# Air pollution from gas refinery through contamination with various elements disrupts semiarid Zagros oak (Quercus brantii Lindl.) forests, Iran

### Description of study areas

IGR plant (33° 42/N, 46° 13/E) is located along the edge of the mountains of Zagros forests and 25 km from Ilam city. Its main activity, to supply gas to the western provinces of Iran, started in 2007. It converts sour gas to sweet gas and also produces various products such as pastil sulfur, ethane, and liquefied gas. The refinery has two chimneys, which release waste gases into the atmosphere. Oak trees are the main tree species of the Zagros forests around the refinery; these are exposed to various air pollutants and different elements from this source. Based on random analysis of exhaust emissions, sulfur dioxide and sulfide hydrogen are the major pollutants emitted from the flare gases of this refinery plant34. The sampling points have an average altitude of about 1000–1250 m and a slope of less than 20%. The climate of the region is semiarid and influenced by Mediterranean winds. The predominant wind direction was west and southwest. The highest and lowest air temperatures were 41.4 °C and − 11.3 °C, respectively. The average annual rainfall was 71.94 mm (http://www.amarilam.ir).

### Samples collection and analyses

All methods were carried out in accordance with the relevant institutional, national, and international guidelines and legislation. Besides they were discussed and approved by the Research Ethics Committee of Tarbiat Modares University. The formal identification of the Quercus brantii Lindl. was performed by H. Dadkhah-Aghdash based on colorful Flora of Iran35. The permissions or licenses to collect Brant oak (Quercus brantii Lindl.) trees in Zagros forests were obtained. A voucher specimen of Brant oaks were collected and deposited at the Herbarium of department of Plant Biology of Tarbiat Modares University.

We studied different distances (1000, 1500, 2000, 2500, and 10,000 m [control]) in an easterly direction from the gas refinery. The map of study area was drawn by software of ArcGIS version of 10.5, https://desktop.arcgis.com (Fig. 5). At each distance, three soil samples taken from the depth of 0–20 cm with a plastic gardening shovel, 30 healthy and mature leaves were collected from a certain height (nearly the middle of the canopy) and the outer canopy of three Brant oak trees in the late spring, summer, and autumn of 2019. These trees with average height and diameter at breast height of 5.5 m and 45 cm were selected randomly. The leaf and soil samples were put into polyethylene bags and transported to the laboratory for analysis36.

In the lab, firstly the leaves were categorized into two types: unwashed leaves and leaves washed with ethylenediaminetetraacetic acid (EDTA) solution to remove some atmospheric dusts and particles deposition. The leaf and soil samples were dried for 10 days until they reached a constant weight at lab temperature. The leaves were grinded and homogenized, soils were sieved with ASTM mesh (DAMAVAND, Iran) with a diameter of 2 mm and homogenized.

To determine the pH and electrical conductivity (EC) of soils, 2 g of the soil samples were shaken in 10 ml of double-distilled water with a ratio of 1:5; after 1 h, the pH and electrical conductivity (EC) of the solution were measured by a digital pH meter (Fan Azma Gostar Company, Iran) and EC meter (Sartorius, PT-20, USA). The analysis of the particle sizes of the soil was carried out using the hydrometer method and texture class was determined with a soil texture triangle37.

According to different U.S.EPA protocols that were modified by following references, the soil and leaf samples were prepared and dissolved. The digestion of soil samples was conducted with a mixture of concentrated HF–HClO4–HNO338. Approximately 0.5 g of dry soil sample was digested with 10 mL of HCl on a hot plate at ~ 180 °C until the solution was reduced to 3 mL. Approximately 5 mL of HF (40%, w/w), 5 mL of HNO3 (63%, w/w), and 3 mL of HClO4 (70%, w/w) were then added and the solution was digested. This process was continued with adding 3 mL of HNO3, 3 mL of HF, and 1 mL of HClO4 until the silicate minerals had fully disappeared. This solution was transferred to a 25 mL volumetric tube, and 1% HNO3 was added to bring the sample up to a constant volume for the element’s determinations. After filtering the digested samples, the concentrations of sulfur (S), arsenic (As), chromium (Cr), copper (Cu), lead (Pb), zinc (Zn), manganese (Mn), and nickel (Ni) were measured via inductively coupled plasma mass spectrometry (ICP-MS,7500 CS, Agilent, US). The procedures of quality assurance and quality control (QA/QC) were performed.To quantify element contents from soil samples, external standards with calibration levels were used. The precision and the repeatability of the analysis were tested on the instrument by analyzing three replicate samples.

According to Liang et al.39 leaf samples were acid digested and sieved powder samples were placed in the acid-washed tubes and 10 mL of 65% nitric acid was added to it. The solution was placed at room temperature overnight (12 h) after that, it was placed for 4 h at 100 °C and then 4 h at 140 °C until the solution color was clear. After cooling, the solution was diluted by deionized water to 50 mL and then passed through Whatman filter paper until 25 mL of the filtrate volume was provided. Each sample was digested three times and the average of measurements is reported. Total plant elements were measured by using the ICP-MS (7500 CS, Agilent, US). A control sample was also used beside each sample to determine the background pollution during digestion. To confirm the accuracy of the methodology and to ensure the extraction of trace elements from the leaf samples, the standard solution of each studied elements was used.

### Measuring of pollution levels of different elements in soils and leaves

For assessment of contamination levels (concentration) of different elements in soils and trees, common indices of pollution including geoaccumulation index (Igeo), pollution index (PI), pollution load index (PLI), enrichment factor of plants (EFplant), bioconcentration factor (BCF), air originated metals (AOM ), metal accumulation index (MAI) were used.

Igeo was calculated using the following (Eq. 1):

$${text{I}}_{{{text{geo}}}} = log_{2} left[ {{text{C}}_{{text{n}}} / 1.5{text{ B}}_{{text{n}}} } right]$$

(1)

where Cn is the measured concentration of the element n, Bn is the geoaccumulation background for this element and 1.5 is a constant coefficient used to eliminate potential variations in the baseline data40. The Igeo classifies samples into seven grades: < 0 for practically unpolluted; 0–1 for unpolluted to moderately polluted; 1–2 for moderately polluted; 2–3 for moderately to strongly polluted; 3–4 for strongly polluted; 4–5 for strongly to extremely polluted; and > 5 for extremely polluted30.

The first PI is expressed as (Eq. 2):

$${text{PI }} = {text{ C}}_{{text{i}}} /{text{S}}_{{text{i}}}$$

(2)

where Ci is the concentration of element i in the soil (mg kg−1) and Si is the soil quality standard or reference value for element i (mg kg−1). The PLI for different elements is calculated via the (Eq. 3):

$${text{PLI}} = left( {{text{PI}}_{{1}} times {text{ PI}}_{{2}} times {text{ PI}}_{{3}} times cdots times {text{PI}}_{{text{n}}} } right)^{{{1}/{text{n}}}}$$

(3)

The PLI of soils is classified as follows: PLI < 1 is unpolluted, 1 < PLI < 2 is unpolluted to moderately polluted, 2 < PLI < 3 is moderately polluted, 3 < PLI < 4 is moderately to highly polluted, 4 < PLI < 5 is highly polluted, and PLI > 5 is very highly polluted.

Plant EF is calculated as (Eq. 4):

$${text{EF}}_{{{text{plant}}}} = {text{C}}_{{{text{plant}}}} /{text{ C}}_{{{text{control}}}}$$

(4)

where Cplant and Ccontrol are element concentrations (mg kg−1) in tree leaves at the polluted site and the control site, respectively. A value of EF > 2 indicates that a tree is enriched with the specific element41.

BCF, which indicates the ability of plants to accumulate different elements from the soil7,42, is calculated as (Eq. 5):

$${text{BCF }} = left( {left[ {text{C}} right]_{{text{L}}} / left[ {text{C}} right]_{{text{S}}} } right)$$

(5)

where [C]L and [C]S are, respectively, the concentration of different elements in leaf and soil samples. Values of BCF > 1, BCF < 1, and BCF = 1 imply that the trees are accumulators, excluders, and indicators, respectively, for different elements39.

AOM, used to determine the number of different elements in leaves originating from ambient air18, are calculated by the (Eq. 6):

$${text{AOM }}left( % right) = left( {left( {{text{C}}_{{{text{unwashed}};{text{leaves}}}} } right) – left( {{text{C}}_{{text{washed leaves}}} } right)} right) / left( {{text{C}}_{{text{unwashed leaves}}} } right) , * , 100$$

(6)

where (c) is the concentration of studied element.

Oak tree leaves had different abilities to accumulate atmospheric studied elements. MAI was calculated as (Eq. 7):

$$text{MAI}=1/text{n}sum_{text{j}=1}^{n}text{Ij}$$

(7)

where n is the total number of studied elements, and Ij is the sub-index of variable j, calculated by dividing the mean concentration (x) of each element by its standard deviation43,44.

### Data analysis

Before analysis of variance, the normality of the data distribution was checked by SAS version software using Shapiro–Walk and Kolmograph-Smirnov tests. The logarithmic conversion was used to the normality of the data in Excel. Different properties of soils such as pH and EC for two factors including seasons in three levels (spring, summer, and autumn) and distance in five levels (1000, 1500, 2000, 2500, and 10,000) with three replicates were analyzed in a completely randomized design. Analysis of variance was performed by two-way ANOVA with SAS software and the comparisons of means were conducted with Duncan’s test at p < 0.05. The differences among the distance and season factors were studied using a resemblance matrix for the different elements (S, Cr, Cu, Mn, Ni, Pb, As, and Zn). The resemblance is the general term in the PRIMER software used to cover (dis)similarity or distance coefficients between all pairs of the samples.

Following this, elements matrices were log x + 1 transformed, and the resemblance matrix was built using the Euclidean distance. Elements were analysed using nonmetric multidimensional scaling (NMDS) and the Kruskal stress formula (minimum stress: 0.01) for visualizing the level of similarity. Statistical analyses were conducted using PRIMER 6 software45,46.