Soil contamination in nearby natural areas mirrors that in urban greenspaces worldwide

Soil contamination undermines biodiversity, ecosystem functions, and human health, challenging multiple UN Sustainable Development Goals. While urban areas are expected to show greater contamination from vehicle emissions, industry, pesticides, and waste, contaminants can disperse via aerial transport, runoff, and littering, potentially affecting nearby natural ecosystems. Some contaminants (e.g., certain metal(loid)s and antibiotic resistance genes, ARGs) also have natural origins. However, previous efforts often examined single contaminants within limited environmental gradients, leaving unclear how contamination in urban greenspaces compares to adjacent natural areas at a global scale and which human vs. natural factors best explain their distribution. The authors hypothesized that human activities drive a spillover of contaminants from cities to nearby natural areas and that soil microbes, due to their sensitivity and functional roles, can serve as indicators of multi-dimensional contamination. They conducted a global paired survey across six continents to: (i) compare levels of metal(loid)s, pesticides, microplastics, and ARGs between urban greenspaces and adjacent natural areas; (ii) identify environmental and socio-economic factors associated with contamination; and (iii) assess links between contamination and microbial functional traits related to stress resistance, nutrient cycling, and pathogenesis. They also contrasted urban sites with remote maritime Antarctica to contextualize contamination levels.

Prior studies have documented that contaminants of concern (metal(loid)s, pesticides, microplastics, ARGs) can disperse via atmospheric transport, waste disposal/littering, and surface runoff, affecting areas beyond direct sources. Some contaminants have natural geochemical or biological origins and may be enriched irrespective of anthropogenic inputs. Most previous work focused on single contaminant classes and narrower climatic or environmental gradients, often lacking simultaneous measurement and direct urban–natural comparisons. Local and regional dispersion of specific contaminants (e.g., pesticides, microplastics) has been reported, including atmospheric transport to remote regions like Antarctica. Human pressures (population density, management practices, socio-economic context) are implicated in contamination patterns, but their relative importance across ecosystems remains insufficiently quantified.

Design and sites: A standardized global field survey sampled 56 paired urban greenspaces and adjacent natural/semi-natural areas (112 ecosystems total) across six continents and 17 countries. Natural areas were typically ~20 km from urban sites and free of evident historical human impacts; urban greenspaces were established parks and large residential gardens. Climate across sites spanned 210–1577 mm mean annual precipitation and 3.1–26.4 °C mean annual temperature. Sampling: At each ecosystem, a 30 m × 30 m plot with three parallel transects was established. Surface soil (top 5 cm) was collected at three points and composited; three composite samples per plot yielded 336 composite samples across 112 plots. Three additional composite samples were collected from remote maritime Antarctica. Soils were sieved (<2 mm); subsamples were air-dried for chemistry/contaminants and frozen at −20 °C for microbiome analyses. Contaminants measured: Four groups were targeted: (1) metal(loid)s (As, Cd, Cr, Cu, Pb, Hg, Ni, Zn) measured in all 336 composites; (2) 46 pesticide residues measured in 54 selected plots; (3) microplastics (shape and polymer type) measured in 64 selected plots; (4) 285 ARGs quantified in all 336 composites. Subsets for pesticides and microplastics were chosen to cover biogeographic gradients due to analytical costs and sample requirements. Environmental and socio-economic data: Site plant cover, soil properties (pH, total C and N, soil organic C, total P), climate (WorldClim 2013–2021 MAP, MAT), and management (mowing, irrigation, fertilization) were recorded. Socio-economic indicators included regional/city population size/density, GDP, and HDI. Laboratory analyses: Metals (As, Cd, Cr, Cu, Pb, Ni, Zn) via microwave digestion (HNO3+HF) and ICP-OES; Hg via aqua regia digestion and CVAFS. Pesticides: accelerated solvent extraction followed by HPLC–MS/MS (QTrap 5500), covering 46 analytes; LOQs 0.064–36 µg kg−1 depending on compound. Microplastics: Fenton oxidation to remove organic matter (<40 °C), NaCl density separation (repeated four times, aided by CaHCO3), filtration, visual counting (5 mm to 20 µm), polymer identification via Raman spectroscopy (Open Specy library). Rigorous contamination controls (glassware, filtered solutions, procedural blanks). ARGs: DNA extraction with PowerSoil; high-throughput qPCR targeting 285 ARGs across major antibiotic classes plus 16S rRNA as reference. Metagenomics: Shotgun sequencing on 54 composites (27 pairs). Quality-trimmed reads mapped to BacMet and MG-RAST/SEED databases; functional annotation summarized (SEED level 3) with SUPER-FOCUS; relative abundances calculated. Derived indices and statistics: Individual contaminants were standardized to 0–1 across sites and averaged within each category to form category indices; a multi-contamination index averaged the four categories for sites with all measurements (n=48). Response ratios (%) compared urban vs. paired natural values. Differences were also tested using nested PERMANOVA (adonis, 999 permutations) with location as strata. Drivers of contamination were assessed via structural equation modeling (AMOS 21; paths among socio-economic factors, climate, soil, plant cover, management) and linear mixed-effects models with model averaging (glmulti; importance by summed Akaike weights). Associations between contaminants and microbial functional genes were analyzed using OLS regressions (multi-contamination vs. functional gene proportions), threshold analyses (number of contaminants over defined maxima), and co-occurrence networks (Spearman correlations visualized in Cytoscape).

• Across continents, urban greenspaces and adjacent natural areas exhibited similar levels of multiple soil contaminants (metal(loid)s, pesticides, microplastics, ARGs), as shown by comparable category indices and response ratios. • Metal(loid)s: Pb, Ni, Cd, As showed similar levels between paired ecosystems, while urban soils had higher Hg, Zn, Cu, and Cr. Using Finnish soil contamination guidelines, 42% of urban sites and 36% of natural sites exceeded the lower limit for As contamination. Average metal(loid) levels in remote Antarctic soils were below guideline limits. • Pesticides: Residues (fungicides, herbicides, insecticides) were detected widely; 63% of surveyed natural ecosystems contained residues. No significant difference in pesticide occurrence between urban and natural soils was detected. The analyte list did not include several globally common actives (e.g., glyphosate, glufosinate, paraquat, 2,4-D), so true occurrence may be underestimated. • Microplastics: Ubiquitous in both urban and natural soils with similar contents between paired ecosystems. Mean abundance was 916.2 items kg−1 (range 166.7–3482.3 items kg−1). Dominant shapes were fibers; dominant polymers were polypropylene and polyester, with similar shape/polymer distributions in both ecosystem types. • Antibiotic resistance genes (ARGs): ARGs were ubiquitous across sites; urban soils supported greater ARG richness than natural soils, though the relative abundances of specific ARG classes varied geographically. Antarctic soils had significantly lower ARG richness. • Remote Antarctica: Soils contained measurable contaminants; microplastic concentrations and polymer/signature types in Antarctic surface soils were similar to those in surveyed urban greenspaces, consistent with long-range transport and local sources (e.g., research stations, tourism). • Drivers: Human-associated factors were essential in explaining contamination patterns. Population density was the most important socio-economic correlate of microplastics; GDP and HDI showed negative associations with metal(loid)s and microplastics. Management practices, especially fertilization, were positively associated with pesticide and ARG abundances. These findings held across SEM and mixed-effects modeling. • Microbial functional traits: Higher contamination (particularly multi-contamination) was associated with increased relative abundances of genes related to stress resistance (e.g., multidrug resistance, metal resistance), and pathogenicity (e.g., Listeria prophage genes), and decreased proportions of genes involved in phosphorus metabolism, iron acquisition/transport, and DNA repair. Specific associations included positive correlations between Hg and streptothricin-resistance genes. • Multi-contaminant effects: The number of contaminants exceeding defined thresholds correlated with shifts in functional gene profiles, suggesting cumulative impacts on the soil functional microbiome.

The study demonstrates that contamination in nearby natural areas mirrors that in urban greenspaces at a global scale, supporting the hypothesis of anthropogenic spillover of contaminants across ecosystem boundaries via atmospheric deposition, runoff, and other vectors. Comparable levels of pesticides, microplastics, and many metal(loid)s in natural and urban soils challenge assumptions that natural areas are less impacted by human activities. The associations between human factors (population density, management) and contamination reinforce the role of societal drivers, while negative associations with GDP/HDI suggest that wealthier regions may better mitigate certain contaminants. The observed links between multi-contamination and microbial functional traits indicate potential shifts toward stress resistance and pathogenicity alongside reductions in functions related to key nutrient cycles and DNA repair, implying risks to soil health and ecosystem services. Although causality cannot be inferred from observational data, the consistent patterns across continents and contaminant classes underscore the need for risk assessments that consider multiple contaminants simultaneously and for strategies that reduce diffuse pollution from urban sources to protect adjacent natural ecosystems.

This cross-continental, paired-ecosystem assessment reveals that soils in natural areas adjacent to cities are as contaminated as urban greenspaces across multiple contaminant classes. Human activities, notably higher population density and site management, are significant correlates of contamination levels, while higher GDP/HDI relate to lower burdens of certain contaminants. Multi-contamination is associated with shifts in the soil microbiome toward stress resistance and pathogenicity, potentially compromising ecosystem functions. The study highlights an expanding global contaminant footprint reaching even remote regions such as Antarctica. Future research should include broader contaminant spectra (e.g., pharmaceuticals), long-term monitoring, and controlled experiments to establish causal pathways, quantify bioavailability and risks, and evaluate mitigation strategies at landscape scales.

• Sampling for pesticides (54 plots) and microplastics (64 plots) covered subsets of sites due to analytical costs and sample requirements, leading to unbalanced datasets across contaminant categories. • The pesticide panel did not include several widely used actives (e.g., glyphosate, glufosinate, paraquat, 2,4-D), likely underestimating true pesticide burdens. • Microplastics extraction and visual counting may underestimate small particles; despite contamination controls, methodological limits remain. • Observational analyses (SEM, correlations) do not establish causality; unmeasured local factors may influence contamination patterns. • ARG detection is limited by qPCR target panels and detection limits; gene abundance does not directly equate to functional activity. • Antarctica comparisons are based on a small number of composite samples. • Bioavailability and toxicity were not directly assessed; risks depend on soil properties and contaminant speciation.