Source identi fi cation and risk analysis of potentially toxic elements ( PTEs ) in rainwater runoff from a manganese mine ( south central Hunan , China )

Potentially toxic elements (PTEs) in manganese ore areas are prevalent in rainwater runoff and pose a major threat to human health. In this study, field investigation and geostatistical analysis methods of positive matrix factorization (PMF) and geographic information systems (GIS) were used to systematically study the pollution in rainwater runoff from a manganese mining area in Xiangtan, China, to evaluate source contributions for the health risk assessment of PTEs. The average concentrations (mg/L) of six PTEs were: 0.3357 (Mn), 0.0450 (Ni), 0.0106 (Cu), 0.0148 (Zn), 0.0068 (Cd) and 0.0390 (Pb). The coefficients of variation (CV) for Mn and Zn were >180% and >130%, with the other analytes having values below 70%. The GIS and PMF analysis produced more refined spatial source apportionments, including mining, smelting, transportation, agricultural production and natural sources. The results of the health risk assessment showed that the non-carcinogenic risk was negligible, and the carcinogenic risk was potentially dangerous but acceptable for both adults and children. In addition, the children’s total carcinogenic risk value was greater than that of adults, highlighting their vulnerability. This study demonstrates the potential of PMF to provide a framework to spatially prioritize treatment objectives within the mining region to improve environmental conditions.


INTRODUCTION
A number of potentially toxic elements (PTEs) have been highlighted as priority contaminants globally and have received widespread attention due to their persistence and toxicity to humans and organisms (Liu et al. ). Prolonged exposure to an environment containing PTEs can cause a wide range of diseases; for example, excessive manganese (Mn) in the human body may lead to mental illness, as well as more serious Parkinson's disease (Li et al. ). High environmental levels of lead (Pb) can lead to high blood Pb Among them, PMF has only recently been applied to identify sources of PTEs (Huang et al. ; Xiao et al. b). In addition, the spatial power of geographic information systems (GIS) and statistical analysis (using SPSS) allow data outliers to be accurately distinguished. In combination with this, identifying the hot spot locations of high PTE concentrations at a specific location can aid in the reduction of uncertainty and the cost of assessment. This approach has been used to determine spatial structure characteristics and pollution levels and perform source analysis in a number of cases, including assessment of surface sediments, soil organic matter variation, the distribution of sulfur species in paddy soils, pollutant metal and microbial community responses in soil, quality of water in gulf zones and metals in mine soil (Herbert which is therefore an overlooked topic in many locations.
In this study, we describe a detailed investigation of PTE content in rainwater runoff from a large-scale manganese mining site using a combination of PMF and GIS mapping.
Xiangtan is an old industrial base in Hunan Province (China) which possesses rich mineral resources, including manganese. Since 1913, large quantities of manganese ores have been exploited, and long-term mining, tailings production and smelting activities have led to very serious environment degradation. Ren et al.
() previously determined the average concentrations of six PTEs: Mn, nickel (Ni), copper (Cu), zinc (Zn), cadmium (Cd) and Pb in rainwater runoff in the region, and attempted source identification using PCA. That study lacked true spatial resolution, highlighting varied enrichment between PTEs and limited source identification to three factors.
The objectives of the research described in this study are to: (1) build a spatially refined hot spot map of pollution in surface waters in the region; (2) apply an analysis using PMF to evaluate potential sources of six PTEs; (3) extrapolate environmental conditions to assess the health risks of six PTEs to local residents. This study intends to improve the use of site-specific information to understand the extent of impacts and provide an improved framework to manage, prevent and control further pollution, thus providing a scientific basis for health risk assessment in the water environment.

Study area
Hongqi mine is located at Xiangtan city (111 58 0 ∼113 05 0 E, 27 21 0 ∼28 05 0 N), Hunan Province, south central China, at an average altitude of 97.39 m. The mine covers 2.6 km 2 in a mountainous area. The climate type is subtropical monsoon and moist, with average temperatures of 16.7-17.4 C; the maximum wind speed can be up to 17 m/s. Precipitation is abundant, and the annual precipitation is 1,200-1,500 mm, with between 60 and 80% falling between April and October (Fang et al. ; Jiang et al. ). In this study, the sampling points were distributed at intervals of 0.04-0.06 km 2 to make a total of 43 sampling points. These were divided into four regions: S1-S6 and S43 are Region 1, S7-S23 is Region 2, S24-S30 is Region 3, and S31-S42 is Region 4. The distribution of the sampling points is shown in Figure 1.

Sample selection and analysis
In September-October 2018, 0.5-3 L rainwater runoff water samples were taken at each sample point and transferred to polypropylene containers. A total of five sub-samples were collected at each location based on a 5 m radius circle (points at east, west, south, north, and center) and mixed into a composite sample. The containers were packed in a black plastic bag with ice to ensure the freshness of the water samples. Subsequently, the water samples were shaken until homogeneous water samples were obtained, followed by a natural settlement time of 20-30 min, then the samples were siphoned and filtered through a 0.45 μm filter membrane. The filtrate was acidified to pH <2 using hydrochloric acid and nitric acid for future analysis (Ren et al. ). The detection limits (μg/L) for Ni, Pb and Cd were 0.054, 0.05 and 0.012 respectively. In order to control the accuracy of the sample analyses, the standard reference material of the China National Standards Research Center: GBW (E) 080194 was used, and each sample was subjected to reagent blank and three repeated tests. The recovery rates of the target PTEs in the standard references were good and ranged between 94 and 106%.
In addition, a standard recovery experiment was performed to verify the precision of the test method (Ren et al. ). The recovery rates of Mn, Ni, Cu, Zn, Cd, and Pb were between 96.2% and 103.9%.

Health risk assessment
Health risk assessment is defined as the process of estimating the probability of occurrence of events and the possible magnitude of adverse health effects over a specified time period (Lim et al. ). For the assessment of the health risks of PTEs in the water environment, direct intake of rainwater as drinking water and skin absorption from rainwater on the human body are usually considered (Zeng et al. ). Locally relevant factors were used in the application of a standard risk assessment model (USEPA ), where the exposure doses for direct intake (ADD ingestion ) and skin absorption (ADD dermal ) were as follows (with ADD being the chronic daily intake of PTEs in μg/kg/day): where C w is the concentration (μg/L). IR is the intake rate (L/day); 2.0 for adults and 0.64 for children in this study (Wang et al. ). ABS g is a gastrointestinal absorption factor; in this study, they were 6.0%, 4.0%, 57%, 20%, 5.0% and 11.7% for Mn, Ni, Cu, Zn, Cd and Pb, respectively (Wang et al. ). EF is the exposure frequency (day/year), which was 350 in this study (Zeng et al. ). ED is the duration of exposure (year); 70 for adults and 6 for children in this study (USEPA ). BW is the residents' weight (kg), according to the survey of local residents; 65 for adults and 20 for children. AT is the non-carcinogenic mean time (day); 25,550 for adults and 2,190 for children in this study (USEPA ). SA is the exposed skin area (cm 2 ); 18,000 for adults and 6,600 for children in this study (Wang et al. Potential non-carcinogenic risks are assessed by hazard quotient (HQ) (USEPA ). HQ is defined as the health hazards due to two paths of exposure (ingestion and skin contact) in a lifetime. HQ > 1 indicates that there may be non-carcinogenic risk, and the risk increases with the increase in HQ value.

HQ ¼ ADD RfD
where RfD is the reference dose obtained from a risk-based centralized database (USEPA ). RfD dermal and RfD ingestion are listed in Table 4.
The hazard index (HI) was used to assess the total potential non-carcinogenic risk caused by different pathways.
HI ¼ X HQ ing þ HQ der As for HQ, when HI > 1, there may be non-carcinogenic risk, and the risk increases as the HQ value increases (USEPA ).
Carcinogenic risk (CR) is defined as the increasing probability that an individual will develop cancer in their lifetime due to chemical exposure in a given situation (Chen & Liao ). For this study, the carcinogenic risk of both Cd and Pb elements through both ingestion and skin absorption was calculated as follows: where ADD i is the daily intake by ingestion or skin absorption. SF i (kg·day/mg) is the slope factor of carcinogens. The values are based on the USEPA risk concentration table: Cd is 6.3 and Pb is 8.5 × 10 À3 . When CR < 1 × 10 À6 , the carcinogenic risk of rainwater runoff to health is negligible. When CR > 1 × 10 À4 , the carcinogenic risk in the local residents is high and unacceptable. When 1 × 10 À6 < CR < 1 × 10 À4 , there is a certain chance of carcinogenic risk for local residents, but it is acceptable.

Statistical analysis
Descriptive statistics of water samples were performed using

RESULTS AND DISCUSSION
Water contamination by PTEs Table 1   The most suitable interpolation applied to one heavy metal could be selected by comparing average error and root mean square error, which was generated from Kriging, inverse distance weighted (IDW) and radial basis function (RBF) by comparing the average error, the response error range and the root mean square error response sensitivity.
By comparing the three interpolation methods, the results showed that the RBF method was the best for Mn, Ni, Cu and Cd, Kriging interpolation for Zn and Pd. The results are shown in Figure 2. There were obvious differences in the pollution and content of PTEs: the contaminated area of Mn, Cu, Zn and Pb were similar, mainly distributed in the northeast of the study area and related to industrial activities, especially direct mining, smelting, and tailings. Ni was mainly distributed in the south of the study area, where the majority of the regional population lives.
The pollution from Cd was more dispersed, which may be related to intensive agricultural production and residential areas having aggregated inputs from multiple sources.

Source analysis of PTEs
The Pearson correlation analysis is an effective way to explore the relationship between multiple data as an initial   The first factor is dominated by Cu and Pb, with contributions of 72.6% and 28.4%, respectively. According to field investigations, we found higher levels of Cu and Pb in the  . Therefore, factor 1 can be considered to be derived from smelting and traffic emissions.
The second factor is dominated by Mn, with a contribution of 87.3% associated with a small contribution from Ni and Pb. According to field investigations, high concentrations of Mn may be directly derived from manganese ore activities, such as mining and tailings. The association of Ni with Mn may also be derived from the ore as associated elements. The main source of Pb is traffic emissions, which may be related to ore transportation in the mining area. Therefore, the factor 2 can be considered to be derived from mining activities.
The third factor is related to Cd and Pb, and the contribution were 73.7% and 54.7%, respectively. Cd may be derived from some municipal sewage, pesticides and fertilizers, and atmospheric deposition (Rehman et al. ). As an additive, Cd has been widely used as a pesticide and has been associated with phosphate fertilizer in agricultural production in China. However, many scientists insist that atmospheric deposition is the source of Cd accumulation, not the use of pesticides and fertilizers (Yi et al. ).
According to a study in Hunan Province ( Therefore, factor 5 can be considered to be derived from smelting activities.

Health risk assessment of PTEs
Using the health risk assessment method to evaluate the elements in the rainwater runoff in the manganese mining area of Xiangtan city, the non-carcinogenic risks to adults and children under different exposure conditions (Table 4, Figure 4) and carcinogenic risks (  (Table 5). For adults and children, the average carcinogenic risk of rainwater runoff was 7.00 × 10 À5 and  7.94 × 10 À5 , respectively. This indicates that children were more vulnerable than adults to carcinogenic risk. The carcinogenic risk value was lower than 1.00 × 10 À4 for both adults and children, but both were higher than 1.00 × 10 À6 , which indicates that local adults and children may be affected by increased carcinogenic risk, but at an acceptable level. The carcinogenesis risk by ingestion was significantly higher than by skin absorption (  for Zn is likely to be from smelting activities. The health risk assessment showed that the non-carcinogenic risk in this study was low and negligible. Carcinogenic risks, however, were more significant for both adults and children, with children being more vulnerable, especially to the potential ingestion of Cd.
The use of PMF has improved source apportionment that fits with industrial activities in the area and spatial assessment focuses on priority areas for surface water management. The results of this study will provide effective information for further prevention of PTE pollution and provide a scientific basis for health risk assessment in water environments.