Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2024 Nov 15.
Published in final edited form as: Environ Res. 2023 Sep 6;237(Pt 2):117091. doi: 10.1016/j.envres.2023.117091

Fine particulate matter composition in American Indian vs. Non-American Indian communities

Maggie Li 1, Vivian Do 1, Jada L Brooks 2, Markus Hilpert 1, Jeff Goldsmith 3, Steven N Chillrud 4, Tauqeer Ali 5, Lyle G Best 6, Joseph Yracheta 7, Jason G Umans 8,9, Aaron van Donkelaar 10, Randall V Martin 10, Ana Navas-Acien 1, Marianthi-Anna Kioumourtzoglou 1
PMCID: PMC10591960  NIHMSID: NIHMS1930386  PMID: 37683786

Abstract

Background:

Fine particulate matter (PM2.5) exposure is a known risk factor for numerous adverse health outcomes, with varying estimates of component-specific effects. Populations with compromised health conditions such as diabetes can be more sensitive to the health impacts of air pollution exposure. Recent trends in PM2.5 in primarily American Indian- (AI-) populated areas examined in previous work declined more gradually compared to the declines observed in the rest of the US. To further investigate components contributing to these findings, we compared trends in concentrations of six PM2.5 components in AI- vs. non-AI-populated counties over time (2000 – 2017) in the contiguous US.

Methods:

We implemented component-specific linear mixed models to estimate differences in annual county-level concentrations of sulfate, nitrate, ammonium, organic matter, black carbon, and mineral dust from well-validated surface PM2.5 models in AI- vs. non-AI-populated counties, using a multi-criteria approach to classify counties as AI- or non-AI-populated. Models adjusted for population density and median household income. We included interaction terms with calendar year to estimate whether concentration differences in AI- vs. non-AI-populated counties varied over time.

Results:

Our final analysis included 3,108 counties, with 199 (6.4%) classified as AI-populated. On average across the study period, adjusted concentrations of all six PM2.5 components in AI-populated counties were significantly lower than in non-AI-populated counties. However, component-specific levels in AI- vs. non-AI-populated counties varied over time: sulfate and ammonium levels were significantly lower in AI- vs. non-AI-populated counties before 2011 but higher after 2011 and nitrate levels were consistently lower in AI-populated counties.

Conclusions:

This study indicates time trend differences of specific components by AI-populated county type. Notably, decreases in sulfate and ammonium may contribute to steeper declines in total PM2.5 in non-AI vs. AI-populated counties. These findings provide potential directives for additional monitoring and regulations of key emissions sources impacting tribal lands.

Keywords: air pollution, American Indian, environmental health, environmental justice

Graphical Abstract

graphic file with name nihms-1930386-f0001.jpg

Introduction

Fine particulate matter (PM2.5) air pollution is a risk factor for numerous adverse health outcomes, including but not limited to respiratory and cardiovascular diseases (Hystad et al., 2020; Pope et al., 2019; Virani et al., 2020). Racial and ethnic minorities often bear disproportionate burdens of PM2.5 exposure and associated morbidity and mortality risk (Bullard, 1993; Jbaily et al., 2022; Tessum et al., 2019). Given extensive environmental degradation on and near tribal lands across the United States (US) (Fernández-Llamazares et al., 2020; Lewis et al., 2017), air pollution is a critical public health concern in many American Indian and Alaska Native (AI/AN) communities (Shriver and Webb, 2009; Woo et al., 2020). AI/AN face a high prevalence of chronic diseases such as diabetes mellitus and these populations may be more susceptible to the health effects of continued air pollution exposure (Centers for Disease Control and Prevention, 2020). Many American Indian (AI) communities are located in the rural US, where PM2.5 concentrations have not been well characterized (Dewees and Marks, 2017), as the Environmental Protection Agency (EPA) PM2.5 monitors tend to be stationed in more densely populated regions. Moreover, AI communities have historically been excluded from most air pollution epidemiological studies, and our understanding of the human health hazards in these populations is limited.

Our prior work comparing total PM2.5 mass concentrations between AI- and non-AI-populated counties across the contiguous US found a lower percent decline over time in PM2.5 in AI-populated counties and higher levels in recent years compared to other counties (Li et al., 2022). PM2.5 consists of a complex mixture of particles, which vary geographically and temporally and arise from different emission sources (Seinfeld and Pandis, 2016). Comparing levels of PM2.5 components could help elucidate the contributions of specific air pollutants and potential sources to total PM2.5 trends. This present study considers the six major PM2.5 components contributing to total PM2.5 mass: ammonium (NH4+), sulfate (SO42−), nitrate (NO3−), black carbon (BC), organic matter (OM), and mineral dust. NH4+ primarily originates from fertilizer use and livestock production (Dentener and Crutzen, 1994; Schiferl et al., 2014). SO42− and NO3− are secondary pollutants resulting from the oxidation of sulfur dioxide and nitrous oxides and mainly originate from anthropogenic fossil fuel combustion, wildfires, and volcanoes (Park, 2004). BC and OM can originate from any combustion process but mainly from biomass burning and transportation emissions, and oxidation of OM compounds can further lead to the formation of secondary organic pollutants (Briggs and Long, 2016; Lin et al., 2010). Mineral dust consists of a mixture of aluminum, silicon, calcium, iron, and titanium (van Donkelaar et al., 2019) and arises from wind erosion of particles in deserts, other landscapes (Prospero, 1999) and construction derived dust.

To evaluate trends in compositional PM2.5 over time in AI communities, we compared concentrations of NH4+, SO42−, NO3−, BC, OM, and mineral dust from 2000–2017 in the contiguous US between AI- and non-AI-populated counties using predicted estimates from a satellite-based chemical transport model. We hypothesized that trends in specific PM2.5 components, namely sulfate, organic matter, and mineral dust, would be differential by AI-populated county type, given the locations of AI-populated counties across the contiguous US. Furthermore, given varied PM2.5 composition by location, we investigated geographic differences in PM2.5 component trends by county type through stratifying by rural status and climate region.

Methods

Our analysis included 3108 study counties, derived from counties and county equivalents across the 48 contiguous US and the District of Columbia. We used a three-criteria scheme to classify study counties with substantial AI populations as “AI-populated counties” as in our prior study (Li et al., 2022). Counties were classified as AI-populated if they fit one or more of the following criteria: (1) reported greater than 5% self-identified population of AI/AN in the 2010 Census (“census” classification), (2) contained at least 20% areal overlap with a federally-recognized tribal entity (“Tribal entity” classification), or (3) were previously classified as a rural AI county in a nationwide analysis using k-means clustering (“rural cluster” classification) (Wallace et al., 2019). We included counties that did not meet these criteria as “non-AI-populated” in our analysis. In earlier work, we described the breakdown of AI-populated counties by the three classification schemes in more detail (Li et al., 2022).

We compared airborne concentrations of SO42−, NH4+, NO3−, BC, OM, and mineral dust from 2000–2017 in counties defined as AI- vs. non-AI-populated. We conducted our analysis at the county level to provide findings relevant for informing regulatory policymaking.

Annual PM2.5 Component Concentrations

We extracted airborne concentrations of particle-phase NH4+, SO42−, NO3−, BC, OM, and mineral dust from satellite-based surface PM2.5 models that provide annual predictions of each PM2.5 component at approximately 1 km2 grid resolution across the US (van Donkelaar et al., 2019). These models incorporated satellite retrievals, chemical transport modeling (GEOS-Chem), and ground-based measurements to derive total PM2.5 mass, which were then partitioned into PM2.5 compositional species using a GEOS-Chem chemical transport model simulation. Next, these PM2.5 component estimates were statistically fused with ground monitor data to yield accurate continuous surfaces despite sparse compositional monitor density in many regions of the US. These models exhibit relatively high predictive accuracies. Cross-validated R2 values for the six PM2.5 components range from 0.57 to 0.96 with the strongest for SO42− (R2 = 0.96) and NO3− (R2 = 0.90) and the lowest for OM (R2 = 0.57).

We estimated modeled annual concentrations of the six components separately for every study county by averaging the PM2.5 component concentrations in all grids with their centers contained in each study county.

Covariates

We included county-level population density and median household income from the 2010 Census to examine differences in PM2.5 component concentrations between AI-populated and non-AI populated counties independent of these factors. Given that population density and income distributions are highly skewed across US counties, we included deciles of their distributions as categorical variables in the regression models.

Statistical Analysis

We ran separate linear mixed regression models to compare annual concentrations of the six PM2.5 components in AI- versus non-AI-populated counties. The primary analysis included six regression models for each component separately, with NH4+, SO42−, NO3−, BC, OM, or mineral dust concentrations as the response variable in each model. The predictor variable of interest in our statistical models was binary AI-populated county type. We accounted for potential within-state correlation using random intercepts for each state and for correlation of observations over time in counties using nested random intercepts for counties within states. We adjusted each regression model for calendar year (categorical), population density (deciles), and median household income (deciles). To evaluate whether the trends in PM2.5 components varied over time between AI- vs. non-AI-populated counties, we included interaction terms between county type and each calendar year in the above models. Specifically, we ran one set of models including main effects only and another set additionally including county type and year interaction terms. To assess the presence of statistically significant modification of the AI-populated county-type effect measure over time, we compared the two sets of models using likelihood ratio tests.

We used R Statistical Software, version 3.6.3, to conduct all statistical analyses. The code and datasets used to run the present analyses are publicly available and accessible at: https://github.com/maggie-mengyuan-li/pm2.5_comp_ai.git.

Secondary analysis

We ran a secondary analysis stratifying by rural and non-rural counties using the 2013 National Center for Health Statistics (NCHS) Urban-Rural Classification Scheme for Counties (Rothwell et al., 2014). Rural counties comprised of nonmetropolitan (i.e., micropolitan and noncore) counties defined by the NCHS and non-rural counties included metropolitan counties. Since the relative composition of PM2.5 widely differs by climate and geography over time (Seinfeld and Pandis, 2016), we conducted a secondary analysis of the models containing county type and year interaction terms stratifying by grouped US climate regions as defined by the National Oceanic and Atmospheric Administration (NOAA), using the following categorizations: West/Northwest (combining the West and Northwest climate regions due to low count of AI-populated counties), Northern Rockies and Plains, Southwest, Upper Midwest, South, and East (combining the Ohio Valley, Northeast and Southeast climate regions).

Sensitivity Analyses

We ran a sensitivity analysis additionally adjusting for US climate regions. To account for spatial autocorrelation of PM2.5 component concentrations across counties, we also ran sensitivity analyses using geoadditive modeling with an additional tensor-product smooth term for latitude and longitude of the county centroid. For our geoadditive models, we considered 10 knots as our default for the six components. We examined a varying number of knots (5, 10, and 20) for two components, NH4+ (a relatively spatially homogenous component) and BC (a relatively spatially heterogeneous component).

Results

Of the 3108 counties included in our analysis, we classified 6.4% (n = 199) as AI-populated and 93.6% (n = 2909) as non-AI-populated. Between 2000 and 2017, the mean concentrations for total PM2.5, NH4+, SO42−, NO3−, BC, and OM were lower in AI-populated counties than non-AI-populated counties, while the mean concentration of mineral dust across the study period was higher in AI-populated counties (Table 1). The distribution of population density and household income by AI- vs. non-AI-populated counties are detailed further in Table 1.

Table 1:

Descriptive Statistics for AI- and Non-AI-Populated Counties: United States; unadjusted averages across 2000–2017.

Characteristic County Type
AI Non-AI
Number of counties (n) 199 2909
Mean (SD)
AI population (%) 18.2 (19.9) 0.6 (0.6)
Total PM2.5 (μg/m3) 6.3 (2.1) 8.4 (2.2)
NH4+ (μg/m3) 0.56 (0.34) 0.80 (0.46)
SO42− (μg/m3) 1.4 (0.83) 2.2 (1.15)
NO3− (μg/m3) 0.66 (0.42) 0.86 (0.58)
BC (μg/m3) 0.46 (0.18) 0.63 (0.25)
OM (μg/m3) 2.2 (0.82) 2.8 (0.92)
Mineral Dust (μg/m3) 0.66 (0.34) 0.57 (0.27)
Population Density (per mi2) 41 (116) 276 (1790)
Median Annual Household Income (×$1,000) 40.4 (7.6) 44.4 (11.6)
Open in a new tab

Across the contiguous US, county-level concentrations of NH4+, SO42−, and NO3−, BC, and mineral dust declined between 2000–2017 (Figure 1, bolded dashed lines). Adjusted concentrations of all components were slightly but significantly lower in AI-populated counties on average across the study period (Table 2). Using likelihood ratio tests, we detected statistically significant variation in AI-populated county type estimates over time for all six PM2.5 components.

Figure 1:

Figure 1:

Open in a new tab

Trends in observed and predicted concentrations of six PM2.5 components over the study period (2000 – 2017) for AI and non-AI populated counties. The thin, light-colored teal and brown lines display the observed (raw data) state-specific annual means for AI- and non-AI-populated counties. The bolded dashed teal and brown lines show the observed annual concentrations averaged across all 48 states for AI- and non-AI-populated counties. The bolded solid teal and brown lines indicate the estimated annual concentrations from our linear mixed regression model with county type × year interaction terms, averaged across states for AI- and non-AI populated counties, for decile groups of median annual household income between $40,300 and $42,400 and population density between 22 and 33 individuals per sq mile.

Table 2:

Mean (95% CI) difference in modeled PM2.5 component concentrations (μg/m3), comparing AI populated vs. non-AI-populated counties (2000 – 2017).

Primary analysisa Adjusted for
climate regionb 		Adjusted for
latitude/longitudec 		Stratified by Rural Statusd
(Secondary Analysis)
Rural Non-rural
NH4+ −0.05 (−0.07, −0.03) −0.05 (−0.07, −0.03) −0.01 (−0.02, 0.00) −0.04 (−0.06, −0.02) −0.04 (−0.10, 0.02)
SO42− −0.06 (−0.10, −0.02) −0.06 (−0.10, −0.02) −0.06 (−0.10, −0.02) −0.06 (−0.10, −0.01) −0.07 (−0.19, 0.05)
NO3− −0.12 (−0.16, −0.08) −0.11 (−0.16, −0.07) −0.03 (−0.05, −0.00) −0.09 (−0.13, −0.05) −0.08 (−0.19, 0.03)
BC −0.02 (−0.03, −0.01) −0.02 (−0.03, −0.01) −0.01 (−0.02, 0.00) −0.02 (−0.03, −0.00) −0.01 (−0.05, 0.03)
OM −0.06 (−0.12, −0.01) −0.06 (−0.12, −0.00) −0.03 (−0.08, 0.01) −0.04 (−0.10, 0.01) −0.09 (−0.26, 0.08)
Dust −0.03 (−0.05, −0.01) −0.03 (−0.05, −0.01) −0.01 (−0.02, −0.00) −0.01 (−0.03, 0.00) −0.01 (−0.03, −0.00)
Open in a new tab
a

Model adjusted for calendar year, population density and income deciles, and random intercepts for counties within states

b

Model additionally adjusted for NOAA climate region

c

Geoadditive model with a tensor term for latitude and longitude

d

Separate analyses for micropolitan and noncore counties (Rural) and metropolitan counties (Non-rural) defined by the CDC NCHS

On average, between 2000 and 2017, NH4+ was 0.05 μg/m3 (95% confidence interval [CI]: 0.03, 0.07) lower and SO42− was 0.06 μg/m3 (95% CI: 0.02, 0.10) lower in AI-populated counties compared to non-AI-populated counties (Table 2). Although NH4+ and SO42− concentrations were lower at baseline in AI- vs. non-AI-populated counties, this difference was attenuated over time through 2010. After around 2011, concentrations were higher in AI-populated counties compared to the other study counties for both NH4+ and SO42− (Figure 2A-B). Using estimated values from the mixed models with interaction terms, we predicted that from 2000 through 2017, NH4+ concentrations decreased by 64.7% in AI-populated counties and by 85.3% in non-AI-populated counties (Figure 1A, bolded solid lines). SO42− concentrations decreased by 39.8% in AI-populated counties and 69.7% in non-AI-populated counties (Figure 1B, bolded solid lines).

Figure 2: Mean difference in annual PM2.5 component concentrations between AI- and non-AI-populated counties: 2000–2017.

Figure 2:

Open in a new tab

The solid line represents the total estimated effect (adjusted for calendar year, random intercepts for counties within states, population density and income deciles, and county type × year interaction) of being classified as an AI-populated county vs. non-AI-populated county on annual concentrations of each PM2.5 component over the study period; the dashed lines show the 95% confidence intervals from the primary analysis. The solid red line represents no difference on average in adjusted component concentrations between AI and non-AI-populated counties.

Across the study period, NO3− was on average 0.12 μg/m3 (95% CI: 0.08, 0.16) lower in AI- compared to non-AI-populated counties (Table 2). Although we detected statistical effect modification over time and consistently lower levels in AI-populated counties across the study period, the maximal difference in NO3− in any given year between AI- and non-AI-populated counties was similar in magnitude to NH4+ and smaller in magnitude than for SO42− (Figure 2C). When we compared component concentrations using estimated values from our primary linear mixed models, NO3− concentrations between 2000 and 2017 decreased by 36.8% for AI-populated counties and 43.1% for non-AI-populated counties (Figure 1C, bolded solid lines).

Concentrations of BC were also lower in AI-populated vs. non-AI-populated counties on average between 2000 and 2017 (Table 2). Although we detected statistically significant effect modification of differences by county type over time, there was no clear pattern in this difference from year to year (Figure 2D). We estimated that BC concentrations decreased by 0.36% in AI-populated counties and by 15.3% in non-AI-populated counties over time (Figure 1D, bolded solid lines).

Between 2000 and 2017, OM and mineral dust concentrations were lower in AI-populated vs. non-AI-populated counties (Table 2). Despite statistically significant changes over time by county type for both components, we did not observe any consistent patterning in changes of this difference across the study period (Figure 2E-F). Estimated concentrations over time reflected a 12.5% increase in OM in AI-populated counties and 2.2% decrease in non-AI-populated counties. There was a 19.3% decrease in mineral dust in AI-populated counties and 17.4% decrease in non-AI-populated counties (Figure 1E-F, bolded solid lines).

Comparing rural to non-rural counties, the relative contributions of SO42− to total PM2.5 were generally higher in non-rural counties, whereas relative contributions of mineral dust to total PM2.5 were generally higher in rural counties (Supplementary Table S1). The median relative contributions of ammonium, nitrate, BC, and OM to total PM2.5 differed by less than 1% between rural and non-rural counties. The overall differences in component levels estimated across the study period between AI- and non-AI-populated counties were similar in magnitude and direction in rural counties compared to the primary analysis; in non-rural counties, the estimates also did not differ substantially from the primary analysis, with wider confidence intervals observed (Table 2). Annual estimated differences over time in mean concentrations for all six components in AI-populated vs. non-AI-populated counties were similarly patterned across rural and non-rural counties to our primary analysis (Supplementary Figure S4-5).

Evaluating county type differences over time across climate regions, the number of AI-populated counties in each stratum varied from 11 in the East (Ohio Valley, Northeast, and Southeast climate regions) to 72 in the South climate region (Supplementary Figures S2-7). In the West/Northwest, Northern Rockies and Plains, and East climate regions, there were no substantial changes in component level differences over time by AI-populated county type (Supplementary Figures S2, S3, S7). In the Southwest and Upper Midwest climate regions, concentrations of NH4+, SO42−, and NO3− were generally lower in AI-populated counties across all study years, but this difference was attenuated over time (Supplementary Figures S4-5). In the South climate region, SO42− and OM concentrations became somewhat higher after 2008 in AI-populated counties, while concentrations of NH4+ were higher during the middle of the study period (2008–2012) in AI-populated counties (Supplementary Figure S6).”

Sensitivity Analysis

When we additionally adjusted for climate region, average differences across the study period and over time did not vary considerably from our primary analysis (Table 2 and Supplementary Figure S6). In our geoadditive models, which included a tensor product between latitude and longitude to account for spatial autocorrelation in component concentrations, findings for all components were also consistent with our primary analysis (Table 2 and Supplementary Figure S7). Estimates did not substantively differ when we used 5, 10, and 20 knots (results not shown).

Discussion

We analyzed differences in six PM2.5 component concentrations from 2000 through 2017 between AI- and non-AI-populated counties across the contiguous US. For most components, levels were lower on average in AI-populated counties vs. non-AI-populated counties. When examining differences in annual concentrations over time, we detected increasing trends in AI- vs. non-AI-populated county type for NH4+ and SO42−. For BC, OM, and mineral dust there were no clear differences over time by county type; For NO3−, the differences between AI- vs. non-AI-populated counties were small; yet consistently lower for AI- vs non-AI-populated counties over the entire study period. Overall, the percent declines over time were higher in levels of NH4+, SO42−, NO3−, and BC in non-AI- vs. AI-populated counties, while for OM and mineral dust they were higher in AI- vs. non-AI-populated counties.

Furthermore, based on results from the climate region-stratified analyses, differences in NH4+ and SO42− over time by county type may be somewhat driven by differences in the Upper Midwest and South climate regions. NH4+ exists as a component of PM2.5 in two primary forms: ammonium sulfate and ammonium nitrate. Ammonium sulfate, which includes SO42−, makes up a substantially higher proportion of total PM2.5 in the Midwestern and Eastern US while ammonium nitrate, which includes NO3−, contributes to a higher proportion of total PM2.5 in the Western US. In previously published work, we estimated overall declines in PM2.5 with steeper declines in non-AI-populated counties. Our findings suggest that ammonium sulfate differences by county type are likely related to total PM2.5 differences by county type in our previous study (Li et al., 2022), given the similarity in the magnitude and direction of time trends of NH4+, SO42−, and total PM2.5 differences by county type. Ammonium sulfate levels have drastically reduced overall in non-AI-populated counties across the study period, with particularly steeper declines relative to AI-populated counties estimated after 2011. In contrast, there has been little to no decline in many AI-populated counties. Common sources of ammonium sulfate include sulfur dioxide emissions from coal-fired power plants and steel production, and ammonia from agriculture (Speight, 2017). Mass closures of coal-fired power plants in recent decades in these regions (Midwest and East) could be a factor contributing to differences in NH4+ and SO42− in AI-populated vs. non-AI-populated counties across our study period (Seinfeld and Pandis, 2016).

The existing literature on air pollution in indigenous communities is limited and findings have been mixed. A recent study reported that population-weighted average concentrations of total PM2.5 for Native Americans were lower from 2000–2016 compared to Black, White, Hispanic, and Asian communities across ZIP Code Tabulation Areas (ZCTAs) nationwide (Jbaily et al., 2022). Mortality rates in 2014 attributable to PM2.5 exposure from electricity generation sources were lower among Native Americans compared to other racial and ethnic groups overall across the US, but higher for Native Americans in Oklahoma, a state containing a substantial proportion of AI-populated counties (Thind et al., 2019). Our study also detected lower concentrations of all six components in AI-populated counties on average from 2000–2017. However, none of the previous studies examined differences in trends over time by an AI- vs non-AI classification for total PM2.5 or its components. In our previous study (Li et al., 2022), we also detected overall lower PM2.5 concentrations in AI-populated counties, but a slower rate of decrease in PM2.5 concentrations in AI-populated counties over time resulted in higher PM2.5 concentrations in more recent years. With this present study, we expand our previous findings and show that the concentrations of specific components, namely SO42− and NH4+, reflect a similar pattern of differences over time by AI-populated county type as total PM2.5 concentration.

Stronger adverse health outcomes have been detected for exposure to specific PM2.5 components. Studies have found that exposure to NH4+, SO42−, and NO3− is associated with an increased risk of asthma exacerbation and other respiratory effects (US EPA, 2019). Higher effects were explicitly estimated in regions with high correlations between these components and total PM2.5 (US EPA, 2019). One study observed an increase in all-cause mortality risk from exposure to higher levels of NH4+ and decreased risk from higher levels of NO3−, across multiple cities in different countries (Masselot et al., 2022). Increased exposure to SO42− was associated with a higher risk of cardiopulmonary mortality (Brook et al., 2010), and exposure to sulfur originating from coal combustion was positively correlated with coronary heart disease risk (Thurston et al., 2016). AI/AN populations face significantly higher burdens of cardiovascular disease and diabetes compared to white Americans and other racial and ethnic groups (Hutchinson and Shin, 2014). Existing studies have found that persons with diabetes appear to display proinflammatory responses to short- and long-term exposure to air pollution (Jacobs et al., 2010). Higher levels of air pollution were associated with higher rates of CVD incidence, hospitalizations, and mortality for persons with diabetes (Hart et al., 2015; Pereira Filho et al., 2008; Pinault et al., 2016). Thus, AI/AN populations may be at a higher risk for PM2.5- and component-specific CVD-related adverse health outcomes. Substantial barriers to accessing quality healthcare, such as geographic isolation and lack of transportation access, are prevalent in rural AI/AN communities, further exacerbating health burdens (Cromer et al., 2019). Characterizing trends in PM2.5 components and identifying those with the largest differences between AI- vs. non-AI-populated counties is critical for maximally efficient policies to protect public health equitably.

Limitations

First, the ecological design of this study limits our ability to draw conclusions and comparisons at the community or individual levels. Second, uncertainties in the predicted estimates of PM2.5 components, if correlated with temporal changes in concentrations, may induce some bias in the estimated effects of AI-populated county type on PM2.5 concentrations. Third, due to the scarcity of monitoring data of speciated PM2.5 in predominantly rural and American Indian communities, we are unable to compare patterns of ground-level concentrations between AI-populated vs. non-AI-populated counties; since this analysis incorporates modeled predictions of PM2.5 component concentrations, potential differences in predicted PM2.5 model performance in AI- vs. non-AI-populated counties given fewer monitors in highly rural areas could influence the validity of our findings and conclusions. Future work currently under development to begin to address this present limitation includes planned residential monitoring of PM2.5 components in American Indian communities in the Southwest and Great Plains. Fourth, analyzing annual concentrations of PM2.5 components cannot capture any seasonal variability in levels of various components, which could fluctuate substantially. Fifth, while broad in scope, our method of classifying AI-populated counties may not capture counties with substantial AI/AN populations. In particular, the census definition in our classification scheme for AI-populated counties incorporated data from the 2010 Census. Data from the 2020 Census showed an 86.5% increase in the self-reported AI/AN population, indicating a historical undercount of AI/AN in previous census reports (ICT, 2021). As of 2022, the majority of AI/AN lived in metropolitan areas (U.S. Department of Health and Human Services Office of Minority Health, 2023). Although we conducted a secondary analysis stratifying by rural status and did not detect substantive differences from our primary analysis findings, our county-level analysis is unable to assess exposure burdens among the large urban AI population and thus cannot generalize exposure patterns among the AI population as a whole. AI-populated counties are not randomly distributed across the US and certain climate regions have no or very few AI-populated counties, limiting the statistical power of the climate region stratified analyses; thus, these results may not be wholly representative in estimating differences between AI-populated vs. non-AI-populated communities regionally. Further research conducted at finer resolutions could potentially investigate differences in concentrations associated with distinct Indigenous cultural regions and environmental regions. There is also potential for residual confounding by socioeconomic status by factors other than household income, population density, or climate region. Finally, we could not compare levels of other components, such as trace metals, that pose substantive risks to human health. Nonetheless, the components we examined are the largest contributors to total PM2.5 mass.

Conclusions and Policy Implications

Air pollution in tribal areas is a widely known issue among many AI communities. Our previous work highlighted slower declines in total PM2.5 over time in AI- vs. non-AI-populated counties (Li et al., 2022). The examination of PM2.5 component trends revealed that specific components, namely NH4+ and SO42−, may be contributing to these slower declines in AI-populated counties. Future work investigating PM2.5 emissions in AI-populated counties could elucidate sources of air pollution impacting tribal lands and areas with large populations of AI people. Additional monitoring is also needed to further validate the ground accuracy of analyses using modeled data. The resourcefulness of tribal air programs in data collection and management invites many opportunities for collaboration and mutual learning among researchers, federal and state agencies, and tribal governments. These collaborations could aid in addressing the critical gap in air pollution research to ensure that tribal populations are being considered when informing the National Ambient Air Quality Standards. The establishment of stable funding mechanisms and institutional infrastructure can continue to improve tribal air pollution monitoring and regulatory efforts.

Supplementary Material

1

  • PM2.5 components levels on average were lower in AI-vs. non-AI-populated counties

  • NH4+ and SO42− levels lower at baseline, higher over time in AI-populated counties

  • NO3− levels lower throughout study period in AI-populated counties

  • Ammonium sulfate reflect similar trends to total PM2.5 by AI-populated county type

Acknowledgements

Support for this work was provided by the National Institute of Environmental Health Sciences, National Institutes of Health (NIH; T32 ES007322, R01 ES030616, F31 ES034645, P30 ES09089, and P42 ES033719).

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Declaration of interests

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Maggie Li: Conceptualization, Methodology, Software, Formal analysis, Visualization, Investigation, Writing - Original Draft. Vivian Do: Validation, Writing - Review & Editing. Jada L. Brooks: Writing - Review & Editing. Markus Hilpert: Writing - Review & Editing. Jeff Goldsmith: Methodology, Formal analysis, Writing - Review & Editing. Steven N. Chillrud: Writing - Review & Editing. Tauqeer Ali: Writing - Review & Editing. Lyle G. Best: Writing - Review & Editing. Joseph Yracheta: Writing - Review & Editing. Jason G. Umans: Writing - Review & Editing. Aaron van Donkelaar: Data curation, Writing - Review & Editing. Randall V. Martin: Writing - Review & Editing, Data curation. Investigation. Ana Navas-Acien: Conceptualization, Methodology, Formal analysis, Investigation, Writing - Review & Editing, Supervision. Marianthi-Anna Kioumourtzoglou: Conceptualization, Methodology, Formal analysis, Investigation, Writing - Review & Editing, Supervision.

References

  1. Briggs NL, Long CM, 2016. Critical review of black carbon and elemental carbon source apportionment in Europe and the United States. Atmospheric Environment 144, 409–427. 10.1016/j.atmosenv.2016.09.002 [DOI] [Google Scholar]
  2. Brook RD, Rajagopalan S, Pope CA, Brook JR, Bhatnagar A, Diez-Roux AV, Holguin F, Hong Y, Luepker RV, Mittleman MA, Peters A, Siscovick D, Smith SC, Whitsel L, Kaufman JD, 2010. Particulate Matter Air Pollution and Cardiovascular Disease. Circulation 121, 2331–2378. 10.1161/CIR.0b013e3181dbece1 [DOI] [PubMed] [Google Scholar]
  3. Bullard RD, 1993. Race and environmental justice in the United States. Yale J. Int’l L 18, 319. [Google Scholar]
  4. Centers for Disease Control and Prevention, 2020. National Diabetes Statistics Report, 2020 32. [Google Scholar]
  5. Cromer KJ, Wofford L, Wyant DK, 2019. Barriers to Healthcare Access Facing American Indian and Alaska Natives in Rural America. Journal of Community Health Nursing 36, 165–187. 10.1080/07370016.2019.1665320 [DOI] [PubMed] [Google Scholar]
  6. Dentener FJ, Crutzen PJ, 1994. A three-dimensional model of the global ammonia cycle. J Atmos Chem 19, 331–369. 10.1007/BF00694492 [DOI] [Google Scholar]
  7. Dewees S, Marks B, 2017. Twice Invisible: Understanding Rural Native America (Research Note No. 2). First Nations Development Institute. [Google Scholar]
  8. Fernández-Llamazares Á, Garteizgogeascoa M, Basu N, Brondizio ES, Cabeza M, Martínez-Alier J, McElwee P, Reyes-García V, 2020. A State-of-the-Art Review of Indigenous Peoples and Environmental Pollution. Integrated Environmental Assessment and Management 16, 324–341. 10.1002/ieam.4239 [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Hart JE, Puett RC, Rexrode KM, Albert CM, Laden F, 2015. Effect Modification of Long-Term Air Pollution Exposures and the Risk of Incident Cardiovascular Disease in US Women. J Am Heart Assoc 4, e002301. 10.1161/JAHA.115.002301 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Hutchinson RN, Shin S, 2014. Systematic Review of Health Disparities for Cardiovascular Diseases and Associated Factors among American Indian and Alaska Native Populations. PLOS ONE 9, e80973. 10.1371/journal.pone.0080973 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Hystad P, Larkin A, Rangarajan S, AlHabib KF, Avezum Á, Calik KBT, Chifamba J, Dans A, Diaz R, Plessis du JL, Gupta R, Iqbal R, Khatib R, Kelishadi R, Lanas F, Liu Z, Lopez-Jaramillo P, Nair S, Poirier P, Rahman O, Rosengren A, Swidan H, Tse LA, Wei L, Wielgosz A, Yeates K, Yusoff K, Zatoński T, Burnett R, Yusuf S, Brauer M, 2020. Associations of outdoor fine particulate air pollution and cardiovascular disease in 157 436 individuals from 21 high-income, middle-income, and low-income countries (PURE): a prospective cohort study. The Lancet Planetary Health 4, e235–e245. 10.1016/S2542-5196(20)30103-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. ICT, 2021. 2020 Census: Native population increased by 86.5 percent [WWW Document]. ICT. URL https://indiancountrytoday.com/news/2020-census-native-population-increased-by-86-5-percent (accessed 12.22.22). [Google Scholar]
  13. Jacobs L, Emmerechts J, Mathieu C, Hoylaerts MF, Fierens F, Hoet PH, Nemery B, Nawrot TS, 2010. Air Pollution–Related Prothrombotic Changes in Persons with Diabetes. Environ Health Perspect 118, 191–196. 10.1289/ehp.0900942 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Jbaily A, Zhou X, Liu J, Lee T-H, Kamareddine L, Verguet S, Dominici F, 2022. Air pollution exposure disparities across US population and income groups. Nature 601, 228–233. 10.1038/s41586-021-04190-y [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Lewis J, Hoover J, MacKenzie D, 2017. Mining and Environmental Health Disparities in Native American Communities. Curr Envir Health Rpt 4, 130–141. 10.1007/S40572-017-0140-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Li M, Hilpert M, Goldsmith J, Brooks JL, Shearston JA, Chillrud SN, Ali T, Umans JG, Best LG, Yracheta J, van Donkelaar A, Martin RV, Navas-Acien A, Kioumourtzoglou M-A, 2022. Air Pollution in American Indian Versus Non–American Indian Communities, 2000–2018. Am J Public Health 112, 615–623. 10.2105/AJPH.2021.306650 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Lin L, Lee ML, Eatough DJ, 2010. Review of Recent Advances in Detection of Organic Markers in Fine Particulate Matter and Their Use for Source Apportionment. Journal of the Air & Waste Management Association 60, 3–25. 10.3155/1047-3289.60.1.3 [DOI] [PubMed] [Google Scholar]
  18. Masselot P, Sera F, Schneider R, Kan H, Lavigne É, Stafoggia M, Tobias A, Chen H, Burnett RT, Schwartz J, Zanobetti A, Bell ML, Chen B-Y, Guo Y-LL, Ragettli MS, Vicedo-Cabrera AM, Åström C, Forsberg B, Íñiguez C, Garland RM, Scovronick N, Madureira J, Nunes B, De la Cruz Valencia C, Hurtado Diaz M, Honda Y, Hashizume M, Ng CFC, Samoli E, Katsouyanni K, Schneider A, Breitner S, Ryti NRI, Jaakkola JJK, Maasikmets M, Orru H, Guo Y, Valdés Ortega N, Matus Correa P, Tong S, Gasparrini A, 2022. Differential Mortality Risks Associated With PM2.5 Components: A Multi-Country, Multi-City Study. Epidemiology 33, 167–175. 10.1097/EDE.0000000000001455 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Park RJ, 2004. Natural and transboundary pollution influences on sulfate-nitrate-ammonium aerosols in the United States: Implications for policy. J. Geophys. Res 109, D15204. 10.1029/2003JD004473 [DOI] [Google Scholar]
  20. Pereira Filho MA, Pereira LAA, Arbex FF, Arbex M, Conceição GM, Santos UP, Lopes AC, Saldiva PHN, Braga ALF, Cendon S, 2008. Effect of air pollution on diabetes and cardiovascular diseases in São Paulo, Brazil. Braz J Med Biol Res 41, 526–532. 10.1590/S0100-879X2008005000020 [DOI] [PubMed] [Google Scholar]
  21. Pinault L, Tjepkema M, Crouse DL, Weichenthal S, van Donkelaar A, Martin RV, Brauer M, Chen H, Burnett RT, 2016. Risk estimates of mortality attributed to low concentrations of ambient fine particulate matter in the Canadian community health survey cohort. Environmental Health 15, 18. 10.1186/s12940-016-0111-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Pope CA, Lefler JS, Ezzati M, Higbee JD, Marshall JD, Kim S-Y, Bechle M, Gilliat KS, Vernon SE, Robinson AL, Burnett RT, 2019. Mortality Risk and Fine Particulate Air Pollution in a Large, Representative Cohort of U.S. Adults. Environmental Health Perspectives 127, 077007. 10.1289/EHP4438 [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Prospero JM, 1999. Long-range transport of mineral dust in the global atmosphere: Impact of African dust on the environment of the southeastern United States. Proceedings of the National Academy of Sciences 96, 3396–3403. 10.1073/pnas.96.7.3396 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Rothwell CJ, Madans JH, Arispe IE, 2014. National Center for Health Statistics 81. [Google Scholar]
  25. Schiferl LD, Heald CL, Nowak JB, Holloway JS, Neuman JA, Bahreini R, Pollack IB, Ryerson TB, Wiedinmyer C, Murphy JG, 2014. An investigation of ammonia and inorganic particulate matter in California during the CalNex campaign. Journal of Geophysical Research: Atmospheres 119, 1883–1902. 10.1002/2013JD020765 [DOI] [Google Scholar]
  26. Seinfeld JH, Pandis SN, 2016. Atmospheric Chemistry and Physics : From Air Pollution to Climate Change. Wiley. [Google Scholar]
  27. Shriver TE, Webb GR, 2009. Rethinking the Scope of Environmental Injustice: Perceptions of Health Hazards in a Rural Native American Community Exposed to Carbon Black. Rural Sociology 74, 270–292. 10.1111/j.1549-0831.2009.tb00392.x [DOI] [Google Scholar]
  28. Speight JG, 2017. Industrial Inorganic Chemistry, in: Environmental Inorganic Chemistry for Engineers. Elsevier, pp. 111–169. 10.1016/B978-0-12-849891-0.00003-5 [DOI] [Google Scholar]
  29. Tessum CW, Apte JS, Goodkind AL, Muller NZ, Mullins KA, Paolella DA, Polasky S, Springer NP, Thakrar SK, Marshall JD, Hill JD, 2019. Inequity in consumption of goods and services adds to racial–ethnic disparities in air pollution exposure. PNAS 116, 6001–6006. 10.1073/pnas.1818859116 [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Thind MPS, Tessum CW, Azevedo IL, Marshall JD, 2019. Fine Particulate Air Pollution from Electricity Generation in the US: Health Impacts by Race, Income, and Geography. Environ. Sci. Technol 53, 14010–14019. 10.1021/acs.est.9b02527 [DOI] [PubMed] [Google Scholar]
  31. Thurston GD, Burnett RT, Turner MC, Shi Y, Krewski D, Lall R, Ito K, Jerrett M, Gapstur SM, Diver WR, Pope CA, 2016. Ischemic Heart Disease Mortality and Long-Term Exposure to Source-Related Components of U.S. Fine Particle Air Pollution. Environmental Health Perspectives 124, 785–794. 10.1289/ehp.1509777 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. U.S. Department of Health and Human Services Office of Minority Health, 2023. Profile: American Indian/Alaska Native [WWW Document]. URL https://minorityhealth.hhs.gov/omh/browse.aspx?lvl=3&lvlid=62 (accessed 5.3.23).
  33. US EPA, 2019. Integrated Science Assessment for Particulate Matter (December 2019). [PubMed]
  34. van Donkelaar A, Martin RV, Li C, Burnett RT, 2019. Regional Estimates of Chemical Composition of Fine Particulate Matter Using a Combined Geoscience-Statistical Method with Information from Satellites, Models, and Monitors. Environ. Sci. Technol 53, 2595–2611. 10.1021/acs.est.8b06392 [DOI] [PubMed] [Google Scholar]
  35. Virani SS, Alonso A, Benjamin EJ, Bittencourt MS, Callaway CW, Carson AP, Chamberlain AM, Chang AR, Cheng S, Delling FN, Djousse L, Elkind MSV, Ferguson JF, Fornage M, Khan SS, Kissela BM, Knutson KL, Kwan TW, Lackland DT, Lewis TT, Lichtman JH, Longenecker CT, Loop MS, Lutsey PL, Martin SS, Matsushita K, Moran AE, Mussolino ME, Perak AM, Rosamond WD, Roth GA, Sampson UKA, Satou GM, Schroeder EB, Shah SH, Shay CM, Spartano NL, Stokes A, Tirschwell DL, VanWagner LB, Tsao CW, On behalf of the American Heart Association Council on Epidemiology and Prevention Statistics Committee and Stroke Statistics Subcommittee, 2020. Heart Disease and Stroke Statistics—2020 Update: A Report From the American Heart Association. Circulation 141. 10.1161/CIR.0000000000000757 [DOI] [PubMed] [Google Scholar]
  36. Wallace M, Sharfstein JM, Kaminsky J, Lessler J, 2019. Comparison of US County-Level Public Health Performance Rankings With County Cluster and National Rankings: Assessment Based on Prevalence Rates of Smoking and Obesity and Motor Vehicle Crash Death Rates. JAMA Netw Open 2, e186816. 10.1001/jamanetworkopen.2018.6816 [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Woo SHL, Liu JC, Yue X, Mickley LJ, Bell ML, 2020. Air pollution from wildfires and human health vulnerability in Alaskan communities under climate change. Environ. Res. Lett 15, 094019. 10.1088/1748-9326/ab9270 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS1930386-supplement-1.docx (4.2MB, docx)

RESOURCES