Environmental patterns are imposed on the population structure of Escherichia coli after fecal deposition

Abstract

The intestinal microbe Escherichia coli is subject to fecal deposition in secondary habitats, where it persists transiently, allowing for the opportunity to colonize new hosts. Selection in the secondary habitat can be postulated, but its impact on the genomic diversity of E. coli is unknown. Environmental selective pressure on extrahost E. coli can be revealed by landscape genetic analysis, which examines the influences of dispersal processes, landscape features, and the environment on the spatiotemporal distribution of genes in natural populations. We conducted multilocus sequence analysis of 353 E. coli isolates from soil and fecal samples obtained in a recreational meadow to examine the ecological processes controlling their distributions. Soil isolates, as a group, were not genetically distinct from fecal isolates, with only 0.8% of genetic variation and no fixed mutations attributed to the isolate source. Analysis of the landscape genetic structure of E. coli populations showed a patchy spatial structure consistent with patterns of fecal deposition. Controlling for the spatial pattern made it possible to detect environmental gradients of pH, moisture, and organic matter corresponding to the genetic structure of E. coli in soil. Ecological distinctions among E. coli subpopulations (i.e., E. coli reference collection [ECOR] groups) contributed to variation in subpopulation distributions. Therefore, while fecal deposition is the major predictor of E. coli distributions on the field scale, selection imposed by the soil environment has a significant impact on E. coli population structure and potentially amplifies the occasional introduction of stress-tolerant strains to new host individuals by transmission through water or food.

FIG. 1.

Contour plots of soil variables. (A) Contour plot of elevation. Black lines represent elevation contours, and black text indicates the contour values (10-m areal resolution). Points represent soil sample locations in May, July, and October 2008. The Rhinebeck silt loam area is shaded tan, and the Hudson silt loam area is unshaded. (B, C, and D) Contour plots of soil pH, %M, and %OM, respectively. Black contours represent the trend surface output from universal kriging accounting for a gradient in the data mean. Black text indicates the contour values. Colored points represent the measured value for each soil sample.

FIG. 2.

Principal coordinate analysis (PCA) of average genetic distance estimates from 501 ClonalFrame dendrograms. Sphere colors indicate subpopulation membership of isolates. Purple spheres depict isolates from long branches. Lines connect isolate positions to sequence collection centroids. Translucent ellipses represent 67% confidence interval clouds around the genetic distribution of sequence collections. Field E. coli isolates (n = 353) consisted of soil and fecal isolates from this study. Global E. coli isolates (n = 438) consisted of sequence types of diverse clinical, food, and environmental specimens from the GenBank and shigatox.net databases.

FIG. 3.

Venn diagrams from variance partitioning of subpopulation genetic structures between edaphic, landscape, temporal, and spatial variables. Variables included in partitions are listed in each panel. The central rectangle represents interactions between the three nonspatial variance partitions and the spatial partition. Nonspatial interactions among edaphic, landscape, and temporal variables (E × L × T) are indicated near the lower left corners of diagrams. Proportions of variance explained by each partition are displayed, with the significance of F tests annotated as follows: *, P < 0.05; **, P < 0.01; and ***, P < 0.001. (A) Legend; (B) all soil E. coli isolates; (C) clade B1A; (D) clade B1B; (E) ECOR B2; (F) ECOR D; (G) ECOR E; and (H) clade ET-1.