Urban soil microbiomes exhibit taxonomic and functional potential for enhanced contaminant cycling

Urbanization is a leading cause of global biodiversity loss, but its effects on soil microorganisms and biogeochemistry remain uncertain. Although urban soil microbiomes are influenced by common anthropogenic processes, generalizable patterns in how urbanization shapes their composition and functional potential are lacking. Identifying such patterns is essential for understanding cities’ roles in biogeochemical processes and managing continued urban expansion. We re-analyzed soil metagenomic sequences from the Global Urban Soil Environment Ecology Network (GLUSEEN) to identify coordinated changes in microbial taxonomy and functional potential and to determine core taxa and metabolisms across five global cities spanning multiple ecoregions and management regimes. We identified 111 gene annotations representing core urban soil functions that were present across all cities but rare in reference soils (<5% occurrence). Across gradients from low to high urban land use, nitrogen, trace and heavy metal, glycerol, and fructose/fructan metabolic processes were associated with more highly urbanized areas. Core urban microbial taxa, defined by their overrepresentation in urban land uses relative to reference soils, included diverse bacterial groups and a more constrained set of methane- and nitrogen-cycling archaea. These urban-associated taxonomic and functional signatures may serve as indicators of altered nutrient and contaminant cycling and provide a foundation for monitoring soil health and informing management strategies in rapidly urbanizing landscapes. More broadly, this work generates testable hypotheses and improves our ability to predict and manage how urban soils influence ecosystem services, contaminant dynamics, and global environmental change.

Despite urban areas covering only 1% of global land, 70% of the world’s population is projected to live in cities by 2050 (1, 2). Human activity in these urban centers is reshaping ecological and evolutionary processes, altering species distributions, interactions, and ecosystem function (3). Research on urbanization’s effect on ecosystems is becoming more common, but knowledge of urban soil microbiomes remains limited despite their central role in nutrient cycling, contaminant transformation, greenhouse gas production, and other processes that regulate global biogeochemistry (4, 5). Because soil microorganisms mediate many of these ecosystem services, understanding how urbanization alters microbial communities and their functional potential is critical for predicting and managing the ecological consequences of continued urban expansion.

Rapidly increasing urbanization and associated changes in land-use are responsible for biodiversity loss of macroorganisms (i.e., animal and plant communities) and for increased pollution, both of which may subsequently impact urban biogeochemistry and soil microbiome function (6, 7). Land clearing for agriculture and urban expansion, for instance, have resulted in fragmentation of forest biomes across the world (8). This practice also leads to increased pollutants, like nitrate and ammonium, which in turn, change soil biogeochemical properties and microbiome function (9). Additionally, industrial activities can lead to heavy metal and organic contamination in urban soils, in particular polycyclic aromatic hydrocarbons, that are toxic to humans and are added to soil through processes like burning fossil fuels and waste incineration (10, 11). Finally, urban land use and management practices like fertilization, irrigation, and the installation of impervious surfaces increase nutrient inputs into soils via runoff (12). Nitrogen runoff in particular can enhance microbial denitrification processes that release nitrous oxide – a potent greenhouse gas (13, 14).

In parallel, soil microorganisms (i.e., bacteria, archaea, and fungi) are major conductors of global carbon and nutrient cycles. Soil biogeochemical properties that are impacted by urbanization like moisture, temperature, pH, and carbon and nutrient concentrations have a strong effect on microbial communities (15–19). Microbial diversity is also dependent on interspecies and plant-microbe interactions (20–22), which have been shown to increase the relative abundance of genes involved in greenhouse gas emissions in urban soils (23). More generally, urban soil microbiomes have been shown to exhibit homogenization of bacterial communities compared to surrounding natural ecosystems (23, 24).

The extent to which the urban environment drives soil microbiomes towards a common state (i.e., microbiome composition and function; urban convergence hypothesis) versus supporting a robust, diverse microbial community (i.e., the urban mosaic) remains debated (25). Within the dataset re-analyzed here, for example, it was reported that soil biogeochemical properties like organic carbon and total nitrogen content converged due to anthropogenic factors (25). Other studies have also seen the effect of urban convergence on soil properties, such as soil particle size and soil carbon content (26). However, due to the plethora of land covers and uses contained within different metropolitan areas, it remains to be explored whether or not urbanization has a uniform effect that will drive microbial communities towards greater biodiversity loss or greater taxonomic and functional potential. Further, understanding the extent to which urbanization acts upon microbial taxa versus the functional traits they contain is important for representing the impacts of urban population growth on global biogeochemistry. Standardized spatial observations distributed within and across cities are key to resolving these questions.

Determining a core urban soil microbiome is essential to study patterns associated with anthropogenic factors. Broadly defined, a core microbiome is a set of shared taxa and/or functional potential across all samples in a dataset or a group within a dataset (27–30). In urban ecosystems, identifying a core microbiome can reveal microbial taxa and functions that consistently respond to urbanization despite differences in climate, geography, and land-use history, providing insight into the mechanisms by which cities influence ecosystem processes (31–34). Therefore, pinpointing specific taxa and functional potential in a core microbiome is the initial step in identifying microbial consortia that can sustain ecological function in urban soils and in forecasting how the community will react to disturbances. Methods to determine a core microbiome vary with experimental factors such as spatial or temporal scales, and can leverage occurrence-based data, relative abundance data, or a combination of relative abundance and occurrence data (29, 30). Here, we define the core urban soil microbiome as organisms and functional genes that are distributed across global urban land uses and are also scarce or absent from more pristine reference soils.

We use metagenomic sequences from soils collected by the Global Urban Soil Environment Ecology Network (GLUSEEN; 35) to investigate the taxonomic and functional diversity of the urban soil microbiome. We aim to examine (1) if soil taxonomic and functional diversity are conserved across cities worldwide, (2) if changes in taxonomy and functional potential are coupled, and (3) if there are ubiquitous (“core”) taxa and functions that are characteristic of urban soils regardless of climate and geography. Previous work with this dataset by Epp Schmidt et al. (36, 37) has shown some consistency in archaeal and bacterial communities, with archaeal convergence being driven by ammonia- and methane-oxidizing organisms and genes and bacterial convergence being driven by nitrite-oxidizing organisms. However, Epp Schmidt et al. (36, 37) did not explore aspects of urban microbiomes related to contaminants and other carbon and nitrogen cycling processes. We further extend this work by reporting concomitant changes in taxonomy and functional diversity within and across cities; extracting the core urban microbiome across all cities; and uncovering a newly reported set of widespread soil genomic potential for the tolerance and/or cycling of urban pollutants.

Dataset description

We used publicly available metagenomic sequencing data from GLUSEEN to investigate the urban soil microbiome in 89 soil samples across five global cities: USA: Baltimore (n = 16), Finland: Helsinki (n = 17), Finland: Lahti (n = 18), Hungary: Budapest (n = 18), and South Africa: Potchefstroom (n = 20) (Figure 1). Soils spanned three levels of urbanization that encompass a gradient of common anthropogenic impacts. A full description of the number of samples per land use per city is available in Supplementary Table 1. Further details on land use classifications, soil collection, and metagenomic sequencing are described by Pouyat et al. (25) and Epp Schmidt et al. (37).

Briefly, soils were collected prior to 2017 from surface soils (0–10 cm) as part of GLUSEEN, an international collaborative effort designed to examine the effects of urbanization on soil properties and ecological processes across diverse metropolitan regions. GLUSEEN sampling protocols, experimental design, and site-selection criteria were developed through a series of workshops held in 2013 and 2014 at the University of Helsinki (Lahti, Finland) and Johns Hopkins University (Baltimore, USA), with the goal of enabling standardized comparisons among urban ecosystems worldwide. The network employs a common experimental matrix based on gradients of physical disturbance and management intensity that define six urban soil habitat types broadly identifiable across cities. These habitat types represent a continuum of anthropogenic influence, ranging from native reference ecosystems with minimal urban impacts to remnant habitats exposed to urban environmental pressures, and highly altered turfgrass and ruderal habitats characterized by intensive management and/or substantial physical disturbance.

For the present study, soils were analyzed from five metropolitan regions spanning a broad range of climatic zones, including boreal-hemiboreal regions in Helsinki/Lahti, Finland; humid-subtropical regions in Baltimore, USA; continental regions in Budapest, Hungary; and semiarid regions in Potchefstroom, South Africa. Together, these cities encompass substantial variation in temperature, precipitation, urban development patterns, and population size, ranging from just over 100,000 to nearly 2 million residents. Within each city, samples were collected from remnant habitats (e.g., forest or grassland patches), turfgrass habitats (e.g., managed public greenspaces), and ruderal habitats associated with heavily disturbed or fill soils, as well as nearby reference sites representative of native ecosystem conditions. These habitat classes capture variation in site grading, soil disturbance, irrigation and fertilization practices, intensity of human use, vegetation cover, and broader urban environmental influences such as atmospheric pollution and habitat modification. The inclusion of standardized habitat types and paired reference sites across geographically distinct cities enables both local and global comparisons of urbanization effects on soil microbial communities and their functional potential.

DNA extraction and sequencing were performed by GLUSEEN. DNA was extracted from soil samples using the PowerLyzer PowerSoil DNA Isolation Kit (MoBio Laboratories, Carlsbad, CA, USA) following the manufacturer’s protocol with minor modifications described previously (37). DNA concentrations were quantified using a Qubit 2.0 Fluorometer (Invitrogen, Carlsbad, CA, USA), normalized to 0.2 ng μL - ¹, and prepared for shotgun metagenomic sequencing using the Nextera XT DNA Library Preparation Kit (Illumina, San Diego, CA, USA). Indexed libraries were pooled and sequenced on an Illumina HiSeq 3000 platform at the Oregon State University Center for Genome Research and Biocomputing (Corvallis, OR, USA). Samples with insufficient sequencing depth (<3 million reads) after the initial run were resequenced in a second pooled library.

Raw sequence data were processed using the Metagenomic Rapid Annotations using Subsystems Technology (MG-RAST) pipeline. Because tagmentation generated fragments approximately 1.5–2 kb in length, forward and reverse reads were annotated independently and merged following annotation. Quality control included removal of plant-derived sequences, dereplication of artificial duplicates, and dynamic trimming with a minimum Phred score threshold of 15 and a maximum of five bases below this threshold. Functional and taxonomic annotations were obtained using MG-RAST with default parameters, except that a more stringent maximum e-value cutoff of 1 × 10 - ¹ 5 was applied. Samples used for this publication are available at MG-RAST (39; https://www.mg-rast.org/mgmain.html?mgpage=project&project=mgp16346).

Taxonomic and functional inference of metagenomes

Metagenomic sequences from each soil were retrieved from MG-RAST and processed using the Joint Genome Institute (JGI) Integrated Microbial Genomes and Microbiomes (IMG/M) standard workflows (40). Briefly, reads meeting the QC standards were assembled into contigs using BBDuk version 38.79, bbcms version 38.44 and metaSPADES version 3.13.0 (41, 42). Read-based taxonomy was assigned to assembled contigs using the genome-based taxonomy database, and assembled contigs were further annotated against the Kyoto Encyclopedia of Genes and Genomes (i.e., KEGG Orthology, KO; (43–45) as part of standard JGI workflows (46, 47). While read-based taxonomy is a useful tool for determining microbial clades via metagenomics, future work that assesses taxonomy by sequencing selected portions of conserved marker genes may be a useful complement to the results presented here (40, 48, 49).

After sequence processing, we summarized taxa and KO for each soil sample into matrices by counting sequences in each soil assigned to every unique taxon (order-level due to inconsistencies in taxonomic annotations extending beyond microbial orders) or KO. Counts in each matrix were normalized using a relative log expression (RLE; 50) implemented in the R package ‘DESeq2’ (51). KO were mapped to pathways using the ‘KEGGREST’ package in R (version 4.3.1) (52). Code is available on GitHub (https://github.com/izabelstohel/urban_soil_paper).

Statistical analysis

All statistical analyses were implemented in R (version 4.3.1) with base statistical packages and with the ‘vegan’ and ‘usedist’ packages (53, 54). We examined how taxonomic diversity varied across land uses and cities by calculating alpha- (richness and Shannon Diversity Index) and beta-diversity (Bray-Curtis dissimilarity), followed by ANOVA and Tukey’s honestly significant difference (HSD) post hoc test or by permutational multivariate analysis of variance (PERMANOVA) to test for statistically significant differences in alpha- and beta-diversity, respectively. Diversity metrics were independently calculated based on the distribution of microbial orders (for microbial taxonomy) and KO (for microbial functional potential). Then, we individually compared how taxonomic and functional composition changed with geographic distance, using the Haversine dissimilarity matrix within and across cities, using Mantel tests. Relative changes in taxonomic and functional composition were assessed using Mantel test of Bray-Curtis dissimilarity for microbial taxonomy vs. functional potential.

Additionally, we visualized data in the following ways. For visualizations of read-based taxonomy, we first generated phylogenetic trees using phyloT v2 (55), an online tree generator based on the Genome Taxonomy Database. Then, we visualized the tree in R using the packages, ‘ggtree’ (56), ‘treeio’ (57), and ‘ggnewscale’ (58). We generated heatmaps using the ‘pheatmap’ package (59). Distributions of shared and unique taxa and KO across cities were visualized using the ‘ComplexUpset’ package in R (60, 61). All other visualizations were generated with ‘ggplot2’ (62).

To identify core and auxiliary components of the urban soil microbiome, taxonomic and functional datasets were evaluated separately. For taxonomic analyses, core urban taxa were defined as microbial orders detected across urban land uses in all five cities. Taxa were not required to be absent from reference habitats, as most were detected across all habitat types yet showed consistent enrichment in urban land uses relative to reference habitats.

Because functional potential is substantially more diverse and redundant than taxonomic composition, with many metabolic functions distributed across multiple phylogenetic groups, functional data were analyzed using several complementary approaches. First, we identified KO shared among urban land uses across all five cities, regardless of their occurrence in reference soils, to characterize broadly conserved functions of urban soil microbiomes. To identify functions more uniquely associated with urban environments, we further examined KO annotations that occurred in urban soils but were rare in reference soils (<5% occurrence). Two levels of stringency were applied: (i) high-confidence urban-associated functions, defined as KO detected in urban land uses across all five cities and in fewer than 5% of reference samples, and (ii) other urban-associated functions, defined as KO detected in at least one urban soil sample but in fewer than 5% of reference samples. The latter category was retained primarily for identification of potentially novel urban-associated microbial functions.

Microbial diversity of urban soils

When comparing diversity metrics across cities, the Baltimore soil microbiome appeared to be distinct. The taxonomic diversity of soil microbiomes in Baltimore was both higher (alpha) and distinct (beta) compared to other cities (Figures 2A, B). Excluding reference soils, taxonomic richness was nearly two-fold higher in Baltimore soils than all other cities (261.46 vs. 120.57-172.20, P < 0.001, F = 9.25) and Shannon diversity followed a similar trend (H’ = 3.77 vs. H’ = 0.80-1.87, P < 0.001, F = 29.29, Figure 2A). Baltimore soils also had lower functional richness compared to Budapest and Potchefstroom (mean = 2,189.636 vs. 3,440.714 and 3,325.067 unique KO per sample, P = 0.021 and 0.040, respectively, F = 3.01) and the lowest Shannon diversity index when compared to all other metropolitan areas (H’ = 7.110, H’ range = 7.669-7.816, P < 0.001, F = 9.24, Figure 2A). Taxonomic and functional composition in the Baltimore soil microbiome was also distinct when compared with other cities (PERMANOVA, both with and without reference soils P < 0.001, Figures 2B, C).

Highly urbanized land uses in most cities contained distinct soil microbiomes relative to reference soils. Though there was no significant change in taxonomic or functional alpha- diversity metrics within urban land uses (ruderal, remnant, turf) and only slight differences in alpha- diversity between urban and reference land uses (Figure 2); soil microbiomes associated with the most highly urbanized land use (ruderal) were distinct from reference soils with respect to taxonomy and/or functional potential in Baltimore, Potchefstroom, and Helsinki (PERMANOVA, P = 0.02, 0.03, and 0.09, F = 5.21, 2.90, and 2.38 for taxonomy and P = 0.02, 0.17, and 0.01, F = 1.68, 1.42, and 2.70 for functional potential, respectively).

We then tested relationships between microbial diversity (taxonomic and functional) and geographic position, and between taxonomic and functional dissimilarity, to assess the extent to which taxonomic and functional attributes of urban microbial communities were coordinated through space (Figure 3). When examining soils from all cities collectively, both taxonomic and functional dissimilarity increased with geographic distance (P = 0.01), an effect that was present both with and without the inclusion of reference soils. For each city individually, neither taxonomic nor functional diversity varied significantly with geographic distance within urban land uses in any city (P > 0.10). We obtained similar results when including reference soils, where the only significant correlations with distance were found in Baltimore (P = 0.04, 0.05 for taxonomy and functional potential, respectively). However, taxonomic and functional composition were partially coupled across all cities collectively (Mantel test, all P = 0.01, with and without reference soils, Figure 3F) and within Budapest, Helsinki, and Potchefstroom (Mantel test, all P = 0.01, with and without reference soils, Figure 3C).

Core urban microbial taxa and metabolic potential

While previous analyses of GLUSEEN soils identified broad patterns in urban-associated microbial functions, including nitrogen and methane cycling, antibiotic resistance, and heavy metal metabolism (25, 37), our study provides substantially greater taxonomic and functional resolution. Specifically, we identify globally distributed core urban taxa and 111 urban-associated core gene annotations, allowing us to evaluate their potential ecological roles and relationships to urban environmental conditions. This analysis expands upon prior work by highlighting specific metabolic pathways and candidate genes associated with contaminant cycling, nutrient transformations, and other urban-associated processes, thereby generating a suite of novel hypotheses and targets for future investigation.

Two hundred fifty-seven microbial orders were common to soil microbiomes across all cities (hereafter, ‘core’ organisms, Figure 4A). Although most of these orders were also contained within reference soils, they exhibited clear increases in abundance from reference to more urbanized soils (Figure 5; Supplementary Figures S1, S2). Among core urban microbial orders, eight bacterial and two archaeal orders had greater than 50% of their total abundance contained within ruderal soils (the most urban land use type, unclassified Candidatus Saccharibacteria, unclassified Crenarchaeota, Caldisericales, Clostridiales, Desulfurellales, Petrotogales, Micrococcales, Archaeoglobales, Chlamydiales, and unclassified Firmicutes), with low abundance in reference soils (0-15% for eight orders and 17-29% for the remaining two orders; Supplementary Table 2). Eleven more orders (incl. seven archaeal orders) had more than 50% of their total abundance contained within turf soils, with low corresponding abundance in reference soils (<12% of total abundance contained within reference soils, Aeromonadales, Caulobacterales, Burkholderiales, Candidatus Nitrosocaldales, Methanopyrales, unclassified Thermoplasmata, Nitrosopumilales, unclassified Aigarchaeota, unclassified Calescamantes, Thermoplasmatales, unclassified Thaumarchaeota). Overall (regardless of presence in reference soils), the most abundant shared orders of bacteria included Enterobacterales, Burkholderiales, Lactobacillales, Bacillales, Rhizobiales, Micrococcales, Caulobacterales, Acidobacteriales, Clostridiales, and Pseudomonadales.

Core urban archaeal orders were generally more evenly distributed across reference and urban land uses, however, several taxa still appeared to be more prevalent in urbanized soils. No core orders were found only in urban land uses, but 7 of 18 core archaeal orders were at least twice as abundant in ruderal vs. reference soils: Methanopyrales, Nitrosopumilales, Nitrososphaerales, Thermoproteales, Thermoplasmatales, Archaeoglobales, and Methanocellales. Overall (regardless of presence in reference soils), Nitrososphaerales was by far the most abundant shared archaeal order, followed by other nitrogen- and methane-cycling (Nitrosopumilales, Methanosarcinales, Methanomicrobiales, Methanobacteriales) and salt-tolerant archaea (Halobacteriales, Haloferacales, Natrialbales). Interestingly, after excluding reference soils, Baltimore soils displayed five times as many auxiliary (i.e., not in core microbiome) microbial orders as any other city (25 vs. five or less in all other cities; Figure 4A).

Further, we found 3,667 KO shared across urban land uses in all cities (Figure 4B). The 10 most abundant KO in urban soils, regardless of whether they were also detected in reference soils, encoded proteins that are associated with generalized cell functions (K07347 outer membrane usher protein, K01992, K02004, and K01990 ABC-2 type transport system permease and ATP-binding proteins, K02529 LacI family transcriptional regulator, galactose operon repressor, K03088 polymerase sigma-70 factor ECF subfamily, K01652 acetolactate synthase I/II/III large subunit, K07497 putative transposase), notably including two metal-dependent functions: K02014 (iron complex outer membrane receptor protein) and K02035 (peptide/nickel transport system substrate-binding protein).

When narrowing these potential functions to those not widely contained within reference soils, our results point to a core set of 111 gene annotations that were found in urban land uses in all five cities and in less than 5% of reference soils (Figure 6). In particular, 29 KO were common to urbanized land in all cities but were not found in any reference soil. Of these, 12 were found in at least 25% of soils from the most urbanized land use type (ruderal). This included four genes associated with nitrogen cycling (K02591 nifK, nitrogenase molybdenum-iron protein beta chain, K02586 nifD, nitrogenase molybdenum-iron protein alpha chain, K02588 nifH, nitrogenase iron protein NifH, K12256 spuC putrescine-–pyruvate transaminase). In addition to the two Mo-dependent genes related to N cycling (nifK and nifD), five other genes in this group are closely associated with heavy metals, contaminants, and/or microbial stress responses (K05297, rubredoxin-NAD+ reductase; K18601 aldehyde dehydrogenase; K07231 putative iron-regulated protein, K03332, fruA, fructan beta-fructosidase, K01960, pycB pyruvate carboxylase subunit B). The three remaining genes were associated with common cellular functions (K19123 casA, CRISPR system Cascade subunit CasA, K03041, rpoA1, DNA-directed RNA polymerase subunit A’, K10726, replicative DNA helicase Mcm).

If we further extend this analysis to urban soil KO found in less than 5% of reference samples and at least one urban soil, we generate an additional 82 KO for consideration, of which 34 are found in more than 25% of soils from the most urbanized land use type (Supplementary Table 2). Notable KO include: K08352 (thiosulfate reductase/polysulfide reductase chain A, in 74% of ruderal soils); K13794 (LysR family transcriptional regulator, regulatory protein for tcuABC, in 70% of ruderal soils); K00869 and K00938 (mevalonate kinase and phosphomevalonate kinase, in 48% and 39%); K02587 and K02229 (nifE, nitrogenase molybdenum-cofactor synthesis protein and cobG, precorrin-3B synthase, in 44% and 26%); K00520 and K08365 (mercuric reductase and MerR family transcriptional regulator, mercuric resistance operon regulatory protein, in 44% and 26%); K18833 (MFS transporter, DHA3 family, multidrug efflux protein, in 35%); K09819 (manganese/iron transport system permease protein, in 26%); K00596 (2,2-dialkylglycine decarboxylase (pyruvate), in 30%); genes related to fructose/fructan (K10554, frcA fructose transport system ATP-binding protein and K00895 diphosphate-dependent phosphofructokinase, in 30% and 30%), K16876 (2-furoate-–CoA ligase, in 26%); genes related to glycerol contamination (K17325 glycerol transport system ATP-binding protein in 30%, K00712 tagE poly(glycerol-phosphate) alpha-glucosyltransferase in 26%, K17322 glpP glycerol transport system permease protein in 26%), signatures of viral infection (K01186, sialidase-1 in 26%, K19046, casB, CRISPR system Cascade subunit CasB, in 39%), some uncharacterized functions, and several functions related to amino acid transport and metabolism.

Despite having the fewest auxiliary functional genes (i.e., genes not included in the core microbiome described above), Baltimore soils harbored a distinctly unique microbiome enriched in genes associated with heavy metal resistance and toxin processing, in contrast to other cities where auxiliary functions were dominated by common cellular processes. Excluding reference soils, Lahti soils had the most auxiliary microbial KO at 2895, followed by Budapest (2190), Helsinki (2104), Potchefstroom (2053), and Baltimore (1122). While the auxiliary microbiome of Baltimore soils contained the fewest potential functions of any city, it was also mostly unique in comparison to other cities. The most abundant of these auxiliary potential functions is thought to convey heavy metal resistance (K15725, outer membrane protein, heavy metal efflux system). Other highly abundant auxiliary functions in Baltimore soils were also associated with metals or other toxins, such as: K01578 MLYCD, malonyl-CoA decarboxylase, K15855, exo-1, 4-beta-D-glucosaminidase, K01577 oxc, oxalyl-CoA decarboxylase, K04108 4-hydroxybenzoyl-CoA reductase subunit gamma, and K01301 N-acetylated-alpha-linked acidic dipeptidase. In contrast, the auxiliary microbiome of other cities was mostly comprised of common cellular processes (K01223, K02761, K07345, K02073, K03475).

Urban centers are increasing in extent and population density, in turn disturbing soil microbiomes and the biogeochemical processes that they mediate (63). Soil microbial functions are an essential component of global biogeochemical cycles, as microorganisms regulate soil C storage and respiration (64). Here, we used 89 soil metagenomes from the GLUSEEN network to evaluate patterns in the taxonomic and functional potential of urban soil microbiomes. We extend the work of Epp Schmidt et al. by revealing a suite of core urban microorganisms and metabolisms related to (1) trace and heavy metals, (2) fructose/fructan, (3) viral infection, and (4) glycerol; in addition to N cycling processes that have been reported previously. These novel findings present a basis for understanding and modeling the ways in which urban biogeochemical cycles are distinct from natural ecosystems and for sustainably managing growing population centers (65).

Taxonomy and functional potential of urban soil microbiomes

We found biogeographical patterns in the urban soil microbiome that align with ecological processes such as microbial niche diversification (urban mosaic hypothesis) and environmental filtering (urban convergence hypothesis). There was a great array of functional potential relative to the amount of taxonomic diversity present in the urban soil microbiome (Figure 2A). High functional diversity lends support for the urban mosaic hypothesis, which proposes that microbial diversity increases with urbanization as a result of interacting selective pressures like land use, nutrients, pollution, and other factors on microbiomes (63, 66). Under the urban mosaic hypothesis, potential microbial functions, and in some cases taxonomic composition, expand to occupy new niche space opened by anthropogenic activity. For example, Mo et al. (67); and Scholier et al. (66) reported increased soil microbial diversity in anthropogenic areas compared to natural areas. Taxonomic similarities, on the other hand, have been more widely reported across cities (68–71). This provides evidence towards the urban convergence hypothesis, where physical characteristics of the urban environment drive microbial similarities, and supports the existence of a core urban microbiome. In our study, we observed comparatively low taxonomic diversity in urban soils (Figure 2A) and a clustering of microbiomes from highly urbanized land uses across cities (Figures 2B, C) that lends support for common microbial taxa and functional potential across five vastly different cities. This is consistent with findings in urban greenspaces where despite greater functional diversity, microbial taxonomies are relatively homogenized (23, 67).

However, because urban habitats in the GLUSEEN design encompass multiple land-use types spanning a continuum of disturbance and management intensity, whereas reference sites represent a more narrowly defined habitat, we acknowledge that some of the observed differences in taxonomic and functional diversity may reflect differences in habitat heterogeneity in addition to urbanization itself. Future studies incorporating a broader range of reference habitats and greater sampling intensity across land-use types may help disentangle the effects of urbanization from underlying environmental heterogeneity.

Interestingly, Baltimore appeared to have a distinct soil microbiome relative to all other urban centers. It had the highest taxonomic diversity but also the lowest functional diversity as compared to other cities (which had similar diversity, Figure 2A). Baltimore was also the only city that showed a sizable increase in taxonomic diversity in more urban (turf and ruderal) vs. more pristine (reference and remnant) soils (Figure 2A). We posit that the influence of environmental factors on soil microbiomes may be particularly important in Baltimore, which has a long history of industrialization, leaky sewage infrastructure, and high pollution levels (72–74). For example, Baltimore has been shown to display a dramatic urban heat island effect with increased temperatures along an urban gradient (75) and has reportedly high levels of lead and other pollutants (72, 76). Each of these factors may both exert unique selective pressures on microbial functions and vary substantially at the local scale. They may also lead to the emergence of genetic potential that is not well-described in microbial communities (i.e., microbial dark matter), which would result in the suppression of detectable functional potential observed in this study. While the reasons driving different patterns in microbial taxonomic and functional diversity in Baltimore are unknown, its distinctive patterns suggest further investigation into the unique taxonomic and functional potential its soil microbiomes contain.

Spatial variation in functional potential and taxonomic composition across the urban soil microbiome

While the mixture of diverse land uses within cities may obscure patterns based on space alone, the low magnitude of compositional changes within and across cities further underscores the existence of a core urban microbiome (Figure 3). Differences between urban and reference soils suggest that these core organisms and functions are distinct from more natural environments. We found that urban soil microbiome composition and functional diversity tended to be consistently structured through space, consistent with previous work on this dataset (36). For each individual city, neither taxonomic nor functional diversity varied significantly across spatial distances (Figures 3A, B). Across cities, both taxonomic and functional dissimilarity showed some correlation with spatial distances, (r = 0.34 and r = 0.35, respectively; (Figures 3D, E). This variation in microbial community composition across cities was likely due to overarching drivers such as soil mineralogy, pH, weather, and other abiotic factors that are known to influence soil microbiomes (16, 19, 77, 78).

Interestingly, microbial taxonomic composition and functional potential were partially decoupled from each other and showed different biogeographical relationships (Figures 3C, F). Taxonomic composition was more variable than functional potential through space in some cities (as indicated by flatter slopes on Figure 3), as it was virtually uncorrelated to functional potential in Lahti and Baltimore. Microbial functional potential is expected to be somewhat correlated to taxonomic composition due to phylogenetic conservation of microbial traits, which can lead to similar responses to environmental factors like substrate availability and climatic variables (79). However, if ecological selection operates on microbial genomic elements that are exchanged between organisms or if selection operates on unrelated genes and organisms, shifts in microbial taxonomy may not be coordinated with changes in genomic potential (18, 80–82). We posit that the diversity of factors such as available substrates, carbon sources, and/or pollutants in the urban environment may lead to variation in gene presence and abundance despite taxonomic convergence. Given the usage of microbial taxa as biomarkers of ecosystem health (4) and as engineering solutions (83), the disconnect between taxonomy and potential function observed here for urban soils is an important consideration in sustainable management practices.

Core vs. auxiliary taxa and metabolisms in global cities

Examining the core urban microbiome provides insight into the microorganisms that are likely to mediate the biogeochemical cycling of key nutrients, greenhouse gases, and pollutants in urban soils. Bacterial orders that were ubiquitous across cities and urban land covers relative to reference soils were typified by microorganisms usually associated with plant health, rhizosphere functions, and antibiotic and/or contaminant production and consumption (Figure 5; Supplementary Figures S1, S2). For example, Enterobacterales and Rhizobiales reside in the root systems of plants and some members have been found to maintain plant health under abiotic stress (84, 85). Bacillus, a common microorganism in the Bacillales order, is also ubiquitous in soils and promotes plant health through nutrient cycling and pathogen protection (86, 87). With regard to antibiotic and contaminant metabolism, some members of the Enterobacterales order (e.g., Enterobacteria) harbor increased antibiotic resistance genes in urban centers, while others are human pathogens (e.g., Salmonella, Escherichia, Shigella). Burkholderiales have also been shown to participate in pentachlorophenol degradation, a chemical used as a pesticide that persists in soil and has been coupled to nitrogen, sulfate, and iron reduction (88–90); while Clostridiales and Caulobacterales are commonly associated with a wide range of contaminants (91, 92). Like bacteria, ubiquitous urban archaea have been previously associated with nitrogen and methane cycling and/or salt tolerance (12, 93–95). While we cannot directly assess the coupling of these urban organisms with trace metals, pollutants, and/or harsh conditions using the data available in this study, their high plasticity and potential to acquire new genes (87, 96, 97) may make them a key focus for future research.

We also report core soil microorganisms with greater than 50% abundance in ruderal soils and corresponding low abundance in reference soils as organisms for deeper investigation (Figure 5; Supplementary Figures S1, S2). We focus on ruderal soils because they represent the most urbanized land cover type investigated here. First, members of Saccharibacteria have been reported to be key components of activated sludge (98, 99) and are often found in soil and soil-like environments. Members of Saccharibacteria are able to metabolize diverse compounds, putatively conveying an ability to thrive in a polluted urban environment (99). Saccharibacteria are also common constituents of the human oral microbiome which may also explain its presence in urban environments that have high human populations (100). Members of Clostridiales and other Firmicutes are also commonly associated with contaminant degradation, including polycyclic aromatic hydrocarbons, polychlorinated biphenyls, and phthalates that are of key interest to urban ecosystems because of their detrimental impacts on human health and long residence times (93, 101–104). Finally, many core urban soil microorganisms that were enriched in ruderal soils are thought to be thermophilic, including Caldisericales, Desulfurellales, and Petrotogales, in addition to having associations with various environmental contaminants. These organisms may therefore be more tolerant to higher temperatures associated with urban heat islands than their mesophilic counterparts (105). The prevalence of these taxa and other taxa with related reported functions in urban soils are suggestive of an urban soil microbiome that may be at least partially adapted to pollutants and extreme temperatures common to urban areas.

When extending our analysis to include urban turf (50% of total abundance contained in turf soils with low corresponding abundance in reference soils), we note an overrepresentation of archaea, along with some of the previously discussed bacterial clades associated with contaminant cycling (e.g., Caulobacterales, Burkholderiales). Archaea may be particularly well-suited to turf soils due to their abilities to thrive in extreme environments. Urban turf is subjected to a unique set of stressors including fertilization, irrigation, mowing, chemical applications for weed control and other uses, and high temperatures from lack of tree canopy cover. Indeed, many of the core archaeal clades enriched in turf are associated with nitrogen cycling (Candidatus Nitrosocaldales, Nitrosopumilales, unclassified Thaumarchaeota), anaerobic metabolisms (Methanopyrales), and/or heat tolerance (unclassified Thermoplasmata, Thermoplasmatales, unclassified Aigarchaeota; 106–111).

When considering core functions encoded by the urban soil microbiome, we found a plethora of metal-dependent genes and/or genes related to metal cycling (Supplementary Table 2). We highlight a suite of genes that were found in all cities, were absent from reference soils, and occurred in at least 25% of ruderal soils (Supplementary Table 2). Many iron-related functions fell in this category including: K05297, rubredoxin-NAD+ reductase, containing a central iron atom, and K07231, a putative iron-regulated protein, K09819, a manganese/iron transport system permease protein. In a study by Goff et al. (112), multi-metal contamination led to disruption in iron homeostasis, essential for most microbes, which could be a consequence of interest for urban environments (113). Furthermore, although N-cycling genes have been reported previously, we note the dependence of N-cycling genes in the core urban soil microbiome on molybdenum, in particular, nifK, nifD, and nifE. Molybdenum is a known urban contaminant with sources including industrialization and fertilization (114–116). These genes display a functional potential for adaptive response to an anthropogenic environment where greater concentrations of metals and contaminants are present.

When we extended our analysis to genes found in all cities and present in more than 25% of ruderal soils and less than 5% of reference soils, we found further evidence for the role of industrial processes in structuring the urban soil microbiome. Both phsA and psrA met this threshold and are involved in thiosulfate reductase, which is important in cycling of metals in the environment (117). Specifically, phsA is required for hydrogen sulfide production, which is toxic to both humans and the environment (118, 119). Additionally, MerA (K00520 mercuric reductase), encoded by the mer operon, was found in 43% of ruderal samples. MerA is a protein encoding a subunit in the mercuric reductase enzyme, allowing organisms to survive in the presence of high mercury (120). Within the same operon, MerR (K08365), a mercuric resistance operon regulatory protein, was found in ruderal soils at 26% (121). Fossil fuel combustion has led to increased mercury pollution in the environment, which is expected to continue to rise, and is also highly toxic to humans and the environment (122–124).

We also found a suite of genes associated with fructose and glycerol metabolism, and with cellular infection and replication, in more than 25% of ruderal soils and less than 5% of reference soils. While these genes are central to basic metabolic functions in many microorganisms, their disproportionate presence in urban soils raises questions surrounding additional functions these genes may serve in more polluted environments. Fructose, for example, is commonly implicated in plant-microbiome interactions in agricultural and natural soils (125–128). However, the urban environment includes additional sources of fructose, such as food waste enriched in high fructose corn syrup (129–132) and industrial byproducts including furfural (133), that may place a greater importance on fructose cycling in urban soils than more natural soils. We specifically highlight fruA (K03332), frcA (K10554), pfp (K00895), and hmfD (K16876), as potential areas for future research. Likewise, glycerol is a major component of manufactured products like fertilizers, soaps, cosmetics, antifreeze solutions, and pharmaceuticals, in addition to its natural occurrences (134–136). In particular, glpT (K17325), glpP (K17322), and tagE (K00712) may be disproportionately involved in the cycling of these urban waste products in urban soils versus central metabolic processes. Finally, we elucidate three cellular maintenance genes that could be signatures of heightened viral activity in urban soils – K19123 casA, CRISPR system Cascade subunit CasA, K03041, rpoA1, DNA-directed RNA polymerase subunit A’, K10726, and replicative DNA helicase Mcm. While the exact reason for the enrichment of these genes in urban soils is unknown, soil viral ecology is rapidly progressing to the forefront of microbial ecology due to the potential for soil viral impacts on human health and soil biogeochemistry (137–141). The enrichment of putative viral signatures in urban soils further underscores the importance of new investigations into soil viral ecology within urban centers.

Despite many commonalities in urban soil microbiomes, Baltimore soils strikingly contained some unique potential functions that may signify a greater impact of anthropogenic processes on microbial communities in this city. Baltimore’s unique functional potential supported diverse genes associated with metal cycling and aromatic degradation (Supplementary Table 2). Genes unique to Baltimore include K15725, the czcCBA efflux system, which is found to play a role in resistance to cobalt, zinc, and cadmium (142), and K04108 4-hydroxybenzoyl-CoA reductase, which is found in anaerobic phenol metabolism for benzene degradation (143). We therefore posit that cities that have a relatively recent legacy of industrial pollution and fossil fuel combustion, like Baltimore, may harbor organic contaminants (144–146) that drive distinct soil functional potential relative to cities with more distant industrial eras or stronger environmental regulations.

Future directions

Our research brings to light several exciting avenues for future research into urban ecology and biogeochemistry. Efforts to understand core urban microorganisms and functions will benefit from additional sampling campaigns to capture spatial and temporal fluctuations within microbial communities and from methodologies that describe soil microbiome composition and function in more detail. In particular, transitioning from read-based taxonomy to amplicon-based methodologies, or increasing sequencing depth, will likely improve the breadth and resolution of taxonomic identification, providing deeper insights into the microorganisms driving urban soil biogeochemistry. Investigating microbial gene expression through metatranscriptomics or metaproteomics, particularly focusing on the functions identified in this study, will also deepen our understanding of microbial roles within urban ecosystems. Finally, expanding sample collection to encompass a broader range of cities and intra-city locations, as well as temporal fluctuations, will ensure a more comprehensive and representative dataset. These advances can not only refine current ecological models but also provide a robust foundation for hypothesis-driven research and ecosystem management strategies.

Studies into the unique organisms and functions contained within the urban soil microbiome are essential as the human population continues to expand. Here, we conducted a re-analysis of 89 urban and reference soil microbiomes using metagenomic sequences from GLUSEEN. Building upon previous work by Epp Schmidt et al. (36, 37), we revealed a set of core microorganisms associated with the urban soil environment and uncovered a set of potential biogeochemical functions that were overrepresented in urban soils relative to more pristine environments. These included genes associated with industrial byproducts, trace and heavy metal cycling, and glycerol and fructose/fructan metabolism; as well as N-cycling genes previously suggested by Epp Schmidt et al. (36, 37). Our results also demonstrate that both geography and urbanization shape urban soil microbiomes. Geographic differences among cities contributed to variation in microbial taxonomic composition and functional potential, while urbanization produced consistent taxonomic and functional signatures that were observed across cities spanning diverse climates, ecoregions, and management regimes. Taken together, these findings highlight the interplay between local environmental context and shared urban pressures in structuring soil microbial communities. Overall, our synthesis reveals unique and ubiquitous features of the urban soil microbiome that provide a basis for further studies on the effects urbanization has on soil and human health.