Comparative Chloroplast Genomics and Machine Learning Authentication of Bletilla Species (Orchidaceae)

The genus Bletilla comprises horticulturally and medicinally valuable orchids in East Asia. This study integrated comparative chloroplast genomics, codon usage bias analysis, and machine learning to develop a molecular authentication system for B. striata and its adulterants. Complete chloroplast genomes of six species were assembled and analyzed using IRscope, MISA, CodonW, and phylogenomic tools. Comparative analyses revealed conserved quadripartite structures, A/T-biased SSRs, and elevated non-coding variation, providing genomic resources for population genetics and marker development. All species exhibited a universal A/U-ending codon preference, reflecting conserved evolutionary selection pressures. Three barcoding regions (ITS2, matK , ycf1 ) were evaluated via phylogenetic and machine learning analyses (BLOG, WEKA). All barcodes discriminated authentic B. striata , with BLOG achieving 100% accuracy. This work establishes a multi-omics framework for germplasm authentication and management, contributing to conservation genetics, germplasm development, and breeding of this horticultural orchid. The pipeline provides a transferable model for molecular identification in other non-model horticultural and forestry species.