Phylogenetic analysis
Total genomic DNA was obtained from the fresh leaves of the C. pingtaoi and R. brevipedunculatum Y. Wan with the plant genomic DNA kit (DP305, Tiangen, Beijing, China), and then sent to Novogene (Tianjin, China) for library (350 bp) preparation for genome skimming sequencing. Paired-end (150 bp) sequencing was conducted on an Illumina HiSeq 2000, generating 10 Gb raw data of this sample. After quality controlling of the raw data by fastp v.0.23 (Chen et al. 2018), 3 Gb paired reads were extracted for plastid and nuclear ribosome assembly with GetOrganelle v.1.7 (Jin et al., 2020). The plastid genome ofApocynum venetum L. (MT313688) and a continuous sequence (18S-ITS1-5.8S-ITS2-26S) of the ribosome genome of Asclepias albicans S.Watson (JN665082), and the nrETS of Calciphila galgalensis (LN896997) and Cynanchum adalinae K.Schum. subsp.adalinae (LN897003) were used as the reference respectively. Geneious Prime 2019 (https://www.geneious.com) was employed to annotate and extracte three plastid DNA markers (one spacers of trnL-F , and two introns of rps16 and trnL ) and two nuclear DNA regions (nrITS: internal transcribed spacer and nrETS: external transcribed spacer), and these data were upload to GenBank (https://www.ncbi.nlm.nih.gov/).
The samples of 58 species were used for molecular analysis, including 51 species of Cynanchum , and seven samples were taken as the out groups, including Pentatropis madagascariensis Decne., P. nivalis (J.F.Gmel.) D.V.Field & J.R.I.Wood and Pentatropis sp. which belonged to Tylophorinae and Calciphila galgalensis Liede & Meve (including two samples), Calciphila gillettii Liede & Meve, and Calotropis procera (Aiton) W.T.Aiton which belonged to Asclepiadinae (Khanum et al. 2016) (Appendix 1). The sequences of three plastid DNA markers and two nuclear DNA regions were aligned separately in MEGA v.7.0.26 using MUSCLE with manual adjustment (Kumar et al., 2016), and then concatenated into a super matrix in Geneious Prime 2019.The incomplete sequences were filled with missing data. The maximum likelihood (ML) was performed under RAxML-HPC2 8.2.12 on XSEDE (Stamatakis, 2014) through the CIPRES portal (Miller et al., 2010) by using GTRCAT model and 1 000 bootstraps, and the trees is viewed in Figtree 1.4.2 (Rambaut, 2012).