DNA extraction, library construction, and Illumina sequencing
A total of 93 specimens were whole genome sequenced at the Geogenetics Sequencing Core (GLOBE Institute, University of Copenhagen, Denmark). Total genomic DNA was extracted from approximately a 1–3 cm2 piece of dried (herbarium mounted) material from a single leaf. In cases of small leaves, a single whole leaf was used. The leaf material was ground to a powder using TissueLyser II (Qiagen, Hilden, Germany) and DNA was extracted using the DNeasy Plant Mini Kit (Qiagen, Hilden, Germany) following manufacturer’s protocol. The amount of extracted DNA was estimated using the Qubit double-stranded DNA High-Sensitivity Assay kit (ThermoFisher Scientific, USA).
A total of 4.6–110 ng DNA was fragmented using the 96 plate set-up in a Covaris LE220-plus (Covaris, Woburn, USA) to a target insert size of 300 bp and 32 µl of fragmented DNA (4.2-100 ng) was used to build shot-gun sequencing libraries using the BEST protocol (Carøe et al., 2018). The final library concentrations were evaluated with quantitative PCR (Mx3005; Agilent Technologies) and libraries were amplified (25 µl reaction volume) with the desired number of PCR cycles (Table S1) adding indices to the P5 and P7 end of the libraries according to Carøe et al. (2018). A total of 41–47 compatible libraries from the same or different projects were pooled in equimolar concentrations and each pool was paired-end sequenced (100bp or 150 bp; Table S1) on one lane of Illumina NovaSeq6000. For 24 samples additional sequencing was performed (Table S1). Libraries were re-built as above and pooled in equimolar concentrations and paired-end sequenced (150 bp) on one lane of Illumina NovaSeq6000.
For 24 samples (Table S1) single indexed libraries were built using the NEBNext Ultra II FS DNA Library Prep Kit (Illumina) adjusting the fragmentation time to 25 min for a target insert size of 160 bp and PCR amplified with 12 cycles. The resulting libraries were pooled in equimolar concentrations and single-end sequenced (80 bp) on one lane of Illumina HiSeq4000 (Table S1).
Raw reads were quality filtered in AdaptorRemoval v.2.3.1 (Schubert, Lindgreen, & Orlando, 2016) removing adaptor sequences, ambiguous bases (N), and consecutive bases with low quality (Q<20) from both the 5’ and 3’ termini. Trimmed reads shorter than 25 bp and with more than 20 ambiguous bases after the quality filtering were discarded. All other settings were default. Finally overlapping pair-end reads were collapsed and all paired-end reads without a mate were discarded.