September 23, 2023


It's Your Education

Global homogenization of the structure and function in the soil microbiome of urban greenspaces


Urban greenspaces, such as parks and residential gardens, are critically important for promoting mental and physical well-being and for reducing morbidity and mortality (1). According to the United Nations, 68% of the global population will live in cities by 2050, increasing the environmental and social stresses for the billions of humans living in urban areas (2). Urban greenspaces make up most of the open spaces available for recreational activities in urban areas such as sport and social engagement and play important roles in curbing pollution, reducing noise, and lowering air temperatures (13). Urban greenspaces also provide habitat for a myriad of plants and animals including a wide range of bird species (e.g., pigeons), terrestrial and arboreal mammals (e.g., squirrels), and diverse above- and below-ground invertebrates (3, 4). Similarly, soils in urban greenspaces are also home to a diverse community of microbes, including archaea, bacteria, fungi, and protists (46). This soil biodiversity plays important roles in maintaining ecosystem services such as soil fertility, plant health, productivity, and waste decomposition (7). Moreover, human exposure to soil microbes has been shown to be beneficial to human health by promoting effective immunoregulation functions and reducing allergies (8). However, some soil microbes (e.g., Mycobacterium, Listeria, and Fusarium spp.) can also have substantial negative consequences for the sustainability of urban greenspaces and for animal, human, and plant health. For instance, some soil microbial taxa harbor antibiotic resistance genes and could potentially influence the health of people and animals by reducing our ability to fight human diseases (9, 10). Unfortunately, unlike with urban birds, plants, and mammals, we have a very limited understanding of the identity and function of the soil organisms inhabiting our urban parks and gardens. Previous microbial studies in urban greenspaces have focused on specific locations (e.g., New York’s Central Park) (4, 5), specific groups of soil organisms (e.g., mycorrhizal fungi) (6), or specific microbial functions (e.g., denitrification) (11). However, we currently lack a global perspective on the soil microbiome of urban greenspaces and how these soil microbiomes compare to those found in nearby natural ecosystems. A comprehensive evaluation must consider multiple cities with contrasting populations, vegetation, and climates and include a wide range of functional and taxonomic groups of soil organisms. Improving our understanding of soil organisms associated with urban parks is a critical step toward building a better understanding of the role of these fundamental organisms in controlling the functioning and health of urban environments and toward managing the sustainable development of urban greenspaces.

Here, we report results of the first global field assessment of the soil microbiome in urban greenspaces, aiming to (i) compare the diversity and community composition of microbial taxonomic and functional groups of urban greenspaces with those of nearby “natural” ecosystems; (ii) identify the soil microbial taxa (bacteria, archaea, protists, and fungi) that are consistently found to be residents of urban greenspaces across the globe; (iii) evaluate the socioeconomic and environmental factors linked to the structure of the soil microbiome of urban greenspaces; and (iv) assess the microbial functional attributes characterizing soils in urban greenspaces including genes associated with pathogenesis, greenhouse gas emissions, nutrient cycling, and abiotic stress.


To achieve these goals, we carried out a field survey across 112 sites from 17 countries spanning six continents (Fig. 1A, table S1, and fig. S1). We used an experimental design that included paired urban and nearby natural sites located across 56 globally distributed locations (statistical “blocks”; Fig. 1A, table S1, and fig. S1). In each location (e.g., Cape Town, South Africa; fig. S1), we established a 30 m by 30 m plot in a representative urban greenspace (e.g., an urban forest or lawn in a city park; fig. S1) and a nearby, relatively undisturbed natural ecosystem resembling the original vegetation representative of these locations (e.g., a natural grassland; fig. S1). Urban greenspaces were located mostly in public parks and large residential gardens and comprised a mixture of open areas with lawns, scattered trees, patches of shrubs, and associated flowerbeds. We refer to these ecosystems as urban greenspaces hereafter (1). Natural areas near these cities were unmanaged or seminatural forests, shrublands, and grasslands, many of them relict ecosystems. Urban greenspaces and natural ecosystems strongly differ in vegetation structure and supported lower levels of plant species richness (fig. S2 and table S2). Agricultural lands were not included in this study. Our database comprises a wide range of urban areas including relatively small (population < 50,000; e.g., Alice Springs, Australia), medium (population, 50,000 to 1 million; e.g., Ljubljana, Slovenia), and large cities (population > 5 million; e.g., Beijing, China; table S1), and spanned a large range in vegetation types (forests, grasslands, and shrublands) and climatic regions (arid, temperate, tropical, and cold ecosystems; table S1). Our sampling was explicitly designed to account for the spatial heterogeneity in our plots (fig. S1; Materials and Methods). At each plot, we collected three composite soil samples (to 7.5 cm in depth) along with information on the dominant vegetation type (e.g., forest), vegetation structure in three 30-m transects (plant species richness, plant cover, and proportion of locations with ectomycorrhizal dominant plant species; table S2 and fig. S1) and management practices (irrigation, fertilization, and mowing). A total of 336 composite topsoil samples were analyzed for this study as detailed in Materials and Methods.

Fig. 1 The diversity and structure of the soil microbiome in urban greenspaces across the globe.

(A) The locations of the 56 surveyed cities included in this study along with their respective human populations. The names of some of the largest cities are included in this figure (see table S1 for details on all cities). (B) The changes [site-level response ratios, mean ± 95% confidence interval (CI)] in microbial diversity and the proportional abundances of significant taxa, comparing urban greenspaces to nearby natural ecosystems (n = 56 response ratios). Points above the dashed line indicate that the diversity or relative abundances of those taxa are higher in urban greenspaces compared to the corresponding natural areas near each city. Complementary figures showing results for other groups of organisms can be found in figs. S7 to S10. Phylotypes, ASVs; MG-RAST, metagenomics rapid annotation using subsystem technology.

To achieve our first aim, we characterized the biodiversity [number of amplicon sequence variants (ASVs); taxa that share 100% sequence similarity in the targeted gene region] and community composition (proportional abundances of ASVs, phyla/class, and functional groups) of bacteria, archaea, protists, and fungi in the soil samples from 56 paired cities and natural vegetated areas using ribosomal RNA (rRNA) gene amplicon sequencing (Materials and Methods). Diversity within each group was calculated at an equivalent sequencing depth across all samples using plot-level ASVs based on three soil composite samples per plot (Materials and Methods). We compared the diversity and proportion of soil organisms using two statistical approaches that explicitly considered our sampling using a blocked design (paired urban and natural ecosystems): (i) response ratios (i.e., changes in the values of microbial attributes from natural to cities; see Materials and Methods for correlations with other choices of response ratios) and (ii) nested permutational multivariate analysis of variance (PERMANOVA) comparing urban to natural ecosystems (56 paired locations; see Materials and Methods for details).

Our analyses show that urban greenspaces harbor soil microbiomes that are distinct from adjacent natural ecosystems (Figs. 1 to 8 and figs. S3 to S10). Urban greenspaces harbored communities of soil protists and bacteria that were, on average, 12 and 17%, respectively, more diverse than the adjacent natural ecosystems, with no significant differences in the richness of fungi and archaea (Fig. 1B and fig. S3). However, our analyses also revealed that, at the global scale, urban greenspaces tended to host more homogeneous microbial communities across cities than those found across natural ecosystems (Betadisper P < 0.001; Fig. 2, A and B, and figs. S4 to S6). Specifically, microbial communities were between 25% (protists) and 101% (fungi) more homogeneous in urban greenspaces than in nearby natural ecosystems (+39% for bacteria and +83% for archaea; Fig. 2B). In other words, our analyses show a greater similarity in the community composition of archaea, bacteria, fungi, and protists across the 56 globally distributed urban greenspaces than across the corresponding natural ecosystems (Fig. 2B). This result was maintained across contrasting geographic, climatic, and vegetation contexts (fig. S6). We further show that patterns in soil community similarity for different soil organisms (archaea, bacteria, fungi, and protists) are highly correlated across communities, both across cities and across natural ecosystems (Fig. 3). This finding suggests that the homogenization effects of urban greenspaces and the potential environmental drivers of these effects are shared across distinct groups of soil organisms. Human-induced land-use changes such as the conversion of forests to pasture have been reported to result in the biotic homogenization of soil microbial communities (12). Moreover, the importance of land-use intensification (e.g., livestock grazing) in generating cross-site multitrophic homogenization has been previously reported at larger spatial scales (13). Likewise, homogenization of bacterial communities has also been reported in dust samples collected from urban areas compared with more rural areas across the United States (14). Our results, emphasizing a global convergence (reduced geographic variation) in the soil microbiome of urban greenspaces, are consistent with the effects of urbanization on bird (15), plant, and invertebrate communities (16). Together, our findings provide novel evidence that urban greenspaces are important hot spots for the local (alpha) diversity of some microbial groups such as bacteria and protists worldwide. However, we also show that the geographic variation in microbial community composition is typically lower across cities worldwide than across natural ecosystems, suggesting a global homogenization in the soil microbiome of urban greenspaces.

Fig. 2 Global homogenization in the soil microbiome of urban greenspaces.

(A) The soil community composition heterogeneity of soil microorganisms in natural ecosystems and urban greenspaces. Boxes include median and 25th/75th percentile of the distances to the group centroid derived from betadisper (vegan R package). Asterisks indicate significant differences in compositional heterogeneity based on permutation test for homogeneity of multivariate dispersions (Materials and Methods). (B) Information on the within-group (urban or natural) community similarity for archaea, bacteria, fungi, protists, and metagenomics. Dots represent the average similarity (Bray-Curtis) in the soil community composition of each natural/urban site compared with the rest of natural/urban sites, respectively. In these panels, higher community similarity values indicate that the soils harbor communities that are more similar in composition, compared with the rest of the sites, across the 56 surveyed locations. The solid lines show mean values (n = 112 urban greenspaces and natural ecosystems). Asterisks indicate significant differences in nested PERMANOVA analyses using a block design as described in the Materials and Methods. (C) The relationship between soil pH and within-group (urban or natural) community similarity for archaea, bacteria, fungi, and protists. **P < 0.01; *P < 0.05.

Fig. 3 Relationships between the within-group (urban or natural) community similarity of archaea, bacteria, fungi, and protists.

(A to F) Dots represent the average similarity in community composition of each natural/urban site compared with the rest of natural/urban sites, respectively. In these panels, higher community similarity values indicate that the soils harbor communities that are more similar in composition, compared with the rest of the sites, across the 56 surveyed locations (n = 56 urban greenspaces or natural ecosystems). **P < 0.01; *P < 0.05.

Fig. 4 Microbial phylotypes comprising the urban greenspace soil microbiome.

(A) Two phylogenetic trees showing the diversity of prokaryotic and eukaryotic organisms identified as being characteristic of urban greenspaces. See table S3 for the complete list of taxa and their representative sequences. (B) The proportion (mean ± SE) of the microbial phylotypes comprising the urban greenspace microbiome across continents, climates, and vegetation types (mean ± SE; n in brackets). ns, not significant; **P < 0.01. P values are based on nested PERMANOVA analyses using a block design as described in the Materials and Methods. EcM, ectomycorrhizal.

Fig. 5 Socioeconomic, management, and environmental drivers of the soil microbiome in urban greenspaces.

(A to L) Selected relationships between the proportional abundances or diversity of soil organisms and site characteristics across urban areas (n = 56 urban greenspaces). USD, U.S. dollars; MAT, mean annual temperature; MAP, mean annual precipitation.

Fig. 6 Spearman correlation analyses aiming to identify the most important socioeconomic factors, management practices, and environmental drivers of the taxonomic and functional properties of the soil microbiome of urban greenspaces (n = 56 urban greenspaces).

Statistically nonsignificant correlations (P > 0.05) are shown in white. Total P and N = Soil total P and N.

Fig. 7 The functional attributes of the soil microbiome in urban greenspaces across the globe.

(A to C) The changes (site-level response ratios) in the proportion of selected functional genes associated with human diseases and antibiotic resistance (A), nutrient cycles (B), and abiotic stress (C) from natural to urban ecosystems (mean ± 95% CI; n = 56 response ratios). Points above the dashed line indicate a positive response ratio, with the selected gene category or soil property being relatively more abundant in urban greenspace soils compared with the corresponding “natural” sites.

Fig. 8 Selected relationships between socioeconomic, management, and environmental drivers and the proportional abundances or richness of selected functional genes across urban greenspaces (A to E) (n = 56 urban greenspaces).

Soils from urban greenspaces harbored communities of archaea, bacteria, fungi, and protists distinct from those found in natural ecosystems, with urban greenspaces consistently supporting a significantly greater proportion of Gammaproteobacteria, Deltaproteobacteria, Bacteroidetes, Gemmatimonadetes, Ascomycota, Chlorophyta, and Amoebozoa (Fig. 1D and figs. S7 to S10). These phyla/classes include multiple fast-growing organisms (e.g., members from Gammaproteobacteria and Bacteroidetes) that could take advantage of the often fertilized and irrigated conditions found in urban ecosystems (Fig. 1D). The Amoebozoa group, for example, includes a wide variety of species that feed on bacteria and could more likely thrive in urban systems because of irrigation practices more often providing the water films needed to sustain their activities (Fig. 1D). Urban greenspaces also supported a higher proportion of Chlorophyta, important photosynthetic organisms, which often colonize bare soils in urban greenspaces (Fig. 1D) (17). Our results further indicate that the soils in urban greenspaces have a lower proportion of ectomycorrhizal and ericoid mycorrhizal fungi (but not arbuscular mycorrhizal fungi) compared with those from adjacent natural areas (Fig. 1D and figs. S7 to S10). These results are consistent in plots dominated by both ectomycorrhizal (EcM) and no-EcM plants (table S2 and fig. S11). The negative impact of urban environments on ectomycorrhizal fungi has been previously reported, at a local scale, in Southern Finland (6). Our findings further indicate that urban greenspaces support a higher proportion of fungal parasites and plant pathogens that are often economically important pests (Fig. 1D and figs. S8 and S10) (18).

Identifying the members of the soil microbiome that commonly reside in urban greenspaces is fundamental for the appropriate management of these habitats, as revealed by studies on macro-organisms (19). Here, we identified the indicative members of the soil microbiome of urban greenspaces across the globe (urban greenspaces taxa hereafter). We focused on those microbial phylotypes (ASVs) that (i) were relatively ubiquitous (>25% of all cities), (ii) were classified as “species indicators” of greenspaces using the algorithm of the “indicspecies” R package (see Materials and Methods), and (iii) showed statistically significantly higher proportions in urban rather than natural environments (see Materials and Methods for a more detailed description of these analyses). On the basis of these three criteria, we identified a total of 539 phylotypes (i.e., 488 bacteria, 47 protists, 2 archaea, and 2 fungi from a total of 142 genera) characterizing the soil microbiome of urban greenspaces (Fig. 4A and table S3). These patterns were consistent across geographical regions, climates, and vegetation types (Fig. 4B). The urban greenspace taxa included important fungal and oomycete plant pathogens such as Fusarium intricans, Pythium rostratifingens, and Pythium uncinulatum, fungal decomposers such as Mortierella elongata, archaeal nitrifiers such as Nitrososphaera sp., bacteria such as Streptomyces and Pseudomonas spp., and multiple species of bacteria-feeding amoebae (table S3 for sequences). Our results suggest that, similar to what has been found for birds (e.g., pigeons) and plants (e.g., roses), many comparable microbial species thrive in urban ecosystems across the globe.

To improve the sustainable management of urban greenspaces, we need to understand how environmental and socioeconomic changes will affect important structural and functional attributes of the soil microbiome. Here, we identified the most important socioeconomic factors [in terms of per capita gross domestic product (GDP) and population density], vegetation structure (plant diversity, plant cover, and presence of ectomycorrhizal dominant plants), park management practices (irrigation, fertilization, and mowing), and environmental factors (climate, solar radiation, total plant cover, and soil properties) associated with the soil microbiome of urban greenspaces (n = 56 urban greenspaces). Socioeconomic and climatic factors such as GDP per capita and high temperatures could influence the soil microbiome of urban greenspaces by increasing environmental stress associated with disturbance, pollution, and climatic stress. As expected, we found that soil properties and climate significantly influenced the soil microbiome across cities. For instance, we found a well-established association between soil pH and bacterial richness (20), a positive correlation between the proportion of Cercozoa and mean annual precipitation (21), and a positive relationship between bacterial and protist richness (Figs. 5 and 6 and figs. S12 to S14) (22). Likewise, soil pH was also the most important factor associated with the changes in the soil community similarity across cities, which mostly supported curvilinear relationships (Fig. 2C). Plant diversity was negatively related to bacterial and archaeal community similarity, as well as to the proportion of urban greenspace taxa (Fig. 6).

Our findings show that economic metrics, park management practices, and climate are important factors associated with the richness and community composition of the soil microbiome in urban greenspaces (figs. S12 and S13). Our results indicate that warmer and more irrigated cities supported a greater proportion of fungal plant pathogens (Figs. 5 and 6 and figs. S12 and S13) and revealed that warmer and more densely populated cities also had a lower proportion of symbiotic ectomycorrhizal fungi—a pattern that could also be influenced by a reduction in ectomycorrhizal hosts at lower latitudes (23). Warming and disturbance have been shown to regulate the distribution of fungal pathogens and mycorrhizal organisms in natural terrestrial environments worldwide (24). In addition, more affluent cities were characterized by soils with a greater proportion of Nitrososphaeria (often involved in nitrification) and potentially rapid-growing Gammaproteobacteria (e.g., Pseudomonas sp.) (see other examples in Fig. 5 and figs. S12 and S13). These patterns are likely the result of fertilizer applications to greenspace soils, given that similar taxonomic shifts have been observed in grasslands worldwide receiving elevated nutrient inputs (25). Thus, our findings suggest that changes in temperature, as well as differences in important socioeconomic and management factors, are largely associated with the soil microbiome of our greenspaces.

To gain a deeper understanding of the functional attributes of the urban greenspace soil microbiome, we obtained shotgun metagenomic data from a subset of 54 soil samples (one of the soil composite samples from a selection of 27 cities and their corresponding natural ecosystems). This selection covered a wide range of cities from contrasting climates and populations and 17 countries from both hemispheres (table S1). Consistent with our results for bacterial richness, soils from urban greenspaces typically had greater functional gene richness than adjacent natural terrestrial ecosystems (Fig. 1C and fig. S15). We found further evidence of global homogenization of the soil functional microbiome of urban greenspaces (Fig. 2), with functional gene profiles from urban greenspaces being more homogeneous across cities worldwide than the variability among natural ecosystems (Fig. 2, B and C, and fig. S15)—a pattern likely associated with the higher soil pH in urban greenspaces compared with natural ecosystems (Fig. 2 and figs. S12 and S14).

The microbial communities in urban greenspaces had distinct functional gene profiles (Figs. 1C and 7, A to C, and figs. S4 to S6 and S15). We further investigated well-known functional genes associated with pathogenesis, greenhouse gas emissions, nutrient cycling, and abiotic stress (Fig. 7) and found that urban greenspaces had a higher proportion of plant and human pathogens (2629), greenhouse gas emission, and nitrogen and phosphorus cycle genes. For instance, soils from urban greenspaces had a higher proportion of genes associated with Mycobacterium virulence. This was especially important in the most alkaline soils [Figs. 6 to 8 and (15)] and is in agreement with our observation that one of the most abundant species in the soil microbiome of urban greenspaces is a member of the genus Mycobacterium (table S3). Although most soil mycobacteria are nonpathogenic (28), others are known to be important pathogens of humans and animals and can cause important respiratory infections (28, 29). Moreover, compared with natural ecosystems, urban greenspaces had a greater proportion of genes associated with Listeria and diphtheria toxins, Vibrio pathogenesis islands, and key antibiotic resistance genes (e.g., β-lactamases in Streptococcus, which includes penicillin) (Fig. 7A and fig. S15) (26, 27), which could all potentially influence human health. The proportion of genes coding for Vibrio pathogenesis islands was higher in more affluent cities and that of genes coding for β-lactamases in Streptococcus in warmer, more densely populated cities (Fig. 4C and fig. S12). We also found higher proportions of viral genes in urban greenspaces than in natural ecosystems, particularly in fertilized greenspaces (Fig. 7A and fig. S15). We note that gene annotations are approximate, and we do not know whether these particular genes detected may affect human health outcomes. Therefore, extrapolating and linking the occurrence of particular soil microbial genes to human health needs to be further investigated in the future to better understand how those microbes found in urban soils may affect human health.

Our results also indicate that soils from urban greenspaces included a higher proportion of bacterial genes associated with N and P cycling, likely associated with the fact that nitrogen and P are important fertilizers in urban greenspaces (from atmospheric deposition and direct fertilizer application). We also found a greater proportion of genes associated with archaeal methylotrophic methanogenesis (30) and denitrification processes, especially in irrigated and mowed greenspaces (Figs. 5 to 7B and fig. S15). This is important because it suggests that urban greenspaces could potentially be important sources of greenhouse gas emissions (methane and nitrous oxide) to the atmosphere (11). However, direct measurements of these soil processes are required to determine how these patterns in gene abundances might relate to actual process rates in urban greenspaces. Last, we found a higher proportion of gene copies associated with bacterial tolerance to alkaline or saline conditions (Figs. 5 to 7C and fig. S15), which are typical of many urban greenspaces (fig. S14). Together, these findings suggest that the soil microbiome of urban greenspaces supports a wide variety of potentially pathogenic organisms and microbes associated with important soil biogeochemical conditions and processes. The potential consequences of these relationships justify more detailed future investigations to better understand the functioning of urban soils and their contributions to environmental and human health.

In summary, we found that urban greenspaces are important hot spots of local soil microbial taxonomic and functional diversity but also support a global homogenization in the structure and function of the soil microbiome. More specifically, we show that urban soils across the globe harbor more similar microbiomes than would be expected from comparable analyses of soils from adjacent natural ecosystems. Our analyses indicate that soils from urban greenspaces are characterized by higher proportions of fast-growing bacteria, algae, nitrifiers, and important plant pathogens, which were particularly dominant in the warmer, more affluent, and more intensively managed greenspaces. Last, our results indicate that the urban greenspace microbiome harbors a greater proportion of genes associated with greenhouse gas emissions (denitrification and methanogenesis), as well as elevated proportions of genes associated with human pathogens and antibiotic resistance (e.g., β-lactamases), which may potentially have important implications for human health. Together, our study represents the first global assessment of the structure and functional attributes of the soil microbiome of urban greenspaces worldwide.

Acknowledgments: We would like to thank C. Walsh and R. Ochoa-Hueso for advice on bioinformatics and statistical analyses. We also thank M. Martin for revising the English of the manuscript. In addition, we thank J. Owojori for connecting us with our sampling collaborator in Nigeria, A. R. Bamigboye. Funding: M.D.-B. and this project were supported by a 2019 Leonardo Grant for Researchers and Cultural Creators, BBVA Foundation (URBANFUN) and by the BES grant agreement no. LRB171019 (MUSGONET). M.D.-B. is also supported by a Ramón y Cajal grant from the Spanish Ministry of Science and Innovation (RYC2018-025483-I). N.F. was supported by grants from the U.S. National Science Foundation (DEB1556090 and DEB1542653). L.T. acknowledges support from Norway-Baltic collaboration grant EMP442 and Estonian Science Foundation grant PRG632. B.K.S. acknowledges a research award by the Humboldt Foundation and funding from the Australian Research Council (DP190103714). F.A. is supported by ANID FONDECYT 11180538 and 1170995. S.A. is funded by ANID FONDECYT 1170995 and ANID ANILLO ACT192027. F.B. and J.L.M. acknowledge support from the Spanish Ministry and FEDER funds for the project AGL2017-85755-R, the i-LINK+ 2018 (LINKA20069) from CSIC, as well as funds from “Fundación Séneca” from Murcia Province (19896/GERM/15). C.P. acknowledges support from the Spanish State Plan for Scientific and Technical Research and Innovation (2013–2016), award reference AGL201675762-R (AEI/FEDER, UE). M.B. acknowledges support from a Juan de la Cierva Formación grant from the Spanish Ministry of Economy and Competitiveness (FJCI-2018-036520-I). T.P.M. would like to acknowledge contributions from the National Research Foundation of South Africa and cities involved in the South African survey. Slovenian coauthors were supported by the research project J4-1766 “Methodology approaches in genome-based diversity and ecological plasticity study of truffles from their natural distribution areas” and the Research Program in Forest Biology, Ecology, and Technology (P4-0107) of the Slovenian Research Agency. J.D. and A. Rey acknowledge support from the FCT (IF/00950/2014 and SFRH/BDP/108913/2015, respectively), as well as from the MCTES, FSE, UE, and the CFE (UIDB/04004/2020) research unit financed by FCT/MCTES through national funds (PIDDAC). J.P.V. acknowledges financial support from SERB (Science and Engineering Research Board) (EEQ/2017/000775) India. J.-Z.H. and H.-W.H. are financially supported by Australian Research Council (DP170101628). The BBVA Foundation accepts no responsibility for the opinions, statements, and contents included in the project and/or the results thereof, which are entirely the responsibility of the authors. Author contributions: M.D.-B. developed the original idea of the analyses presented in the manuscript and coordinated all field and laboratory operations. Field data were collected by M.D.-B., D.J.E., Y.-R.L., A.R.B., J.L.B.-P., J.G.I., T.P.M., C.S., P.T., E.Z., J.P.V., L.W., J.W., T.G., M.B., G.F.P.-B., T.U.N., A.L.T., X.-Q.Z., J.D., A. Rodriguez,, X.Z., F.A., S.A., C.P., and A. Rey. Laboratory analyses were done by M.D.-B., N.F., H.-W.H., J.-Z.H., F.B., J.L.M., and L.T. Bioinformatic analyses were done by N.F., B.S., J.-T.W., B.K.S., and C.C.-D. Statistical analyses were done by M.D.-B. The manuscript was written by M.D.-B. and edited by N.F. and D.J.E., with contributions from all coauthors. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. The data associated with this study can be found in the following:; DOI: 10.6084/m9.figshare.12930986.