The global distribution and environmental drivers of the soil antibiotic resistome

Abstract

Background: Little is known about the global distribution and environmental drivers of key microbial functional traits such as antibiotic resistance genes (ARGs). Soils are one of Earth's largest reservoirs of ARGs, which are integral for soil microbial competition, and have potential implications for plant and human health. Yet, their diversity and global patterns remain poorly described. Here, we analyzed 285 ARGs in soils from 1012 sites across all continents and created the first global atlas with the distributions of topsoil ARGs.

Results: We show that ARGs peaked in high latitude cold and boreal forests. Climatic seasonality and mobile genetic elements, associated with the transmission of antibiotic resistance, were also key drivers of their global distribution. Dominant ARGs were mainly related to multidrug resistance genes and efflux pump machineries. We further pinpointed the global hotspots of the diversity and proportions of soil ARGs.

Conclusions: Together, our work provides the foundation for a better understanding of the ecology and global distribution of the environmental soil antibiotic resistome. Video Abstract.

Keywords: Antibiotic resistance; Global change; Global scale; Human health; Mobile genetic elements.

Fig. 1

Location of the 1012 sites included in this study

Fig. 2

Proportion and richness of antibiotic resistance genes (ARGs) across global biomes and continents. A includes the distribution histogram for the proportion and richness of ARGs. B and C include the mean (± SE, number of sites/ecosystems in brackets) of the proportion and richness of soil ARGs across global biomes and continents, respectively. Each site is considered a statistical replicate in the PERMANOVA analyses. The proportion of ARGs was determined as the average standardized relative abundance of 285 individual ARGs. Details of the global biome classification can be found in Supplementary Table 2. Temp., temperate. Trop., tropical. We grouped our data to ensure high enough resolution for all ecosystem sub-types

Fig. 3

ARG composition across global biomes and continents. A shows the proportion (%) of ARG types across global biomes and continents (n for each biome in brackets). B shows the ubiquity (%) of dominant ARGs. Dominant ARGs are those that are abundant (top 20% of abundance), ubiquitous (occurred at > 50% of sites), and present in at least 2/3 of the biomes surveyed. Details on the global biome classification used can be found in Supplementary Table 2. Temp., temperate and Trop., tropical. An additional visualization of the community composition of all ARGs across global biomes and continents can be found in Fig. 4

Fig. 4

Community composition of ARGs across continents (A) and global biomes (B). NMDS analysis (Bray-Curtis) summarizing the community composition information (relative abundance) of 285 ARGs across different continents and global biomes (see Supplementary Table 2)

Fig. 5

Drivers of the proportion (A) and richness (B) of topsoil ARGs globally. A and B include structural equation models assessing the direct and indirect effects of environmental factors on the proportion and richness of ARGs. The proportion of ARGs was determined as the average standardized relative abundance of 285 individual ARGs. We grouped the different categories of predictors (climate, soil properties, vegetation, and MGEs) in the same box for graphical simplicity (these boxes do not represent latent variables). Variables within these boxes are allowed to covary, with the exception of elevation and spatial dissimilarity, which constituted our degree of freedom. Numbers adjacent to arrows are indicative of the effect size of the relationship. Only significant effects (P < 0.05) are plotted. Supplementary Tables 4–5 show the full SEM. F, forests; G, grasslands; S, shrublands. MAT, mean annual temperature. PSEA, precipitation seasonality. TSEA, temperature seasonality. There was a nonsignificant deviation of the data from the model (χ² = 0.10, df = 1; P = 0.75; RMSEA P = 0.93; bootstrap P = 0.71). C includes selected scatter and boxplots showing the regression between environmental factors and soil ARGs. Red lines are Loess regressions. MGEs includes both richness and proportions. Y-axis in C is shown in log scale. Units and acronyms are available in Supplementary Table 3

Fig. 6

Spearman rank correlations between environmental predictors and the proportion and richness of soil ARGs. Acronyms are available in Supplementary Table 3. MAT, mean annual temperature. MAP, mean annual precipitation. PSEA, precipitation seasonality. TSEA, temperature seasonality. No significant correlation is plotted in gray

Fig. 7

A global atlas of the distribution of topsoil ARGs. A and B represent the present global distribution of the proportion and richness of soil ARGs, respectively. Numbers associated with the legend of this figure show standardized proportions and richness of topsoil ARGs. The proportion of ARGs was determined as the average standardized proportion of 285 individual ARGs. Our models returned an R2 of 0.92 (for richness) and 0.86 (for relative abundance). Outlier regions in our global survey (> 97.5% quantile of the chi-squared distribution) were not considered in our global atlas and are plotted in white as no data (see Supplementary Fig. 1)