1. Introduction

Lycoris, a genus within the Amaryllidaceae family, is known for its beautiful flowers and its bulbs as a medicinal plant (Kurita, 1986; Ji and Meerow, 2000). The genus comprises more than 30 species, predominantly distributed in East Asia, with the largest number of species in China (Hsu et al., 1994; Ji and Meerow, 2000; Zhang et al., 2021b; Li et al., 2022). Their amazing flowers have attracted the attention of botanists and gardeners worldwide, and interspecific hybridization has been one of the most important breeding methods for improving the ornamental quality, including flower color, shape, and extending flowering period (Ma et al., 2000). In recent years, China has intensified efforts to conserve and utilize Lycoris genetic resources and has established two national Lycoris germplasm resource banks. The Institute of Botany, Jiangsu Province and Chinese Academy of Sciences (Nanjing Botanical Garden Mem. Sun Yat-Sen) plays an important role in collecting and preserving numerous Lycoris resources from across the world (Wang et al., 2023). However, the widespread morphological variation and high diversity of Lycoris has created challenges in germplasm sorting, development and utilization.

Lycoris species classification has traditionally relied on morphological traits, such as leaf emergence time, flower color and shape, and the lengths of pistil and stamen (Tae and Ko, 1995; Kurita and Hsu, 1996). However, these traits may exhibit plasticity under different environmental conditions, further complicating species delineation, and leading to inconsistent taxonomic outcomes (Kevin, 2007; Liu et al., 2012), e.g., L. houdyshelii, L. albiflora and recently published species (Quan et al., 2013; Lou et al., 2022; Zhang et al., 2022a). For example, the spring-leaf emergence species previously referred to as L. albiflora has recently been re-named L. × jinzheniae (2n = 19) (Zhang et al., 2022b). In addition, Flora of China distinguishes L. houdyshelii from L. straminea based on flower color (i.e., L. houdyshelii has white flowers vs. L. straminea has yellow flowers), however, some accessions of L. straminea also exhibit the same flower color as L. houdyshelii, which can easily lead to confusion between the two species. Chromosome analysis may provide a solution to this type of taxonomic confusion. For instance, previous studies have revealed that L. houdyshelii has 30 or 31 chromosomes, whereas L. straminea has 16 or 19 (Kurita, 1987a; Liu and Hsu, 1989), indicating that chromosome number can help delimit species when morphology is not definitive.

Cytological characteristics are particularly valuable when morphological traits alone are insufficient for accurate species identification (Stace, 2000; Heslop-Harrison and Schwarzacher, 2011; Jang et al., 2024). For instance, comparing the chromosomal features of different individuals or populations can help determine whether they belong to the same species (Kato et al., 2005). Genome size, typically measured by base pair count or DNA content, can vary considerably among species, even within the same taxonomic group (Bennett and Leitch, 2005; Ng et al., 2016). These cytological features not only provide insights into the evolutionary history of species but also reflect their ecological adaptability (Leong-Škorničková et al., 2007; Sánchez-Jiménez et al., 2012). Several studies have reported the chromosome numbers and karyotypes of Lycoris, revealing significant diversity in chromosome number and ploidy (Kurita, 1988; Liu et al., 2012, 2019a). For example, L. aurea possesses 12, 13, 14, 15, 16 or 18 chromosomes (Kurita, 1986, 1987a; Wang et al., 2022). L. traubii has 12, 13 or 14 chromosomes (Kurita, 1987b). This chromosomal complexity complicates species identification and our understanding of interspecific relationships within Lycoris.

DNA sequence-based phylogenetic analysis has become an essential tool in plant taxonomy, and has increased the accuracy of species classification compared to phenotype-based approaches (Uncu et al., 2015; Qian et al., 2023; Zhao et al., 2024). Previous molecular studies have provided valuable insights into the classification of Lycoris species. For example, phylogenetic analysis based on two plastid markers in 29 Lycoris populations revealed evidence of reticulation within the genus (Shi et al., 2014). More recently, complete chloroplast genome sequences have been used to reconstruct the phylogenetic tree and analyze interspecific relationships, as well as the potential maternal donors involved in hybridization events (Zhang et al., 2020, 2021a). By integrating morphological, karyotypic and DNA sequence-based phylogenetic analyses, Quan et al. (2024) identified the origin of two new Lycoris species and demonstrated the role of natural hybridization in lineage diversification. These work contributed significantly to our understanding of species classification and interspecific evolutionary relationships within the genus. However, due to ongoing hybridization, polyploidy and morphological variation, the cytological diversity and precise phylogenetic relationships within the genus remain unclear. In this study, we conducted a comprehensive analysis of chromosome numbers, genome size, and phylogenetic relationships in 64 Lycoris accessions, representing 20 species, one putative species, two variants, and five natural hybrids. These accessions capture the morphological diversity of Lycoris. The results will help to clarify the effect of hybridization and polyploidy on genome size variation and speciation, and provide evidence for infrageneric taxonomy and species delimitation in Lycoris, further enhancing our understanding of the evolutionary processes within this genus.

2. Materials and methods 2.1. Plant material, DNA extraction, and sequencing

A total of 64 Lycoris accessions (Table 1) were collected, representing the observed morphological diversity (Fig. 1–5, left). Of these, 24 accessions have spring-leaf emergence and 40 accessions have autumn-leaf emergence, encompassing 20 species, one putative species, two variants, and five natural hybrids. We based species identification on morphological characteristics, chromosome number, and genome size. Plants were grown at the National Germplasm Resource Bank of Lycoris at the Institute of Botany, Jiangsu Province and Chinese Academy of Sciences (Nanjing Botanical Garden Mem. Sun Yat-Sen). Fresh leaves of 55 accessions (Table 1, except for five natural hybrids and our four published accessions) were collected for genomic DNA extraction and sequencing. Genomic DNA was isolated using DNA Extraction Kit (Huayueyang Biotechnology, Beijing, China). DNA was fragmented by sonication to 350 bp. DNA fragments were then end-polished, A-tailed, and ligated with the full-length adapter for Illumina sequencing, followed by library quality assessment and quantification. After quality control, different libraries were pooled based on the effective concentration and targeted data amount. The 5′-end of each library was phosphorylated and cyclized. Subsequently, loop amplification was performed to generate DNA nanoballs. These DNA nanoballs were finally loaded into flowcell with DNBSEQ-T7 for sequencing in Novogene Bioinformatics Technology Co., Ltd (Beijing, China).

2.2. Chromosome counting

Chromosomes were counted from Lycoris bulbs planted in pots until new roots emerged to a length of 1–2 cm. Fresh root tips were collected in the morning between 10:00 to 10:30 am, and soaked in 2 mM 8-hydroxyquinoline for 1–2 h at room temperature, followed by an additional 1–2 h at 4 ℃. Root tips were then transferred to a fixative solution containing ethanol/acetic acid = 3:1 (v/v) (prepared fresh) for at least 2 h at room temperature. The fixative solution was changed 2–3 times during this period. The fixed root tips were stored at −20 ℃ for chromosome counting. For slide preparation, 4–6 root tips from each sample were washed three times for 10 min in water in a small Petri dish to remove soil remains and the fixative. They were then washed in 2–5 mL 1x enzyme buffer for 10 min. After that, the root tips were transferred into 1–2 mL enzyme solution and digested at 37 ℃ for 50–55 min until the material became soft. The soft root tips were then transferred into the enzyme buffer for further use. The preparation of the root tips, enzyme buffer, and enzyme solution followed Trude's method (Schwarzacher, 2016). A single root tip was transferred to a slide that contained one drop of 45% acetic acid. Under stereo microscope, the root cap was removed and the remaining cells were dispersed using a fine dissecting needle. Slides were cover-slipped and cells were gently squashed (Schwarzacher and Heslop-Harrison, 2000). For each accession, at least three root tips were collected, and three optimal metaphases of each tip were observed. The prepared slide was checked under a phase contrast microscope (Olympus BX43, Tokyo, Japan) and captured using a digital camera (Mshot MSX2, Mingmei photoelectric, Guangzhou, China). The images were uniformly processed for contrast and brightness using Adobe Photoshop CS6 (Adobe Systems, Inc.).

2.3. Genome size estimation by flow cytometry

Fresh leaves were submerged in 0.8 mL of pre-cooled dissociation solution containing 45 mM MgCl₂·6H₂O, 20 mM MOPS, 30 mM sodium citrate, 1% (W/V) polyvinylpyrrolidone PVP 40, 0.2% (v/v) Tritonx-100, 10 mM Na₂EDTA, 20 μL/mL β-mercaptoethanol, and adjusted to pH 7.5. The leaves were swiftly and vertically sectioned using a sharp blade and then kept on ice for 10 min. Subsequently, the suspension was filtered through a 40 μm mesh, resulting in a cell nucleus suspension. Propidium iodide (PI) and RNAase solutions, both with a stock concentration of 1 mg/mL, were added to the appropriate volume of the cell nucleus suspension. The mixture was then stained in the dark for 0.5–1 h, using a working concentration of 50 μg/mL for both the PI staining solution and the RNAase solution (Doležel and Bartos, 2005; Doležel et al., 2007).

To estimate genome size, the DNA C-value was determined using a flow cytometer and tender, fresh leaves of plants. Cycas panzhihuaensis was utilized as the internal standard of genome size calculation (Liu et al., 2022). The suspension of C. panzhihuaensis and each sample were thoroughly mixed to achieve equivalent concentrations. The stained cell nucleus suspension was analyzed using a BD FACScalibur flow cytometer with an excitation set at 488 nm to detect the fluorescence intensity of propidium iodide emission. In each detection, a total of 10, 000 particles were collected, and the coefficient of variation (CV) was maintained within 5%. Fluorescence intensity represents the relative DNA content of the genome, as the amount of propidium iodide intercalated during the staining process is proportional to the DNA amount. By comparing the fluorescence peaks of the sample being tested and the control plant, the ratio of DNA content between the two plants can be determined. Multiplying this ratio by the C-value of the reference plant yields the C-value of the sample being tested. The calculation formula is as follows: DNA content of the sample being tested = DNA content of the reference plant × fluorescence intensity of the sample being tested/fluorescence intensity of the reference sample. Analysis and plotting were performed using Modifit v.3.0.

2.4. Phylogenetic analyses based on chloroplast genomes and nrDNA ITS sequences

Raw DNA sequences were quality-filtered using Trimmomatic v.0.36 (Bolger et al., 2014) to remove adapters, contaminants, and low-quality reads. Chloroplast genomes were assembled using GetOrganelle v.1.7.6 (Jin et al., 2020) with parameters: –R 15 -k 21, 45, 65, 85, 105 –F embplant_pt. The assembled sequences were manually checked in Geneious Prime® 2022.0.1 (https://www.geneious.com/). Nuclear ribosomal DNA internal transcribed spacer (nrDNA ITS) sequences were also extracted using GetOrganelle v.1.7.6 and verified through BLAST against reference databases. A total of 59 accessions were used for phylogenetic analyses of both chloroplast genomes and nrDNA ITS sequences, including 55 newly sequenced accessions and four published sequences. Furthermore, 101 accessions, including 59 accessions mentioned above and 42 published sequences, were used for phylogenetic analysis based on chloroplast genomes. Multiple sequence alignments of complete chloroplast genome sequences and nrDNA ITS sequences were performed using the MAFFT v.7 plugin (Katoh and Standley, 2013) within Geneious Prime. Maximum likelihood (ML) phylogenetic trees were constructed for both datasets using RAxML v.8.2.12 (Stamatakis, 2014) with 1000 bootstrap replications. Narcissus poeticus was used as the outgroup. FigTree v.1.4.3 (https://github.com/rambaut/figtree/releases) was used to visualize the resulting phylogenetic trees.

3. Results 3.1. Chromosome number statistics

We observed chromosome numbers of 64 Lycoris accessions. Each accession is accompanied by a color image (Fig. 1–5, left) to facilitate morphological and cytological identification of the species. The chromosome numbers of 64 accessions were 12, 14, 16, 18, 19, 22, 27, 29, 30, and 33 (Fig. 1–5). A total of 16 chromosomes were identified in 10 accessions, including L. longituba (two accessions), L. longituba var. flava (two accessions), L. chinensis, L. anhuiensis (two accessions) and three putative natural hybrids of L. chinensis (Fig. 1). For L. sprengeri, 22 chromosomes were identified (Fig. 2a and b), whereas 19 chromosomes were observed in four L. × jinzheniae accessions (Fig. 2e–h). L. guangxiensis resembles L. caldwellii but is smaller in size, with yellow petals marked by red bands. We included two putative L. guangxiensis accessions, which were morphologically similar; these putative accessions had 19 chromosomes each (Fig. 2c and d). In contrast, in a similar species, L. caldwellii, 27 chromosomes were identified (Fig. 2i and j), consistent with that of L. squamigera (Fig. 2k). Additionally, L. shaanxiensis (Fig. 2l) and L. incarnata (Fig. 2m and n) exhibited similar morphological characteristics and both possessed 30 chromosomes, representing the species with the highest chromosome count among spring-leaf emergence accessions.

We also analyzed 40 autumn-leaf emergence accessions, encompassing 11 species, one variant, and two natural hybrids (Fig. 3–5). The observed chromosome numbers were 12, 14, 16, 18, 19, 22, 27, 29, 30, and 33, with several species displaying varying chromosome counts. For example, L. aurea had 14 and 16 chromosomes (Fig. 3c–g), L. radiata showed 22 and 33 chromosome numbers (Fig. 5h–p), and L. haywardii displayed 18 and 22 chromosome numbers (Fig. 4b and c). Four L. straminea accessions had 19 chromosomes (Fig. 3h–k), whereas three L. albiflora accessions (Fig. 3l–n) with similar morphological characteristics had 18 chromosomes. Both L. radiata var. pumila (Fig. 5a–g) and L. wulingensis (Fig. 4a) had the same chromosome numbers (2n = 22) (Fig. 4a). L. rosea and L. chunxiaoensis had 33 chromosomes (Fig. 4d–f). The chromosome numbers of L. hubeiensis (2n = 29) and L. houdyshelii (2n = 30) were consistent with previous reports (Kurita, 1987a; Meng et al., 2018b) (Fig. 4g and h). Furthermore, two putative natural hybrids (accessions 4 and 5) had 29 chromosomes (Fig. 4i and j), which was previously unreported. One of these putative hybrids has white petals similar to L. houdyshelii, but features numerous red stripes on the petals (Fig. 4i). The other putative hybrid has big, red flowers (Fig. 4j) and is similar to L. chunxiaoensis and L. radiata.

3.2. Genome size variation

The 64 Lycoris accessions exhibited genome sizes ranging from 18.03 Gb (L. wulingensis) to 32.62 Gb (L. caldwellii) (Table 1). In the autumn-leaf emergence group, L. radiata var. pumila (2n = 22) had small genomes (18.11–18.81 Gb), while L. radiata and L. haywardii (2n = 22) averaged ~18.9 Gb. The genome sizes of five L. aurea (2n = 14 and 16) and two L. traubii (2n = 12) accessions were similar (24.88–25.2 Gb), despite chromosome number differences. The genome size of L. straminea (21.3–21.92 Gb) was close to that of the spring-leaf emergence accession L. chinensis. Polyploids such as L. hubeiensis (2n = 29; 29.7 Gb) and the two natural hybrids (29.82 and 30.76 Gb) had larger genomes, but triploids (2n = 33), such as L. radiata (26.14–28.05 Gb), L. chunxiaoensis (27.09 Gb) and L. rosea (27.92 Gb), showed reduced sizes relative to chromosome count. In addition, the genome size of L. houdyshelii (2n = 30) was 27.50 Gb (Table 1). In the spring-leaf emergence group, L. sprengeri had the smallest genome (19.4 Gb), while L. chinensis (21.19 Gb) was smaller than three related hybrids (accessions 1–3; 22.69–22.88 Gb). The genome sizes of L. longituba, L. longituba var. flava and L. anhuiensis ranged from 23.01 to 23.43 Gb, larger than that of L. chinensis, although they all have the same chromosome numbers. The genome sizes of four L. × jinzheniae (2n = 19) accessions were approximately 20.5 Gb, smaller than those of two putative L. guangxiensis (2n = 19; 21.8 and 21.81 Gb) accessions. Polyploids such as L. shaanxiensis (2n = 30; 29.08 Gb), L. incarnata (2n = 30; 29.59 and 29.67 Gb), L. squamigera (2n = 27; 31.82 Gb), and L. caldwellii (2n = 27; 31.73 and 32.63 Gb) had the largest genomes. Coefficients of variation for both the internal standard and sample peaks were less than 5 % for all tested samples. Histograms of fluorescence intensities are provided in Supplemental File 1.

3.3. Correlation between chromosome number and genome size

Fig. 6 provides a clearer and more intuitive understanding of the correlation between chromosome number and genome size. Leaf emergence is one of the important characteristics of genus classification. In this study, accessions with chromosome numbers 12, 14, 18, 29 and 33 belong to the autumn-leaf emergence type (Fig. 6a, blue dots). In contrast, L. squamigera and L. caldwellii, which have 27 chromosomes, belong to the spring-leaf emergence type. Interestingly, accessions with chromosomes 16, 19, 22, and 30 include both spring- and autumn-leaf emergence accessions (Fig. 6a). For L. aurea and the putative L. traubii, genome sizes did not change significantly with their chromosome number. Among the species with 16 chromosomes, L. aurea had a larger genome size than the spring-leaf emergence species such as L. chinensis, L. longituba, L. longituba var. flava, and L. anhuiensis. The accessions with 19 chromosomes also contain spring- and autumn-leaf emergence types, with genome sizes concentrated around 21 Gb (Fig. 6a). Linear correlation of 64 accessions revealed a weak positive correlation (R² = 0.3026) between genome size and chromosome number (Fig. 6a). Considering the complex hybridization events within the genus, it is more meaningful to classify accessions into diploids and hybrid aneuploids for correlation analysis. Among the 35 diploid accessions (2n = 12, 14, 16, 18, and 22), genome size was negatively correlated with chromosome number (Fig. 6d, R² = 0.8575), likely due to differences in chromosome morphology. Species with 2n = 12, 14, and 16 generally possessed relatively long chromosomes, in contrast to the shorter rod-shaped chromosomes observed in L. radiata and L. sprengeri (2n = 22) (Fig. 1–5). Conversely, in 20 aneuploid accessions derived from natural hybridization, genome size was positively correlated with chromosome number (Fig. 6b, R² = 0.8484). Notably, the strongest correlation was found in 23 accessions with chromosome base number x = 11 (2n = 22 and 33) (Fig. 6c, R² = 0.9841).

3.4. Inter- and intraspecific phylogenetic relationships of Lycoris

Phylogenetic analyses of chloroplast genomes and nrDNA ITS sequences from 59 Lycoris accessions revealed significant nuclear-plastid incongruence, while sharing phylogenetic signals (Fig. 7). Both plastid and nuclear phylogenies clustered L. chinensis, L. longituba, L. longituba var. flava and L. anhuiensis into one clade. The nuclear phylogeny split the nine diploid/triploid L. radiata and seven L. radiata var. pumila into two clades, however, the plastid phylogeny clustered these into a polyphyletic clade (red font in Fig. 7). Nuclear data supported a monophyletic grouping for the five chromosomes divergent L. aurea accessions, which were closely related to L. traubii and L. albiflora. However, in plastid phylogeny, L. aurea accessions were distributed across three distinct clades, displaying a significant nuclear-plastid incongruence (green font in Fig. 7). In addition, the plastid phylogeny clustered the two L. haywardii accessions together, showing a close relationship with the triploid species L. rosea and L. chunxiaoensis; however, the nuclear phylogeny separated these accessions, with one accession (2n = 18) closely related to L. sprengeri, while the other accession (2n = 22) clustered within a clade containing L. radiata var. pumila, L. radiata (2n = 22 and 33), L. rosea and L. chunxiaoensis.

Several hybrid species demonstrated complex phylogenetic patterns. For example, the four L. straminea accessions were consistently distributed across distinct clades in both trees (light blue font in Fig. 7). A similar pattern was observed in the two L. caldwellii accessions. The four L. × jinzheniae accessions showed closer relationships in the plastid phylogeny than in the nuclear tree. L. incarnata formed an independent plastid clade distant from L. shaanxiensis, L. squamigera, and L. caldwellii, but showed nuclear affinity with the latter two. L. houdyshelii exhibited associations with L. straminea (plastid) and L. radiata var. pumila. In addition, the expanded plastid phylogeny of 101 accessions (Supplementary File 2) maintained consistent relationships for the 59 studied accessions but also revealed some complexities associated with recently published species. For example, L. longifolia (published in 2022) formed a monophyletic group with L. aurea accession 2 (2n = 16) from this study, showing close affinity to published L. aurea sequences (MN831471) and accession 5 (2n = 14) (green font in Supplementary File 2). Three L. insularis (published in 2022) accessions failed to form a monophyletic group in the plastid phylogeny, with one accession (OP034614) constituting a monophyletic clade with L. sprengeri (OP034620), showing the complex phylogenetic relationships between these two species.

4. Discussion 4.1. Inter- and intraspecific variation in chromosome number and genome size

Flow cytometry analyses revealed exceptionally large genomes across Lycoris, ranging from 18.03 to 32.62 Gb in 64 Lycoris accessions. Large genome sizes and chromosome variation have also been found in other genera of the Amaryllidaceae family, such as Nerine (Zonneveld and Duncan, 2006), Hippeastrum (Poggio et al., 2007), Narcissus (Marques et al., 2012). Our Lycoris dataset represents the most comprehensive karyological resource for the genus, confirming previously reported base numbers of x = 6, 7, 8, and 11 (Kurita, 1986; Huang et al., 2011; Meng et al., 2018a). Our finding that one L. haywardii accession has 2n = 18 chromosomes (Fig. 4c) contradicts research that previously reported this species as 2n = 22 (Hsu et al., 1984). We report that L. rosea (2n = 3x = 33) is triploid, contrasting with previous diploid records (Liu and Hsu, 1989). Additionally, we provide the first karyotype for L. chunxiaoensis (2n = 33) and have identified five putative hybrids (2n = 16, 29), three of which (2n = 16; Fig. 1h–j) likely derive from L. chinensis crosses.

Chromosomal diversity in Lycoris has been attributed to hybridization and polyploidization (Nishikawa et al., 1979; Liu and Hsu, 1989). While ancient whole-genome duplication (WGD) drives karyotypic diversification in some angiosperms (Soltis et al., 2015; Van de Peer et al., 2017), no conclusive evidence of recent lineage-specific WGD events have been reported in Lycoris. In Amaryllidaceae, Leucocoryne exhibits WGD-driven genome expansions, whereas polyploid Nothoscordum shows monoploid reductions (Pellicer et al., 2017), highlighting the heterogeneous genomic trajectories within the family. This suggests that the unusually large nuclear genomes of Amaryllidaceae (including Lycoris) may be the result of alternative mechanisms, such as the fusion of gametes with different ploidy levels, gains or losses of single chromosomes, chromosome fission or fusion (dysploidy) or post-polyploid diploidization (PDD) (Heslop-Harrison and Schwarzacher, 2011; Mayrose and Lysak, 2021). Of these, PDD through descending dysploidy is an important process that may generate multiple base chromosome numbers via structural rearrangements, thereby contributing to genetic and taxonomic diversity (Mandáková and Lysak, 2018; Plačková et al., 2024). Similar mechanisms have been observed in other Amaryllidaceae species, for example, chromosome changes in Allium have been shown to occur through demi-polyploidization and aneuploidy (Babin and Bell, 2022), and two major nuclear lineages in Hippeastrinae have been distinguished by different chromosome numbers (García et al., 2017). In Lycoris, the week proportional relationship between genome size and ploidy level suggests that some variation in chromosome numbers may arise from post-polyploid diploidization processes or loss of repeat sequences, rather than large-scale genome expansion. Notably, although the loss of chromosomes represents a substantial karyotypic alteration, the accompanying genome size variation was smaller than expected. For instance, the rare intraspecific variation in genome size observed in L. aurea accessions (despite differing chromosome counts) implies limited post-polyploid genome restructuring and structural evolution. Therefore, we propose that chromosomal evolution of Lycoris may have been driven by hybridization-mediated aneuploidy, dysploidy or PDD.

4.2. The impact of genome size and chromosome variation on adaptive evolution

As one of the inherent properties in plants, genome size carries a distinct evolutionary history and has considerable ecological significance, influencing both distribution and growth traits (Leitch and Leitch, 2013). Environmental pressures have been shown to drive adaptive evolution through genome size modulation (Petrov, 2001; Nevo, 2001), as exemplified by adaptive genome size variations in Lilium species (Du et al., 2017). As a plant endemic to East Asia, Lycoris exhibits a narrow and species-specific distribution. For example, the triploid L. radiata is distributed throughout much of the species' range in China and has a broader distribution compared to diploid populations, which are primarily restricted to the central and eastern regions of the country (Liu et al., 2019a). In addition, phenological differentiation in leaf emergence patterns correlates with spatial distribution: spring-leafing species predominate in southeastern China, whereas autumn-leafing variants are less frequent in southwestern areas (Ji and Meerow, 2000).

The traits of genome size, ploidy and chromosome number could be unifying characters explaining plant population structure and distribution, but mechanisms influencing genome size have yet to be addressed except for polyploidy (Pandit et al., 2014). In Lycoris, no strong significant correlation was observed between genome size and chromosomes across all accessions, but when they were grouped by ploidy, a high correlation was observed in the diploid and aneuploid groups, especially in accessions with a base number of 11 chromosomes (2n = 22 and 33). Because the base number of chromosomes in diploids varies, we speculate that different correlations may reflect different evolutionary mechanisms, such as autopolyploidization through somatic genome duplication (linear genome size increase in accessions of 2n = 22 and 33; Fig. 6c) and allopolyploidization via interspecific hybridization (aneuploid; Fig. 6b). Although polyploidy is recognized as a key driver of genome size evolution, its long-term adaptive consequences remain ambiguous. Studies on polyploids have shown that rapid genome changes can occur after hybridization due to chromosome rearrangement, the accumulation of repetitive DNA sequences or low-copy genes (Ozkan et al., 2001; Biscotti et al., 2015; Liu et al., 2019b; Zhang et al., 2023; Heslop-Harrison et al., 2023). Similar patterns of genome dynamism driving species diversification occur in ferns and lycophytes (Fujiwara et al., 2023), suggesting a conserved evolutionary mechanism across plant lineages. Therefore, considering the distribution and leaf-emergence characteristics of Lycoris, we hypothesize that the diverse morphology and cytology was produced by hybridization of diploid ancestral populations. During the hybridization process, factors such as parental combination, homologous and heterologous chromosome recombination, or accumulation of repeated DNA sequences, together likely led to the huge genome size of Lycoris species and complex chromosome numbers. These factors also affect the species diversity of Lycoris and led to different species distribution differences and environmental adaptability.

4.3. Phylogenetic relationships and taxonomic controversies in Lycoris

The phylogenetic relationships within the genus Lycoris have long been a subject of debate. Although numerous studies have utilized various nuclear and plastid loci to explore these relationships (Shi et al., 2006, 2014; Zhang et al., 2021a), due to the limited number of sampled species, many phylogenetic relationships constructed by single or a few genes do not fully reflect the true relationships within the genus. Plastid genomes have been widely utilized in phylogenetic reconstruction (Zhou et al., 2024), but few studies have integrated plastid and nuclear datasets to analyze Lycoris phylogeny across a broad range of taxa. In this study, we employed both plastid and nuclear sequences for phylogenetic reconstruction and uncovered significant incongruence between the two datasets. The nuclear phylogeny supported the monophyly of most naturally occurring diploid species, including L. longituba, L. aurea, L. sprengeri, L. radiata, and L. radiata var. pumila, whereas the plastid phylogeny failed to resolve these species as a single clade, particularly for L. aurea and L. radiata var. pumila (Fig. 7). It is noteworthy that the plastid genome is haploid, uniparentally inherited, and non-recombining, whereas the nuclear ITS region is biparentally inherited and subject to concerted evolution (Schulte et al., 2009; Zhang and Ma, 2024). As a result, discrepancies between these markers are expected, similar patterns of nuclear-plastid discordance have been observed in other plant groups (Pessoa et al., 2022). This incongruence likely suggests evolutionary processes such as gene flow, chloroplast capture, or incomplete lineage sorting (ILS) in the genus, which are particularly common in hybrid polyploid species (e.g., L. incarnata, L. squamigera, and L. shaanxiensis) showing strong phylogenetic discordance. The relatively low resolution of the ITS tree likely reflects limited sequence variation across the sampled taxa, though this alone may not fully explain the observed discordance. Incorporating additional nuclear loci in future studies would likely enhance phylogenetic resolution and offer deeper insights into species relationships and evolutionary processes such as hybridization or incomplete lineage sorting, which may be obscured when relying solely on cpDNA or single-locus nuclear data. Nonetheless, close relationships among certain taxa, such as L. chinensis, L. longituba, and L. anhuiensis were consistently well-supported, with evidence suggesting sister–group relationships. Previous studies have been controversial about whether L. haywardii is a hybrid. Kurita (1986) suggested it was a hybrid between L. sprengeri and L. radiata var. pumila, whereas Shi et al. (2014) considered it a diploid species. This study identified an L. haywardii accession with 18 chromosomes, which showed a close relationship with L. sprengeri, but it was an independent branch, indicating that it was likely a diploid species involved in speciation.

In addition, our phylogenetic results challenged the distinctiveness of two newly published species. Plant speciation generally proceeds through allopatric or sympatric processes (Butlin et al., 2008). Lycoris insularis, proposed as an allopatric species diverging from L. sprengeri based on geographic distribution and differences in perianth tube morphology, was supported as monophyletic in previous plastid phylogenies (Zhang et al., 2022a). However, our study revealed that L. insularis (OP034614) and L. sprengeri (OP034620) clustered within a single clade (pink font in Supplementary File 2), with minimal chloroplast divergence. This result calls into question their morphological separation and taxonomic independence. Similarly, L. longifolia, a recently described species endemic to Sichuan Province, was clustered as a monophyletic group together with L. aurea 2 (2n = 16) and was sister to L. aurea (MN831471) and L. aurea 5 (2n = 14) (Fig. 7a). This close genetic affinity suggests incomplete plastid divergence and that L. longifolia should be treated as an uncertain species, possibly representing a geographic subspecies of L. aurea, rather than a distinct species. L. aurea, which exhibits extensive morphological and karyotypic variation (2n = 12, 14, 15 and 16), has been previously proposed as a species complex (Wang et al., 2022). Here, we further supported it as a species complex. In addition, nuclear-plastid discordance suggests its potential role as a hybrid ancestor of closely related species (e.g., L. hubeiensis and L. albiflora) in different branches of the plastid phylogenetic tree. L. traubii, morphologically similar to L. aurea (Kurita, 1987b), formed a well-supported, monophyletic lineage in our analysis, with accessions representing the 2n = 12 cytotype. This supports its recognition as an independent species. Notably, both L. traubii accessions analyzed in this study were collected from mainland China, expanding its known range beyond previous records limited to Japan and Taiwan Province of China (Kurita, 1987a). These findings underscore the need for updated field surveys and broader genomic sampling to refine the taxonomy and biogeography of the genus.

4.4. Putative pathway of speciation in Lycoris

This study accurately identified Lycoris germplasms through chromosome counting. It represents the most comprehensive phylogenetic analysis of Lycoris to date and provides crucial insights into the mechanisms that have driven Lycoris speciation. Based on the chromosome number base (x = 6, 7, 8 and 11) of naturally distributed diploid species, we inferred the chromosome composition of hybrid species (Fig. 8a). The putative pathway of speciation suggests that species with 16, 18 and 19 chromosomes may have originated from hybridization events between ancestral species with chromosome constitutions of 8 + 8, 11 + 7, and 11 + 8, respectively. Conversely, some polyploid species likely arose through chromosome duplication events, as seen in species 2n = 27 = 11 + 8*2, 2n = 29 = 7 + 11*2, 2n = 30 = 8 + 11*2, 2n = 33 = 11 + 11*2 (Fig. 8a). By integrating phylogenetic relationships with chromosomal composition, we further developed parental origin hypotheses for hybrid species (Fig. 8b–e). Our hypotheses posit that L. anhuiensis may be derived from hybridization involving L. chinensis, L. longituba, and/or L. longituba var. flava (Fig. 7, Fig. 8b). L. straminea (2n = 19) likely originated from hybridization between ancestors with base chromosome numbers of 8 and 11. However, phylogenetic analyses revealed that its four accessions are separated in different clades, suggesting that their genomes may have originated from distinct ancestral subgenomes. Similar patterns have been observed in ancient allopolyploid hybridizations within Malvaceae, where processes such as chromosomal translocation and genome fusions between divergent lineages contribute to complex genome structures (Sun et al., 2024). Therefore, further chromosomal structure analysis (i.e., fluorescent in situ hybridization) will be essential for elucidating the origin and mechanisms of allopolyploid or polyploid species formation in Lycoris. Plastid and nuclear phylogenies respectively placed L. houdyshelii (2n = 30) in clades with L. straminea and L. radiata var. pumila, suggesting that L. houdyshelii (2n = 30 = 8 + 11*2) may have formed through allopolyploidization while retaining the duplicated chromosome set of L. radiata var. pumila. Furthermore, diploid L. radiata var. pumila and L. radiata can hybridize with L. aurea to produce autumn-leaf emergence species, such as L. albiflora, which contrasts with previous reports suggesting a parentage of L. traubii and L. radiata (Kurita, 1987a). Previous studies found that hybridization between L. radiata (2n = 22) and L. aurea (2n = 14) produced L. hunanensis (2n = 18 = 11 + 7) and the allopolyploid L. hubeiensis (2n = 29 = 11*2 + 7) has also been produced from this combination (Meng et al., 2018b; Quan et al., 2024), which our findings further corroborate.

Polyploid speciation represents a common mode of sympatric speciation, where hybridization between sympatric species with different chromosome numbers or morphologies generates new lineages (Coyne, 2007). The processes that have produced some Lycoris polyploid hybrids (e.g., 2n = 27 = 8*2 + 11, 2n = 30 = 11*2 + 8) appear particularly complex, as neither nuclear nor plastid phylogenies can unequivocally identify their parental origins, implying potential involvement of multiple ancestral species through successive hybridization and polyploidization events (Fig. 7 and 8d). For several species with 33 chromosomes, both autopolyploid and allopolyploid pathways may exist. Phylogenetic analyses revealed a close relationship between diploid/triploid L. radiata and L. radiata var. pumila, with strong correlations between genome size and chromosome numbers, supporting the autopolyploid origin of triploid L. radiata. The newly reported L. chunxiaoensis and L. rosea (2n = 33), clustered with 22-chromosome species (L. sprengeri, L. haywardii, L. radiata, and L. radiata var. pumila), suggesting these as potential progenitors, though the specific chromosome duplication event remains unclear (Fig. 8e, dashed lines). According to the putative pathway of speciation, triploid L. radiata is considered as an autopolyploid and eight other hybrid species as allopolyploids, reinforcing the significant roles of hybridization and genome doubling in Lycoris speciation.

5. Conclusion

This study reveals the remarkable diversity and complexity in the morphology, cytology and phylogeny of Lycoris. Our integrated analyses suggest that chromosome number variation corresponds more closely with morphological classifications than plastid phylogenies. However, significant nuclear-plastid discordance highlights the reticulate nature of Lycoris evolution, which has complicated species delimitation based solely on morphology. For instance, our results support treating the recently described species L. insularis and L. longifolia as geographic populations of L. sprengeri and L. aurea, respectively. Based on current findings, we propose a putative pathway of speciation in Lycoris driven by multiple hybridization and polyploidization events, with allopolyploidy likely serving as a major evolutionary force. Further analyses of chromosomal structure will be crucial for elucidating the mechanisms underlying allopolyploidy formation, and will provide essential insights into the genetic diversity, speciation processes, and adaptive evolution of this genus.

Acknowledgements

We thank the National Germplasm Bank of Lycoris for providing the plant materials, which were collected with suitable permissions and supplied from the germplasm center for research use, following all ethical requirements. This work was supported by the Scientific Fund of Nanjing Botanical Garden Men. Sun Yat-Sen (JSPKLB202519); Jiangsu Provincial Crop Germplasm Resource Bank (Lycoris) (JS-ZW-K04); Forestry Science and Technology Popularization Demonstration Project of the Central Finance [Su(2024)TG06].

CRediT authorship contribution statement

Xiaochun Shu: Writing – original draft, Resources. Ruisen Lu: Writing – original draft, Software, Formal analysis. Pat Heslop-Harrison: Writing – review & editing. Trude Schwarzacher: Writing – review & editing. Zhong Wang: Resources, Formal analysis. Yalong Qin: Visualization. Ning Wang: Formal analysis. Fengjiao Zhang: Writing – review & editing, Writing – original draft, Project administration, Methodology, Data curation, Conceptualization.

Data availability statement

The chloroplast genome sequences of Lycoris with GenBank accession numbers used in the text were downloaded from NCBI (https://www.ncbi.nlm.nih.gov/).

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.pld.2025.06.010.