Silica-dried tissue samples of Memecylon leaves were collected in the field, mainly from Sri Lanka, India, Singapore, Thailand, Philippines, Indonesia, Seychelles, Madagascar, and South Africa. Additional samples of Memecylon and an outgroup, Mouriri, were taken from herbarium specimens. Total genomic DNA was extracted following a modified CTAB extraction protocol (Jantzen et al., 2020). Target-capture was employed to enrich genomic regions of interest for the rest of the samples. All accessions were processed according to a workflow described in detail in Jantzen et al. (2020). Cleaned target-capture reads were assembled with the Burrows-Wheeler alignment version of HybPiper v.1.3.1 using the template sequences used for probe design as references (Johnson et al., 2016). Genome-skims were assembled using a modified HybPiper script. The postprocessing scripts in the HybPiper were used to retrieve intron regions flanking targeted exons and the supercontigs (Johnson et al., 2016). The supercontigs were individually aligned using MAFFT v7.215 (Katoh and Standley, 2013). The alignments were trimmed using trimAl v1.2 (Capella-Gutiérrez et al., 2009). Gene trees were generated for each of the individual gene alignments using maximum-likelihood (ML) analysis performed with RAxML v8.2.12 (Stamatakis, 2014), assuming a GTRGAMMA model. Genes (i.e. 100 supercontigs) were concatenated, and a partition scheme was provided using Phyx v8.2.0 (Brown et al., 2017). ML analysis was performed following the above RAxML parameters for the concatenated dataset. To infer a coalescent species tree, the 100 optimal gene trees generated from RAxML were input to ASTRAL-III v5.0.3 (Zhang et al., 2018). As informative branch lengths are required for downstream analyses, the resulting ASTRAL-III topology was constrained on the ML analysis. A time-calibrated phylogeny was produced using two fossils and two secondary calibrations from Berger et al. (2016). The calibrated Memecylon phylogeny was used to elucidate its biogeography and ancestral character-states.

In the alignment of low copy nuclear loci, the sequences of M. fruticosum M. elaeagni, M. lateriflorum, M. lanceolatum, M. magnifoliatum, and M. symplociforme contained a significant amount of missing data due to low success in target capture for these samples. However, these samples were retained in the analysis, because some of them showed important relationships in the resulting phylogeny. In the morphological data matrix, some data are missing because these specimens were sterile.