Image-based cultivar identification remains challenging in perennial medicinal crops because cultivar-specific datasets are often small and morphological differences can be subtle. This study investigated whether a Siamese network-based hybrid learning framework could support cultivar classification and image-derived phenomic similarity analysis in Panax ginseng under low-data conditions. A total of 347 images representing 22 cultivars across four above-ground image acquisition categories were evaluated using stratified five-fold cross-validation. The framework jointly optimized class-weighted cross-entropy and contrastive losses, and five ImageNet-pretrained backbone architectures were assessed across contrastive margins. ConvNeXt-Tiny with a margin of 0.50 achieved the highest five-fold mean performance, with an accuracy of 64.31 ± 7.23% and a macro-F1 score of 61.66 ± 6.63%. UMAP visualization indicated qualitative reorganization of the embedding space after fine-tuning, while Grad-CAM++ localized model responses mainly to plant structures, including leaf, stem, and fruit regions, rather than broad background areas. Hierarchical clustering of cultivar embeddings further suggested structured phenomic relationships, with cosine distance and average linkage yielding a cophenetic correlation of 0.81 ± 0.08 and Kendall’s τ-b of 0.61 ± 0.06. These findings support the potential of hybrid classification and metric learning as a complementary tool for extracting interpretable phenomic representations from limited ginseng image datasets. The framework may provide supporting image-based evidence alongside conventional morphological cultivar assessment. However, the observed cultivar relationships should be considered exploratory and require validation across environments, developmental stages, independent datasets, and genetic information before broader biological interpretation.

Background: Cirsium nipponicum, a pharmaceutically valuable plant from the Asteraceae family, has been utilized for over 2000 years. Unlike other thistles, it is native to East Asia and found exclusively on Ulleung Island on the Korea Peninsula. Despite its significance, the genome information of C. nipponicum has remained unclear. Methods: In this study, we assembled the genome of C. nipponicum using both short reads from Illumina sequencing and long reads from Nanopore sequencing. Results: The assembled genome is 929.4 Mb in size with an N50 length of 0.7 Mb, covering 95.1% of BUSCO core groups listed in edicots_odb10. Repeat sequences accounted for 70.94% of the assembled genome. We curated 31,263 protein-coding genes, of which 28,752 were functionally annotated using public databases. Phylogenetic analysis of 11 plant species using single-copy orthologs revealed that C. nipponicum diverged from Cynara cardunculus approximately 15.9 million years ago. Gene family evolutionary analysis revealed significant expansion and contraction in genes involved in abscisic acid biosynthesis, late endosome to vacuole transport, response to nitrate, and abaxial cell fate specification. Conclusions: This study provides a reference genome of C. nipponicum, enhancing our understanding of its genetic background and facilitating an exploration of genetic resources for beneficial phytochemicals.

Korean ginseng (Panax ginseng C. A. Meyer) is a perennial medicinal crop highly sensitive to waterlogging stress. Although excess soil moisture is known to induce oxygen limitation in the rhizosphere, the resulting metabolic constraints may precede visible shoot symptoms such as wilting, chlorosis, or necrosis. To examine the molecular responses under sustained high soil moisture before overt shoot necrosis, one-year-old ‘Yunpoong’ plants were exposed to ambient waterlogging at 45–55% volumetric soil water content (VSWC), while control plants were maintained at 25% VSWC. Phenotypic observations and hyperspectral imaging were conducted immediately after treatment initiation (week 0) and at weeks 1, 2, and 3, whereas whole-plant samples for RNA-seq were collected at weeks 1–3. Hyperspectral imaging and normalized difference vegetation index (NDVI) analysis indicated limited temporal changes in canopy-level spectral traits during the treatment period. In contrast, transcriptomic analyses revealed time-dependent changes in whole-plant gene expression. A total of 6448 unique differentially expressed genes (DEGs) were identified, of which 4369 were specific to week 3. Gene Ontology (GO) enrichment analysis suggested a stepwise adaptive pattern consisting of early priming, transient stabilization, and long-term remodeling. Hypoxia-, jasmonic acid-, and lignin-related processes were enriched at week 1, water deprivation and protein quality control responses were adjusted at week 2, and glycolysis, phosphate starvation, and repression of growth-related processes became prominent at week 3. Weighted gene co-expression network analysis (WGCNA) and STRING-based protein–protein interaction (PPI) analysis further identified functional modules associated with RNA processing, photosystem regulation, proteostasis, and central carbon, energy, and phosphate metabolism. Integrated transcriptome and network analyses prioritized six candidate genes, GAPC2, MDH, PGI1, PPC1, Lhb1B1, and RD21A, for further functional validation in relation to prolonged ambient waterlogging responses.