"Jumping Jack": Genomic Microsatellites Underscore the Distinctiveness of Closely Related Pseudoperonospora cubensis and Pseudoperonospora humuli and Provide New Insights Into Their Evolutionary Past

Abstract

Downy mildews caused by obligate biotrophic oomycetes result in severe crop losses worldwide. Among these pathogens, Pseudoperonospora cubensis and P. humuli, two closely related oomycetes, adversely affect cucurbits and hop, respectively. Discordant hypotheses concerning their taxonomic relationships have been proposed based on host-pathogen interactions and specificity evidence and gene sequences of a few individuals, but population genetics evidence supporting these scenarios is missing. Furthermore, nuclear and mitochondrial regions of both pathogens have been analyzed using microsatellites and phylogenetically informative molecular markers, but extensive comparative population genetics research has not been done. Here, we genotyped 138 current and historical herbarium specimens of those two taxa using microsatellites (SSRs). Our goals were to assess genetic diversity and spatial distribution, to infer the evolutionary history of P. cubensis and P. humuli, and to visualize genome-scale organizational relationship between both pathogens. High genetic diversity, modest gene flow, and presence of population structure, particularly in P. cubensis, were observed. When tested for cross-amplification, 20 out of 27 P. cubensis-derived gSSRs cross-amplified DNA of P. humuli individuals, but few amplified DNA of downy mildew pathogens from related genera. Collectively, our analyses provided a definite argument for the hypothesis that both pathogens are distinct species, and suggested further speciation in the P. cubensis complex.

Keywords: downy mildew; evolution; genotyping; host specificity; obligate pathogens; oomycete; speciation.

FIGURE 1

Population structure inferred using STRUCTURE. Genotyped dataset was split either by the preattributed species (A), six groups that reflected the sample geographic origin and age [(B) Pseudoperonospora cubensis: historical North American (“PcUSAh”), historical Old World (“PcEh”; Europe and South-East Asia), current North American (“PcUSAc”), current European (“PcEc”); P. humuli: current European (“PhEc”) and historical origin (“PhEh”)], or the affected host plants (C). The results were visualized using the most supported number of inferred genetic clusters (D) as per the Evanno method (Evanno et al., 2005).

FIGURE 2

Discriminant analysis of principal components of the genotyped Pseudoperonospora cubensis and P. humuli separates the taxa and visualizes the substructure within. The pathogen groups are coded the same way as in Figure 1. The analysis was optimized by 1,000 permutations of the dataset across principal components analyzed from 2 to 48 toward the principal components used for the projection (PCA = 20). Two alleles contributing the most and used for the projection are identified along the respective axes, with % of total variance explained also indicated. Insert: pairwise population genetic distances (Prevosti, 1974) are reticulated with distances between each split indicated on the branches, and bootstrap support for the splits > 70% is indicated in bold, based on 1,000 permutations of the dataset.

FIGURE 3

Genetic distances indicate Pseudoperonospora cubensis, but not P. humuli, host specialization. Counts for specimens analyzed from each affected host plant species are indicated next to each species nameplate (n). Bootstrap values >70% are indicated; those were calculated based on 1,000 permutations of the genotyped dataset.

FIGURE 4

DIYABC analyses of the evolutionary relationship between Pseudoperonospora cubensis and P. humuli. Left column: best scenarios for each dataset; right column: respective model checking (PCA of priors, posteriors, and the observed dataset). (A) “Species” dataset with multispecific outgroup; (B) “three-subpopulations” dataset with multispecific outgroup; (C) sequence matrix across six genomic simple sequence repeats. For either analysis, 1,000,000 pseudo-observed datasets (PODs) were generated using ranges of prior indices. The subsequent analyses utilized 1% (n = 10,000) of the PODs closest to the observed dataset as per within- and among-subpopulation indices. The comparative analyses of both scenarios provided the support for each regarded evolutionary scenario (D, direct; L, logistic; 95% CI given). t_n, time to split into the coalescent (generations); numbers below groups: effective population sizes; insert number: mutation rate.

FIGURE 5

Alignment of the sequencing results for genomic simple sequence repeat (gSSR) PC021 [GT]₆ allows clear distinction between Pseudoperonospora cubensis and P. humuli, as visualized in BioEdit (Hall, 1999). Sequence labels include preattributed subpopulation code (see Figure 1), isolate code, and the host plant identifier (Supplementary Table 1). Occasional nucleotide substitutions in the SSR flanking regions were found across all six sequenced gSSRs (Supplementary Table 7).