Origin and diversification of Xanthomonas citri subsp. citri pathotypes revealed by inclusive phylogenomic, dating, and biogeographic analyses

Abstract

Background: Xanthomonas citri subsp. citri pathotypes cause bacterial citrus canker, being responsible for severe agricultural losses worldwide. The A pathotype has a broad host spectrum, while A* and A^w are more restricted both in hosts and in geography. Two previous phylogenomic studies led to contrasting well-supported clades for sequenced genomes of these pathotypes. No extensive biogeographical or divergence dating analytic approaches have been so far applied to available genomes.

Results: Based on a larger sampling of genomes than in previous studies (including six new genomes sequenced by our group, adding to a total of 95 genomes), phylogenomic analyses resulted in different resolutions, though overall indicating that A + A^W is the most likely true clade. Our results suggest the high degree of recombination at some branches and the fast diversification of lineages are probable causes for this phylogenetic blurring effect. One of the genomes analyzed, X. campestris pv. durantae, was shown to be an A* strain; this strain has been reported to infect a plant of the family Verbenaceae, though there are no reports of any X. citri subsp. citri pathotypes infecting any plant outside the Citrus genus. Host reconstruction indicated the pathotype ancestor likely had plant hosts in the family Fabaceae, implying an ancient jump to the current Rutaceae hosts. Extensive dating analyses indicated that the origin of X. citri subsp. citri occurred more recently than the main phylogenetic splits of Citrus plants, suggesting dispersion rather than host-directed vicariance as the main driver of geographic expansion. An analysis of 120 pathogenic-related genes revealed pathotype-associated patterns of presence/absence.

Conclusions: Our results provide novel insights into the evolutionary history of X. citri subsp. citri as well as a sound phylogenetic foundation for future evolutionary and genomic studies of its pathotypes.

Keywords: Biogeography; Divergence dating; Genome evolution; Phylogenomics; Recombination.

Fig. 1

ML tree (model: GTR + I + R2, where “R” means free rate model) based on concatenation of the unicopy data set (1785 genes), with clades zoomed to the right. Ancestral area reconstruction at each node is presented (highly probable ancestral states for nodes discussed in the main text are highlighted, with text color matching state color). “r/m” values correspond to the relative probabilities of a site being altered due to recombination relative to mutation (i.e., r/m = 1.0 means a random site is equaly probable to have suffered recombination or mutation), as obtained in ClonalFrameML

Fig. 2

LCB-based ML trees (branches not proportional to actual lengths), either keeping all sites (left), or after removing LCBs with significant signs of recombination (right). Only ingroup is shown. Arrows point to node support associated with the smallest clade containing both A (including A2) and A^w

Fig. 3

Inference of populations and distances between them according to different analyses. a Centroids of populations according to DAPC analysis; b Consensus network based on splits present in at least 0.05 of the 161 LCB gene trees for the 34-taxa set

Fig. 4

Reconstruction of ancestral hosts at nodes, with pie charts representing the likelihood of inferred states. To the right, best number of inferred populations (k = 3) according to BAPS v6.0, where each genome (individual horizontal bars) has a probability of pertaining to each of the three populations (represented by its proportion of yellow, red, and blue colors)

Fig. 5

Dating analyses summary. Top: the eight tests performed, each changing a parameter. Bottom left: tMRCAs of the root (= start of diversification of the XCP group), with box borders corresponding to 95% HPDs. Bottom right: times of origin and diversification of XCC (95% HPDs)

Fig. 6

Heatmap of presence/absence of the 44 genes with variable pattern of pathogenicity/virulence (among a larger set of 120 genes), across the 95 genomes (ML tree shown to the left)

Fig. 7

Schematic view of the main results regarding evolution of X. citri subsp. citri. The lineage originated ~ 16.0–46.0 thousand years ago (kya), with an associated event of host switch from Fabaceae to Rutaceae, within the Indian Subcontinent. A* and A2 likely share a great portion of ancestral polymorphism, whereas A and A^w had a larger impact from recombination (“Rec”) on their genetic varibility (especially in A, the generalist pathotype) prior to each respective diversification. Colors of the most common haplotypes in each lineage are the same as in previous figures (except for A2, which due to its high genetic similarity to A* according to BAPS v6.0, is also shown in blue). Dotted lines correspond to minor genetic contributions from given haplotypes (as detected in BAPS v6.0) or inferred from the ML-unicopy phylogeny. Fabaceae images obtained and modified from the Encyclopedia of Life database: “Clitoria ternatea” (https://eol.org/pages/47317701; copyright: Vinoth Kumar Rajalingam; license: cc-by-nc-4.0), and “Cajanus cajan (L.) Millsp.” (https://eol.org/pages/643268; copyright: Andres Hernandez S.; license: cc-by-nc-sa)