The divided-genome model of Niehus, Mitri, Fletcher and Foster (Nature Communications, 2015) predicts that immigration of genotypes lacking a locally beneficial trait enables horizontal selective sweeps in microbial communities: when the per-generation migration rate mm does not overpower selection (m<sm<s, with ss the trait's selection coefficient) and the horizontal transfer rate rr of the trait is small but nonzero (10−610^{-6}–10−410^{-4}), the beneficial locus spreads across diverse genomic backgrounds on timescales of about 10410^{4}–10610^{6} generations, reducing diversity at the swept locus relative to the rest of the genome so that, ecologically, genes rather than species occupy niches. A decade later, gene-specific sweeps have been documented in natural systems, but the model's mechanism itself remains unvalidated: no study has jointly quantified rr, mm, and ss in a natural community, tested the predicted regime boundary (m<sm<s), sweep timescales, and locus-versus-background diversity signatures under controlled sampling designs, or discriminated the migration-plus-transfer mechanism from competing explanations of the same genomic signatures. The problem is to settle whether this mechanism operates in nature.

Background

Horizontal gene transfer (HGT) is the acquisition of genetic material across lineage boundaries rather than by vertical descent, via transformation (uptake of free DNA), transduction (phage-mediated transfer), or conjugation (cell-to-cell plasmid transfer); in microbes it is a dominant source of evolutionary innovation, notably including antibiotic-resistance genes. Classical bacterial population genetics predicts that when recombination and transfer are rare, a strongly beneficial mutation spreads together with the genome that carries it in a genome-wide (clonal) selective sweep, erasing diversity across the whole genome — the assumption of the ecotype model of bacterial speciation, in which each adaptive event seeds a new ecologically specialized clonal lineage. Population-genomic studies of natural communities (for example, marine Vibrio populations examined by Shapiro et al. 2012) instead found gene-specific sweeps: loci under selection sweep through a community while the rest of the genomes remain diverse. Reconciling this with measured recombination rates, which appear too low under standard population-genetic arguments, is an unresolved puzzle of microbial population genomics, closely tied to the interpretation of the pangenome — the partition of a microbial species' gene pool into a conserved core genome and a variable accessory genome. Niehus et al. (2015) proposed a mechanistic explanation with an eco-evolutionary model of a community in an ecologically focal patch (for example a nutrient particle, a host, or a patch selecting for resistance to a locally present antibiotic). By deterministic ordinary-differential-equation analyses and coalescent arguments (the coalescent is the backward-in-time reconstruction of the genealogy relating the sequences sampled from a population), they showed that immigration of genotypes lacking the focal trait transforms the effect of transfer: incoming non-carriers act as fresh recipients, so horizontal transfer of the trait is greatly amplified even for realistic rates 10−6<r<10−410^{-6}<r<10^{-4} per generation. The horizontal sweep requires migration not to overpower selection (m<sm<s: too much migration swamps adaptation), transfers most effectively at intermediate selection strengths, and proceeds on timescales of about 10410^{4}–10610^{6} generations (months to years for fast-growing bacteria). The signature is a high diversity ratio DR=Dbg/DfDR=D_{bg}/D_{f} — high background-genome diversity DbgD_{bg} relative to low diversity DfD_{f} at the focal locus — and, crucially, this signature is observable only under patch-focused sampling, while sampling one species across patches instead makes transfer appear to generate accessory-genome diversity. The authors thus predicted that microbial genomes can be mosaics of loci with distinct ecologies (a 'divided genome'), so that ecological species concepts need not map onto whole organisms. They also characterized a two-compartment extension for pre-adapted migrants: when the external environment surrounding the patch is large, returning migrants rarely carry the trait and the original predictions hold; when it is small, patch and environment collapse into a single effectively closed population with little horizontal transfer. Several competing mechanisms can produce the same gene-specific sweep signatures: negative frequency-dependent selection on linked deleterious loci (selection in which a genotype's fitness depends on its own frequency; Takeuchi et al. 2015), which however requires low basal recombination; recombination barriers between ecologically diverging populations; soft sweeps from neutrally pre-transferred standing variation (the adaptive allele already sits on several genomic backgrounds before selection begins, so the sweep does not pass through a single clone); and, added later, metacommunity structure with moderate HGT (a metacommunity is a set of habitat patches connected by dispersal; Pompei et al. 2023). Because these mechanisms make overlapping predictions about single-locus diversity patterns, distinguishing them — and establishing the migration-plus-transfer mechanism as the (or a) cause of gene-specific sweeps in nature — requires joint quantification of the key rates and discriminative field tests.

Problem Statement

Determine whether the migration-assisted horizontal gene sweep mechanism of the divided-genome model operates in natural microbial communities, and distinguish it from competing explanations of the same genomic signatures. Concretely: (1) jointly estimate, for natural communities in ecologically focal patches, the baseline per-generation horizontal transfer rate rr of naturally sweeping loci, the effective per-generation immigration rate mm of genotypes lacking the trait, and the selection coefficient ss of the swept trait, with uncertainty quantification and explicit identification of the source environment of migrants; (2) test the model's discriminative predictions — that locus-specific sweeps across diverse genomic backgrounds occur only when migration does not overpower selection (m<sm<s), proceed on timescales of order 10410^{4}–10610^{6} generations, and produce reduced diversity at the focal locus relative to the background genome (high diversity ratio DR=Dbg/DfDR=D_{bg}/D_{f}) under patch-focused sampling but not under background-genome-focused sampling; and (3) discriminate this mechanism against the main alternatives for gene-specific sweeps — negative frequency-dependent selection on linked loci, recombination barriers between ecotypes, soft sweeps from standing variation, and metacommunity structure with moderate transfer — using predictions or data that separate the mechanisms rather than a single-locus signature, which is known to be insufficient. An answer that shows the mechanism is absent or dominated by specific alternatives in nature, with the same rate quantification and discriminative power, resolves the problem just as fully as one that confirms it.

References

  1. Niehus R, Mitri S, Fletcher AG, Foster KR. Migration and horizontal gene transfer divide microbial genomes into multiple niches. Nature Communications 6, 8924 (2015). doi:10.1038/ncomms9924, https://doi.org/10.1038/ncomms9924
  2. Takeuchi N, Cordero OX, Koonin EV, Kaneko K. Gene-specific selective sweeps in bacteria and archaea caused by negative frequency-dependent selection. BMC Biology 13, 20 (2015). doi:10.1186/s12915-015-0131-7, https://doi.org/10.1186/s12915-015-0131-7
  3. Shapiro BJ, Friedman J, Cordero OX, Preheim SP, Timberlake S, Szabó G, Polz MF, Alm EJ. Population genomics of early events in the ecological differentiation of bacteria. Science 336, 48–51 (2012). doi:10.1126/science.1218198, https://doi.org/10.1126/science.1218198
  4. Pompei S, Bella E, Weitz JS, Grilli J, Cosentino Lagomarsino M. Metacommunity structure preserves genome diversity in the presence of gene-specific selective sweeps under moderate rates of horizontal gene transfer. PLOS Computational Biology 19, e1011532 (2023). doi:10.1371/journal.pcbi.1011532, https://doi.org/10.1371/journal.pcbi.1011532
  5. Wolff R, Garud NR. Gene-specific selective sweeps are pervasive across human gut microbiomes. Nature 650, 710–717 (2025; online first 17 December 2025, print issue February 2026). doi:10.1038/s41586-025-09798-y, https://doi.org/10.1038/s41586-025-09798-y
  6. Baumdicker F, Kupczok A. Tackling the Pangenome Dilemma Requires the Concerted Analysis of Multiple Population Genetic Processes. Genome Biology and Evolution 15, evad067 (2023). doi:10.1093/gbe/evad067, https://doi.org/10.1093/gbe/evad067
  7. Järvenpää M, Gutmann MU, Vehtari A, Marttinen P. Gaussian process modelling in approximate Bayesian computation to estimate horizontal gene transfer in bacteria. The Annals of Applied Statistics 12, 2228–2251 (2018). doi:10.1214/18-AOAS1150, https://doi.org/10.1214/18-AOAS1150
  8. Zhu S, Hong J, Wang T. Horizontal gene transfer is predicted to overcome the diversity limit of competing microbial species. Nature Communications 15, 800 (2024). doi:10.1038/s41467-024-45154-w, https://doi.org/10.1038/s41467-024-45154-w
  9. Shoer S, Reicher L, Zhao C, Pollard KS, Pilpel Y, Segal E. Pangenomes of human gut microbiota uncover links between genetic diversity and stress response. Cell Host & Microbe 32, 1744–1757 (2024). doi:10.1016/j.chom.2024.08.017, https://doi.org/10.1016/j.chom.2024.08.017
  10. Bonham KS, Wolfe BE, Dutton RJ. Extensive horizontal gene transfer in cheese-associated bacteria. eLife 6, e22144 (2017). doi:10.7554/eLife.22144, https://doi.org/10.7554/eLife.22144
  11. Whelan F. How the social lives of bacteria affect their pangenome. Essays in Biochemistry (2026). doi:10.1042/ebc20250039, https://doi.org/10.1042/ebc20250039