Phylogenetic analyses
In total 55 sequences obtained from 14 taxa were analysed, 41 of them were newly generated sequences provided from eight Tulipa taxa (six species and two subspecies) collected from wild populations in Kosovo and 14 sequences were obtained from GenBank (Table 1). The ITS sequences for Tulipa ulophylla (HF952978), T. tschimganica(HF952976), T. sylvestris subsp. sylvestris (HF952974),T. suaveolens (MK334468), T. julia (HF952964), T. gesneriana (MK335217, MK335224) and the trnL-trnF sequences forT. ulophylla (HF953003), T. tschimganica (HF953001),T. sylvestris subsp. sylvestris (HF952999), T. suaveolens (HF952998), T. julia (HF952989) were obtained from GenBank. The trees were rooted using Lilium martagon (obtained from GenBank: trnL-trnF KF850988 and ITS KX865057) as outgroup.
ITS and trnL-trnF sequence of most of taxa were amplified and then sequenced from three specimens for each species, while the T. kosovarica (locality Goriç) and T. luanica (locality Qafë Prush) were amplified and sequenced successfully from two specimens per species. Because of the failure of the amplification of some specimens (ITS T8 and T10; trnL-trnF T9, T16 and T18), some species were represented by only one or two sequences.
The sequences were aligned using MEGA X software (Kumar et al. 2018). The datasets were loaded into PAUP 4.0 (Swofford, 2002) and then analysed as separate (ITS and trnL-trnF ) and combined (ITS +trnL–trnF ) datasets. For ITS analyses, in total 29 sequences were aligned to determine sequence statistics, 21 of them were newly generated and eight were obtained from GenBank, while fortrnL-trnF statistical analyses included 26 sequences (20 newly generated and six obtained from GenBank) (Table 2). The combined (ITS +trnL–trnF ) dataset included 24 sequences obtained from the taxa which were present in both datasets and were used for compatibility testing (ILD test). The ILD test revealed no incongruence between ITS and trnL-trnF sequences (p = 0.82). Phylogenetic trees were constructed using Neighbor-Joining (NJ), Maximum Parsimony (MP) and Maximum Likelihood (ML) methods, conducted in PAUP 4.0 (Swofford, 2002). In the phylogenetic trees, bootstrap scores <50% were not taken in consideration, scores between 50% and 74% were defined as weak support, scores between 75% and 89% as moderate support and scores >90% BS as strong support.