The hazardous oxygenated polycyclic aromatic hydrocarbons (OPAHs) originate from combustion (primary sources) or post-emission conversion of PAHs (secondary sources). We evaluated the global distribution of up to 15 OPAHs in 195 mineral topsoils from 33 study sites (covering 52°N-47°S, 71°W-118°E), to identify indications of primary or secondary sources of OPAHs. The sums of the (frequently measured 7 and 15) OPAH concentrations correlated with those of the Σ16EPA-PAHs. The relationship of the Σ16EPA-PAHs concentrations with the Σ7OPAHs/Σ16EPA-PAHs concentration ratio (a measure of the variable OPAH sources) could be described by a power function with a negative exponent <1, leveling off at a Σ16EPA-PAHs concentration of ca. 400 ng g^-1. We suggest that below this value, secondary sources contributed more to the OPAHs burden in soil than above, where primary sources dominated the OPAHs mixture. This was supported by a negative correlation of the Σ16EPA-PAHs concentrations with the contribution of the more readily biologically produced highly polar OPAHs (octanol-water partition coefficient, log K_OW <3) to the Σ7OPAHs concentrations. We identified mean annual precipitation (Spearman-r = 0.33, p <0.001, n = 143) and clay concentrations (r = 0.55, p <0.001, n = 33) as important drivers of the Σ7OPAHs/Σ16EPA-PAHs concentration ratios. Our results indicate that at low PAH contamination levels, secondary sources contribute considerably and to a variable extent to total OPAH concentrations, while at Σ16EPA-PAHs contamination levels >400 ng g^-1, there was a nearly constant ratio of Σ7OPAHs/Σ16EPA-PAHs (0.08±standard error 0.005, n = 80) determined by their combustion sources.

The concentrations of 16 EPA-PAHs and up to 15 OPAHs were taken from the literature as listed in Table 1 or specifically analyzed for this study using the method briefly described below. In the various cited studies, from which the data originated and in our new sampling in Germany and Switzerland, topsoils were sampled in a way representative for the A horizon (Argentina, China, Slovakia), at 0-0.05 m depth (Switzerland [two sites], Thailand), at 0-0.1 m depth (Brazil, Germany, Uzbekistan, Switzerland [one site]), or 0-0.15 m depth (Switzerland [one site]). The soil samples were frozen in field-fresh state or air/freeze-dried, sieved <2 mm, and stored at ‑20°C before analysis. pH was measured with a glass electrode in H₂O, 0.01 M CaCl₂, or 1 M KCl at a soil:solution ratio of 1:2.5. Clay concentrations were determined with the pipet method after removal of carbonates with HCl, organic matter with H₂O₂, and Fe oxides with dithionite-citrate-hydrogencarbonate, and wet sieving <50 μm. An aliquot of each soil sample was milled and the total carbon (TC) and total nitrogen (TN) concentrations were determined with an elemental analyzer (vario EL cube, Elementar Analysensysteme GmbH, Langenselbold, Germany). The inorganic C (IC) concentration of each sample was also determined from an aliquot of soil after combusting soil organic carbon (C_org) in a muffle oven (550°C, 2 h). The C_org concentration was quantified as the difference between the TC and IC concentrations. All available soil properties are summarized in Table S2. Although our data set did not cover all continents, we consider it as global because it covers a wide range of latitudes and longitudes (Fig. S1).

At all study sites, at least the concentrations of the 16 EPA-PAHs, which we used here, and of up to 15 OPAHs (see Table S1) in topsoils, European Reference Material (ERM-CC013a), and procedural blanks (inert sorbent, Isolute HM-N, Biotage, Sweden or diatomaceous earth, Dionex, Sunnyvale, CA, USA) were determined using previously published methods (Bandowe & Wilcke, 2010; Bandowe et al., 2010; Lundstedt et al., 2014). In brief, soils (10-20 g) were mixed with Isolute HM-N or diatomaceous earth and transferred into 33-mL accelerated solvent extractor (ASE) cells. Each sample was spiked with a mixture of deuterated PAHs (six to 11 compounds) and two deuterated OPAHs. Each sample was extracted twice by pressurized liquid extraction with an ASE (ASE 200, Dionex, Sunnvale, CA, USA). Dichloromethane was used for the first extraction followed by acetone:dichloromethane:trifluoroacetic acid (1%) [250:125:1 v:v:v] for the second extraction. An acidified solvent mixture was used in the second step to improv the extraction of hydroxyl-PAHs and carboxyl-PAHs which were part of the target compound list in some of the previous studies (Bandowe & Wilcke, 2010; Bandowe et al., 2010; Bandowe et al., 2011), but were not considered here. The instrumental conditions for the ASE were the same as in Bandowe & Wilcke (2010), were more details of the used method are explained

Extracts from each sample were combined, passed through Na₂SO₄, spiked with hexane and rotary evaporated until <1 mL remained. Extracts were transferred to 3 g silica gel (10% deactivated) in an 8-mL glass column. Each sample was eluted with (a) hexane:dichloromethane (5:1 v:v) and (b) dichloromethane followed by acetone. Fractions a and b, which contained PAHs and OPAHs, respectively, were collected in separate flasks. Each flask was spiked with drops of toluene, and then rotary evaporated to <1 mL before being transferred to 2-mL vials to determine the concentrations of target PACs. PAHs and OPAHs were determined in two different runs with a gas chromatograph-mass spectrometer (Agilent 7890 A GC coupled to Agilent 5975 C mass spectrometer, Agilent, Santa Clara, CA, U.S.A.). The GC-MS was operated in the electron ionization mode with selected ion monitoring of target PACs. As a check of the accuracy of our analytical procedure, we simultaneously analyzed aliquots of the European Reference Material ERM-CCO13a (Polycyclic Aromatic Hydrocarbons in Soil) from the Federal Institute of Materials Research and Testing (BAM), Berlin, Germany. Procedural blanks (inert bulk sorbent: Isolute HM-N or diatomaceous earth) were also extracted and analyzed with the same methods as the samples and reference materials. Concentrations of target compounds were determined by the internal standard procedure. The average mass of target compounds in blanks was deducted from that in the samples before calculating the final concentrations per dry mass of extracted soil. Further details of quality control procedures are specified in previous papers (Bandowe & Wilcke, 2010; Bandowe et al., 2010; 2019; Lundstedt et al., 2014). Table S1 lists the names and abbreviations of all analyzed compounds. Results of our quality control procedures are reported in the Supporting Information and Table S2 reports the OPAH and PAH concentrations in all considered individual soil samples along with important soil properties.

The sum of the concentrations of 16 PAHs defined by the U.S. Environmental Protection Agency (EPA) as priority pollutants is Ʃ16EPA-PAHs and the sums of the concentrations of the seven OPAHs (i.e., 1-indanone, 1,4-naphthoquinone, 1-naphthaldehyde, 2-biphenyldicarboxaldehyde, 9-fluorenone, 1,2-acenaphthenequinone, and 9,10-anthracenedione), which have been determined in all samples, and 15 OPAHs (Table S1) are ∑7OPAHs and ∑15OPAHs, respectively. To roughly convert all pH values measured in H₂O or 0.01 M CaCl₂ to the pH in 1 M KCl, we lowered the pH (H₂O) by 1 unit and the pH (0.01 M CaCl₂) by 0.5 units accounting for the different ionic strength of the various commonly used equilibrium solutions. Because the data were usually not normally distributed and could not be brought to normal distribution by square root or ln(x+1) transformations, we used the nonparametric Spearman correlation analysis in the software package STATISTICA (Statsoft Inc., Tulsa, OK, U.S.A.). Octanol water partition coefficients (K_OW) for OPAHs (Table S1) were estimated according to the US EPA KOWWINTM module of Estimation Programs Interface Suite™ for Microsoft® Windows, v 4.11. U.S. EPA, Washington, DC, USA. https://www.epa.gov/tsca-screening-tools/epi-suitetm-estimation-program-interface, verified on 04/03/2019.

If cells are empty, data is not available.