Supporting Information
Additional supporting information may be found in the on-line version of this article.
Table S1 Information for all samples including collection location, taxonomic and genotype labels, and data used to infer ploidy
Table S2 . Polymorphic sites in the intron of the chloroplast gene rpl16 . Positions are based on alignment to GenBank accession KF853027 (Scataglini et al. 2014).
Figure S1 Histogram of pairwise genetic distances between samples within (A ) fine-textured P. vaginatum , (B ) Coarse-textured P. vaginatum , and (C )P. distichum . Dotted red line indicates distance threshold for identifying clonal genotypes.
Figure S2 (A ) The basic workflow of triploid_comp.py. (B ) An illustration of how triploid_comp.py counts reads at a locus for a hypothetical triploid genotype. Because GBS reads start at the same location due to the restriction cut site, reads representing each allele at a locus are expected to be identical to each other. Therefore, reads not identical to the most common two read types are considered to be sequencing errors and ignored (i.e. not included in calculating ratios or percentages of reference reads). Due to the ~2:1 ratio of reference reads this hypothetical example could represent a triploid with two haploid copies of the P. vaginatum subgenome and one from the unknown progenitor; however, a single locus could have a skewed ratio due to chance so it is necessary to look at the trend genome wide to make a definitive assessment.
Figure S3 Principal component analysis (PCA) showing the genetic separation of all sampled Paspalum subgroup Disticha accessions into three well defined groups as well as three admixed individuals (two of which are clonal and overlap). The color of each individual indicates its genome-wide heterozygosity rate (reported as the individual inbreeding coefficient F). Darker shades of blue indicate lower heterozygosity.
Figure S4 ADMIXTURE plots showing population assignment of 33 unique diploid P. vaginatum genotypes at K=2 (A ) and K=3 (B ). Putative fine-coarse admixed individuals are indicated with stars. (C ) Cross-validation error for K = 1-7.
Figure S5 Number of SNP markers where heterozygous calls are unique to given group or shared between groups. The relatively small number of loci that are heterozygous in both coarse-textured P. vaginatum and P. distichum but not fine-texture P.vaginatu m (light blue region) indicates that P. distichumand coarse-textured P. vaginatum likely do not share a non-P. vaginatum subgenome.
Figure S6 The raw counts of the two most common reads (where one is identical to the reference and the other is alternate) covering heterozygous SNPs in four coarse-textured P. vaginatum samples representing the most common diploid hybrid genotype (coarse_03, represented by DG076) (A ) and the three triploid genotypes: coarse_01(accession DG117) (B) , coarse_02 (accession DG107) (C ) and coarse_04 (accession DG120) (D) . Red lines indicate expected ratio of reference to alternate reads for diploids at 1:1 and reciprocal triploids at 1:2 and 2:1. A bias toward either 1:2 or 2:1 in the triploids reflects having one or two haploid copies of theP . vaginatum subgenome respectively.