Global genotype flow in Cercospora beticola populations confirmed through genotyping-by-sequencing

Abstract

Genotyping-by-sequencing (GBS) was conducted on 333 Cercospora isolates collected from Beta vulgaris (sugar beet, table beet and swiss chard) in the USA and Europe. Cercospora beticola was confirmed as the species predominantly isolated from leaves with Cercospora leaf spot (CLS) symptoms. However, C. cf. flagellaris also was detected at a frequency of 3% in two table beet fields in New York. Resolution of the spatial structure and identification of clonal lineages in C. beticola populations using genome-wide single nucleotide polymorphisms (SNPs) obtained from GBS was compared to genotyping using microsatellites. Varying distance thresholds (bitwise distance = 0, 1.854599 × 10-4, and 1.298 × 10-3) were used for delineation of clonal lineages in C. beticola populations. Results supported previous reports of long distance dispersal of C. beticola through genotype flow. The GBS-SNP data set provided higher resolution in discriminating clonal lineages; however, genotype identification was impacted by filtering parameters and the distance threshold at which the multi-locus genotypes (MLGs) were contracted to multi-locus lineages. The type of marker or different filtering strategies did not impact estimates of population differentiation and structure. Results emphasize the importance of robust filtering strategies and designation of distance thresholds for delineating clonal lineages in population genomics analyses that depend on individual assignment and identification of clonal lineages. Detection of recurrent clonal lineages shared between the USA and Europe, even in the relaxed-filtered SNP data set and with a conservative distance threshold for contraction of MLGs, provided strong evidence for global genotype flow in C. beticola populations. The implications of intercontinental migration in C. beticola populations for CLS management are discussed.

Fig 1. Principal component analysis of 333 Cercospora spp. isolates collected from Beta vulgaris genotyped through genotyping-by-sequencing (GBS).

SNPs (n = 7,431) obtained through GBS detected two distinct clusters later identified as C. cf. flagellaris (triangles) and C. beticola (circles) using multi-locus sequence typing.

Fig 2. Recurrent multi-locus lineages (MLLs) shared among Cercospora beticola populations.

Circles represent MLLs shared among Hawaii (HI), Michigan (MI), New York (NY), and Europe (EUR), with circle sizes proportional to MLL frequencies. The vertical axes show the MLLs detected in the microsatellite (A) and single nucleotide polymorphism (SNP) data sets generated through genotyping-by-sequencing (B [strictly filtered], C [relaxed-filtered data set 1], and D [relaxed-filtered data set 2]). MLLs detected using microsatellites are indicated in bold and italic font. When the same MLL was detected in a SNP data set, the original MLL number was replaced with the microsatellite MLL number to allow comparisons between markers. SNP MLLs that included some, but not all, of the individuals in a microsatellite MLLs are indicated with an asterisk.

Fig 3

Relationships between the (A) Number of multi-locus lineages (MLLs); (B) Clonal fraction; and (C) Simpson’s complement index of genotypic diversity for Cercospora beticola populations estimated using 12 microsatellites (SSR) and single nucleotide polymorphisms (SNPs) generated using genotyping-by-sequencing. Values were estimated using the strictly filtered SNP data set (filled triangles), relaxed-filtered SNP data set 1 (open circles) and relaxed-filtered SNP data set 2 (open squares).

Fig 4. Relationships between indices of population differentiation.

(A) Jost’s D, (B) Pairwise Nei’s G_ST, and (C) Pairwise F_ST between Cercospora beticola populations estimated using 12 microsatellites (SSR) and single nucleotide polymorphisms (SNPs) identified using genotyping-by-sequencing. Values were estimated using the strict SNP data set (filled triangles) and relaxed-filtered SNP data set 1 (open circles). Values estimated using the relaxed-filtered SNP data set 2 were almost identical to those obtained from data set 1.

Fig 5

Discriminant analysis of principal components (DAPC) for Cercospora beticola populations from Hawaii (HI), Michigan (MI), New York (Farms 1 and 2; Fields 3 and 5), North Dakota (ND), and Europe using (A) microsatellite, (B) strictly filtered and (C) relaxed-filtered SNP data sets generated using genotyping-by-sequencing.

Fig 6

Assignment of Cercospora beticola isolates from Hawaii (HI), Michigan (MI), New York (Farms 1 and 2; Fields 3 and 5), North Dakota (ND) and Europe to three clusters detected through Bayesian clustering analysis of (A) microsatellite, and single nucleotide polymorphism (SNP) data sets generated by genotyping-by-sequencing using (B) strict filtering and (C) relaxed filtering. Each bar represents one individual and the bar height indicates estimated membership fraction of each individual in the inferred clusters.