Assessment of Sources and Pollution Level of Airborne Toxic Metals through Foliar Dust in an Urban Roadside Environment

Concentrations of 19 elements (Al, Fe, Ca, K, Mg, Na, S, Ti, Ba, Sr, Zn, V, Cu, Mn, Cr, Pb, Ni, Co, and Cd) in foliar dust samples were determined from 6 different road-side locations of Bilaspur city (Chhattisgarh), India. Principal component analysis (PCA) indicated the significance of vehicular activities followed by sources such as firework events and other industrial/regional/transboundary sources in foliar dust in the area of study. Risk assessment of metal levels in foliar dust was performed using several indices based on the data collected from different sites. The geo-accumulation index ( Igeo ) analysis indicated foliar dust was moderately and extremely polluted with S and Cd, respectively, while practically unpolluted with most other elements (Al, Fe, Ca, K, Mg, Na, Ti, Ba, Sr, Zn, V, Cu, Mn, Cr, Pb, Ni, and Co). The values of pollution ( I POLL ) index and contamination factor (CF) of Cd indicated a high pollution level. Comparable results were found for the ecological risk (Er i ) of Cd (above 320) with a very high Er i at all sites. In addition, the overall Er i index ( RI ) of foliar dust at all sites was very high due to a greater Cd contribution.


INTRODUCTION
Pollution levels in the urban environment can be assessed by analyzing the structure and metal concentration of suspended dust.The toxic metal pollution in urban environments arises from many sources such as atmospheric deposition (dry and wet), degradation of vehicular parts and fluids, emission of particulate matter (PM), biological load (fallen leaves from roadside plants containing deposited airborne toxic metals) and road surface paint degradation (middle of the road) (Gajbhiye et al., 2019;Cheng et al., 2011;Liu et al., 2010).Pollution of the roadside environment is caused by the emission of PM from vehicular exhaust particles, lubricant oil residue, wear and tear of vehicle parts (break lining, plating etc.), fragmentation of tires, bushings and bearings, decomposition of batteries, cracking of bumpers, and engine exhaust (Bhandarkar, 2013).Therefore, vehicular activities are major sources of toxic metal emission in roadside environments and are mainly responsible for the contamination of roadside dusts and soil in urban areas (Karbassi et al., 2015).Toxic metals are released through vehicular exhaust and accumulated further via mixing with suspended dust.
However, several researchers used foliar dust to assess pollution levels in specific areas (Simon et al., 2014;Ugolini et al., 2013).Based on these results, in general, the source of dust metals was mainly due to anthropogenic factors (airborne sources such as vehicles and industry) (Gajbhiye et al., 2016a, b).Because foliar dust also contains very fine PM (up to respirable suspended particulate matter (RSPM)), it can easily enter the human body (Kim et al., 2017;Sgrigna et al., 2015).Hence, the presence of toxic metals in ambient dust can be very hazardous for human health and the surrounding environment (Kim et al., 2017).Qiu et al. (2009) studied S and several toxic metals including Pb, Cd, Cr, Cu, and Zn in foliar dust originated from power plants and industrial, commercial, and traffic areas in urban Huizhou (China).The statistical analysis of foliar metal data (Zn, Cu, V, Co, Pb, Cr, and Ni) showed toxic metals (Cr, Pb, and Co) were emitted from vehicular sources in Guangzhou, China (Zheng et al., 2013).Mori et al. (2015) examined foliar dust to determine roadside air pollution and reported the presence of 21 metals in the ambient air of Pescia, Italy.Similarly, air contaminants (Pb, Zn, Mn, Ni, Fe, Ca, Ba, S, and Sr) in foliar dust were also investigated in an urban area of Debrecen, Hungary (Simon et al., 2014).
In recent years, contamination levels and risks associated with toxic metals have attracted much attention and several methods are being used to evaluate those (Men et al., 2018).For instance, geo accumulation index (Igeo), contamination factor (CF), and ecological risk (Er i ) factors are generally used to evaluate contamination levels (Karbassi et al., 2008;Qingjie et al., 2008).The sampling of roadside dust and soil are important methods for assessment of toxic metal pollution in roadside environments (Ghanavati et al., 2019;Liu et al., 2019;Jadoon et al., 2018;Dehghani et al., 2017).In recent years, road dust and roadside soils have been used to analyze ecological risk through derivation of different pollution indices based on toxic metals pollution in many cities such as based on road-dust in Urumqi City, China: Wei et al., 2009;Ulsan, Korea: Duong and Lee, 2011;Shanghai, China: Zhang et al., 2013;Baotou, China: Xu et al., 2015;Rafsanja, Iran: Aminiyan et al., 2018;Nsukk, Nigeria: Mama et al., 2020 and road-soil in Qinghai-Tibet plateau: Yan et al., 2013;Ahvaz City, Iran: Ghanavati et al., 2019.Some researchers also used sediments and sewage sludge to assess the contamination level (Kirat and Aydin, 2018;Wei et al., 2016;Huang et al., 2011).Although it can be a very useful medium, foliar dust has not been used to estimate the ecological risk posed by airborne toxic elements.
In the present study, airborne toxic metals in foliar dust on numerous plants growing naturally at different roadside locations of a subtropical region in the city of Bilaspur (C.G.), India were determined.To explore the basic characteristics of pollution occurring in the urban roadside environment, a variety of metals that bound with the foliar dust were measured from different plant species using inductively coupled plasma -optical emission spectrometry (ICP-OES).Based on concentration data from the different sites, risk assessment of airborne toxic metals was conducted using several key pollution indices (geo-accumulation index (Igeo), contamination factor (CF), ecological risk (Er i ) and Er i index (RI).Igeo and CF are used to assess the contamination level of different metals in foliar dust by comparing their concentration in samples and geochemical background values.Er i represents the toxicity of metals which accounts for both the toxic response and CF.In contrast, RI estimates the potential risk which is mainly based on the toxicity of metals at a particular site.Previously, these indices have been employed to assess the level of metal pollution in sediments, soil and road dust.In the present study, these indices were also used to evaluate the factors and processes affecting toxic metal pollution that occurs in the form of foliar dust.

1 Study Sites
Bilaspur is the second largest city in the Chhattisgarh state of India.The city is administered by Municipal Corporation with a metropolitan population of 616,000 (Source: Bilaspur, India Metro Area Population, 2021).The geographical location of Bilaspur is 22°05 N 82°09 E/22°09 N 82°15 E with an average elevation of 262 m (860 ft) (Gajbhiye et al., 2016c).The climate of the city is pleasant in winter and generally hot and humid in summer.The temperature varies approximately from 9°C to 45°C.The monsoon generally arrives in the second week of June and continues until September.Average annual rainfall in the city is 1,259 mm (Source: TBC, IGKV, Bilaspur).Random sewer projects and unplanned road-ways are responsible for regular traffic jams in the city.Big industrial areas (e.g.Sirgitti, Tifra and Silpahari) are located in and around Bilaspur.Moreover, large thermal power plants of the National Thermal Power Corporation (NTPC) and mining areas of South East Central Coalfields Limited (SECL) are located around the city.
In this study, 6 different roadside sites were selected for the collection of foliar dust.To this end, emphasis was given to represent diverse number and nature of traffic activities.(1) Turkadih Arpa Bridge near Koni: a it is a bypass road which connects three national highways (NH 130 with NH 130A and NH 45).Most of the heavy vehicles passes through it with a reduced speed, (2) near Koni residential roadside area: it was affected by both heavy and light vehicular traffic which connect residential area to main road, (3) roadside in front of Guru Ghasidas Central University, Bilaspur (GGU roadside) gate: affected by moderate vehicular activities; (4) Near the Bilasa Park Gate: on the road side affected by moderate vehicular activities, (5) Agricultural college gate: near to site 4, and (6) Seepat Chowk: a major traffic square in the city affected by heavy vehicle congestion and impacted by residential/commercial activities (Fig. 1).

2 Sampling
At each study site, plant species, especially trees/ shrubs, were selected for the collection of dust depending on their availability, ease of sampling and height (Table S1).Consequently, a total of 35 dust samples were col-lected from 6 different study sites: 4 from site 1 (Annona squamosa, Calotropis procera, Citrus limon and Primula pulverea), 10 from site 2 (Annona squamosa, Pongamia pinnata, Bambusa bambos, Butea monosperma, Capparis zeylanica, Ficus religiosa, Hemidesmus indicus, Alangium lamarckii, Senna siamea and Alstonia scholaris), 11 from site 3 (Ailanthus altissima, Alstonia scholaris, Antigonon leptopus Alba, Carissa carandas, Ficus benghalensis, Mitragyna parvifolia, Paulownia tomentosa, Ricinus communis, Saraca asoca, Alangium lamarckii and Senna siamea), from site 4 (Artocarpus heterophyllus, Gmelina arborea, Psidium guajava, Senna siamea and Syzygium cumini), from site 5 (Alstonia scholaris, Butea monosperma and Mitragyna parvifolia) and 2 from site 6 (Butea monosperma and Mangifera indica).The foliar dust samples from each site were collected within 1 day to minimize temporal effects.This time represents the dry weather conditions near the end of the winter season.For foliar dust, leaf samples were collected at a height of around 1.5 m above ground level at all sites to match the ambient height and to minimize road dust re-suspension during sample collection.Three leaves (corresponding to three replicates) from each plant were collected at once toward the road side face in ziploc plastic bags and brought to the laboratory.

3 Sample Preparation
The leaves of individual plant were immersed in mL of deionized water (using a 200-mL beaker) and mixed firmly for five minutes.The water with the suspended dust was kept on a hot plate at 100°C for evaporation and recovery of the dust and 0.5 g of dust sample was separated for digestion.Sample preparation for the analysis of elemental species was performed using a microwave extraction system (Milestone, START D, USA) with nitric acid solution (US EPA Method 3051A) (Fig. 2).The dust samples were soaked in digestion reagent (HNO 3 , 10 mL) in a Teflon (perfluoroalkoxy, PFA) vessel and digested under extreme pressure and temperature conditions conditions using the following procedures: (i) Temperature and pressure of the microwave digestion system were increased up to 180°C and 100 bar, respectively, for 10 min, with 1,000 W power; (ii) The temperature and pressure were steadily maintained for 15 min under above conditions with the same power; (iii) Teflon vessels were cooled down to reduce the temperature and pressure.Digestion vessels were cleaned by operating 2 blank runs using only 10 mL HNO 3 followed by washing with ultrapure water and finally drying with air.Extracted solutions were then transferred into a volumetric flask (25 mL volume) after filtration with Whatman PVDF syringe filter (0.45 μm) and the total volume was brought up to 25 mL with ultrapure water.As men-tioned in method 3051A, it may not reflect the total metal content for certain metals (Al, Ba, Cr, Fe, Mg, and V: they require addition of HCl to be comparable with method 3050).Therefore, total mineralization methods with more aggressive acids (such as microwave-aquaregia + HF) can also be used, however, with proper care, as it can also underestimate the concentration of metals such as Pb, Al, Ca, Fe, Mg, and Ba (Chen and Ma, 2001).

5 Statistical Analysis
The Normality of data were checked by a Shapiro-Wilk test and visual inspection of normal Q-Q plot through SPSS v16.0 (Razali and Wah, 2011).The summary statistics of this normality test is shown in the form of Table S2 and Fig. S1.As such, data were not normally distributed in most cases.As 19 metals of diverse nature were quantified in present study, it was important to identify their major sources.Principal component analysis (PCA) is frequently applied in environmental pollution research to assess relative information regarding pollution sources (Li et al., 2017;Tang et al., 2017).Moreover, PCA results in a considerable reduction in the number of variables and the detection of structure in the relationships of different variables.As such, PCA was applied using all the data of different metals measured at six sites as variables through XLSTAT 2018.The PCA was run based on the correlation matrix (Pearson (n)) type which standardized and scaled up the data.As there was not any missing data, do not accept missing data option was selected.For extraction of components, Eigen value (above 1), scree plot, and the total variance explained (above 70%) were considered.The bi-plots obtained were correlation biplots.

6. 1 Igeo
As the primary index, Igeo was first used to determine the metal pollution level in each dust sample (Müller, 1969) and was calculated according to equation (Eq.1): where C n represents the measured concentration of ele-ment n and B n represents the background value of the element in fossil argillaceous samples (average shale).B n was considered the concentration of metals in the earth's crust (Taylor and McLennan, 1995).The B n values (mg/ kg) of metals were 80,400 for Al, 35,000 for Fe, 30,000 for Ca, 28,000 for K, 13,300 for Mg, 260 for S, 28,900 for Na, 3,000 for Ti, 550 for Ba, 350 for Sr, 71 for Zn, 60 for V, 25 for Cu, 600 for Mn, 35 for Cr, 20 for Pb, 20 for Ni, 10 for Co and 0.098 for Cd (Taylor and McLennan, 1995;Taylor, 1964).As the background value of metals in the foliar dust was not assessed in this study, the background concentration values of metal in earth crust was considered to calculate the ecological risk index.The background matrix correction due to terrigenous effects was factor 1.5 to reduce the effects of potential variations in the control values and lithographic diversity (Lu et al., 2009).The Igeo is classified as follows: Igeo≤0 (practically unpolluted), 0<Igeo≤1 (unpolluted to moderately polluted), 1< Igeo≤2 (moderately polluted), 2< Igeo≤3 (moderately to strongly polluted), 3<Igeo≤4 (strongly polluted), 4<Igeo≤5 (strongly to extremely polluted) and Igeo>5 (extremely polluted).The above methodology adopted to estimate the Igeo considered mean metal concentrations in earth crust due to unavailability of soil metal data from the study area.However, consideration of the actual concentration of target metals in soil samples of study area at 100 cm depth will be more representative and realistic.

6. 2 I POLL Index
To assess the intensity of metal contamination in foliar dust, pollution index (I POLL ) was calculated according to Eq. 2 by Karbassi et al., (2008): C n = metal concentration in foliar dust, B n = geo-chemical concentration of metal in earth crust.The Müller Igeo equation was modified into I POLL without correction factor (1.5) for calculating the intensity of metal contamination more precisely (Karbassi et al., 2008).I POLL is used to measure the effect of the metal pollution present in the lithosphere (Esmaeilzadeh et al., 2016).I POLL is classified according to Igeo categories.

6. 3 CF
The CF is the ratio between the concentrations of each metal in the samples and baseline or background value Eq. 3 (Hakanson, 1980).C n is the measured concentration of metal in the sample and B n is the geochemical background concentration of the metal (Taylor and McLennan, 1995).The level of pollution can be categorized based on CF values: CF≤1 low contamination, 1≤CF ≤3 moderate contamination and CF≥3 high contamination.

6. 4 The Hakanson Potential RI Method
Er i was proposed by Hakanson (1980) to quantitatively determine the potential Er i of a given contaminant using the following Eq.4: where Er i is for a given substance, Tr represents the toxic response factor of a single element pollution (Pb Xu et al., 2015); Ba = Zn = Mn = Ti = 1 (Wei et al., 2016;Zheng-Qi et al., 2008); Cd = 30 (Hakanson, 1980) and CF represents the contamination factor.The Hakanson potential RI is thus calculated according to Eq. 5: where RI represents the potential for a given region/site.The pollution category of Er i and RI values is given in Table S3.

RESULTS AND DISCUSSION
A total of 19 elements were detected in foliar dust samples.The concentrations of the elements determined from foliar dust are shown in Table 1.

1 Concentrations of Metals/Trace Elements at Different Sites
We quantified 19 elements (18 metals/transition metals/ alkali metals/alkaline earth metals and 1 non-metal [S]) in foliar dust samples.Their concentrations ranged from 1.53±0.37mg/kg for Co to 177,728±32,793 mg/kg for Fe.Subsequently, the metals were divided into 2 groups depending on their inclusion in hazardous air pollutant (HAP) listing: (i) metals/elements other than HAP list and (ii) metals in the US EPA HAP list.

1. 1 Metals/Elements other than HAP List of EPA
As shown in Table 1a, when the general pattern was compared for each site, the metal concentrations at sites 1 showed the order of Ca>Fe>Al>Mg>K>S>Na >Ti>Zn>Ba>Sr>V>Cu, respectively.Moreover, site 2 and 4 also showed the similar pattern as Ca>Fe> Mg>Al>K>S>Na>Ti>Zn>Ba>Sr>V>Cu, respectively.At site 3, the order was changed mainly due to changes in K, S, Ba, and Zn.Moreover, at other sites (site 5 and 6) the pattern of Fe>Mg>K>S>Ti>Ba was similar, however, changed for other elements.V and Cu showed the lowest concentration among all sites except site 6.

1. 2 Metals in the US EPA HAP List
(Mn, Cr, Ni, Co, Pb, and Cd) As shown in Table 1c, the metal concentrations followed the order of Mn>Cr>Pb>Ni>Cd>Co at site 1.The results obtained at sites 2, 3, and 4 followed the similar order of metal concentrations Mn>Cr>Pb> Cd> Ni> Co.At sites 4 and 6, the relative order was changed due to Pb, Cr, and Cd.In contrast, the Mn and Co showed maximum and minimum concentrations, respectively.

2 Comparison with Previous Studies
The mean concentrations of metals determined in this study were compared with concentrations reported in numerous previous studies from Calcutta, India (Chatterjee and Banerjee, 1999), Hangzhou, China (Lu et al., 2008), Huizhou, China (Qiu et al., 2009), Vienna, Austria (Simon et al., 2011), Miskolc, Hungary (Simon et al., 2016) and different locations in Bilaspur, India (Gajbhiye et al., 2016a, c).The mean Al concentration (12,352 mg/ kg) was more than 7 times higher than the Al concentration (1,683 mg/kg) in foliar dust of P. acerifolia leaf in Miskolc, Hungary.The mean Fe concentration (57,181 mg/kg) was more than 26 times higher than previously reported from other locations of Bilaspur, India (2,062 mg/kg) and Vienna, Austria (2,136 mg/kg) and 2 times higher than reported in Calcutta (21,800 mg/kg).The location of sites and traffic density is directly reflected on metal concentration in foliar dust (Zhang et al., 2017).As such, in present study, significantly high vehicular activities compared to previous study might have contributed in significant deposition and enrichment of Fe in foliar dust.If we compare the concentration of metals across different sites in the present study, it reflects the influence of environmental factors such as traffic activity.For instance, as Site 6 was main traffic square of city with the highest traffic load, it showed the highest concentration   The mean concentration of metals (Mn, Cr, Pb, Ni, Co, and Cd) in the HAP category also showed highly variable patterns.The mean Mn level (138 mg/kg) in the present study was significantly lower than in previous studies.For example, Mn level was more than 2 times lower than in Bilaspur, India (392 mg/kg) and Miskolc, Hungary (364 mg/kg).In contrast, the mean Cr concentration (24.9 mg/kg) was nearly 2 times higher than in Bilaspur, India (15.2 mg/kg) and Miskolc, Hungary (12 mg/kg).Pb (17.4 mg/kg) was slightly lower than in Bilaspur (21.3 mg/kg) and Vienna, Austria (18 mg/kg).The mean Ni concentration (4.95 mg/kg) was 7 times lower than in Calcutta (38 mg/kg).If we compare the concentration of metal sin HAP category across different sites, it reflects the influence of variable source activities (mainly traffic activity).For instance, Site 6 showed the highest concentration of HAP metals (Pb, Ni, Cr, Mn, Co, and Cd) similar to other metals among all the sites (Table 1).Moreover, the micrometeorological conditions can also affect the concentrations of these metals up to a certain degree.

3 Sources Apportionment and Factors Affecting the Distribution of Metals
If the Eigen values above 1 is considered, there were four principal components extracted (Table S4).However, as shown in the scree plot, the cumulative variance explained by two principle components were 71.8% (Fig. 3), only two principle components (PC1 and PC2) were considered.As shown in Table 2, PC1 was dominated by Cr, Co, V, Cu, Cd, Pb, Mn, Fe, Al, Ti, and Ba explaining 48.8% of total variance.The presence of metals such as Cr, Co, Cu, Cd, Mn, and Pb and with high factor loadings/correlations of 0.962, 0.945, 0.896, 0.893, 0.880, and 0.856, respectively in PC1 indicated their likely ori- gin from corrosion in diesel/gasoline vehicles (braking, wearing of rust) use of fuel, lubricant, corrosion of metallic and engine parts, breakdown of vehicle tires and body parts, and vehicular exhaust (Gajbhiye et al., 2019;Wu and Lu, 2018;Bourliva et al., 2017;Li et al., 2017).In the present study, all samples (foliar dust) were collected from the surface of leaves located near roadsides.Therefore, depending on the location of the sites and activities at the source, the primary source was apparently vehicular activities.The results of present study can be compared with previous studies conducted at sites affected by similar source characteristics.For instance, in present study, concentration of Cr (24.7 mg/kg) was considerably two times higher than previous studies such as Miskolc, Hungary (12 mg/kg) and Bilaspur, India (15.2 μg g -1 ) (Gajbhiye, 2016a;Simon et al., 2016).The mean Co concentration of present study was (2.88 mg/kg) nearly 3 orders of magnitude lower than those in Calcutta, India (9.73 mg/kg) and Guangzhou, China (8.09 mg/kg).However, it was relatively higher than those of Coimbatore, India (2.03 mg/kg) (Subpiramaniyam et al., 2021;Liang et al., 2019;Chatterjee and Banerjee, 1999).The mean concentration of Cu (13.1 mg/kg) in present study was much lower than those of Calcutta, India (269 mg/ kg); Guangzhou, China (225 mg/kg); Outer-Ring Highway of Shanghai, China (191.8 mg/kg) (Liang et al., 2019;Yin et al., 2014;Chatterjee and Banerjee, 1999).
The mean concentration of Cd (9.91 mg/kg) in foliar dust was around 4 time higher than those of Hangzhou, China (2.62 mg/kg) and Panzhihua, China (2.70 mg/kg) (Yang et al., 2016;Lu et al., 2008).In case of Pb, it was (17.4 mg/kg) 5 and 17 order of magnitude lower than those in Miskolc, Hungary (98 mg/kg) and Panzhihua, China (306.95 mg/kg), respectively (Simon et al., 2016; Yang et al., 2016).The PC2 showed 23.05 % of total variance with Zn, K, Sr, Ti, and S, as main contributors with a factor loading/ correlation of 0.875, 0.800, 0.719, 0.689, and 0.680, respectively.There are a number of studies that reported a significant increase in elements such as Zn, Sr, Ti, K, and S in urban PM due to firework events (Dickerson et al., 2017;Baranyai et al., 2015).The study area covered in this study is also affected by significant firework events during several festivals throughout the year.Moreover, there is a usual practice of huge road processions in religious, political, and marriage events during which high amount of firework products are combusted for celebrations.As such, the high loadings of these metals in PC2 can be attributed to firework events at least in partial sense.However, when the mean concentration of metals of present study was compared with the metals emitted by fireworks events in previous studies, it was found that mean concentration of Sr (23.2 mg/kg) and Ti (69.7 mg/ kg) was 20 and 600 order of magnitude higher than those of Coimbatore, India (Sr: 1.24 mg/kg; Ti: 0.10 mg/kg) (Subpiramaniyam et al., 2021).In contrast, mean concentration of S (1,425 mg/kg) and K (1,697 mg/kg) was 6 and 12 order of magnitude lower than those of 9,417 mg/ kg and 22,159 mg/kg south east part of Debrecen, Hungary, respectively (Baranyai et al., 2015).Alternatively, mean concentration of Zn (26.5 mg/kg) was 12 orders of magnitude lower than those of Debrecen, Hungary (346 mg/kg) and Vienna, Austria (311 mg/kg) (Baranyai et al., 2015;Simon et al., 2011).Hence, it indicated that the factors other than vehicular activities such as firework events along with industrial (e.g., presence of power plants for sulfur emission)/commercial and/or regional/transboundary pollution could also potentially affect the distributions of metals at roadside locations in urban environment (Yadav et al., 2018;Ismail et al., 2017).
The foliar dust samples were collected from different plants growing near a given site.Thus, when the concentration of a given metal was compared across different plants at a given site, considerable variation was observed.Most toxic metals (e.g., metals listed as HAPs, such as Mn, Cr, Co, Ni, and Pb) showed the average variation in their concentration was 2-fold or less across different plants from different sites, except Cd where it was more than 4 times at site 6.The occurrence of significantly high level of Cd at site 6 is consistent with comparable concentrations found in foliar dust samples (23.5 to 34.3 mg/kg) near the same site in other study (Tiwari and Pandey, 2016).Site 6 (Seepat chowk) is a major traffic square of the city and adjacent to Nehru chowk where NH130A and NH 49 originate from NH 130 and connects Bilaspur City to other states of India.Hence, this site has significantly high traffic density and heavy vehicular load which might have contributed in the significantly high deposition and enrichment of Cd in foliar dust samples.It has been reported that traffic and industrial emissions were two main sources of Cd including with other HAP listed metals in suspended dust around road side environment (Hou et al., 2019;Wang et al., 2010).Emission of Cd results from the gasoline use, aging of automobile tires, car body wear, and brake lining wear (Weckwerth, 2001).Moreover, Cd can also possibly be emitted from paints used on the external surface of nearby public facilities (Li et al., 2018).Moreover, the comparison of other metals (Al, Fe, Ca, Mg, and V) showed a more than 10-fold difference between their concentrations at site 3. Morphological variation in plant leaves has been reported to influence the retention and accumulation of dust on their surface (Gajbhiye et al., 2019;Chen et al., 2017).The deposition of foliar dust tends to remain temporary on leaf surfaces due to climatic conditions (rainfall, humidity, wind speed and direction, season, and temperature) (Schleicher et al., 2011;Nowak et al., 2006).Hence, the climatic conditions can also exert direct impacts on metal concentration through foliar dust.

4. 1 Igeo
Toxic metal contaminants in foliar dust were assessed with Igeo.The Igeo was calculated for all metals at different sites (Table 3a).Based on the Müller Igeo index, all metals (Al, Fe, Ca, K, Mg, Na, Ti, Ba, Sr, Zn, V, Cu, Mn, Cr, Pb, Ni, and Co) showed values in the Igeo≤0 category (practically unpolluted) except S and Cd.The highest Igeo values were found for Cd at all sites (Igeo>5), indicating extremely polluted conditions.Moreover, site 6 showed a maximum Igeo value (59.6) for Cd which represented a high traffic density square.The average Igeo for Cd (20.3) was approximately 20 times higher than road dust in Beijing, China (1.06; Tang et al., 2013) and in Huainan, China (1.00; Tang et al., 2017).The Igeo value for S ranged from 1.11 to 1.36 at sites 1 through 5, indicating a moderately polluted environment (1<Igeo ≤2).Similarly, the Igeo value for Fe (1.02) showed moderate pollution only at site 6 (1<Igeo≤2).Based on Igeo, the foliar dust of selected sites was mainly affected by Cd along with S. The Igeo values for these elements are mainly influenced by anthropogenic activity (traffic activity).The high Igeo for S can be correlated with SO 2 pollution in the area with many large thermal power plants.

4. 2 I POLL Index
The I POLL index for Al, Fe, Ca, K, Mg, Na, Ti, Ba, Sr, Zn, V, Cu, Mn, Cr, Pb, Ni, and Co at all sites ranged from 0.001 to 0.60 and showed unpolluted to moderate pollution status (I POLL below 1) except at site 6 (with an I POLL of 1.53 for Fe showing moderate pollution status relative to other metals) (Table 3).However, I POLL index for S (except site 6) was in moderately polluted category.In present study, I POLL of Cd in foliar dust was 30 times higher than those of Anzali Wetland sediments (0.92) and Hamedan soil samples (0.59) (Esmaeilzadeh et al., 2016;Mohammadpour et al., 2016) (Table 3).It reveals that foliar dust was extremely polluted in Cd.

4. 3 CF
Based on the CF computations, a list of elements (Fe, Ca, Mg, V, Cu, Cr, Pb, and S) showed moderate contamination (1<CF≤3) across different sites (Table 3).Regarding S, CF ranged from 1.99 (site 6) to 6.77 (site 5) with an average of 5.48, indicating a high level of S contamination (CF≥3).Fossil fuel emissions resulting from traffic activity, coal combustion residues and power plants can be considered as main sources of S in the form of SO 2 pollution in urban environments (Medunić et al., 2016;Zhang et al., 2012;Streets and Waldhow, 2000).In addition, the Cd CF ranged from 46.6 (site 3) to 297 (site 6) with an average of 101, corresponding to significantly high Cd contamination (CF>3).These values were significantly higher than for Cd CF in the soil (12.99) in the Tiexi District of China (Sun et al., 2010).In a previous study of Bilaspur, the enrichment factor of Cd in foliar dust ranged from 775 to 1,428 suggesting a strong source signature comparable to the present study (Gajbhiye et al., 2016c).

4. 4 Assessment of Potential Er i of Metals in Foliar Dust
The concentration of metals in foliar dust is a serious concern for the maintenance of roadside environments (flora, fauna and human health).The Hakanson method provides a quantitative evaluation for the potential Er i of different contaminant levels (Hakanson, 1980).This assessment method is a relatively rapid, simple and standard method to assess the degree of pollution caused by toxic metals as well as their adverse effects on the surrounding biological environment (Yan et al., 2013).The Er i factor and RI of 11 metals in foliar dust are summarized in Table 3d.All metals were in the category of low potential Er i with an Er i <40 (Ti, Ba, Zn, Mn, V, Cu, Cr, Pb, Ni, and Co) except for Cd.However, the Er i for Cd ranged from 1,399 (site 3) to 8,908 (site 6) with a mean value of 3,035 indicating significantly high Er i at all sites.The mean value of Cd (Er i ) was 90 and 20 times higher than the reported values in Huainan, China (94.8: Tang et al., 2017) and Xian, China, respectively (469: Li et al., 2017).The RI denotes the sum of Er i values for all met-als at a site.RI values for all sites ranged from 1,406 to 8,931 indicating a significantly high Er i .Most of the RI values were affected by a single toxic metal, Cd.

CONCLUSIONS
The concentrations of metals assessed in foliar dust samples of present study were significantly different across different plants showing differential retention capacities.Based on the results of present study, Cd and S were the most significant pollutants in foliar dust samples among all target species.Moreover, the concentration data for all metals when examined using different pollution indices indicated that most study sites were contaminated by toxic metals (especially Cd).This study also suggests that foliar dust can be used as feasible and cost effective medium to assess the presence of toxic metals in ambient air.Moreover, the results of risk assessment give direct indication towards the health risks which can be posed by PM bound toxic metals.Further research should be directed to assess the relative effect between different sources and mechanisms that control the transport, fate and behavior of these toxic metals in different matrices, especially living entities such as plants and animals.

Fig. 2 .
Fig. 2. A schematic diagram showing major steps of methodology in this study.

Fig. 3 .
Fig. 3.The results of PCA in form of different plots considering all data of concentrations of 19 metals.

Table 1 .
Mean concentration of metals in foliar dust across different sites (unit = Al, Fe, Ba, V, Cu, Mn, Cr, Pb, Ni, Co, and Cd)in comparison to other sites (Table1).The mean Ca concentration in the present study (38,520 mg/kg) was

Table 2 .
The results of principle component analyses using all the data of metal concentrations measured from 6 study sites (Values in bold are larger in magnitude).

Table 3 .
Various indices of metal pollution levels determined from foliar dust samples at different study sites

Table S1 .
Details of sampling sites for the analysis of metal concentrations in foliar dust and associated plant species.

Table S2 .
Details of summary statistics of Normality test based on Shapiro-Wilk test.

Table S3 .
Relationships among ecological risk (Er i ) factors, risk index (RI) and pollution degree.

Table S4 .
The results of PCA considering all data of concentrations of 19 metals.

Table S5 .
Concentration of 19 metals measured across different sites and shown by plant species.

Table S5 .
Continued.(c)Metals on US EPA hazardous air pollutant (HAP) list