The speciation and hybridization history of the genus Salmonella

Abstract

Bacteria and archaea make up most of natural diversity, but the mechanisms that underlie the origin and maintenance of prokaryotic species are poorly understood. We investigated the speciation history of the genus Salmonella, an ecologically diverse bacterial lineage, within which S. enterica subsp. enterica is responsible for important human food-borne infections. We performed a survey of diversity across a large reference collection using multilocus sequence typing, followed by genome sequencing of distinct lineages. We identified 11 distinct phylogroups, 3 of which were previously undescribed. Strains assigned to S. enterica subsp. salamae are polyphyletic, with two distinct lineages that we designate Salamae A and B. Strains of the subspecies houtenae are subdivided into two groups, Houtenae A and B, and are both related to Selander's group VII. A phylogroup we designate VIII was previously unknown. A simple binary fission model of speciation cannot explain observed patterns of sequence diversity. In the recent past, there have been large-scale hybridization events involving an unsampled ancestral lineage and three distantly related lineages of the genus that have given rise to Houtenae A, Houtenae B and VII. We found no evidence for ongoing hybridization in the other eight lineages, but detected subtler signals of ancient recombination events. We are unable to fully resolve the speciation history of the genus, which might have involved additional speciation-by-hybridization or multi-way speciation events. Our results imply that traditional models of speciation by binary fission and divergence are not sufficient to account for Salmonella evolution.

Keywords: Salmonella; evolution; genomics; hybridization; speciation; taxonomy.

Fig. 1.

Phylogenetic tree of 73 Salmonella strains based on all shared core genes. The balanced minimum-evolution phylogenetic tree was constructed using FastME (see the Methods section). The 11 phylogroups are indicated above their ancestral branch; Enterica groups A and B are also indicated. Bootstrap-based branch support values are indicated next to the nodes. The scale bar corresponds to 0.01 nucleotide substitutions per character.

Fig. 2.

Coancestry matrix of 73 Salmonella genomes, computed using ChromoPainter. Each cell of the coancestry matrix is coloured according to the corresponding coancestry value (colour scale on the right), i.e. the amount of genetic material copied from a donor genome (columns) to a recipient genome (rows), with dark blue corresponding to 0 % and dark red corresponding to 10 %. The 73 strain names as well as the 12 phylogroups are indicated on the left.

Fig. 3.

Cumulative curves of gene-by-gene distances between selected pairs of genomes. (a) Comparisons with Enterica (group B, serovar Schwartzengrund CVM19633). The arrowhead shows that 20 % (0.20, y-axis) of the genes of an Enterica B strain have less than 1 % (0.01, x-axis) divergence to Houtenae B. (b) Comparisons with Houtenae B (2193/78). The arrowhead shows that 5 % of the VII genome and 6 % of Houtenae A have less than 0.1 % divergence with Houtenae B. (c) Comparisons with Salamae A (1268/72). (d) Comparisons with Arizonae (CDC 129–73).

Fig. 4.

Coancestry matrix between nine unrelated genomes, computed using ChromoPainter. Each cell of the coancestry matrix is coloured according to the amount of genetic material copied from a donor genome (columns) to a recipient genome (rows).

Fig. 5.

Treemix analysis of 12 genomes representative of phylogroup diversity. The black arrowhead indicates the position of the ancestor contributing to extant Houtenae A, Houtenae B and VII lineages. The red arrows indicate gene fluxes inferred by Treemix from one position on the tree to another.