Biogeography of soil bacteria and archaea across France

Abstract

Over the last two decades, a considerable effort has been made to decipher the biogeography of soil microbial communities as a whole, from small to broad scales. In contrast, few studies have focused on the taxonomic groups constituting these communities; thus, our knowledge of their ecological attributes and the drivers determining their composition and distribution is limited. We applied a pyrosequencing approach targeting 16S ribosomal RNA (rRNA) genes in soil DNA to a set of 2173 soil samples from France to reach a comprehensive understanding of the spatial distribution of bacteria and archaea and to identify the ecological processes and environmental drivers involved. Taxonomic assignment of the soil 16S rRNA sequences indicated the presence of 32 bacterial phyla or subphyla and 3 archaeal phyla. Twenty of these 35 phyla were cosmopolitan and abundant, with heterogeneous spatial distributions structured in patches ranging from a 43- to 260-km radius. The hierarchy of the main environmental drivers of phyla distribution was soil pH > land management > soil texture > soil nutrients > climate. At a lower taxonomic level, 47 dominant genera belonging to 12 phyla aggregated 62.1% of the sequences. We also showed that the phylum-level distribution can be determined largely by the distribution of the dominant genus or, alternatively, reflect the combined distribution of all of the phylum members. Together, our study demonstrated that soil bacteria and archaea present highly diverse biogeographical patterns on a nationwide scale and that studies based on intensive and systematic sampling on a wide spatial scale provide a promising contribution for elucidating soil biodiversity determinism.

Fig. 1. Sampling design.

Map of France and the systematic sampling grid (16 × 16 km) of the French Soil Quality Monitoring Network (RMQS) (16).

Fig. 2. Representativeness of bacterial and archaeal phyla in French soils.

Left: The proportion of sampling sites where phyla were present. Right: The average relative abundance of the phyla. The four groups were determined by ascendant hierarchical clustering (fig. S1). The statistical differences in the phyla distributions are indicated in table S1.

Fig. 3. Relative abundance of the 47 main genera.

The main genera presented more than 0.5% of sequences on average and occurred in more than 75% of sites. The genera were classified across the bacterial and archaeal phyla. “Others” corresponds to the sum of all genera within each phylum representing fewer than 0.5% of sequences.

Fig. 4. Mapping of abundance of the most dominant bacterial and archaeal phyla across France.

For each map, d is the range in kilometers estimated by the model, and R² corresponds to the correlation between the predicted and measured values. The quality parameters and the model types are detailed in table S2.

Fig. 5. Comparison of maps phylum/genera.

(A) Four examples of a phylum for which the spatial distribution is consistent with its major genus. (B) Two examples of a phylum for which the spatial distribution represents the cumulative distributions of all genera belonging to the phylum. The complete set of maps is available in fig. S4.

Fig. 6. Variance partitioning of the microbial phyla across France according to environmental and spatial parameters.

(A) The 20 microbial phyla are ranked from the most to the least abundant. The explained variance corresponds to the sum of the adjusted R² values of the significant parameters within the contextual groups (soil physicochemical parameters, land management, spatial descriptors, climate, interactions between soil physicochemical properties and land management). The threshold for statistical significance was set at 0.01. Missing values indicate that no variable of the related group was retained in the model. (B to D) Contribution and effect of environmental parameters, land management, and spatial descriptors (PCNM at medium and coarse scales) on the distribution of bacterial and archaeal phyla. The colors depict the direction of the standardized partial regression coefficients (green, positive effect; red, negative effect). The height of the shape and the values indicate the percentage of variance explained by environmental parameters [for (B) and (D), proportions are comparable between boxes] and the coefficient of the standardized partial regression of each land management type. For this last effect, the coefficients are relative to a reference level grouping of 60 samples unclassified in the four types (C). The explained variance represents the respective significant contribution of each variable and was calculated by considering all other variables using partial regression models and adjusting the R² values.