Five sites were sampled to characterize their physico-chemical properties in the Grand Reims area (Figure 1), which is an intercommunal structure located in the Grand Est Region (France) with 300,699 inhabitants in 2019 and a population density of 211/km². Agriculture in this area fall into four main activities which are vineyard, beets and alfalfa. Beet cultures are valorized locally in several products such as sugar, ethanol and alcohol. The area of the sampling site is a large beet agricultural zone crossed by active traffic roads, close to the Reims airfield, agricultural facilities and concrete industry. Several villages with a population between 500 and 1,000 inhabitants are surrounding the sites in a 2 km radius in addition to the city of Reims located 5 km west. All these activities may act as indirect sources of pollutants especially combustion residues. The cultures were spread with the effluents of a sugar factory at a 600 m³·ha⁻¹ (60 mm) maximum rate with a minimum of 2 years between each spreading on the same parcel. Land application with such effluents started in 1978. Only the data focusing on the nutritious capacity of the effluents were available. The concentrations of targeted pollutants were not known. Soil samples were provided by the association of agronomic spreading monitoring (ASAE, France). The sampling was focused on the rich organic content layer of the soil where the pollutants can accumulate. The topsoil was collected with a shovel until 25 cm depth and stored in a closed polyethylene bag at 4°C before use. The soils were characterized for particle size distribution without decarbonation, pH, cation exchange capacity (CEC), organic carbon (OC), and CaCO₃ contents according to French and International standard methods, following NF X31-107, ISO 10390, NF X 31-130, ISO 14235, and ISO 10693, respectively (AFNOR, 1995; AFNOR, 1998; AFNOR, 1999; AFNOR, 2003; AFNOR, 2005). Organic matter (OM) content was calculated as follows: % OM = 1.72% OC.

Methanol and hexane (HiPerSolv Chromanorm), as well as nylon filter (0.2 µm) syringes and filter supports were purchased from VWR. The standards benzo(k)fluoranthene, benzo(b)fluoranthene, benzo(a)pyrene, indéno(1,2,3-c,d)pyrene, benzo(g,h,i)perylene, anthracene, fluoranthene, naphtalene, benzene, toluene, ethylbenzene, xylene (o,m,p), diethylphtalate, tributyltin, 4-octylphenol, 4-tert-octylphenol, 4-n-nonylphenol were of the highest purity and purchased from Sigma Aldrich.

Soil extraction was performed using a Thermo Scientific Ethos Easy microwave extractor with a Fast-24 rotor. The samples in glass tubes were subjected to a 1,800 W power to reach 110°C in 10 min, this temperature was then maintained during 10 min. PAH and BTEX compounds were extracted from 2 g of soils in 10 mL of hexane; for diethylphtalate, tributyltin and phenolic compounds, 5 g of soils in 10 mL of methanol were used. After microwave extraction the liquid phase was filtered through a nylon membrane and stored in a glass tube. The solvent was then evaporated to dryness in a centrifugal concentrator Genevac HT Series (SP Scientific, Warminster, PA, United States). The manufacturer programs “very low boiling point” and “medium boiling point” were used for hexane and methanol extraction respectively. The temperature was set at 40°C with a rotation speed of 1,400 rpm. The evaporation process started with a decrease in pressure from the atmospheric pressure to full vacuum in 1 h and was maintained for 6 h. The residue was dissolved in 500 µL of methanol and put inside a micro-vial for GC-MS analysis.

The target compound concentrations were determined using a Thermo Scientific TRACE 1300 GC system coupled to a Thermo Scientific ISQ Series quadrupole mass spectrometer equipped with an electronic impact source (Low activity). The separation was achieved by injecting 0.5 µL of the sample in a Thermo Scientific TR5-5MS column (30 m, I.D. 0.25 mm, film 0.25 µm). Data were processed with the XCalibur™ 2.2 software. Injection was made in splitless mode. The transfer temperature was set at 250°C with a source at 230°C. The dwell time was set at 0.5 s for each ion. Single Ion Monitoring (SIM) mode was used with ion quantification and identification as listed in Table 2. The temperature program was from 90°C to 300°C with a ramp of 5°C/min and from 40°C during 8 min–300°C with a ramp of 40°C/min for PAH and BTEX, respectively. For the other compounds, it was from 90°C to 300°C with a ramp of 50°C/min. Each soil sample was analyzed three times and a blank extraction without soil was inserted in each extraction batch to check for potential contamination during extraction process.

The concentrations of target compounds were calculated by an external standard calibration method. A calibration curve with five points was obtained using the pure standard in hexane for HAP and BTEX and methanol for alkylphenol, phtalate and TBT. The limits of quantification (LOQ) was calculated based on the signal to noise approach defined in the ICH Harmonized Tripartite Guideline (2005) by dividing the concentration found in extracted samples by a tenth of the signal-to-noise ratio measured in the chromatogram. Finally, the soil contaminant concentrations were calculated by applying a yield factor on each concentration measured after extraction. The extraction yield factors of adsorbed pollutant were calculated by spiking in triplicate a pristine soil in water with a known amount of each contaminant (100 ng per g of soil) and shaking for 24 h to ensure that equilibrium adsorption is reached. After filtration through a nylon membrane (0.2 µm), the liquid fraction was analyzed. The adsorbed amount (deduced by difference between the introduced concentration (100 ng·g⁻¹) and the remaining concentration after shaking) was extracted from the solid fraction by MAE and analyzed to determine the extraction yield. The method was validated by following exactly the same protocol meaning by spiking in triplicated another pristine soil with a known amout of each contaminant. Then the sampled soil was analyzed using the validated method.

Standard deviation, linearity and coefficient of determination were calculated with Excel 2016 datasheet using build-in function (LINEST and STDEV.S) for linear regression. The LOD and LOQ was calculated with the signal to noise approach according to ICH Harmonized Tripartite Guideline (2005). The noise was determinated graphically on the chromatogram for each compounds then the LOQ was calculated as the equivalent concentration of ten times the noise and the LOD as the equivalent concentration of three times the noise.

The soil physico-chemical properties are reported in Table 3. The five soils were very similar in texture: PYZC8, PYZB2, PYZC7, and NOZH22 were loam, and NOZD69 was silt loam. They displayed a basic pH (≈8.4) which is characteristic of calcareous soils of the Champagne-Ardenne region and in accordance with their high CaCO₃ content (between 44% and 73%). Their OC content was relatively high, ranging between 2.9% and 3.3%, and their CEC was comprised between 8.1 and 13.4 cmol·kg⁻¹. The lack of diversity in soil physico-chemical properties and composition makes extrapolation of data difficult, however a previous study of extractability of organic pollutants using MAE with a wide variety of matrix type has shown that the main soil constituant influencing the extractability of pollutants is total organic carbon (Báez et al., 2003).

The performance of the method is summarized in Table 1. Extraction yields ranged from 44% to 96%. The extraction of diethylphtalate, octylphenols, nonylphenols, and toluene was almost quantitative, the one of BTEX compounds ranged from 72% to 96%, while half of the PAH and tributhyltin cation remained adsorbed or lost by degradation during the extraction process. The extraction procedure was intended to be as simple as possible leading in some cases to relatively low extraction yield. However, in each case, the accuracy of the extraction yields was validated by the reproducibility of the extraction allowing an extrapolation of the exact content. In most cases, the LOQ were in the range of a tenth of a ng·g⁻¹ (few hundreds of ppt). In the case of phenolic compounds, the LOQ were significantly higher, around 50 ng·g⁻¹. The LOQ values obtained in this paper for the analysis of BTEX and PAH in soils are similar to those reported in previous studies also measured by GC-MS but after solid-phase microextraction (SPME) extraction for BTEX (Franco et al., 2015) or solvent extraction for PAH (Maliszewska-Kordybach et al., 2009). A significant decrease of the detection limits could be reached for alkylphenols by including a derivation step with N-methyltrifluoroacetamide (Zhang et al., 2009) and an alkylation step for tributyltin (Crnoja et al., 2001). Khosravi and Price, 2015 obtained a better quantification limit of less than 1 pg·g⁻¹ for diethylphtalate using an ASE-SPE-GC-MS method with a higher volume of sample concentrated and transferred in the injector during the analysis. Although other methods have better quantification limits especially when the methods are specifically designed for a single class of compound adding derivation and purification steps (e.g., for alkylphenols) the advantage of our method is the ability to analyze all the target compounds with a small number of preparation steps thus reducing the global cost of analysis. Nevertheless, this strategy leds to a reduction of the performance for some compounds by balancing selectivity and extraction efficiency but allows to reach the quantification limit fixed by the regulation for the whole list of compounds.

The results of soil analyses are reported in Table 4. PAH and BTEX (except benzene) were identified in all samples at relatively high concentrations, reaching levels higher than 100 ng·g⁻¹ in the case of benzo(b)fluoranthene, indeno(1,2,3-c,d)pyrene, benzo(g,h,i)perylene, and fluoranthene. The total concentration ranged from 406 to 1,237 ng·g⁻¹ for ∑8PAH and from 21 to 65 ng·g⁻¹ for ∑BTEX. In France there are no mandatory values defining concentration limits of these pollutants in soil. Concentration limits only exist for ground-water and solid waste. Usually, concentrations are discussed according to the local context or the mean background concentrations. Various field campaigns evidenced the presence of PAH compounds in European soils at amounts depending on the anthropic activities in the area. For example, in Switzerland, concentrations of 98–219 ng·g⁻¹ were found in different forest sites (Bucheli et al., 2004), whereas these concentrations could reach 575 ng·g⁻¹ in agricultural soils impacted by anthropic activities (Berset and Holzer, 1995). Sometimes, very high contents of PAH were found like in Poland with concentrations reaching 6,680 ng·g⁻¹ in soil less than 5 km from the emission source typically industries and urban areas (Maliszewska-Kordybach et al., 2009). In France, a model of the PAH contamination distribution based on 549 uncultivated soil samples throughout the country showed that HAP remained below the detection limits (<5–50 ng·g⁻¹) in most samples (Villanneau et al., 2013). However, in highly contaminated sites (mining, metallurgy and coal burning activities) individual HAP reached 100–1,000 ng·g⁻¹. The authors evidenced that fluoranthene, pyrene and phenanthrene concentrations where geographically correlated with the contamination levels in the northeastern part of France due to historical industrial activities. The concentrations of each PAH in the contaminated soils were between 15 and 35 ng·g⁻¹. In another study, higher values ranging from 190 to 760 ng·g⁻¹ were detected in unamended cropland influenced by urban activity and possibly impacted by direct deposition from agricultural engine particles (Gateuille et al., 2014). In the present study, the individual PAH concentrations measured (n.d to 429 ng·g⁻¹) were in the line with European agricultural soils considering its localization (located in the northeastern part of France and near urban and agricultural areas). In agricultural soil, PAH may impact crops growth or germination and invertebrates reproduction. Exposition studies on terrestrial plants (Sinapsis alba, Trifolium pratense, and Lolium perenne) with several PAHs found EC₂₀ values from 30 to 650 μg·g⁻¹ for the most sensitive values which is above the total of measured PAH (Sverdrup et al., 2003). Survival and reproduction studies of springtail Folsomia fimetaria L., in contaminated soil showed a typical LC₅₀ and EC₅₀ of 21 and 11 μg·g⁻¹ respectively with no-observed-effect concentration of 8.6 μg·g⁻¹ for the most sensitive PAH (Sverdrup et al., 2001) which is also above the total of measured PAH.

Concerning the BTEX, few data are available in the literature concerning their concentration into agricultural soils because they are considered to be mainly present in the atmospheric compartment due to their volatility. BTEX are emitted directly by leakage, stock manipulation or sludge application, but they can reach the soil compartment after their emission into the atmosphere by industry, wastewater treatment plant and vehicle exhausts. For benzene, a predicted environmental concentration of 0.02 ng·g⁻¹ based on all the releases into the environment was determined in non-contaminated soils or without direct emission at vicinity (EU, 2008). Taking into account the deposition through the atmosphere, the predicted concentrations in agricultural soils increased in the range 1.21–129 ng·g⁻¹ (EU, 2008). In the present study, the benzene concentration in the five studied soils was lower than the quantification limit (0.2 ng·g⁻¹) which was consistent with field studies conducted in uncontaminated soils (Ellison et al., 1997). Concerning toluene, modelisation based on emission calculated from traffic and industry was used to estimate its concentration in air, water and soils at local and regional scale (Hansen et al., 2003). According to this model the expected background concentration of toluene in agricultural was 0.1 ng·g⁻¹ and the mean concentration in agricultural soils located near different industrial sites was 7 ng·g⁻¹. In addition, in the case of field studies without any direct pollution context, toluene soil concentrations of 1 ng·g⁻¹ and 2.1–5.8 ng·g⁻¹ were reported in the Netherland (Sloof and Blokzijl, 1987) and in Poland near petrol station or allotment garden (Zygmunt and Namiesnik, 2001), respectively. In this study, the toluene concentration ranged from 5 to 55 ng·g⁻¹. As for ethylbenzene and xylene data in agricultural soils are very scarce and mostly related to direct spill incident via leakage with relative high concentrations in soils. However, using a simple box model, the concentration of ethylbenzene coming from atmospheric depositions was estimated at 0.07 ng·g⁻¹ in agricultural soils and until 39 ng·g⁻¹ in local areas (EU, 2007). In our studied soils, the ethylbenzene measured concentrations were in the lowest part of this range (around 2 ng·g⁻¹); xylene concentrations were between 1 and 9 ng·g⁻¹ which is relatively low compared to the concentrations detected in United States soils ranging from 1 to 190 ng·g⁻¹ (Agency for Toxic Substances and Disease Registry (ATSDR), 2007). Thus, according to the cited examples, the measured concentrations of BTEX in the present study is in the same order of level as the measured value in the literature.

The other contaminants, alkylphenols, phthalates and tributyltin were not detected in the studied soil samples. They are surfactants and plasticizers introduced by contaminated wastewater applications or direct deposition from agricultural treatments. Their concentration are thus related to wastewater contamination and agricultural practices. Tributyltin can also be introduced by rainfall leading to very weak concentrations (<1 ng·g⁻¹ Sn) (Huang et al., 2004). These compounds are known to persist several days before their degradation in well oxygenated soils (Topp and Starratt, 2000; Hoch, 2001) or deeper migration in soils (Sopheak et al., 2015). As the study focused on topsoil, deeper migration in soil cannot be ruled out. Moreover, this study only focused on regulated pollutants without taking into account their degradation products that may impact the ecosystems. Further experiments will be performed, including these degradation products, to assess their occurrence and impact. Alternatively, other data could indicate a historical contamination such as crop health, ecosystem description, soil fertility or soil microbial content. In any case our study shown that no gradient or trace is left on the topsoil.