Global distribution of shared DNA segments reveals selection on recent transfers.

(A) Schematic of pairwise haplotype sharing metric: For each position in the core genome, we compute the fraction of strain pairs from different hosts that have identical genotypes across a window of synonymous sites (Section 5.2 in S1 Text). (B) Observed sharing landscape for B. vulgatus (middle panel); separate comparisons are performed for strains from the same clade (blue, Δℓ≈1,500 synonymous sites ≈ 10 kb) or different clades (orange, Δℓ≈220 synonymous sites ≈ 1.5 kb). The top panel shows the average synonymous divergence computed in sliding windows of size Δl = 3,000. These landscapes reveal regions of elevated sharing across hosts (e.g., shaded region) that cannot be explained by local reductions in diversity. Red shaded region indicates the within-host recombination event in Fig 5B and 5C. The bottom panel shows analogous sharing landscapes from neutral simulations (Section 5.3 in S1 Text), which display more even rates of sharing across the genome. Gray lines denote 100 simulation runs with the same parameters, while the blue line highlights 1 typical run. (C) Sharing landscape for E. rectale, computed for pairs of strains in different hosts (top) and co-colonizing strains from the same host (bottom). (D) Heterogeneous sharing landscapes across 27 species. Blue points show the coefficient of variation of the sharing probability across the genome for all species with sufficient between-host comparisons. B. vulgatus (within clade) and E. rectale are highlighted as pink triangles. Gray points show analogous values derived from neutral simulations across a range of parameter values (Section 5.3 in S1 Text); each point denotes the mean of 100 simulation runs, while lines show the standard deviation. The data underlying this figure can be found in https://doi.org/10.5281/zenodo.10304481.