Foliar Surfaces as Dust and Aerosol Pollution Monitors: An Assessment by a Mining Site

Abstract

Recent studies in the southwestern United States have shown that smelting processes and mine tailings emit heavy metal(loid)s that are distributed via wind dispersion to nearby communities. With increased attention regarding the effect of air pollution on environmental health, communities have begun to use citizen/community-based monitoring techniques to measure the concentration of metal(loid)s and evaluate their air quality. This study was conducted in a mining community to assess the efficacy of foliar surfaces as compared to an inverted disc (frisbee) to sample aerosol pollutants in ambient air. The assessment was conducted by evaluating As, Pb, Cd, Cu, Al, Ni, and Zn concentrations versus distance from a former smelter, statistical and regression analyses, and enrichment factor calculations compared to similar sites worldwide. Both the foliar and frisbee collection methods had a decrease in metal(loid)s concentration as a function of distance from the retired smelter. Statistical calculations show that the collection methods had similar mean concentrations for all of the metal(loid)s of interest; however, the tests also indicate that the frisbee collection method generally collected more dust than the foliar method. The enrichment factors from both collection methods were comparable to similar studies by other mining areas referenced, except for aluminum. Since there is evidence of enrichment, correlation between methods, and citizen/community science potential, these efforts show promise for the field. Further studies should consider alternating the types of plant used for foliar collection as well as collecting samples on a more frequent basis in order to sufficiently categorize results based on meteorological conditions.

Graphical Abstract

1. Introduction

Monitoring air pollution is necessary to mitigate its effects on public health and air quality. Compared to natural and anthropogenic sources of dust such as desertified lands, capped landfills, and construction sites, mining emissions pose an especially high threat to environmental and public health due to the high potential of contaminant concentration and emission of particulates (Csavina et al., 2012). Contaminant-laden aerosols from mining activities may contain harmful amounts of toxic species such as arsenic (As), lead (Pb), and cadmium (Cd), which can be discharged into nearby soils, waterways, and the atmosphere (Alloway, 1995; Jung, 2001; Navarro et al., 2008). This is of particular concern for arid and semi-arid regions that cover approximately one-third of the global land area (Seinfeld and Pandis, 2016), because toxic species have an increased propensity to be transported to downwind sites via wind dispersion of dust and aerosol particles (Johnson et al., 1994; Stovern et al., 2016).

A natural laboratory to examine mining emissions and dust contamination is the southwestern United States owing to the region having higher airborne dust concentrations than any other in North America (e.g. Field et al., 2010) and a large concentration of mines. Extensive research in recent years in Arizona and northern Mexico have shown that heavy metals and metal(loid)s are efficiently emitted from smelting processes and mine tailings (Camacho et al., 2011; Csavina et al., 2014) and subsequently enter the human body via inhalation and ingestion, the latter of which is especially prevalent with infants (Loh et al., 2016) and via the consumption of foods grown in contaminated soils Ramírez-Andreotta et al., 2013a, 2013b; Avila et al., 2017; Manjón and Ramírez-Andreotta, 2019). Ongoing and site-specific monitoring of contaminant-laden aerosols is of great concern, particularly since under the Clean Air Act particulate matter is measured and regulated based upon the concentration of particles of a certain diameter range, not the chemical composition (Environmental Protection Agency [EPA], n.d.).

There has been a proliferation of research and development focused on low-cost air quality monitors owing largely to the growing recognition of the deleterious impacts of both indoor and outdoor air pollution (e.g., Li et al., 2020). Techniques vary widely from optical particle counters (Crilley et al., 2018) to using photographs for estimating PM_2.5 (Pudasaini et al., 2020). An important application of sensor development is citizen-based monitoring (e.g. Jovasevic-Stojanovic et al., 2015). In this case, it is hypothesized that plant leaves can be used as collection devices for dust and aerosol particles. Foliar dust collection has been tested in several past works, and factors that were shown to impact dust capture and retention ability include meteorology, leaf shape, trichome density, branch and leaf density, and types of tree canopy (Zambrano García et al., 2009; Gajbhiye et al., 2016; Yang et al., 2016).

This study aims to assess whether dust passively collected on plant leaves (foliar dust) can serve as a low-cost air monitor and indicator of metal(loid)-laden aerosols. Inverted-disc deposition gauges (inverted frisbees) (Hall and Upton, 1998; Vallack, 1995; Stovern et al., 2016) were deployed and co-located with a pre-selected plant species near a rural, legacy mining town with wind-blown dust that is enriched with heavy metal(loid)s. Both the frisbees and plant leaves were collected, and dust was analyzed for the following metal(loid)s of concern: As, Pb, Cd, Cu, Al, Ni, and Zn. If proven successful, this simple, straightforward technique is broadly applicable to many sites where air monitoring is desired and sampling resources are limited.

2. Methods

2.1. Study and Site Description

This research is associated with the University of Arizona (UA) Gardenroots project (https://gardenroots.arizona.edu/), which assesses residential environmental quality of communities neighboring resource extraction activities through a co-created citizen science design (Ramírez-Andreotta et al, 2015; Sandhaus et al., 2019; Manjón et al., 2020). This project focuses on Superior, Arizona, a mining town with a population of approximately 3,017 (U.S. Census Bureau, 2019). The town currently is considered a legacy mining town, having been home to the Silver Queen and more recently the Magma mines. The latter was officially discontinued in late 2018 when the various structures associated with the copper smelter, such as the smoke stack, were removed. The town of Superior is shaped like a sideways “L”, nestled around a mountain range to the north and east. There is another mountain and elevated terrain to the southwest. The former smelter was located at the elbow of the “L” with houses situated both above and below the building. The elevation in the town is mostly uniform (Figure 1). Superior is approximately 110 km east and north of the major cities of Phoenix and Tucson, respectively, and qualifies as a remote site whereby emissions associated with the legacy mining site can be studied without significant interference from other major sources of nearby anthropogenic pollution (Figure 1). A major source of pollution concerning the community is windblown dust contaminated with elements such as As and Pb, as is the case with nearby mining areas in Arizona (Csavina et al., 2012). Based on local observations and historical knowledge, community champions reached out to the UA’s National Institute of Environmental Health Sciences’ Superfund Research Program in 2018 with environmental quality concerns.

Figure 1.

Map of the study region showing locations of the samples collected and the former smelter. (Google, n.d.).

As part of Gardenroots, 20 participants completed a 2-hour training on 27 July 2019 to learn how to properly collect soil, water, and dust samples from self-selected areas, mainly comprised of residential areas. Samples were collected between 27 July 2019 and 29 August 2019 as part of a community movement to understand the environmental impact that past mining has had on their area. Sixteen of the trained participants submitted frisbee dust samples and ten participants submitted foliar samples. For each individual participating in the foliar dust collection, two leaves were submitted for data analysis. Since one individual submitted frisbee and foliar samples from two different locations, there was a total of 17 frisbee samples and 22 foliar samples. For further details regarding the community recruitment, trainings, and retention procedures, the reader is referred to past works (Ramírez-Andreotta et al., 2015; Sandhaus et al., 2019).

2.2. Sample Collection

For the frisbee dust sampling, participants were given a rebar mounting pole, a 1 L trace-metal free Nalgene bottle, and a 24.1 cm diameter frisbee-cap (Figure 2a and graphical abstract). The rebar was pounded into the ground until the marked white line was level with the ground surface, allowing the frisbee to sit 0.5 m above ground. Subsequently, an upward 1 L bottle was attached with two zip ties to the pole. To capture and drain all of the precipitation into the bottle, the bottle lid was adhered to the base of the frisbee and a ~1.3 cm hole was drilled into both the frisbee and bottle lid. The frisbee was attached to the sampler bottle by screwing the sample bottle lid tightly to the bottle and left undisturbed for one month. At the end of the month, the frisbee was unscrewed and placed into a zip-loc bag. If any water had accumulated on the frisbee and had not evaporated prior to when the participant collected the sample for submission, the water was poured into the 1 L bottle before the frisbee was placed into the zip-loc bag. Finally, a bottle cap was screwed onto the 1 L bottle before removing it from the mounting pole by cutting off the zip ties. The bottles and frisbees were labeled with the participant number and date of collection.

Figure 2.

Schematic of the sampling devices: (a) Inverted-disc deposition gauge (inverted frisbee) sampling apparatus and (b) peppermint plant used for foliar analysis. (Photographs courtesy of Gardenroots).

For the foliar dust sampling, each participant was given a potted Mentha piperita (peppermint plant) and instructed to keep it outside next to the frisbee sampler for a month (Figure 2b and graphical abstract). The peppermint plant was selected based on previously collected plant leaves submitted by Gardenroots participants in 2015 (Sandhaus et al., 2019) and Environmental Scanning Electron Microscopy observations showing dust particles adhering to the surface of a mint leaf, shown in Figure 3 (Manjón et al., unpublished results). The plant was kept in a medium-shaded area, with morning sun and afternoon shade. Participants were instructed to water the plant once a day to ensure that the soil remained moist and to not let the water hit the leaves. When the samples were collected, the participants extracted two leaves (on average 11.37 ± 9.14 cm²) using tweezers from similar spots on their plants and placed each leaf into a 50 mL trace-metal free plastic centrifuge tube. The tubes were labelled according to participant number, leaf sample, and date of collection. The tubes were then placed into individual zip-loc bags and stored in each participant’s home refrigerator until the samples were ready to be prepared and processed for analysis. To ensure both the plant and frisbee were equally exposed to air movement, participants were instructed to keep the plant and frisbee close together. However potential limitations of the approach is that not all receptors were equally exposed to air movement and kept exactly at the same height. Additionally, it cannot be guaranteed that the two leaves had identical exposure to air movement, were exposing the same amount of foliar surface area during the sampling window, and that during plant watering, dust was not washed off leaves.

Figure 3.

Environmental Scanning Electron Microscopy image of a mint leaf at 1000x magnification showing dust particles with a circular trichome.

On 29 August 2019, participants were instructed to carefully collect the frisbee and foliar dust samples from their residential locations. Special care was taken with all samples during transportation to the designated sample drop-off location in Superior, AZ. Researchers from the UA met community members at the drop-off location.

2.3. Laboratory Processing

The frisbee surface was washed of visible dust with a 2% nitric acid (HNO₃) solution and the effluent was collected in the attached 1 L bottle. Due to a wide variation of dust mass collected on the frisbee, the amount of HNO₃ used for each sample bottle varied. To account for different levels of rainwater, deionized (DI) water was added to each bottle until they contained the same volume. Leaves and bugs were filtered out of each bottle through a 2 mm wire mesh and then the filtered sample was collected in a fresh trace-metal free Nalgene 1 L bottle with a total volume (dust, HNO₃, DI water) between 500–750 mL. From each fresh bottle, the solution was poured 40 mL at a time into a corresponding 40 mL glass vial and placed in the oven at 95°C for 6 hours at a time. This process was repeated approximately every day for four weeks until all the dust from each bottle was collected at the bottom of a vial. Then, 35 mL of 2% HNO₃ was pipetted into each 40 mL vial to dissolve the dust. Due to the dust solidifying at the bottom of the vials, they were then sonicated. The contents of each vial were then transferred by a pipette to a hot block digest tube (Environmental Express HotBlock® 200 SC2050–36 Heating Block), which heated every tube evenly on all sides at 85°C for 3 hours to dissolve the particulate matter into the acid solution. Samples were then submitted to Arizona Laboratory for Emerging Contaminants (ALEC) for trace metal analysis using inductively coupled plasma mass spectrometry (ICP-MS, Agilent 7700X). Instrument detection limit (IDL) values for species of interest are summarized in Table 1.

Table 1.

Summary of instrument detection limits (IDLs) for all species analyzed by the ICP-MS in units of μg L⁻¹. The frisbee and foliar datasets were run at different times, between which the instrument settings were adjusted.

Element	Frisbee	Foliar
Al	0.41	0.05
Fe	0.76	0.96
Ni	0.00	0.00
Cu	0.02	0.02
Zn	0.08	0.04
As	0.19	0.16
Pb	0.00	0.01

For foliar dust samples, 40 mL of a 2% HNO₃ solution was pipetted into each centrifuge tube to wash the leaf’s surface of dust and then agitated by gentle hand rotation for 1 minute. The leaf was then extracted from the centrifuge tube using fresh plastic forceps and placed in a Petri dish. The leaf was photographed against graph paper to calculate leaf surface area. Five mL of solution was transferred to a hot block digest tube for a 3-hour digestion at 85°C with a plastic watch glass on top of the centrifuge tube to prevent sample loss. Subsequently, samples were then removed and cooled at room temperature before being submitted to ALEC for trace metal analysis using ICP-MS.

2.4. Data Analysis

The native units of the frisbee and foliar dust data reported by ALEC were μg L⁻¹. From the ICP-MS analysis, ALEC also calculated and reported a dilution factor, instrument detection limit (IDL), and minimum level of detection (MLOD) for each sample. The MLOD was calculated by multiplying the dilution factor by the IDL. The dust concentration units were converted to μg cm⁻² to express concentration in terms of area of the sampling surface. This was done by multiplying by the reported dilution factor and the volume of the HNO₃ solution (40 mL), and then by dividing by the surface area of the frisbee or leaf, respectively. Values below the MLOD were assigned as half the MLOD (USEPA, 1991). Since each participant submitted two leaf samples, the value used for data analysis and comparison was the average concentration of the two leaves. Distances from sample sites to the former location of the Magma smelter were recorded and used for data analysis purposes.

The relationship between the concentration of metal(loid)s from the frisbee and foliar methods was examined in two ways. The first set of tests were done to analyze the differences between the metal(loid) concentrations by sampling method. These tests included a two-sample t-test, intraclass correlation, and a Bland-Altman plot.

The two-sample t-test was performed first to test the null hypothesis that the average concentration of each metal(loid) was the same for both sampling methods (p < 0.05). An intraclass correlation (ICC) was also performed as another method to quantify the similarity between the frisbee and foliar methods. A high ICC coefficient (close to 1) suggests high similarity between concentrations from the foliar and frisbee methods whereas a low ICC value (close to 0) indicates metal(loid) concentrations were different depending on the method utilized. R Studio was used to perform the ICC analysis with a two-way mixed effects model, where groups – in this case the metal(loid)s – are considered fixed. This test looked at the “absolute agreement” between groups, which measures the extent of the differences in concentration between the two sampling methods that were the same distance from the smelter.

To visually compare the sampling methods, a Bland-Altman plot was generated for each metal(loid). Bland-Altman plots are specifically used to compare two measurement techniques where one technique is considered standard, which is the frisbee sampling method for this experiment (Vallack, 1995). The x-axis represents the average of each metal(loid) for both methods at a specific distance and the y-axis represents the difference between the sampling method concentrations for that distance. Each plot has the average concentration represented as a horizontal line, and the “upper” and “lower” lines represent “limits of agreement”; this means, given that the differences are normally distributed, 95% of the data should lie between these limits.

The second set of conducted tests were linear regression analyses comparing the metal(loid) concentrations of the sampling methods, where the strength of the relationship was quantified with Pearson linear correlation coefficients. Since the number of frisbee and foliar samples did not match, samples were grouped into bins of distances from the former Magma smelter (Table 2) before comparison.

Table 2.

Concentrations (μg cm⁻²) of each contaminant of interest for both sampling techniques. The concentrations are grouped by distance (km) from the former smelter.

Distance	Frisbee							Foliar
	Pb	As	A1	Fe	Cu	Ni	Zn	Pb	As	Al	Fe	Cu	Ni	Zn
0.4 – 0.79	0.010	0.004	1.270	1.436	0.075	0.002	0.057	0.007	0.010	0.603	0.954	0.045	0.002	0.078
0.8 – 0.99	0.007	0.002	1.034	1.081	0.045	0.002	0.050	0.001	0.002	0.064	0.084	0.004	0.000	0.015
1 – 1.49	0.007	0.002	1.188	1.134	0.044	0.002	0.072	0.002	0.002	0.144	0.147	0.005	0.000	0.028
1.5 – 2.0	0.005	0.001	1.134	1.251	0.026	0.002	0.081	0.002	0.002	0.208	0.177	0.007	0.001	0.016
51.8	0.009	0.004	1.438	1.427	0.060	0.003	0.057	0.008	0.003	0.269	0.201	0.015	0.001	0.025

We quantify the degree of dust contamination with the Enrichment Factor (EF) parameter:

$$EF=\left[\frac{C_{n,sample}}{C_{ref,sample}}\right]/\left[\frac{C_{n,baseline}}{C_{ref,baseline}}\right]$$ 1

where C_n is the concentration of the contaminant in units of μg cm⁻². We consider Fe to be the reference species, C_ref, as in past work (Li et al., 2015) that is assumed to have no anthropogenic source. As a reference species, Fe has limitations, but it is likely that it underestimates EF if there are non-crustal sources of Fe. Concentrations used in the ratio in the numerator of Equation 1 come directly from the frisbee and foliar datasets and those used in the denominator are derived from previously reported crustal values (Goldschmidt 1937). A contaminant is considered to have originated from crustal origins for EF values less than 10 and EF values greater than 10 correspond to non-crustal origins, such as anthropogenic sources (Liu et al., 2002). EF values between 10 and 100 signify moderately contaminated emissions while values greater than 100 generally indicate more significant contamination (Li et al., 2015).

3. Results

3.1. Concentration versus Distance Analysis

Concentrations of the individual species measured by the frisbee and foliar collectors is summarized in Table 2 and Figure 4. The frisbee concentrations were generally higher than the foliar concentrations. Each metal(loid) had a slight concentration decrease as distance from the smelter increased. At both 0.41 km and 0.69 km, the foliar collection method had a greater concentration of every species of interest compared to the frisbee method, and this comparison was significantly greater for As. Additionally, Al and Fe were the most abundant in the dust collected by the sampling techniques, with concentrations above 1 μg cm⁻² when collected by the frisbee. For both the frisbee and foliar methods, Pb, As, and Ni had concentrations close to 0 μg cm⁻².

Figure 4.

Concentration (μg cm⁻²) versus distance (km) from the smelter for each metal(loid) of interest. The blue bars represent the frisbee concentrations and the orange bars represent the foliar concentrations.

3.2. Statistical Tests and Regression Results

The cumulative probabilities (p-value) for each metal(loid) from the two-sample t-test are listed in Table 3. Every metal(loid) had a p-value greater than the significance level (0.05); therefore, we fail to reject the null hypothesis for any of the metal(loid)s, meaning that there was statistically no difference between the average concentration for each metal(loid) gathered by the sampling methods. The results from the ICC test listed in Table 3 show that no metal(loid) had an ICC coefficient greater than 0.393, indicating poor agreement between groups.

Table 3.

Statistical analysis test results. The first four rows of the table contain the results from the two-sample t-test. The null hypothesis stated that there was no difference between the frisbee and foliar concentrations for each metal(loid). The last row lists the interclass correlation coefficients (ICC) for each contaminant of interest. Values were obtained using a two-way agreement model in R Studio.

Two-Sample t-Test	Pb	As	Al	Fe	Cu	Ni	Zn
Standard Error	0.00	0.00	0.12	0.18	0.01	0.00	0.01
Degree of Freedom	7	5	8	6	8	6	6
T Statistic	2.19	−0.64	8.30	5.37	3.07	3.53	2.39
P-value	0.97	0.27	1	0.99	0.99	0.99	0.97
ICC Coefficients	0.39	0.36	0.03	0.08	0.30	0.01	−0.11

The Bland-Altman plots for all the metal(loid)s exhibited a difference in concentrations above zero, as is shown in Figure 5, indicating that there was a bias toward one of the methods. For the plot of As, the points were somewhat scattered around (0, 0), suggesting that there was not a strong correlation between the sampling methods. Both Pb and Fe had points that partially lie along a straight line, indicating a difference in means between the sampling methods. If the points scatter to form a sloped line, that suggests that there is a difference in range for the two methods – this is partially the case for Cu, Ni, and Zn. The lower and upper limits of agreement (LoA) visually represent correlation between the two methods, with a narrower range between limits indicating stronger agreement. For all metal(loid)s, the points were within the LoA, indicating that there were no systemic biases. The LoA ranges for Pb, As, and Ni were smaller than those for Al, Fe, Cu, and Zn, meaning there was a stronger correlation between sampling methods for the former group.

Figure 5.

Bland-Altman plots for each metal(loid) of interest. The x-axis represents the average metal(loid) concentration at a specific distance for both methods and the y-axis represents the difference between the concentration for that distance. The upper and lower limits of agreement (LoA) indicate the range in which 95% of the values from the dataset lie.

The relationship between concentrations of the two sampling techniques is displayed graphically in Figure 6. The 17 frisbee and 22 foliar samples were separately grouped into 5 points based on distance from the former smelter: 0.49 – 0.79 km, 0.8 – 0.99 km, 1 – 1.49 km, 1.5 – 2.0 km, and 51.8 km. Binning was conducted to try to keep similar numbers of data points in each bin (Table 2). Each contaminant of interest was fitted with a linear trendline and corresponding slope, y-intercept, and R² value to quantify the relationship between the concentrations. Lead, As, and Fe had slopes greater than 1, with As exhibiting the largest slope (2.48). Aluminum and Cu had slopes close to 1 – 0.69 and 0.79, respectively. The magnitude of the slope of Zn was within that range as well, however, it was negative. Nickel had the smallest slope at 0.19 as well as the lowest R² value of 0.01. Other than Zn, which also had a low R² value of 0.07, the contaminants all exhibited an R² value between 0.25 – 0.79. The y-intercepts for the majority of the contaminants were within ±0.1, except for Al and Fe, which were −0.58 and −1.55, respectively.

Figure 6.

Relationships between the concentration (μg cm⁻²) of different metal(loid)s as measured by foliar (y-axis) and frisbee (x-axis) collectors. Data are grouped into bins (represented by blue circles) of distance from the retired smelter. Orange and green lines represent the 1:1 line and linear best-fit line, respectively.

3.3. Enrichment Factor Analysis

To determine the contamination of the dust emissions from the mine, the EF values were evaluated as a function of distance from the smelter (Table 4). For Al and Ni, both sampling techniques had EF values less than 10, with a range of 0.4 – 0.8 and 0.8 – 2.3, respectively. Copper and Pb exhibited EF values within 10 – 100, as well as the frisbee concentrations for As and Zn. However, As and Zn had foliar concentrations greater than 100, except for an outlier of 80.8 between 1.5 – 2.0 km for As.

Table 4.

Enrichment factor values for each contaminant of interest with respect to distance from the former smelter.

Distance (km)	Number of points		Pb		As		Al		Cu		Ni		Zn
	Frisbee	Foliar	Frisbee	Foliar	Frisbee	Foliar	Frisbee	Foliar	Frisbee	Foliar	Frisbee	Foliar	Frisbee	Foliar
0.4 – 0.79	4	6	25.0	23.4	30.8	101.2	0.5	0.4	30.5	23.1	0.8	1.1	65.2	101.5
0.8 – 0.99	3	4	20.6	40.5	20.7	200.5	0.6	0.4	21.5	23.8	0.9	2.0	60.2	213.9
1 – 1.49	4	6	28.3	31.9	28.6	122.9	0.6	0.5	29.8	19.2	1.6	1.3	224.0	223.4
1.5 – 2.0	5	4	13.8	44.9	10.1	80.8	0.5	0.7	10.7	25.6	0.9	2.3	80.5	161.4
51.8	1	2	21.2	110.7	25.9	165.5	0.6	0.8	21.3	37.6	0.9	1.9	51.2	165.2

4. Discussion

The purpose of the study was to assess whether dust passively collected on plant leaves (foliar dust) can serve as a low-cost air monitor and indicator of metal(loid)-laden aerosols. Phytotechnologies are sustainable solutions to environmental exposure challenges that can improve public health and potentially reduce the burden of disease (Henry et al., 2013) and this is the primary motivation for the work. Phytotechnologies are plant-based approaches used to detect, degrade, remove or contain contaminants in soil, groundwater, surface water, sediments, or air. They can be used as an intervention tool to protect environmental health and have attracted considerable attention due to their low cost and ancillary benefits (Simon et al, 2011; Simon et al, 2014; Wang et al., 2019).

Synthesizing the results from all the statistical tests, the null hypothesis failed to be rejected for any metal(loid) from the two-sample t-test. This suggests that there is insufficient statistical power to state that the concentrations between the two methods were significantly different. In contrast, the ICC results indicated poor agreement between the contaminant concentrations from the frisbee and foliar methods. The Bland-Altman plots indicated a bias toward one collection method and some differences in both the average and range of concentrations for some metal(loid)s. However, the LoAs for each metal(loid) indicated moderate agreement between sampling techniques overall. The Bland-Altman plots and the ICC test results imply that the frisbee apparatus collected higher concentrations of metal(loid)s than the foliar apparatus.

Another determining factor for using foliar as a reliable indicator for dust emissions comes from the correlation between the two sampling techniques. The closer the slope is to 1, the better the match between the frisbee and foliar datasets. Most of the slopes of the contaminants were close to 1 (Figure 6), indicating that the two sampling techniques yielded similar concentrations. Arsenic exhibited the largest slope at 2.48, with foliar concentrations significantly higher than the frisbee concentrations. Nickel had the smallest slope at 0.19, meaning the frisbee concentrations were much greater than the foliar concentrations. Zinc was the only contaminant with a negative slope. Similar to the results of the statistical tests, the y-intercept is used to determine if one sampling method collects higher concentrations than the other. Arsenic and Ni both exhibited y-intercepts of 0, but they also had the smallest contaminant concentrations. Similarly, Fe and Al had the highest concentrations and highest y-intercepts of 0.58 and −1.55. For all of the contaminants except As, Ni, and Zn, the values indicate that frisbees were systematically sampling more. The coefficient of determination, R², indicates the how well the given regression – in this case, linear – fits the data (Figure 6). The greatest R² value was for Pb at 0.79, followed by Cu at 0.71, indicating a good linear fit with the data. All other R² values were low, between 0.01 and 0.45, implying another model could possibly fit the data better. There is a distant outlier for most of the metal(loid)s in Figure 6 and its removal did increase the correlation coefficient for each species. However, that data bin corresponds to the sampling points closest to the smelter. Due to its proximity to the smelter and the high volume of samples comprising the data bin (defined in Table 4), we elected to use the data.

There were no overall trends present for the EF calculations except for a somewhat consistent spike at 1 – 1.49 km for the frisbee dataset. Additionally, in general, there were greater EF values across both methods at the 51.8 km distance when compared to the 2 km distance. Zn had the highest EF values compared to the rest of the contaminants of interest and can be considered moderately to significantly enriched in the dust emissions. Al and Ni had EF values less than 10, indicating their presence in the emissions can be due to crustal sources.

The EF values from this study were compared to three additional ones focused on mining sites: a site in Hayden, Arizona (Sorooshian et al., 2012), Slovakia (Demková et al., 2020), and the Hubei province (Zhou et al., 2020). The studies listed all utilized different reference species for their analyses as shown in Table 5. For all contaminants except Al, the EF values were the on same order of magnitude as the compared studies. In this study, Pb and Zn had higher EF values on average: 13.8 – 44.9 (with the exception of 110.7) and 51.2 – 213.9 compared to 1 – 21 among the three references. Copper and Ni had values lower than the other studies: 10.7 – 37.6 and 0.9 – 2.0 compared to 1 – 32 and 1 – 20, respectively.

Table 5.

Summary of EF calculation details and results for other mining sites.

Reference	EF Values	Site Location	Elements Mined	Extraction Solvent	EF Reference
Demkova 2020	Min, Mid, Max As: 3.16, 31.62, 1000 Cu: 1.58, 6.31, 31.62 Ni: 0.40, 1.58, 19.95 Pb: 0.79, 3.16, 19.95 Zn: 0.63, 12.59, 15.85	Former mining village Nižná Slaná, Slovakia	Iron ores, precious metals (Ag, pure Hg, vermilion)	5 mL of 69% HNO3, 1 mL of 30% H2O2, and 5 mL double deionized water	Based on correlation coefficient analysis
Sorooshian 2012	Average EF for multiple particle diameters (Dp) As: 1–10, with 100+ for Dp < 1.0 μm Cd: 1 – 10 Pb: 1 – 10	Hayden and Winkelman, Arizona, United States (First panel of Figure 4 in Sorooshian 2012)	Copper	15 mL of aqua regia (1.03 M HNO3/2.23 M HCl trace-metal grade)	Sc
Zhou 2020	Range Fe: 7.66 – 16.6 Ni: 0.0 – 0.3 Cu: 12.4 – 30.3 Zn: 13.5 – 20.2 Pb: 10.9 – 20.3	Huangshi city, Hubei province, People’s Republic of China	Steel, bronze	Elemental components were directly analyzed by energy dispersive X-ray fluorescence (PANalytical, Epsilon 5, Netherlands)	Ti

5. Conclusions

This study assessed the efficacy of foliar surfaces as compared to an inverted disc (frisbee) to sample aerosol pollutants in ambient air. Based on the results of this study, there is some statistical evidence to support the claim that foliar dust collects similar metal(loid) concentrations as an inverted disc (frisbee). The primary difference between the methods was the amount of dust collected over a period of a month. Additionally, with the exception of Al and Ni, metal(loid) EF values indicate non-crustal origins, such as anthropogenic sources of metal(loid)s in aerosols. Since there is evidence of enrichment, correlation between the methods, and citizen/community science potential, this study should be repeated with different types of plants, specifically plants that are native to a particular area and further diligence to ensure both receptors are equally exposed to air movement.

Another proposed study modification is more frequent sample collection to reduce the potential for dust to deposit on plant leaves and then get washed or blown away. When binning and plotting the data, we took distance to be the main deposition factor. Although that was how we chose to organize the data, deposition is governed by several other factors, including wind and precipitation. Additionally, homes that are equally distant from the smelter could have different environmental conditions, like the topography of the area, which can also affect the deposition rate. Factors that should be taken into consideration for a new study are season, wind direction, and rain periods; for example, during monsoon season, the likelihood of rain could lead to particulates being partially or completely washed away. A new study with these factors in mind could possibly account for the discrepancies in concentrations between sampling methods.

Highlights.

• Local observations/historical knowledge informed community science program

• Sampling dust/aerosol pollutants in ambient air; foliar surface vs. inverted disc

• Foliar surface collected similar metal(loid) concentrations as an inverted disc

• Inverted disc generally collected more dust than the foliar method

• Anthropogenic enrichment, correlation between methods indicates foliar is promising

Data Availability

The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request. The corresponding author is working to make the data available in a public repository.