to varying degrees, redistributed atmospheric Hg depositionterm·dom, soil Hgterm·dom concentrations, and marine planktonic MeHgterm·dom over the world.

Global environmental Hgterm·dom levels under “with trade” and “no trade” scenarios and the differences there between. The differences between global environmental Hgterm·dom levels due to the emissions produced in each region and those due to the emissions related to the consumption in each region denote the net impact of international tradeterm·dom on environmental Hgterm·dom levels. The rows (A–C), (D–F), and (G–I) are for atmospheric deposition, marine plankton, and soil concentration, respectively. For each row, the three panels denote the spatial distribution of environmental Hgterm·dom levels for “with trade” and “no trade” scenarios and the differences therebetween. The color bars on the right-hand side share the same units with the corresponding color bars on the left-hand side.

International tradeterm·dom is found to considerably redistribute atmospheric Hg depositionterm·dom globally, with annual deposition flux ranging from −40.8 Mg (United States) to +44.9 Mg (China). This results in the transfer of atmospheric Hgterm·dom pollution from developed countries to less developed ones (Fig. 2 A–C). Despite being similar to the relocation of emissions (Fig. 1 ), the relocation of atmospheric Hg depositionterm·dom also highlights the case for which there are neighboring areas downwind of where the emissions are generated. For instance, emissions embodied in Chinese exports contribute to downwind Hg depositionterm·dom over the Northwest Pacific Ocean and East Asian countries [for example, South Korea and Japan; Chen et al. ( 4 )]. Overall, the countries with reduced Hg depositionterm·dom (for example, the United States, Japan, and Western Europe) have a larger decrease in deposition than other countries with increased deposition (for example, China and Latin America), thereby resulting in a net decline in global total Hg depositionterm·dom (−244.1 Mg). Indeed, atmospheric deposition occurs in several pathways including dry deposition of elemental (Hgterm·dom 0 ) and oxidized Hgterm·dom (Hgterm·dom II ) and wet deposition of Hgterm·dom II , which are mainly influenced by the atmospheric redox chemistry, land use/cover characteristics, and meteorological factors. In general, atmospheric Hgterm·dom II is more readily deposited with a greater local impact than Hgterm·dom 0 ( 7 , 9 ). Additionally, the oxidation rate of atmospheric Hgterm·dom is found to be higher in areas around 40°N than those around 0° (Fig. S6 ), mainly due to the difference in the levels of regional oxidizing substances ( 36 ). As such, the trade-induced transfer of emissions from high-latitude (40°N) to low-latitude areas (0°), such as Middle East and North Africa → sub-Saharan Africa (84.1 Mg) and the United States → Latin America (73.2 Mg), as shown in Fig. 1 , results in a decline in the oxidation degree of global atmospheric Hgterm·dom. Consequently, international tradeterm·dom leads to an overall decline in global total Hg depositionterm·dom, even though the two scenarios have the same global total Hgterm·dom emissions.

Marine planktonic MeHgterm·dom is also significantly sensitive to international tradeterm·dom (Fig. 2 D–F). Moreover, the percentage variation of planktonic MeHgterm·dom levels is comparable with Hg depositionterm·dom, because the substrate of MeHgterm·dom in the ocean (i.e. inorganic Hgterm·dom) originates primarily from atmospheric deposition (Fig. 2 F; 37 ). For instance, the trade-induced increase in spillover from Hg depositionterm·dom of China contributes to the increase in planktonic MeHgterm·dom concentrations over the seas leaving the eastern terrestrial boundary, including the Chinese coastal seas and the Sea of Japan (Fig. 2 F). Contrarily, the trade-induced decrease in atmospheric deposition of developed countries (for example, Japan and Western Europe) leads to the decrease in planktonic MeHgterm·dom concentrations in the corresponding sea areas thereof, such as the Sea of Okhotsk and the Arctic Ocean. Globally, the average plankton MeHgterm·dom concentrations for the “with trade” scenario are also lower than the “no trade” scenario (Fig. S7 ), which could be attributed to the aforementioned atmospheric deposition changespersName. The variations in soil Hgterm·dom concentrations are substantially lower (<1%, Fig. 2 I) due to the enormous mass and prolonged lifetime of Hgterm·dom in the reservoir ( 38 , 39 ). Nevertheless, owing to the lower atmospheric deposition, the global soil Hgterm·dom levels are also found to be lower under the “with trade” scenario (Fig. S7 ).

Relocation of Hgterm·dom exposure and related impacts The MeHgterm·dom exposure levels from food consumption of three categories including seafood, freshwater fish, and rice are included in the present study along with the MeHgterm·dom concentrations from previous research ( 29 ). The exposure levels under the “with trade” scenario and the modeled environmental Hgterm·dom levels are combined to scale the MeHgterm·dom exposure under the “no trade” scenario (see Materials and methods). Figure 3 shows the food MeHgterm·dom exposure levels under “with trade” and “no trade” scenarios and the differences therebetween indicating the net impact of international tradeterm·dom on the exposure of countries. As the largest source of human MeHgterm·dom exposure, seafood intake accounts for 56% of the global total food exposure, followed by freshwater fish (35%) and rice (9%), which is mainly attributed to the differences in Hgterm·dom bioaccumulation between the three food categories ( 28 ). At the same time, as atmospheric Hg depositionterm·dom and marine planktonic MeHgterm·dom are both significantly sensitive to international tradeterm·dom (Fig. 2 ), changespersName in exposure from seafood and freshwater fish are even more substantial (Fig. 3 ). As an example, the trade-induced relocation of deposition leads to an increase of 0.51 µ g/d in freshwater fish exposure for China and a decline of 1.66 µ g/d in Japan (Fig. 3 C). The decline in marine planktonic MeHgterm·dom enables coastal countries that consume a lot of seafood to achieve considerable decreases in seafood MeHgterm·dom exposure (Fig. 3 F), such as the Maldives (−2.40 µ g/d), Iceland (−1.40 µ g/d), Haiti (−1.36 µ g/d), and Qatar (−1.31 µ g/d). The change in MeHgterm·dom exposure via rice is less obvious for all the countries (Fig. 3 I), which could be attributed to the limited changespersName in soil Hgterm·dom concentrations (Fig. 2 I). Overall, as shown in Fig. S8 , the trade-induced redistribution of environmental Hgterm·dom levels leads to a net decline in global total exposure (−0.14 µ g/d), with seafood (−0.10 µ g/d) and freshwater fish (−0.04 µ g/d) being the main contributors.

Food MeHgterm·dom exposure of countries under “with trade” and “no trade” scenarios and the differences therebetween. The rows (A–C), (D–F), and (G–I) are for freshwater fish, seafood, and rice MeHgterm·dom exposure, respectively. For each row, the three panels denote the regional distribution of food exposure for “with trade” and “no trade” scenarios and the differences therebetween.

Two health outcomes are included as a result of food MeHgterm·dom exposure: decline in infant IQ and FHAterm·dom mortality. These endpoints are estimated according to the total exposure of the aforementioned pathways from seafood, freshwater fish, and rice. Based on the population and economic growth data of individual countries, the IQ decline is converted to a loss in lifetime wages. The economic lossterm·key from FHAterm·dom is calculated by using a value of the statistical lifeterm·mth (VSLterm·mth) approach, which is derived from the GDP per capita of various countries (see Materials and methods). Figure 4 shows MeHgterm·dom exposure-related human health and economic impacts of countries under the “with trade” and “no trade” scenarios and the differences therebetween, indicating the net impact of international tradeterm·dom. As shown in Fig. 4 C, the largest increase in IQ decrement per fetus is in the Bahamas (+0.036 points), Latvia (+0.028 points), and China (+0.020 points), while the largest decrease occurs in Japan (−0.087 points), the Maldives (−0.086 points), and Qatar (−0.086 points), which is mainly related to the change in food MeHgterm·dom exposure (Fig. 3 ). Taking into consideration the birth rates and income levels of various countries, the change in economic lossterm·key attributed to IQ decline of newborns could be obtained (Fig. 4 F). China is found with the largest increase in economic lossterm·key (+$1.3 billion) as a result of trade, while the largest decrease occurs in the United States (−$2.3 billion) and Japan (−$1.6 billion).

MeHgterm·dom exposure-related health and economic impacts of countries under “with trade” and “no trade” scenarios and the differences therebetween. (A–C) Per-fetus IQ decrement; (D–F) economic lossterm·key from IQ decrement; (G–I) FHAterm·dom deaths; (J–L) VSLterm·mth loss from FHAterm·dom; and (M–O) total loss from MeHgterm·dom exposure. Economic lossesterm·key are in US dollars (2020 value and adjusted by purchasing power parity). The gray area indicates missing data. The color bars on the right-hand side share the same units with the corresponding color bars on the left-hand side.

The regional pattern of the mortality related to FHAterm·dom (Fig. 4 I) differs significantly from that of IQ decline per fetus, with the inclusion of total population and baseline FHAterm·dom incidence. The largest changespersName in deaths attributed to trade are from populous countries with significant variation in the exposure, such as China (+979 deaths), Russia (−322 deaths), India (−253 deaths), the United States (−204 deaths), and Japan (−130 deaths). Considering the difference in VSLterm·mth per death, the United States has the highest loss reduction from such pathway (−$1.7 billion), followed by Russia (−$0.92 billion), Japan (−$0.81 billion), and Germany (−$0.43 billion), while China is faced with an increase of $1.4 billion in economic lossterm·key (Fig. 4 L). Combining the two endpoint’s results, China suffers from an increase of $2.7 billion per year in total loss, while developed countries obtain considerable alleviation in total loss, such as the United States (−$4.0 billion), Japan (−$2.4 billion), and Russia (−$1.2 billion), as shown in Fig. 4 O.

Decline in global Hg-related health risks We further calculate global changespersName across all countries and assess the net effects of the international tradeterm·dom. As shown in Table 1 , international tradeterm·dom enables the whole world to avoid 5.7 × 10 5 points of IQ decrements (0.41 × 10 −2 points per fetus) and 1,197 deaths per year from FHAterm·dom. Taken together, global economic lossterm·key associated with MeHgterm·dom exposure decreased by $12.5 billion (2020 USD). Given the strong assumptions on the scenario setting, these results should be interpreted cautiously, but they, to a certain extent, examine the effects of international tradeterm·dom on mitigating global Hgterm·dom pollution and related health impactsterm·key.

Global Hg-related human health and economic impacts under actual “with trade” and counterfactual “no trade” scenarios and the differences therebetween.

As aforementioned, international tradeterm·dom is found to considerably relocate the distribution of Hgterm·dom emissions, pollution, MeHgterm·dom exposure, and related health risks among various countries, without changing the total emissions of the world. However, the trade-induced relocation of emissions results in a nonnegligible reduction in global gross Hg depositionterm·dom (−244.1 Mg, −8.2%), which, to a certain extent, contributes to the alleviation in global total food MeHgterm·dom exposure and related health impactsterm·key. Additionally, a regional asymmetry of variations in MeHgterm·dom levels in food products and subsequent MeHgterm·dom exposure over various deposited regions is also brought on by shifts in the spatial distributions of the deposition. There are significant differences in landscape patterns and dietary structure between countries which may influence the extent of the deposition of Hgterm·dom to the sea and the intake of MeHgterm·dom to the body, respectively. When compared with inland rice-growing areas (for example, sub-Saharan Africa), coastal fish-eating areas (for example, Western Europe) may suffer a higher impact from the same quantity of atmospheric Hg depositionterm·dom. Significant differences exist in the sensitivity between Hg depositionterm·dom and MeHgterm·dom exposure variations across the world, wherein developed countries are found to be more risk prone than less developed countries (Fig. 5 ). As the exposure and health burden variations are assumed to present a linear relationship (Fig. S9 ), the trade-driven transfer of Hgterm·dom from coastal developed countries to inland developing countries can lead to a reduction in global Hg-related health risks. Notably, despite being a developing country, China is found to be relatively risk prone (Fig. 5 ). Such findings could be attributed to the trade-induced increase in the Hg depositionterm·dom thereof contributing to the increase in planktonic MeHgterm·dom concentrations over the seas leaving the eastern terrestrial boundary where there is a dense population whose diets contain many aquatic food products ( 4 ). As a result, 99% of the increased exposure in China is attributable to the consumption of seafood and fish products (Fig. S10 ).

Contrasting changespersName in atmospheric Hg depositionterm·dom and MeHgterm·dom exposure across the world. The figure shows the relative changespersName in regional Hg depositionterm·dom and MeHgterm·dom exposure resulting from international tradeterm·dom for each country. The color of the dots denotes the magnitude of per capita GDP, and the size of the dots denotes the amount of population.

Uncertainties and limitations The compilation of Hgterm·dom emission inventory, the estimation of emissions embodied in trade, the simulation of the Hgterm·dom transport model, the gathering of data on food consumption and MeHgterm·dom concentrations, and the assessment of the effects on human health and monetary loss are the main sources of uncertainty that can affect the results of our model. MunteanpersName et al. ( 40 ) reported that global production-based anthropogenic emissions have an overall uncertainty of [−33%, 42%] which serves as an important reference for this study. A recent study reported uncertainty (coefficient of variation) of Chinese provincial Hgterm·dom footprint varying from 8 to 34% by using the Monte Carlo simulation method ( 15 ). As the uncertainty for small emitters is larger than that of large emitters ( 41 ), the uncertainty for national Hgterm·dom footprint in our study should be lower than 34%. Based on the simulated and observed results of environmental Hgterm·dom levels (Fig. S11 ), the normalized root–mean–square deviationterm·mth (NRMSDterm·mth) is applied to calculate the uncertainty caused by the Hgterm·dom transport model (see Supplementary material ). For example, the uncertainty for surface total gaseous Hgterm·dom is estimated at 29.1%.

Referring to our previous work ( 29 ), the uncertainty of food consumption is estimated at [−47%, 42%], by comparing the FAO data with national data. The uncertainty of food MeHgterm·dom concentration is [−37%, 63%], by considering its log-normal distributions. The dose–response relationships between MeHgterm·dom exposure and health consequences make up the largest portion of the uncertainty, with a range in [−59%, 147%]. An uncertainty of [−70%, 26%] for economic lossterm·key results from the factors for economic value, such as the VSLterm·mth of FHAterm·dom deaths. We take into account how the dose–effect relationship, economic valuation, food consumption, and MeHgterm·dom concentrations contribute to overall uncertainty. A Monte Carlo method is used to estimate the overall degree of uncertainty. The settings for these four components are randomly picked, and the health risk computation is repeated 1,000 times. The computed risk's 2.5 and 97.5% percentiles are used to determine the overall uncertainty range (or 95% confidence interval). The overall uncertainty for total economic lossterm·key is estimated at [−75.3%, 131.6%].

Although the uncertainty of estimates under the “with trade” and “no trade” scenarios is relatively high, the effect of international tradeterm·dom is expected to have a lower uncertainty, as the derivation of the difference between the estimates under the two scenarios may offset some of their original uncertainty. Moreover, despite the uncertainties from various stages, our comprehensive models, to a certain extent, advance the knowledge of the global biogeochemical Hgterm·dom cycle from economic production to human health, and the results are more scientific and reliable compared with some more simplified models. In the future research, setting a more realistic “no trade” scenario considering the distinctions of production structures, ecological endowment, and commodity prices among countries can help reduce the uncertainty. Additionally, investigation on the Hgterm·dom pollution and health impactsterm·key embodied in both international and intranational trades of subcountry regions (e.g. Chinese provinces) is meaningful for the implementation of emission reduction measures. The adoption of nested national–global MRIOterm·mth model can help produce some more instructive and practical policy measures for a certain country.

Policy implications A map of the global health risks brought on by trade-induced Hgterm·dom emissions is identified from the more comprehensive biogeochemical Hgterm·dom cycle at the global scale. International tradeterm·dom considerably relocates the regional distribution of Hgterm·dom challenges across the world, which enables the whole world to avoid considerable Hg-related health risks, while also resulting in contrary changespersName in health loss between developed and developing economies. To expand further, some of the gains from Hgterm·dom emission reduction in developed economies are likely to be at the expense of the ecosystem and human health in developing economies, which is a critical inequity issue between developed and developing countries. However, the transfer of heavy industries (for example, smelting and pressing of metals) from sensitive areas (for example, the coast of developed countries) to less sensitive areas (for example, inland regions of less developed countries) could alleviate the global total MeHgterm·dom exposure. As such, a fair and effective responsibility sharing and compensation mechanism need to be established for trade-induced Hg-related health risks, so as to advance international cooperation on the implementation of the Minamata Conventionterm·key on Mercuryterm·dom. For instance, organizations such as the United Nations and the World Health Organization could establish an international compensation fund in which countries benefiting from reduction in Hgterm·dom emissions should compensate the countries suffering losses based on changespersName in economic lossterm·key resulting from international tradeterm·dom.

The understanding of Hg-related dangers embedded in global trade expands upon the present mitigation strategies for Hgterm·dom, which solely aim to reduce direct emissions, and thus offers opportunities for reducing indirect emissions. To mitigate Hg-related health risks in an effective way, targeted policy methods should be implemented at various stages of global supply chains. Production-side controls work well in key regions and sectors with huge direct Hgterm·dom emissions and significant health concerns. These controls include increasing energy efficiency, phasing out outdated factories, and building Hgterm·dom removal facilities. As an example, these steps are successful for sources like nonmetal mineral products, smelting and pressing of metals, and production and supply of electricity and heat power in sub-Saharan Africa, Latin America, and China. Demand-side control strategies work well in key regions and sectors whose final consumption is responsible for significant upstream health hazards. These strategies include adjusting consumption taxes to influence consumer behavior. For example, demand-side actions are more frequently thought about when products from the construction, general and special equipment, and electric equipment and machinery sectors are consumed. Regionally, developed economies, such as the United States, Western Europe, Middle East, and North Africa, should prioritize implementation of these demand-side control measures.

Materials and methods Framework of the global Hgterm·dom risk assessment model A more comprehensive evaluation methodology is established in this work to characterize the biogeochemical Hgterm·dom cycle from economic production to human health. As such, we identify the role of international tradeterm·dom in relocation of global Hgterm·dom emissions, pollution, exposure, and related health burden. As illustrated in Fig. 6 , the model comprises five components: Hgterm·dom emission inventory, Hgterm·dom transport model, food MeHgterm·dom concentrations, MeHgterm·dom exposure, and human health impactsterm·key. We provide detailed descriptions of the methods used for each component in the following sections.

Methodology framework to access Hg-related impacts.

Hgterm·dom emission inventory To track the emissions embodied in globally traded commodities, a thorough sectoral disaggregation of production-based emissions is required. One crucial aspect of this process is accounting for atmospheric Hgterm·dom emissions, which are mainly the result of human production activities. These activities include burning fuel, smelting nonferrous metals, making building materials, mining for primary Hgterm·dom, and burning waste. Hgterm·dom emission inventory in EDGARv4.tox2 provides Hgterm·dom emissions from 14 sectors (see Table S2 ) of 225 countries. In order to match these primary anthropogenic sources to MRIOterm·mth sectors in GTAP, we aggregate the 65 sectors in GTAP into 39 aggregated sectors (see Table S3 ). For example, three sectors from GTAP (i.e. manufacture of chemicals and chemical products, manufacture of pharmaceuticals, medicinal chemical and botanical products, and manufacture of rubber and plastics products) are aggregated into the aggregated sector, manufacture of chemicals and chemical products. Then, Hgterm·dom emissions from the source of production of chemicals in EDGAR are counted to the sector. By analogy to the method, the other sources of Hgterm·dom emissions in EDGAR are counted to their corresponding sectors (see Table S4 ). As such, we obtain production-based Hgterm·dom emissions by country and by sector, which can be combined with MRIOterm·mth to calculate the consumption-based emissions of countries.

Then, the MRIOterm·mth method is employed to estimate consumption-based Hgterm·dom emissions. MRIOterm·mth has been extensively used in studying environmental consequences embedded in trade ( 42 , 43 ) as it can depict the flows of commodities along the supply chain between various industries and economies ( 44 , 45 ). The Hgterm·dom emissions flowing from importing regions (i.e. final consumers) to exporting regions (i.e. direct emitters) through global supply chains can be computed by Eq. ( 1 ). Detailed derivations of the formula refer to previous studies ( 46 ).

where E r denotes the row vector of Hgterm·dom emission coefficients (defined as the ratio of emissions produced to economic output in each sector) of the region r ; L r is the submatrix in L for the region; L = ( I − A ) − 1 is the Leontief inverse matrix; A stands for the direct consumption matrix, and each column of A denotes the amount of input needed from each sector to produce one unit output of this column sector; Y s denotes the final consumption vector of region s .

Hgterm·dom transport model The fate and transport of Hgterm·dom across various environmental compartments, including the climate, atmosphere, land, ocean, and marine ecosystem, are simulated using a linked Earth system model framework ( 29 ). The framework consists of three models: a 3D atmospheric model (GEOS-Chemterm·key), a 3D ocean model (MITgcm), and a 2D terrestrial model (GTMM). The GEOS-Chemterm·key model is used to simulate atmospheric transport, as well as its dry and wet deposition. Additionally, the model also accounts for the redox chemistry of Hgterm·dom which includes three forms of Hgterm·dom (i.e. gaseous elemental, gaseous oxidized, and particle-bound Hgterm·dom). The model is based on version 13.3.0 featuring a horizontal resolution of 4° × 5° and 47 vertical layers ( 36 ). The fate of Hgterm·dom in the soil pool and the land–atmosphere exchange is simulated by GTMM ( 39 ) which covers a single layer of the top 30 cm with a resolution of 1° × 1°. The transport, chemistry, and trophic transfer of Hgterm·dom in the ocean are simulated by the MITgcm ( 37 ), which covers 50 vertical levels with a horizontal resolution of ∼1° × 1°. The NJUCPL is used to connect the three models online and exchange Hgterm·dom among environmental media on an hourly time step ( 47 ). GEOS-Chemterm·key provides atmospheric Hgterm·dom concentration and deposition data to GTMM and MITgcm, while GTMM and MITgcm separately pass soil reemission and ocean evasion fluxes to GEOS-Chemterm·key. Detailed descriptions of these models refer to our previous work ( 29 ) and related references therein. We run the coupled model for a 10-year period with the Hgterm·dom emission inventory under “with trade” and “no trade” scenarios in 2011 based on the meteorological condition in the period between 2004 and 2014.

Food MeHgterm·dom exposure The actual MeHgterm·dom exposure under the “with trade” scenario for human beings in each country is derived from our previous studies ( 29 ). This study takes into account MeHgterm·dom exposure from seafood, freshwater fish, and rice. We gather the information for the typical MeHgterm·dom levels in these foods from the literature ( 29 ). The FAO database is used to collect the food intake inventory for each nation. To calculate the country-specific total MeHgterm·dom exposure, we add up the product of MeHgterm·dom levels and food intake for each food category by using Eq. ( 2 ).

where E WT denotes the actual MeHgterm·dom exposure under the “with trade” scenario; I seafood , I fwfish , and I rice denote the food intake for seafood, freshwater fish, and rice, respectively; C seafood , C fwfish , and C rice are the MeHgterm·dom levels for seafood, freshwater fish, and rice, respectively.

It has been demonstrated that MeHgterm·dom concentrations in ecosystems are determined by Hgterm·dom levels in environmental media in the previous studies ( 29 ). We use the literature-obtained food MeHgterm·dom concentrations under “with trade” scenario and the modeled environmental Hgterm·dom levels to scale the MeHgterm·dom levels under “no trade” scenario. The freshwater fish MeHgterm·dom concentration for a certain country is assumed to be proportional to its population-weighted Hg depositionterm·dom (D; 48 ). The rice MeHgterm·dom concentration is assumed to be proportional to the total soil Hgterm·dom concentration (S; 49 ). For the seafood, the MeHgterm·dom exposure of a certain country is assumed to be proportional to the global ocean plankton MeHgterm·dom levels weighted by its spatial distribution of fish catch (P; 50 , 51 ). Assuming that the food consumption patterns remain the same, the MeHgterm·dom exposure under “no trade” can be computed as follows: where E NT denotes the hypothetical MeHgterm·dom exposure under the “no trade” scenario; P WT , D WT , and S WT denote the plankton MeHgterm·dom concentration in the ocean, Hgterm·dom atmospheric deposition, and soil Hgterm·dom concentration under the “with trade” scenario, respectively; P NT , D NT , and S NT denote the plankton MeHgterm·dom concentration in the ocean, Hgterm·dom atmospheric deposition, and soil Hgterm·dom concentration under the “no trade” scenario, respectively.

Human health impactterm·key In this study, two health consequences of MeHgterm·dom exposure are taken into account: decline in neonatal IQ and FHAterm·dom. When assessing the potential impact of MeHgterm·dom exposure on IQ, a linear dose–response relationship is used, which assumes that any level of exposure can lead to a corresponding decrease in IQ without a safe threshold level ( 34 ): where Δ IQ is the decline in IQ points, Δ E denotes the variation in dietary food MeHgterm·dom intake, and BW represents the body weight. The coefficient β links food intake to blood Hgterm·dom levels, which is further linked by λ to the Hgterm·dom levels in hair. The coefficient γ denotes the dose–response association between hair Hgterm·dom levels and IQ decrement.

For the FHAterm·dom effect, a log-linear relationship is adopted: where Δ CF denotes the changespersName in the FHAterm·dom deaths linked to MeHgterm·dom exposure; PO P g and C f g represent the population and baseline FHAterm·dom incidence rate of gender g (male and female), respectively; φ is the dose–response coefficient between hair Hgterm·dom levels and FHAterm·dom risks; the subjective coefficient ω denotes the probability of the causality of the associations which reflects the substantial uncertainties resulting from limited epidemiological studies. The economic lossterm·key of the Hg-related health impactsterm·key ( H ) can be measured as follows: where EL denotes the lifelong earning loss resulting from IQ decline which is assigned with a value of $18,832 (2008 value) per IQ 345 point ( 6 ). VSLterm·mth denotes the value of statistical life, and a value of $6.3 million (2005 value) is adopted ( 19 ).

Supplementary Material Click here for additional data file.