A comprehensive approach to quantify the source identification and human health risk assessment of toxic elements in park dust

In this research, enrichment factor (EF) and pollution load index were utilized to explore the contamination characteristics of toxic elements (TEs) in park dust. The results exhibited that park dust in the study area was mainly moderately polluted, and the EF values of dust Cd, Zn, Pb, Cu and Sb were all > 1. The concentrations of Cr, Cu, Zn and Pb increased with the decrease of dust particle size. The investigation results of chemical speciation and bioavailability of TEs showed that Zn had the highest bioavailability. Three sources of TEs were determined by positive matrix factorization model, Pearson correlation analysis and geostatistical analysis, comprising factor 1 mixed sources of industrial and transportation activities (46.62%), factor 2 natural source (25.56%) and factor 3 mixed source of agricultural activities and the aging of park infrastructures (27.82%). Potential ecological risk (PER) and human health risk (HHR) models based on source apportionment were exploited to estimate PER and HHR of TEs from different sources. The mean PER value of TEs in the park dust was 114, indicating that ecological risk in the study area was relatively high. Factor 1 contributed the most to PER, and the pollution of Cd was the most serious. There were no significant carcinogenic and non-carcinogenic risks for children and adults in the study area. And factor 3 was the biggest source of non-carcinogenic risk, and As, Cr and Pb were the chief contributor to non-carcinogenic risk. The primary source of carcinogenic risk was factor 2, and Cr was the cardinal cancer risk element.


Introduction
In recent years, with the rapid development of urbanization, urban environmental contamination such as dust toxic elements (TEs) pollution is increasingly serious (Jayarathne et al., 2019). Pollution of TEs and the human health risk (HHR) caused by them have gradually become research hot spots (Han et al., 2020;Qadeer et al., 2020;Sabouhi et al., 2020). Due to the characteristics of persistence and high toxicity, TEs gradually accumulate in the environment and pose a potential threat to animals and plants (Jiang et al., 2021). Dust TEs contamination will not only inhibit the growth and development of plants, but also endanger human health through oral, inhalation and skin contact. For instance, under lead stress, photosynthesis, water metabolism and enzyme activity of plants will be restrained (Yang et al., 2015). Cadmium can not only enter the human body through the digestive tract, accumulate in the liver and kidneys, and damage the their functions, but also enter the human body through the air, thereby wrecking the central nervous system of the humanity (Khanam et al., 2020). Generally, the concentrations of TEs in dust are mainly affected by natural and human factors (Ali-Taleshi, et al., 2021;Wang et al., 2016;Zhen et al., 2020). Thus, quantifying the source, potential ecological risk (PER) and HHR of dust TEs are vital to environmental protection and human health.
Dust is a common air pollutant with large specific surface area, strong adsorption capacity and small particle size. It is a momentous carrier of TEs in urban environment (Ali-Taleshi et al., 2020a;Shi & Wang, 2021). Simultaneously, the weight of dust is relatively light, and it is easily suspended in the atmosphere for the second time under the influence of external forces, destroying the ecological environment and endangering human health Heidari et al., 2021). Therefore, systematic analysis of the contamination characteristic, particle size effect and chemical speciation of TEs in dust, and quantification of the sources and risks have great significance in determining the main pollutants and carrying out pollution prevention and control. Positive matrix factorization (PMF) model and Pearson correlation analysis have been widely applied in the source apportionment of dust TEs in recent years (Jiang et al., 2021;Stefano et al., 2021;Ye et al., 2018). The PMF model, as a typical receptor model, can effectively ensure the non-negativity of each factor and its contribution rate, and quantitatively assign the source contribution to TE. Many previous studies have exploited PMF model to identify the sources of TEs in dust and their corresponding contributions (Duan et al., 2014;Xiao et al., 2020). The results revealed that the PMF model can validly discriminate the priority sources of TEs and provide a scientific basis for pollution prevention and control. Accordingly, PMF model was employed in this study to analyze the sources of TEs in park dust.
Wuhan is a central city in the central China and an important industrial base, science and education base, and comprehensive transportation hub in the country. Parks, as places for outdoor recreation and entertainment, are the important places for people exposed to TEs. Moreover, TEs accumulate continuously in park dust due to the influence of urban traffic activities, agricultural activities, industrial activities, urban constructions and the migration of natural elements (Zhao et al., 2020). Therefore, it is necessary to investigate and study TEs pollution in park dust. Previously, scholars have studied the pollution of dust TEs in areas around roads (Safiur Rahman et al., 2019), railway stations  and large factories. However, there were few systematic studies on the particle size, chemical speciation, pollution characteristic and risk assessment of TEs in park dust. Consequently, the intents of this research were (1) to estimate contamination level of park dust TEs through EF and PLI; (2) to explore the relationship between the content of dust TEs and particle size of dust; (3) to research the chemical speciation and bioavailability of dust TEs; (4) to apportion the potential origins of dust TEs with PMF model, Pearson correlation analysis and geostatistical analysis; and (5) to quantify the potential ecological risk and human health risk applying the PER and HHR model based on PMF. The results are useful for revealing the impact extent of human activities (industry, transportation and agricultural activities) on the environment of parks, determining priority contamination sources and providing a scientific basis for the green space planning of city.

Study area
Wuhan (113°41′-115°05′ E, 29°58′-31°22′ N) is the capital of Hubei Province and the only sub-provincial city in the six central provinces of China (Fig. 1). The city has 13 districts with a total area of 8569.15 km 2 . At the end of 2019, the population was 11.212 million (WLHO, 2020). Wuhan is located in the east of Hubei Province, central China. The rivers in the city are vertical and horizontal, and the water area accounts for a quarter of total area in the city. As the economic center and geographic center of China, Wuhan is known as the "Nine-province thoroughfare." It is the largest water, land and air transportation hub in inland China and the shipping core in the middle reaches of the Yangtze River. It is also the principal industrial base in China, with complete industrial systems of steel, automobile, photovoltaic, chemical industry, metallurgy, textile, shipbuilding, manufacturing, medicine and so on. The climate of Wuhan is northern subtropical monsoon (humid) climate, with abundant rainfall throughout the year. It has the characteristics of sufficient heat, rain and heat in the same season, and four distinct seasons. The annual precipitation is 1150-1450 mm, and the rainfall is primarily concentrated from June to August each year, accounting for about 40% of the annual rainfall.

Sample collection and chemical analysis
From October to December 2020, 40 typical parks were selected from Wuhan City as the research objects. According to the actual situation of each park, sampling points were evenly arranged on the roads such as park entrances, resting places, entertainment spots and greenways where tourists often gathered. There were at least 20 split points in each park, and samples from all sampling points in each park were mixed into a sample. The sample weight was about 500 g, and a total of 40 samples of park dust were gathered. To guarantee the representativeness, the samples were collected on sunny and windless days, and there was no rainfall in the week prior to sampling. Dust samples were placed in polyethylene sealed bags and sent to laboratory. Dust samples were placed in the room to dry naturally for a week, then dried in a blast drying oven at 60 °C for 12 h to make them fully dry. Samples were passed through a 2 mm nylon sieve to remove impurities, and then passed through 60-mesh, 100-mesh and 200-mesh sieve in sequence. The improved BCR three-step continuous extraction method was applied to analyze the chemical speciation of TEs in park dust which were divided into weak acid extraction state, oxidizable state, reducible state and residue state (Shiowatana et al., 2001). The dust samples were digested with nitric acid, hydrochloric acid and hydrofluoric acid, and the concentration of TEs (Cr, Mn, Ni, Cu, Zn, As, Cd, Sb and Pb) was determined by an inductively coupled plasma mass spectrometer (ICP-MS) with model NexION manufactured by Perkins Elmer (Ali-Taleshi et al., 2020b;Wang et al., 2019). Standard samples (GSS-1 and GBW07401), parallel samples and blank samples were employed for quality control. The relative deviation of the parallel samples was less than 20%, and the recovery rate of the standard materials was between 92 and 114%. The detection limits of Cr, Mn, Ni, Cu, Zn, As, Cd, Sb and Pb were 2, 0.4, 1, 0.6, 1, 0.4, 0.09, 0.08 and 2 mg kg −1 , respectively.

Enrichment factor (EF)
In this research, EF was applied to calculate the enrichment level of TEs in dust, and the calculation formula was (Maedeh et al., 2021): where (C i ∕B r ) sample represents the ratio of the TE to be measured and the content of the reference element and (C r ∕B r ) background represents the ratio of the background value of target TE to the background value of the reference element. Generally, elements with stable chemical properties and not easily affected by the external environment are selected as reference elements. Familiar reference elements are Mn, Ti, Al, Fe, etc. Mn was selected as the reference element in this research, due to its high content and stable properties in the crust. Besides, the data of soil background value of Hubei province (CNEMC, 1990) were utilized in formula calculation. The enrichment degree of TEs was divided into six levels, specifically: not enriched (EF < 1); mildly enriched (1-2); moderately enriched (2-5); significantly enriched (5-20); highly enriched (20-40); and exceedingly enriched (EF ≥ 40) (Jiang et al., 2020).

Contamination factor (CF) and pollution load index (PLI)
CF and PLI calculated the contamination level of each TE and the comprehensive pollution characteristics of all TEs, respectively (Ihl et al., 2015). And the CF and PLI were circumscribed as follows: where C i is the concentration of TE i and C b is the background value of the TE i . The soil background value of Hubei Province was used as the evaluation standard in this study (CNEMC, 1990). According to the PLI value, TE contamination was carved up 4 levels, namely no contamination (PLI < 1), medium contamination (1-2), severe contamination (2-5) and extreme contamination (PLI > 5) (Huang et al., 2022a).

Potential ecological risk model based on PMF
The PER assessment method and the PMF model were combined to calculate the single-factor potential ecological hazard index E i r and the PER of TEs in different sources. Firstly, the mass contribution of every TE should be computed combine PMF, which were calculated as: where * C l jk is the mass contribution (mg kg −1 ) of kth TE from lth source in the jth sample, n% represents the ratio of kth TE of the jth sample from different sources and C jk is the content (mg kg −1 ) of kth TE of the jth sample.
Afterward, the PERs of dust TEs from different sources were reckoned, and the calculation was as follows: Among them, * E i r l jk is the reckoned single PER of kth TE from lth source in the jth sample; C b is the background value of kth TE; and T i r is the toxicity coefficients of TE. Taking into account the accuracy of the PER assessment results, this study re-adjusted the ecological risk classification standards of dust TEs based on the toxicities and types of TEs. The specific method was as follows: In the formula, UTCV represents the unit toxicity coefficient classification value; l 1 is the first-level threshold values of Hakanson (1980); STC represents the sum of toxicity coefficients of all elements; TC represents the total values of the toxicity coefficients of TEs (the toxicity coefficients of Cr, Mn, Ni, Cu, Zn, As, Cd, Sb and Pb were 2, 1, 5, 5, 1, 10, 30, 5 and 5) in this research; and L 1 represents the adjusted first-level limit value, and then, the next-level boundary value was calculated according to the first-level boundary value; each level boundary value was twice the upper-level boundary value. The PER was the ecological risk standard for reclassification, and grading standards are given in Table S1.

Human health risk model based on PMF
HHR based on PMF was utilized to quantify the carcinogenic and non-carcinogenic risks of dust TEs from different sources and different routes (ingestion, inhalation and skin contact). As the same with PER based on PMF, the first step was to run the formula (4). Later, the average daily exposure doses (ADDs) of TEs from different channels were calculated as follows: The reference values of IngR, EF, ED, BW, AT, InhR, PEF, SA, AF and ABS are given in Table S2.
Non-carcinogenic risk was quantitatively characterized by hazard index (HI), and HI was calculated from the sum of the quotient of the ADD of each TE and its corresponding reference dose (RfD) under different exposure routes. The calculation method was as follows: When HI > 1, it indicates that single TE had noncarcinogenic risk; and when THI > 1, it indicates that TEs calculated as entirety showed non-carcinogenic risk. Otherwise, there was no risk of non-carcinogenic (Huang et al., 2022b). The carcinogenic risk was circumscribed by the product of the ADDs of TEs and the corresponding slope factors (SF). TCRI was the total carcinogenic risk in the region, which was the sum of the CR of the carcinogenic risk of various TEs. TCRI was depicted as formula (13): If CR > 10 −4 , it illustrated that humanity come under the carcinogenic risk of a single TE. If TCRI > 10 −4 , it signified that human was exposed to the overall carcinogenic risk of TEs. If CR and TCRI were between 10 −6 and 10 −4 , it indicated that human was within the acceptable carcinogenic risk range. Else, there was no carcinogenic risk. Furthermore, the parameters of formulas (12) and (13) are given in Table S3.

Concentration characteristics of TEs in park dust
Concentration characteristics of 9 TEs in park dust are given in Table 1. The average concentrations of Cu, Zn, Cd, Sb and Pb were higher than their corresponding soil background values of Hubei Province (CNEMC, 1990), with 1.54, 2.20, 2.24, 1.07 and 1.71 times of background values, respectively. The concentrations of Cr, Mn, Ni and As were lower than their background values of Hubei Province. These demonstrated that Cu, Zn, Cd, Sb and Pb in park dust had accumulated to different degrees. Furthermore, the coefficient of variation values of 9 TEs were between 0.38 and 1.28, among which Cd (1.11) and Cu (1.28) showed high anthropogenic influence. Compared with the risk screening values of level II criterion of the Chinese Environmental Quality Standard for Soils (CEPA, 1995), the mean concentration of dust  (Table S4), the EF values of the sampling points of Ni (7.5%), Cu (70%), Zn (95%), As (35%), Cd (65%), Sb (62.5%) and Pb (90%) were greater than 1, expressing that these TEs were more or less interfered by human factors. Particularly, 2.5% of Cu samples, 7.5% of Cd samples and 2.5% of Pb samples were significant enrichment, and the significant enrichment areas were mainly concentrated in the central and northeastern parts of the study area, indicating that Cu, Cd and Pb in the park dust in these areas conspicuously were affected by anthropogenic activities.
As manifested in Fig. 2b, the mean CF values of Cu, Zn, Cd, Sb and Pb were 1.54, 2.20, 2.32, 1.07 and 1.71, respectively. The samples with CF values of Cu, Zn, Cd, Sb and Pb exceeding 1 showed higher percentages, which were 47.5%, 90.0%, 57.5%, 47.5% and 77.5%, respectively (Table S4). These indicated that Cu, Zn, Cd, Sb and Pb were widely contaminated in the study area, especially Cd pollution was the most serious. On account of the CF, the PLI were computed to synthetically appraise the contamination condition of park dust TEs in research area. The range PLI was 0.05-3.36, and the mean was 1.18 (Fig. 2b), indicating that the research region was moderately contaminated. The ratio of non-pollution, moderate pollution and severe pollution in the study area were 50.0%, 35.0% and 15.0%, respectively, and there was no heavy pollution spot. Consequently, the study area was mainly exposed to moderate contamination, and Cd contamination was the worst.

Effect of dust particle size on TE concentration
Particle size distribution is an important factor affecting the TE concentration of dust. Table 2 presents the effect of dust particle size on TEs concentrations. The concentrations of Cr, Cu, Zn and Pb increased with the decrease of dust particle size, which was concordant with the conclusion that the smaller the dust particle size, the stronger the adsorption capacity, and the higher the TE concentration (Acosta et al., 2011;Bian & Zhu, 2009). Nevertheless, in different particle sizes, the average contents of Mn, Ni and Sb were shown as 150 μm > 75 μm > 250 μm. And the mean content of Cd was 150 μm > 250 μm > 75 μm. The content of As was greatest in dust of 250 μm, followed by 150 μm and then 75 μm. Perhaps this related to multiple factors such as the accumulation of fine-grained substances by iron-containing oxides and TEs emitted by anthropogenic activities. Previous studies can also sustain these results, there was no distinct discrimination of Sb content in different particle sizes (Wang et al., 2006) and Cd had the highest content in coarse particle size dust (Acosta et al., 2011).

Chemical speciation and bioavailability of TEs
The chemical speciation of TEs in dust determines the discrepancy in environmental geochemical activities of TEs and also affects the grade of bio-utilization of TEs. In this research, the chemical speciation It is generally believed that the acid-soluble state of TEs is directly available to organisms, while the reducible and oxidizable states are potentially available to organisms. The higher the ratio of the three chemical speciations, the greater the bioavailability. The bioavailabilities of dust TEs were Zn (78.97%) > Mn (66.40%) > Cu (60.37%) > Pb (56.81%) > Ni (56.10%) > Cd (45.77%) > Cr (17.25%) in the study (Fig. S1).

Pearson correlation analysis
This research conducted correlation analysis of dust TEs to provide a scientific basis for source identification . All TEs data conformed to normal or approximately normal distribution, and Pearson correlation coefficient matrix of 9 TEs is given in Table S6. From Table S6, the correlation coefficients of Mn-Cr (r = 0.695), Ni-Cr (r = 0.669), As-Cu (r = 0.737), Zn-Pb (r = 0.675), Sb-Cu (r = 0.721), Sb-Cd (r = 0.793) and Cd-Cu (r = 0.686) were greater than 0.65, certifying that Mn, Cr and Ni; As and Cu; Zn and Pb; and Sb, Cu and Cd had a stronger correlation, and their sources were probably more similar. Previous researches have shown that Mn, Cr and Sb were generally derived from the parent material . As and Cu were mainly related to agricultural activities (Baltas et al., 2020;Wang et al., 2020). Traffic exhaust emissions and vehicle wears were deemed to be the main sources of Pb and Zn (Adimalla et al., 2020). And other studies indicated that industrial activities would release a large amount of Sb, Cu and Cd (Liang et al., 2017;Soltani et al., 2015).

Source apportionment of dust TEs
The PMF model was utilized to qualitatively identify the source of TE and quantify its contribution to the content of TE in park dust. The model was run 13 times, the residuals of majority TEs were between − 3 and 3, the values of Q robust and Q true were highly correlated, and the converged result showed Yes. And the number of factors was determined according to the stable Q value produced during the running process. Combined with the characteristics of TEs contents in dust, pollution evaluation and related analysis results, 3 factors were finally determined in this study (Fig. 3). The signal-to-noise ratios (S/N) of every TE were higher than 2, and the fitting coefficients (R 2 ) were all Fig. 3 Factor profiles and origin contributions of TEs in park dusts based on PMF greater than 0.75, which ensured the rationality of the model and the excellent fitting effect. And 100 bootstraps runs were executed to ensure the robustness of statistical data.
The first factor was delimited by Cd (93.4%), Pb (59.5%), Zn (61.4%), Cu (54.6%) and Sb (41.8%), explicating 46.62% of three sources (Fig. 3a). The average EF values of Cd, Pb, Zn, Cu and Sb in park dust were > 1, indicating enrichment. In addition, in order to intuitively reveal the spatial distribution of TE pollution from the park dust in the study area, kriging interpolation was performed. Figure  S2 exhibits the consequence of kriging interpolation. The high-value district of Cd, Pb, Zn, Cu and Sb situated in the east of the research region; thus, they may have the similar enrichment methods. The industrial distribution in the eastern part of the study area was relatively dense, with steel, metallurgy, petrochemical and other factories. Studies have donated that industrial activities would release TE particles such as Cd, Pb, Zn, Cu and Sb into the surrounding environment (Cui et al., 2020;Jiang et al., 2021;Sabouhi et al., 2016;Wang et al., 2021;Xiao et al., 2014;Yang et al., 2020). Cd is deemed to be a symbolic element of industrial activities. Substantial Cd was released into the surrounding environment in the form of the three wastes via electroplating, alloy processing, pigments, machinery manufacturing and other industrial machinery activities, ultimately polluted the surrounding environment (Guan et al., 2019). Sb is widely utilized in sundry industries, comprising colored glass processing, hardener in alloy production and flame retardant in textile and plastic production (Huang et al., 2021;Vleeschouwer et al., 2014). Machinery manufacturing and ore smelting discharged Cu-containing pollutants into the environment (Cai et al., 2015). Moreover, the hot spots of Pb and Zn were concentrated along the periphery of traffic routes and densely traffic areas (Fig. S2). Zn is regularly applied as reinforcing agent and activator in the tire manufacturing industry, and corrosion and wear of automobile galvanized parts were also important sources of Zn (Cai et al., 2019a;Mohsen et al., 2021). As an identification element of transportation, Pb chiefly came from the emission of automobile exhaust. Besides, the wear of vehicle engines and brakes, and the use of lead-acid batteries were also the main reasons for the enrichment of Pb in the environment. In summary, factor 1 can be considered as mixed source of industrial activities and transportation activities.
Factor 2 accounted for 25.56% of three sources, with Cr (45.7%), Mn (44.9%) and Ni (43.1%) as the main contributing elements (Fig. 3b). The concentrations of Cr, Mn and Ni were lower than their background values. Besides, the EF analysis results displayed that the average EF values of Cr and Ni were both < 1, indicating no enrichment. All of the above revealed that Cr, Mn and Ni were restricted by low human interference, and more depend on natural factors, such as soil parent materials, rock weathering and other factors. Scholars have discovered that the sources of these TEs were mainly classified as natural sources (Huang et al., 2022a). The study of Micó et al. (2006) showed that the concentrations of Cr, Mn and Ni were principally related to the parent rock factor and the corresponding diagenetic components in the principal component analysis. Amaya et al. (2009) found that soil-forming factors were the main control factors for the concentration of Cr, Mn and Ni in NW Spain. The study of Xiao et al. (2014) on TEs in Guangzhou urban forest park showed that the Mn element in dust was mainly related to natural factor. Guo et al. (2020) analyzed the source of heavy metals in the park dust and found that Ni was related to the soil-forming parent material, and the P value of correlation analysis between Ni and Cr was < 0.01, which was related; thus, both Cr and Ni were derived from natural source. In addition, Fig.  S2 shows that the high-value areas of Cr, Mn and Ni concentrations were in the middle of the study area. The survey found that there were several aluminum manufacturing factories and metal processing factories near the parks in the high value area. The Cr, Mn and Ni released by the processing and manufacturing industries might enter the environment, resulting in a small-scale high-value area. Consequently, natural source was the major source of factor 2.
The factor 3, which accounted for 27.82% of three sources, exhibited great loading of As (71.7%) and Cu (39.86%) (Fig. 3c). EF results revealed that 35% of As samples and 70% of Cu samples were enriched in different degrees, which implied that dust As and Cu interfered by human activities. Meanwhile, the highvalue areas of As and Cu were located around several large parks (Fig. S2); thus, their sources were possibly interrelated to certain activities in the park. Studies have shown that some chemicals containing As and Cu were often applied as raw materials in fungicides, pesticides and herbicides (Huang et al., 2022b). Calcium arsenate and sodium arsenate were widely utilized as herbicides (Chen et al., 2005). Copper sulfate-ammonia complex and copper sulfate were often used in pesticides and fungicides (Cai et al., 2015). Simultaneously, the use of chemical fertilizers (phosphate fertilizers, etc.) with high As concentration were also the reason for the high content of dust TEs in the research area (Cai et al., 2019b). In addition, these parks in this study have been established for a long time, and some facilities have been aging, so the corrosion of some alloy objects in the park and the peeling of wall coatings would release As and Cu into the environment. In conclusion, factor 3 was deemed to be a mixed source of agricultural activities and the aging of park infrastructures.

Potential ecological risk assessment of TEs in park dust
Box plot and spatial distribution map of PER of TEs in park dust are shown in Fig. 4. The E i r of TEs were Cd > > Pb > As > Cu > Sb > Ni > Zn > Cr > M n. Among 9 TEs, the mean E i r of Cd was the largest (73.9), reaching a relatively high ecological risk level. And the average E i r of the others were all less than 30, which were only at a slight ecological risk (Table S7). Perhaps this was due to that the toxicity coefficient of Cd was greater than that of other TEs. The mean PER value of TEs in park dust was 114, according to the risk level, the entire research area belonged to higher risk level. The contributions of three sources of TEs to PER were factor 1 (75.75%) > factor 3 (16.79%) > factor 2 (7.46%). Notably, the PER value of factor 1 (mixed sources of industrial activities and transportation activities) was 86.6, which reached higher ecological risk and was the largest contributor to the PER in the study area. Figure 4 also reveals that the hot spots of PER in the research area were mainly concentrated in developed industrial areas in the east of research region and traffic-intensive areas. In short, industrial and transportation activities were the priority control sources for PER of dust TEs in the parks of the research area. Therefore, priority should be given to monitoring and management of intensive transportation activities and industrial activities such as steel processing, metallurgical industries and petrochemical industries.

Human health risk assessment of TEs in park dust
In this study, the HHR model provided by the US EPA combined with the PMF model was applied to evaluate health risks of 8 TEs (Cr, Mn, Cu, Zn, As, Cd, Pb and Ni) in park dust. The evaluation results are shown in Table 3. As given in Table 3, the noncarcinogenic risk values of adults and children were 3.55E-01 and 5.65E-02, respectively, which were less than 1, indicating that TEs in park dust would not cause obvious non-carcinogenic risks to adults and children. Moreover, whether for adults or children, Cr was the biggest contributor to non-carcinogenic risk, followed by As > Pb > Mn > Ni > Cu > C  Figure S3 shows that the high-value areas of non-carcinogenic risks for children and adults were mainly concentrated around several large parks in the study area. And the non-carcinogenic risks created by various sources were Factor 3 > Factor 1 > Factor 2. Source 3 accounted for 43.14% of total HHR and was the master source of HHR. Therefore, the daily management activities of the park (such as the use of fertilization, weeding and disinfection) require reasonable control. And as the main contributors to the non-carcinogenic risk of adults and children, As, Cr and Pb also need to attract attention.
Regarding cancer risk, the values for adults (1.82 × 10 −5 ) and children (2.78 × 10 −5 ) were both between 10 −6 and 10 −4 , which were within the carcinogenic risk range for the humanity, explaining that the carcinogenic risks of TEs from park dust in the study area to adults and children were both within an acceptable range. The devotions of different TEs to carcinogenic risk were Cr > Ni > As > Pb > Cd, indicating that Cr were the cardinal cancer risk element. This is consistent with the results of the carcinogenic risk assessment of heavy metals in dust from Jiaozuo Park by Han et al. (2020). In addition, the devotion of different sources in carcinogenic risk were Factor 2 > Factor 3 > Factor 1. Factor 2 was the biggest contributor to carcinogenic risk. From Fig. S3, the carcinogenic risk had a small range of high-value areas in the dense distribution area of large parks, indicating that the daily activities of insecticide and pesticide spraying in the park were also important factors leading to carcinogenic risk. Exposed to the same environment, children were more sensitive to dust TEs than adults, and certain particular behaviors of children (for instance, hand-to-mouth and object-to-mouth) lead to higher risk of children being exposed to TEs than adults.
The health risks of adults and children under different exposure routes were calculated, and the results are given in Table S8. Contrasting the three exposure routes, it was found that both the non-carcinogenic risk entropy and carcinogenic risk indexes of children and adults were all expressed as oral ingestion > dermal contact > respiratory inhalation. This indicated that oral ingestion was the chief exposure pathway for non-carcinogenic and carcinogenic risks. Previous studies have also reached the same conclusion that oral ingestion was the master pattern of human exposure to dust TEs (Adewumi & Laniyan, 2020;Behnam et al., 2020;Shah et al., 2020;Shahab et al., 2018).

Conclusion
This research investigated the contamination characteristics, particle size effect, chemical speciation and bioavailability, and sources of TEs in park dust, and quantitatively assessed the ecological and health risks of different sources. The average concentrations of Cu, Zn, Cd, Sb and Pb were higher than their background values of Hubei Province, and the concentrations of Cr, Mn, Ni and As were lower than their background values. The results of EF and PLI demonstrated that study area was mainly exposed to moderate contamination, and Cd contamination was the worst. The statistical results of dust particle size effect exhibited that the smaller the dust particle size, the higher the Cr, Cu, Zn and Pb contents. The concentrations of Cr, Ni, Cu, Cd and Pb in the study area were mainly in the residue state, of which the residue state of Cr accounted for the highest proportion of 82.75%. The bioavailability ranking of TEs was Zn (78.97%) > Mn (66.40%) > Cu (60.37%) > Pb (56.81%) > Ni (56.10%) > Cd (45.77%) > Cr (17.25%). Three sources were determined and apportioned using PMF model, Pearson correlation analysis and geostatistical analysis, namely mixed source of industrial activities and transportation activities (factor 1), natural source (factor 2), and mixed source of agricultural activities and the aging of park infrastructures (factor 3). Factor 1 was the largest potential ecological risk contributor, and Cd was the largest potential ecological risk element. The non-carcinogenic risks of children and adults were all lower than the risk limit. The carcinogenic risks of children and adults were within the risk range. For non-carcinogenic risk, mixed source of agricultural activities and the aging of park infrastructures was the most dangerous source. And natural source was the largest contributor for carcinogenic risk. Consequently, regular maintenance of park facilities and scientific arrangements of agricultural activities (such as the rational use of fungicides, herbicides and chemical fertilizers) are vital to the protection of ecological environment and human health.