Estimating ploidy
Since the study was based on herbarium vouchers, chromosome counting or genome size estimates by flow cytometry were not possible. Instead, we bioinformatically estimated the ploidy of 32 specimens for which we obtained an estimated nuclear genome coverage of at least three following mapping. Data for each specimen were merged and used to estimate ploidy in HMMploidy, a program that combines information of sequencing depth and genotype likelihoods to estimate ploidy levels (Soraggi et al., 2021). A multi-sample mpileup file was generated in SAMtools v.1.10 for all genome scaffolds longer than 10kb using only reads with a minimum mapping quality of 30 and only calling sites with a minimum quality of 30, counting anomalous reads, and setting the maximum per-file depth to 50. Genotype likelihoods were then calculated in HMMPloidy using default settings and ploidy levels were inferred in 10kb windows (total of 5,224 windows). The percentage of 10kb windows supporting each ploidy level (1n-4n; no windows supported a ploidy level larger than 4n) were calculated. Specimens for which at least 60% of the windows supported a single ploidy were assigned to one of four categories: 1n=“likely diploid”, 2n=“diploid”, 3n=“likely polyploid”, or 4n=“polyploid”. The estimates for specimens not assigned to any of these categories were considered uncertain.