Bacterial Clones Hold Still: Population Genetics Model Anchors Epidemiological Tracing

A new review argues that the reliability of molecular epidemiology—the practice of tracing how pathogenic bacteria spread through populations—rests on a single, often unexamined assumption: that bacterial lineages remain genetically stable across both space and time. Writing in Frontiers in Bacteriology, Michel Tibayrenc of MIVEGEC, University of Montpellier, CNRS and IRD in France, sets out the “predominant clonal evolution” (PCE) model, a framework he has developed over decades with the late evolutionary biologist Francisco Ayala. The model holds that bacterial lineages are stable from the macroevolutionary scale all the way down to microevolution, on a scale of a few years and a few kilometers—the domain of what researchers call “measurably evolving pathogens.” That stability, the review contends, is the indispensable precondition that makes epidemiological tracing possible at all.

The discipline of molecular epidemiology was born in the 1970s, when genetic and molecular markers first revolutionized the tracking of pathogenic bacteria. The pioneering studies rested on a simple theoretical basis: identical or very similar bacterial “strains” must be genetically related, and therefore testify to a recent epidemiological diffusion. But those early efforts were hampered by markers that lacked resolving power. That drawback has disappeared with the advent of modern genomic markers, which offer very high resolution. The review traces this marker history in detail, beginning with isoenzymes—different electrophoretic variants of enzymes whose migration patterns in a gel reflect differences in amino acid sequences and, underneath those, differences in the underlying genes. In the 1980s isoenzymes were widely used to study bacterial populations, particularly Escherichia coli, and although they are now largely outdated, the author stresses that “outdated” does not mean “wrong”: past studies remain an invaluable source of information on ancient epidemics, provided their limitations are understood.

Subsequent markers moved from protein phenotypes to the analysis of DNA itself. The polymerase chain reaction amplifies target DNA sequences and underpins a family of typing methods. Random amplified polymorphic DNA (RAPD) uses arbitrarily determined primers and is extremely powerful for exploring genetic variability, but its lack of reproducibility limits routine strain identification, and the data are difficult to interpret in population-genetic terms. Restriction fragment length polymorphism (RFLP) reveals DNA variability using bacterial restriction endonucleases that cut DNA at specific sites; amplification fragment length polymorphism (AFLP) is the selective amplification of such restriction fragments by PCR; and pulsed-field gel electrophoresis (PFGE) separates large DNA fragments produced by a low-frequency restriction enzyme, with strains sharing a pattern referred to as “pulse types.” Microsatellites—short DNA sequences of usually one to four base pairs repeated in a row—are fast-evolving, high-resolution markers found in many organisms, including bacteria.

A pivotal standardization came with multilocus sequence typing (MLST), introduced by Maiden and colleagues in 1998. MLST sequences 450-base-pair portions of a set of usually seven housekeeping genes, and its main advantage is perfect portability: results can be reliably communicated among laboratories simply by emailing sequences. Strains sharing the same combination of alleles are assigned a sequence type (ST); strains sharing six of seven alleles are single-locus variants, and those sharing five are double-locus variants—together forming clonal complexes. MLST remains widely used in bacterial molecular epidemiology. Above it sit whole-genome approaches, made routine by the considerable reduction in sequencing costs over the last 20 years. Whole-genome sequencing identifies single-nucleotide polymorphisms (SNPs), and characterizing individual genotypes by hundreds of thousands of SNPs permits molecular epidemiology with extremely high resolution, down to a time scale of a few years and a spatial scale of a few miles—the “measurably evolving pathogens” concept.

But markers alone provide only raw material, the review argues. Population genetics—the study of genetic variation across space and time within and among populations, emphasizing the population or species as a whole—supplies the interpretive tools, and population genomics extends the same topic to the whole genome. Evolutionary genetics explores the past of species and populations over evolutionary timescales, relying mostly on phylogenetics. Surprisingly, the author observes, recent contributions to bacterial molecular epidemiology draw very little, if at all, on population genetics. The central question such analysis must address is the impact of genetic recombination: the exchange of genetic material between individuals or cells at two or more genetic loci. If recombination is frequent and tends toward panmixia, in which genetic exchange occurs at random, multilocus genotypes are unstable, reshuffled at each generation. Molecular epidemiology, by contrast, requires stability in space and time of the relevant units of analysis—and the PCE pattern ascertains exactly that stability.

Clonality, in the genetical definition used in the review, refers to all situations in which offspring are genetically identical or very similar to their parents—a definition that can correspond not only to mitotic propagation but also to selfing, inbreeding, strong homogamy and several cases of parthenogenesis. Its manifestations are specific: widespread multilocus genotypes over space and time with frequencies incompatible with panmictic expectations; linkage disequilibrium verified by adequate statistics; and a persistent phylogenetic signal at all evolutionary levels, from a microevolutionary scale of a few months or weeks to geological times. Frequent recombination is incompatible with the emergence and maintenance of a statistically detectable phylogenetic signal, because reshuffling multilocus genotypes in each generation erases the signal over the long term. The presence of multilocus bifurcating trees in a bacterial population therefore suggests that recombination is severely restricted or absent, and is considered classical evidence for predominant clonality. Such trees should be based on core genes found among all populations of the species, and on as many loci as possible, since a phylogeny based on a single gene does not reliably reflect the overall evolution of a species—gene trees and species trees are not the same.

Crucially, the PCE model does not claim that genetic recombination is absent in bacteria. Bacteria have “sex”—conjugation, transformation and transduction—and the concept of clonality is perfectly compatible with a low level of genetic transfer, provided it does not erase the persistent phylogenetic signal. The review terms this the “clonality threshold,” beyond which clonal evolution outweighs the long-term effects of recombination, with further data reinforcing the phylogenetic signal once the threshold is reached. The model also postulates that restrained recombination in pathogens is not mainly due to natural selection acting on an otherwise recombining species, as has been proposed for Neisseria meningitidis; rather, limited genetic exchange is a specific evolutionary strategy of micropathogens, driven by in-built properties, possibly as a major adaptation to parasitism that helps escape the “recombinational load” that disrupts favorable multilocus associations. Because some genetic exchange is probably present in all bacterial species, the clusters identified by multilocus bifurcating trees cannot be equated with strict cladistic clades; the author instead uses the term “near-clades”—discrete, stable and widespread phylogenetic lines whose discreteness can be somewhat blurred by occasional genetic exchange, with signal strength assessable by bootstrap analysis.

The review also introduces cautionary concepts. A “clonet” designates a set of isolates identical for a given genetic marker in a species undergoing PCE; if marker resolution is limited, an MLST sequence type or PFGE pulse type may be a clonet rather than a true clone, and higher-resolution markers will reveal additional variability within it. The age of a clonet’s most recent common ancestor can range from a few weeks to hundreds of years depending on the markers used, with considerable epidemiological implications. A further possible bias is the Wahlund effect, extended here to mean any apparent deviation from panmixia resulting solely from physical obstacles such as time or space impeding genetic exchange; pooling samples from disparate collections at different times can mimic clonality signatures. The way to rule this out is to verify that widespread multilocus genotypes, linkage disequilibrium and near-clades remain identical across the whole ecogeographical range of the species, or to look for clonality traits within strictly sympatric populations. An alternative model, epidemic clonality, posits occasional and ephemeral bouts of clonality in an otherwise highly recombining species, in which case clonal traits should vanish within months or a few years; the exact opposite is recombination-driven population structure, or panmixia, in which none of the PCE traits should be observed.

The model is then tested against four major species. Escherichia coli, despite a well-known recombination system and a recombination-to-mutation ratio showing recombination almost equal to mutation, appears at all evolutionary scales to be beyond the clonality threshold, subdivided into clear-cut near-clades—A, B1, B2 and D in the historical ECOR collection—corroborated by isoenzymes, RAPD, RFLP, MLST and whole-genome sequencing. Within the B2 near-clade, the widespread ST131 genotype, a recent arrival, nonetheless shows considerable genetic variability under WGS/SNP typing, subdivided into three subclades that began diverging 60 to 90 years ago. Mycobacterium tuberculosis, the causative agent of tuberculosis, exhibits linkage disequilibrium between MIRU VNTR, IS6110 and microsatellite markers, and is subdivided into six to seven main near-clades with specific ethnic and geographical distributions, with lineage 4 branching into ten sublineages, some globally spread. Neisseria meningitidis, classically considered highly recombining, nonetheless shows persistent clonal complexes and multilocus bifurcating trees; the hyperinvasive NZMenB strain, first identified in the Netherlands in 1980 and now worldwide, divides into three clades with further subdivisions. Staphylococcus aureus displays widespread genotypes including MRSA USA300 and ST239, whose internal Russian-doll structure cannot be attributed to spatial sampling alone.

The practical conclusion is that multilocus genotypes—”strains,” or sequence types under MLST—and lesser near-clades are stable in space and time down to a scale of a few months or a few kilometers, fully enabling epidemiological tracing, with resolution power considerably enhanced by whole-genome sequencing and multi-SNP typing. But the author warns that taking clonality for granted could lead to major errors in other bacterial species that may be highly recombining, such as Helicobacter pylori; verifying the stability of the relevant units of analysis requires the population genetics approach outlined here. The same PCE pattern has been observed, with remarkable similarities, in parasitic protozoa, pathogenic yeasts and fungi, and viruses—so much so that the population structures of Trypanosoma cruzi, the yeast Cryptococcus neoformans and E. coli can be called “genetic twins.” Looking ahead, the author anticipates that routine whole-genome sequencing coupled with artificial intelligence will offer an infinitely more refined picture of bacterial population structures and virtually unlimited resolution for the genetic tracing of epidemics.

Subject of Research: The predominant clonal evolution model and its role in the molecular epidemiological tracing of pathogenic bacteria

Article Title: The contribution of population genetics to the epidemiological tracing of pathogenic bacteria: the predominant clonal evolution model

Article References: Tibayrenc, M. (2026). The contribution of population genetics to the epidemiological tracing of pathogenic bacteria: the predominant clonal evolution model. Frontiers in Bacteriology, 5, Article 1947610. https://doi.org/10.3389/fbrio.2026.1947610

Image Credits: AI Generated

DOI: 10.3389/fbrio.2026.1947610

Keywords: molecular epidemiology, predominant clonal evolution, population genetics, bacteria, linkage disequilibrium, multilocus sequence typing, whole genome sequencing, near-clades, clonality, genetic recombination, Escherichia coli, Mycobacterium tuberculosis