Volcanic air pollution and human health: recent advances and future directions

Volcanic air pollution from both explosive and effusive activity can affect large populations as far as thousands of kilometers away from the source, for days to decades or even centuries. Here, we summarize key advances and prospects in the assessment of health hazards, effects, risk, and management. Recent advances include standardized ash assessment methods to characterize the multiple physicochemical characteristics that might influence toxicity; the rise of community-based air quality monitoring networks using low-cost gas and particulate sensors; the development of forecasting methods for ground-level concentrations and associated public advisories; the development of risk and impact assessment methods to explore health consequences of future eruptions; and the development of evidence-based, locally specific measures for health protection. However, it remains problematic that the health effects of many major and sometimes long-duration eruptions near large populations have gone completely unmonitored. Similarly, effects of prolonged degassing on exposed populations have received very little attention relative to explosive eruptions. Furthermore, very few studies have longitudinally followed populations chronically exposed to volcanic emissions; thus, knowledge gaps remain about whether chronic exposures can trigger development of potentially fatal diseases. Instigating such studies will be facilitated by continued co-development of standardized protocols, supporting local study teams and procuring equipment, funding, and ethical permissions. Relationship building between visiting researchers and host country academic, observatory, and agency partners is vital and can, in turn, support the effective communication of health impacts of volcanic air pollution to populations, health practitioners, and emergency managers.


Introduction
Globally, over a billion people are estimated to live within 100 km of an active volcano (Freire et al. 2019). Volcanic eruptions may cause injuries and fatalities via a range of hazardous phenomena (e.g., pyroclastic density currents, ballistics, lahars, lava flows, and localized accumulations or flows of asphyxiant gases such as CO 2 and H 2 S), affecting communities within tens of kilometers of the vent . Eruptions may also displace large numbers of people temporarily or permanently (Cuthbertson et al. 2020) with cascading health and social impacts including disease outbreaks due to overcrowding, food insecurity, mental health issues, and violence (Connell and Lutkehaus 2017). Airborne volcanic emissions, often referred to as "volcanic air pollution" (Tam et al. 2016;Crawford et al. 2021), can also present chronic, far-reaching hazards which may have harmful and long-lasting effects on populations across large geographic areas . Here, we address the state of knowledge regarding volcanic air pollution and health. This includes a discussion of hazard assessment methods, a summary of reported human health effects, a review of risk assessment, population preparedness and protection practices, and a discussion of emerging themes and future directions.

Volcanic emission hazards
Airborne volcanic emissions comprise variable mixtures of silicate ash, gases (H 2 O, CO 2 , SO 2 , H 2 S, CO, HF, and HCl), volatile metal vapors, and sulfate aerosol, formed through SO 2 gas-to-particle conversion ( Fig. 1; Oppenheimer et al. 2003). Ash can be generated during a variety of eruptive processes and can contain substantial amounts of respirablesized particles (<4 μm diameter) that can penetrate into the lungs (Horwell 2007). The physical and chemical properties of ash can vary significantly across eruptions and with distance (Jenkins et al. 2015). As volcanic gases cool and react in the atmosphere, they may condense into particles and/or adsorb to ash surfaces . Volcanic aerosol particles formed through gas condensation are extremely fine-grained, typically ~0.2-0.5 μm in diameter (Mather et al. 2003). Volcanic particulate matter (PM) thus encompasses a heterogeneous mixture of ash PM and acidic sulfate-and metal-bearing aerosol PM. A further airborne hazard is generated when lava flows into seawater, generating a "laze" (lava + haze) plume that contains HCl, volcanic glass fragments, and various metals ; Fig. 1).
Sulfur gases (in particular SO 2 ), sulfate aerosol, and ash are the most important airborne hazards for population-scale, longer-term impacts and have been shown to affect air quality locally as well as hundreds to thousands of kilometers from source during large fissure or explosive eruptions (e.g., Schmidt et al. 2011Schmidt et al. , 2015de Lima et al. 2012;Durant et al. 2012;Eychenne et al. 2015;Ilyinskaya et al. 2017). Many of the volatile trace elements emitted by volcanoes are classified as metal pollutants by environmental and health protection agencies (e.g., lead, zinc, arsenic, cadmium), 1 and emission rates can reach levels comparable to anthropogenic fluxes from industrialized countries . Near persistently degassing volcanoes, elevated levels of metals have been reported in air, soils, surface waters, and plants (Delmelle 2003), which are common exposure sources for humans (Prüss-Ustün et al. 2011), especially in areas where communities consume catchment or surface water and locally grown crops. Persistent degassing is also the source of fluoride contamination of water resources close to certain volcanoes, notably Ambrym and Tanna, Vanuatu (Cronin and Sharp 2002;Allibone et al. 2012;Webb et al. 2021). Acidified rainfall from persistent degassing can leach lead from plumbing fittings or roofing materials into roof catchment rainwater tanks (Macomber 2020). Ash deposition into water supplies can raise concentrations of fluoride and other potentially toxic elements (e.g., copper, manganese) as well as elements that impart an unpleasant taste or color to the water (Stewart et al. 2006(Stewart et al. , 2020. Volcanic emission components, exposure pathways, and categories of human health effects. Gaseous species are shown in red and particulate species in black. The upper plume represents airborne emissions from an explosive eruption, while the lower plume is generated by effusive activity. A steamy "laze" plume, created by interaction of molten lava with seawater, contains hydrochloric acid, vari-ous metals, and volcanic glass particles. Important plume processes include gas to particle conversion and adsorption of gas onto silicate ash surfaces. Contamination of water supplies by volcanogenic fluoride (from gaseous HF) is proposed to be the major ingestion pathway leading to human dental and skeletal effects.

Hazard and exposure assessment
In an eruption crisis, it is rare for there to be an immediate assessment of the health impact of exposure to volcanic air pollution. With limited resources, health agencies must prioritize ensuring sanitary conditions for evacuated communities and monitoring these communities for infectious disease outbreaks, as well as dealing with casualties. In lieu of data to directly measure the health impact, the physicochemical characteristics of the emissions, along with exposure concentrations and durations, may be assessed to get a first indication of whether they may be hazardous to human health.
For volcanic ash, characteristics that inform whether ash may cause harm if inhaled or ingested include particle size, particle shape, surface area, and the presence of leachable elements. Additional, specific hazards can vary according to magma composition and eruption dynamics. For lava dome-related or intermediate to felsic explosive ash samples, crystalline silica (quartz and its polymorphs) is important to quantify as it is the mineral of greatest health concern in ash due to its capacity to cause disease in industrial settings (Baxter et al. 1999;Greenberg et al. 2007). For mafic samples, reactive surface iron and associated generation of free radicals, which are implicated in respiratory diseases (Kelly 2003), can be determined ). Leachate analyses can determine concentrations of readily soluble elements on fresh ash particles relevant to inhalation or ingestion pathways. These methods may require adaptation for ash from hydrothermal system eruptions which typically contain fluoride in slowly soluble forms Stewart et al. 2020). Ash can also scavenge biologically potent organic pollutants from the atmosphere (Tomašek et al. 2021a). Toxicological assays can be used to assess whether the ash can trigger a biological response, which gives an indication of potential pathogenicity for humans .
The International Volcanic Health Hazard Network (IVHHN) 2 has developed methods and protocols for rapid, standardized screening of ash samples (Le Blond et al. 2009;Horwell 2007;Horwell et al. 2007;Stewart et al. 2020;Tomašek et al. 2021b), which have been applied during various eruption crises. Table 1 presents post-2000 studies that have determined health-relevant characteristics of ash samples and whether they have used IVHHN methods or not. The major challenges associated with ash characterization relate to timely collection of ash samples, prior capacity building and training in suitable laboratories, funding analyses, and shipping of samples, given that transportation is often disrupted during an eruption. In practice, analyses are rarely completed within the days to weeks over which acute exposures may be occurring, so cannot be relied upon to inform decision-making. Thus, in advance of future eruptions, the hazard could be informed by study of archived ash samples from historic eruptions (Hillman et al. 2012;Horwell et al. 2010bHorwell et al. , 2017Damby et al. 2017).
Exposure to volcanic emissions rarely occurs in clean atmospheres, raising concerns about co-exposures of volcanic emissions and existing air pollution, particularly in urban areas. Preliminary work on these combined hazards indicates that the specific mixture may be important, with a heightened pro-inflammatory response (in laboratory in vitro tests) reported for simultaneous exposure to respirable ash and diesel exhaust particles (Tomašek et al. 2016) but not for ash and complete gasoline exhaust (Tomašek et al. 2018).
Real-time monitoring of airborne gas and PM concentrations can be used as a proxy for assessing population exposure during eruptions, for persistent degassing, and for posteruption ash resuspension episodes . Indoor and outdoor measurements may be made via fixed monitors or portable sensors. Ambient air quality limits exist for airborne contaminants common to volcanic emissions such as PM 10 , PM 2.5 , and SO 2 , and monitoring data can be used to help alert both healthy and sensitive populations. However, air quality monitoring equipment is not installed at many volcanic locations, and installing instrumentation following eruption onset can present significant challenges (Felton et al. 2019). This can hinder agencies in making evidence-based decisions on community protection. An additional challenge to characterizing volcanic air pollution is that SO 2 and PM concentrations can vary significantly over short distances and durations (Holland et al. 2020). This issue has received significant attention recently with the introduction of low-cost fixed networks and hand-held, portable sensors that augment higher accuracy but costly regulatory air quality monitoring. These low-cost PM and SO 2 sensors perform reasonably well for monitoring volcanic air pollution in communities, as demonstrated during the Kīlauea 2018 eruption (Whitty et al. 2020;Crawford et al. 2021) and in Iceland (Gíslason et al. 2015). Air quality forecast models can complement ambient air monitoring and now play an important role in informing the public about current and predicted levels of volcanic pollution in some locations (Barsotti 2020;Holland et al. 2020).

Assessment of health effects
Post-2000 clinical and epidemiological studies conducted on communities affected by volcanic emissions are presented in Table 2. Collectively, these studies support pre-2000 findings, from studies conducted predominantly at Mount St. Helens, Soufrière Hills, and Sakurajima, that exposures to airborne volcanic emissions can exacerbate      Bioreactivity assays are divided into acellular tests (laboratory tests of particle reactivity without cells), in vitro tests (with cellular models), and in vivo tests (with animal models) 2 Ash-leachate studies are categorized by the leachant used: water, gastric (intended to mimic the chemistry of the gut), and SLF (synthetic lung fluid, which mimics the chemistry of the airways) 3 Study conducted using standardized IVHHN methods Non-infectious respiratory disease (NIRD) healthcare visits: Consistently, the annual NIRD rates increased significantly in areas exposed to ashfall when compared to NIRD rates in non-exposed areas. As ash thickness increased so did annual rates of NIRD Epidemiological descriptive comparative study over 17 years (15 exposed and 2 nonexposed) of 625 ashfall events Moderate year or years symptoms of pre-existing lung conditions (reviewed in Horwell and Baxter 2006). However, very few of these studies have followed populations longitudinally using the timeframes needed (on the order of decades) for longlatency diseases such as pneumoconioses or cancers to manifest. Further, situations that produce chronic exposure to ash are rare, with the best-documented examples being the 15-year cumulative exposure to ash from Soufrière Hills volcano, Montserrat (Baxter et al. 2014) and the eruption of Sakurajima volcano, Japan, with frequent ash exposures since the 1970s (reviewed by Hillman et al. 2012). Overall, major knowledge gaps remain about whether chronic exposures can trigger development of potentially fatal cardiorespiratory diseases and also whether chronic health effects can result from acute exposures.
Beyond respiratory health, studies of human exposure to volcanic emissions have also reported on ocular, dermal, and cardiovascular effects, gastroenteritis, birth outcomes, dental fluorosis, acute injury, and increased use of healthcare services and medications (summarized in Table 2). Documented instances of human fluorosis associated with active volcanism are rare globally (D'Alessandro 2006; Table 2) but may be under-reported. Mental health has also received attention, but few studies have addressed specific mental health impacts resulting from exposure to volcanic air pollution.
The evidence base is weak on which characteristics of volcanic or other air pollution sources are responsible for the observed negative health outcomes. Routine monitoring does not cover all species of concern (e.g., metal pollutants and aerosol acidity). In the past decade, many air pollution studies in cities with traffic-related emissions have shown the importance of fine particulate matter, particularly PM 2.5 , in the development of certain acute and chronic health conditions (respiratory, including lung cancer, and cardiovascular, in particular) and daily mortality. A future challenge will be to determine whether this applies to volcanic PM, as the World Health Organization has concluded that these outcomes relate to geogenic as well as anthropogenic particulate exposures (World Health Organization 2013).
Currently we have little clinical evidence of whether chronic exposures to volcanic crystalline silica can trigger silicosis or lung cancer (World Health Organization 1997). Some toxicology studies have indicated biological mechanisms for silica-related diseases (Lee and Richards 2004;Damby et al. 2018b). However, there is also evidence that volcanic crystalline silica may not be particularly toxic . Geochemical and crystallographic studies indicate that, as with other forms of silica dusts (International Agency for Research on Cancer 1997; Donaldson and Borm 1998), there are inherent characteristics of the silica, and external factors, which may dampen its toxicity, such as chemical (e.g. aluminum) impurities in the crystal structure or the presence of the crystalline silica within an occluding complex mineral matrix Damby et al. 2014;Nattrass et al. 2017).
Conducting high-quality studies on health effects is challenging during an eruption crisis, and the need is often secondary to emergency response. Consequently, important opportunities to study population exposures and health impacts have been missed. Furthermore, many countries with frequent volcanism do not routinely gather public health statistics, or they may have low-quality population registers and no exposure monitoring in place. These conditions make health assessment and follow-up even more challenging. It is also extremely difficult to follow a cohort of people over decades, especially if exposures of study participants are curtailed due to evacuation or permanent migration following the eruption. Obtaining funding for longitudinal studies and having the long-term support of local healthcare professionals and facilities are also great challenges.

Risk assessment and management
Increased knowledge about the hazards posed by volcanic emissions now enables risk assessments (also known as Health Impact Assessments; HIA) to be conducted prior to, or during, eruptions. To date, three such assessments have been published: Hincks et al. (2006), on crystalline silicarich ash exposures on Montserrat; and Schmidt et al. (2011) and Heaviside et al. (2021) on SO 2 /sulfate exposures from a future Laki-style eruption. Mueller et al. (2020a) recently reviewed the potential for conducting HIAs in volcanic locations to predict future morbidity and mortality due to ash exposures from eruptions, given knowledge of eruption scenarios, baseline health data, and expected exposures. They concluded that, given the scarcity of published clinical/epidemiological studies and exposure data from eruptions, the application of outdoor urban air pollution risk estimates (concentration-response functions) to eruption scenarios was the best way to estimate the impact from volcanic ash exposures. Local climate, socioeconomic status, and quality of healthcare facilities also influence vulnerability and should be included in risk calculations.
Progress is being made in integrating atmospheric, volcanological, and medical information for real-time risk management. For example, detailed modeling of volcanic plume chemistry and transport from the 2014 to 2015 Holuhraun eruption informed exposure assessment (Carlsen et al. 2021a). At Kīlauea, characterizing vog (SO 2 and aerosol concentrations) has led to improved exposure assessments for studies seeking to understand vog health impacts (Tam et al. 2016).
Civil protection exercises for volcanic eruptions are now starting to include volcanic emissions (Holland et al. 2020;Witham et al. 2020). Such preparedness steps will help to identify where risks from volcanism need to be balanced against other local background issues and environmental hazards.
Due to the knowledge gaps, especially those related to the health effects of chronic exposures (e.g., to crystalline silica), a precautionary approach is generally taken to the management of health risks. Many agencies around the world will advise communities to reduce their exposures to volcanic air pollution. Little data exists on the efficacy of intervention strategies (air purifiers, dehumidifiers, or air conditioners) on indoor air quality in a volcanic environment. However, recent studies have provided an evidence base for the efficacy of wearing personal respiratory protection to reduce exposure to volcanic ash Steinle et al. 2018). The finding that industry-certified N95-style masks are most effective but hard to source and afford has led to some humanitarian organizations donating or crowdfunding such masks . However, many government agencies distribute less-effective stockpiled masks, raising important ethical questions about the morality and legality of providing suboptimal protection (McDonald et al. 2020;McDonald and Horwell 2021). Provision of information on intervention effectiveness that is specific to local climates and cultures can help address such concerns. For example, IVHHN has produced informational products on protection from volcanic emissions, including on how to fit facemasks 3 . In Hawai'i, the advice has been tailored to local community lifestyles and published on a dedicated "vog dashboard" 4 that is a single, freely accessible source of information, supported by multiple agencies. In multiple locations, ash, gas, and aerosol dispersion forecasts are linked to health information and advice for ongoing eruptions (Businger et al. 2015;Shiozawa et al. 2018;Barsotti et al. 2020). In Iceland, volcanic air pollution forecasts have been broadcast via radio and television and are available online (including social media) ).

Emerging themes, knowledge gaps, and future directions
In general, few studies of health hazards and impacts are conducted relative to the number of eruptions that occur globally. Since 2001, the Global Volcanism Program 5 has reported 124 eruptions of VEI ≥3, while Table 2 reports 48 primary medical studies (at 23 volcanoes) assessing physical health effects of volcanic emissions. However, most of these studies were conducted in advanced-economy countries, notably the USA, Japan, and Iceland. Indonesia, with a 2021 population of ~277 million 6 and recent sustained and/or major eruptions of Merapi, Sinabung, Agung, and Semeru volcanoes, is notably under-represented, with a single clinical case study (Trisnawati et al. 2015). This inequality in attention, which relates to resources, opportunity, contacts, politics, and historical legacy, has meant that the health impacts of many major and sometimes longduration eruptions near large populations have gone completely unstudied. Additionally, with a few exceptions (e.g., Kīlauea, Holuhraun, and Miyakejima) where multiple studies of the health effects of exposure to SO 2 and sulfate aerosol are reported ( Table 2), effects of prolonged degassing have received little attention, relative to explosive eruptions, despite the chronic exposures and likely health effects.
A major research direction must be the development of methods for accurate exposure assessment. Further improvement of meteorological and dispersion models can help calculate ground-based pollutant concentrations at higher spatial and temporal resolution. Refining input parameters, plume models, and dynamic boundary layer representation, or incorporating advanced mathematical models such as Large Eddy Simulation, may also lead to much improved modeled concentrations Burton et al. 2020, Holland et al. 2020Filippi et al. 2021). Limitations in the accuracy and speciation of ground-level concentrations from models or space-based instruments will require the continuation of ground-based in situ measurements. Installation of networks of low-cost gas and particulate sensors is becoming increasingly feasible with a proliferation of technology in the past decade 7 . Such networks provide exciting opportunities for collaborative science with local communities. However, there are challenges for deployment during crises in terms of procurement and delivery in humanitarian situations where agencies have other priorities, and transport and other critical infrastructure networks may be disrupted. Currently, the utility of low-cost sensors is much greater when they are benchmarked against reference-grade instruments, which may not be available, even regionally. Future improvements in sensor accuracy, calibration, and reliable global satellite internet may contribute to better exposure assessment (Kizel et al. 2018;Crawford et al. 2021).
We also foresee that air pollution research, in general, will move beyond a reliance on PM mass concentrations to assess impact and towards an understanding of the distinct PM chemical constituents, including metals and organic compounds, as well as towards physicochemical (e.g., surface area) or biological (e.g., oxidative potential) exposure metrics.
Interactions between volcanic eruptions and the ambient atmosphere and climate are an important future research direction with respect to health impacts. Ambient conditions influence the atmospheric dispersion and lifetime of volcanic emissions (for example, the sulfur gas-to-particle conversion rate; Gíslason et al. 2015), and ash remobilization in arid, windy climates may prolong population exposure (Jarvis et al. 2020). The consequences of global climate change for volcanic emission hazards are poorly understood but likely appreciable; for example, predicted weakening of Pacific trade winds will affect dispersion of emissions in Hawai'i and Vanuatu (Collins et al. 2010).
The greatest overall barrier to advancing our understanding of volcanic air pollution effects on human health is the scarcity of epidemiological and clinical studies. To facilitate future studies, and support risk management, especially where local syndromic surveillance is absent, standardized epidemiological protocols (Mueller et al. 2020b) and crisis response resources 8 have recently been developed. Instigating such studies will be facilitated by continued co-development of standardized protocols, supporting local study teams and procuring equipment, funding, and ethical permissions. Relationship building between visiting researchers and host country academic, observatory, and agency partners is vital for preserving host countries' intellectual property and ensuring beneficial research outcomes for impacted communities. In turn, this can support the effective communication of health impacts of volcanic air pollution to populations, health practitioners, and emergency managers. gand for graphic design of the manuscript figure. We also thank two anonymous reviewers and John Ewert of the U.S. Geological Survey for their review comments, which have improved this manuscript. Any use of trade, firm or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.
Author contribution CS formulated the initial proposal. CS and DED coordinated the content. CS, DED, CJH, and TE wrote parts of, and extensively revised, the manuscript. EI, IT, AS, HKC, EM, BML, PJB, and CW wrote part of the manuscript and/or assisted with the preparation of tables and figures (Table 1 IT; Table 2 BML). SC contributed to conceptual development and review. All authors read and approved the final manuscript.
Funding CS acknowledges funding from New Zealand's Resilience to Nature's Challenges National Science Challenge. IT acknowledges the support received from the Agence Nationale de la Recherche of the French government through the program "Investissements d'Avenir" (16-IDEX-0001 CAP 20-25). AS acknowledges funding from Natural Environment Research Council grants NE/S00436X/1 and NE/ T006897/1.

Conflict of interest
The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.