Acta Biologica Sibirica

Introduction

Gueldenstaedtia Fisch. is the smallest genus of the Fabaceae family and contains only four species worldwide, including G. guangxiensis W.L. Sha & X.X. Chen, G. henryi Ulbr., G. monophylla Fisch., and G. verna (Georgi) Boriss. (Zhu 2004; Xie et al. 2016; Baasanmunkh et al. 2022, 2026a). The native range of this genus is distributed from the Sino-Himalayan region to Mongolia, Siberia, and South Korea (Seol et al. 2025). In the past, Zhu (2004) first revised the genus, including the global distribution, illustration, taxonomic notes, and nomenclature. Later, Zhu (2005) studied the pollen and seed morphology of this genus. The morphology of Gueldenstaedtia is characterized by perennial herbs, thick and woody roots, and flowers that are pink, purple, or yellow (Zhu 2004). The morphological characteristics of Gueldenstaedtia are very similar to the genus Tibetia (Ali) H.P. Tsui; however, several morphological characters distinguish it from Tibetia by its two upper calyx lobes being free, and stipules free and adnate, as well as pollen grains that are 3-colporate, psilate, and spheroidal (Zhu 2004, 2005). In addition, Gueldenstaedtia and Tibetia were separated from each other approximately 17.23 million years ago, which is congruent with the Himalayan orogeny and the uplift of the Tibetan Plateau in the mid-Miocene (Xie et al. 2016). Traditionally, Gueldenstaedtia was placed in the subtribe Astragalinae of the tribe Galageae (Polhill 1981; Zhu 2004). Later, it has been transferred to the subtribe Chesneyinae of the tribe Caraganeae based on nrDNA and plastid markers (Duan et al. 2016). In recent, Duan et al (2021) upgraded monophyletic subtribe Chesneyinae (Chesneya Lindl. ex Endl., Gueldenstaedtia and Tibetia) to tribe Chesneyeae. In recent phylogenomic framework, monophyly of Chesneyeae is well-supported and placed sister to Caraganeae and Hedysareae based on genome skimming approach (Wang et al. 2025).

Recently, several phylogenetic studies have investigated Gueldenstaedtia and Tibetia based on partial molecular markers (Zhang et al. 2015; Xie et al. 2016; Matiz-Ceron et al. 2022) and complete chloroplast genomes (Son and Choi 2022; Byun et al. 2025). However, none of them included G. monophylla because of lacking genetic data. A few studies have examined the reproductive biology, genetic variability, and conservation of this species (Selyutina et al. 2017, 2021; Baasanmunkh et al. 2022).

G. monophylla is a perennial herb with short stem, pink flowers, orbicular or orbicular-reniform leaves, and thin root (Fig. 1). It grows on dry stony and debris slopes, rocks, and alkaline waterside pebbles in Mongolia (western part) and Russia (Altai, Tuva) (Zhu 2004; Baasanmunkh et al. 2022). In Mongolia, the conservation status of this species has been assessed as Near Threatened (Baasanmunkh et al. 2022).

The chloroplast genomes (plastomes) of angiosperms have been widely used for various analyses, such as phylogeny, plastome evolution, and DNA molecular markers, because of their generally conserved structure, uniparental inheritance, and moderate substitution rates (Wang et al. 2018; Liao et al. 2021; Shanker et al. 2022; Nyamgerel et al. 2023, 2024; Oyuntsetseg et al. 2024a, 2024b; Qin et al. 2025; Baasanmunkh et al. 2026b). However, some angiosperm lineages exhibit extensive structural variation, including gene loss, intron loss, and genome rearrangements (Dobrogojski et al. 2020). The inverted repeat-lacking clade (IRLC) of legumes represents one of the most structurally dynamic groups, characterized by the loss of one copy of the inverted repeat region, frequent gene and intron losses, and numerous rearrangements such as inversions and translocations, making it an important model for studying chloroplast genome evolution (Sabir et al. 2014; Choi et al. 2019; Moghaddam et al. 2022, 2023; Seol et al. 2025). Within this context, recent plastome analyses of Gueldenstaedtia verna (Georgi) Boriss. revealed several notable structural features, including the loss of one IR region, complete loss of rps16, loss of an intron of rpoC1 and clpP genes, four inversions (6–50 kb), and one repositioning (Son and Choi 2022). These findings highlight the importance of comparative plastome analyses for improving our understanding of chloroplast genome evolution and phylogenetic relationships within Fabaceae.

This study aims to: (i) characterize the plastome of G. monophylla and examine structural genomic features, including gene loss, intron loss, inversion, and repositioning and (ii) reconstruct the phylogenetic relationships among Gueldenstaedtia and the closely related genera.

Materials and methods

Plant material, DNA extraction, sequencing, assembly and annotation

The fresh leaves of G. monophylla were collected from natural populations in Khovd city, Khovd Province, Mongolia (48.0535° N and 91.6732° E) on 04 Jun 2025. The specimen was deposited in the Herbarium of the National University of Mongolia (contact person: Batlai Oyuntsetseg, email: oyunaa@num.edu.mn). Genomic DNA was extracted from silica gel-dried leaf material using the CTAB method (Doyle and Doyle 1987). A whole-genome sequencing library was constructed from total genomic DNA using the TruSeq DNA Nano Kit protocol and sequenced on a Next-Seq 500 platform (Illumina, USA). Raw paired-end reads were quality filtered by FastQC v.0.11 (Andrews 2010), adapter trimmed by fastp (Chen et al. 2018) with default settings, and reads were assembled by GetOrganelle v1.7.5 (Jin et al. 2020) with the default k-mer settings (e.g., 21, 45, 65, 85, 105, 127) as implemented in the Canonical Ubuntu 2026.04 operating system. To validate the assembly, the clean reads were mapped back to the assembled circular chloroplast genome using BWA-MEM, and the mapping depth was assessed with SAMtools (Li and Durbin 2009). The assembly and orientation were confirmed using the BLAST tool of NCBI, and graphical views were checked using Geneious Prime® 2026.0.2. Chloroplast genome annotations were performed using Geseq Annotation of Organellar Genomes (Tillich et al. 2017) to predict the location of genes, and tRNAs were identified using tRNAscan-SE (Lowe and Chan 2016). A circular chloroplast genomic map was visualized using the Chloroplot online software (Zheng et al. 2020).

Comparative chloroplast genome analysis

Chloroplast genome rearrangements of all available species within Chesneyeae were identified using collinear blocks in Mauve v.2.4.0 (Darling et al. 2010) with Caragana Fabr. from Caraganeae as reference. To compare relative synonymous codon usage (RSCU) of Chesneyeae species, the consensus exons of the protein-coding were separately extracted and screened by MEGA11 (Tamura et al. 2021). RSCU <1 indicated a codon used less frequently than expected, whereas RSCU >1 indicated a codon used more frequently than expected. The codon usage distribution of Chesneyeae species was visualized using the Heatmapper tool with average linkage clustering and Euclidean distance methods (Babicki et al. 2016). Selective pressure analysis was conducted between Gueldenstaedtia species and measured by the nonsynonymous (Ka) and synonymous (Ks) substitution in consensus exons using KaKs_Calculator 2.0 (Wang et al. 2010) with calculation mode NY and genetic code 11 (bacterial and plant plastid code). Genes with Ka/Ks ratios greater than one were considered to be under positive selection, whereas those with ratios less than one were considered to be under purifying selection. Statistical significance was assessed using Fisher’s exact test, with P < 0.05 considered statistically significant.

Phylogenetic analysis and divergence time estimation

For the phylogenetic analysis, protein-coding gene sequences of four Gueldenstaedtia and 26 species belonging to other genera of the Fabaceae were obtained from the GenBank database based on previous studies (Zhao et al. 2021; Son and Choi 2022; Byun et al. 2025). The protein-coding sequences of the genomes were concatenated and aligned using MAFFT v.7.388 (Katoh and Standley 2013) as implemented in Geneious Prime® 2024.0.5. The alignment was filtered to remove ambiguously aligned regions using GBlocks v.0.91.1 (Talavera and Castresana 2007). The best-fitting model for nucleotide substitutions was determined using the Akaike Information Criterion in jModelTest v.2.1.10 (Darriba et al. 2012). Maximum likelihood (ML) analysis was performed using RaxML v.8.2.11 (Stamatakis 2014) with 1000 bootstrap replicates under the GTR+G+I model. Bayesian Inference (BI) analysis was performed using MrBayes v.3.2.2 (Ronquist et al. 2012), with two independent runs of four simultaneous chains, executed for 5,000,000 generations using the Markov chain Monte Carlo algorithm. Trees were sampled every 5000 generations, and the first 25% were discarded as burn-in. The trees were summarized using a 50% majority-rule consensus to estimate posterior probabilities (PPs). The GTR+G model was selected for BI analysis. The resulting phylogenetic trees were visualized using FigTree v.1.4.2 (Rambaut 2012).

Divergence time was estimated using BEAST v.2.7.8 (Bouckaert et al. 2019) with 69 protein-coding genes (PCGs) from nine species of Chesneyeae and 15 additional species from closely related genera within Fabaceae. A GTR replacement model was implemented with gamma distributions across four rate categories and estimated using a Yule Prior. For divergence time analysis, three calibration points were established based on a previous study (Zhao et al. 2021): four-tribe crown at 44.25 (40.31–49.23) million years ago (MYA), three-tribe crown at 39.46 (35.07–44.24) MYA, and Astragaleae crown at 20.92 (19.13–22.92) MYA. MCMC chains were run for 100 million generations, and sampled every 10,000 generations. The effective sample size (ESS) was evaluated using Tracer v.1.7.2 (Rambaut et al. 2018), with all parameters having ESS values greater than 200, indicating sufficient sampling. The first 10% of trees were discarded as burn-in, and maximum clade credibility trees were generated using TreeAnnotator v.2.7.6 in the BEAST package. These were visualized using Figtree (Rambaut 2012), and the results were summarized using the mean height and 95% high posterior densities (HPDs).

Results

Chloroplast genome feature of Gueldenstaedtia monophylla

A total of 97,835,743 paired-end reads were generated, of which 97,784,497 high-quality clean reads remained after quality filtering. De novo assembly generated a single contig with 100% coverage, and the minimum and average read mapping depth of the assembled genome were 548× and 13,867.81×, respectively. The plastome of G. monophylla was 118,980 bp in length and exhibited a single inverted repeat (IR) region, representing an IRLC-type plastome structure (Fig. 2). The plastid genome contained 110 genes, which comprised 76 protein-coding genes (PCGs), 30 tRNA genes, and four rRNA genes. Eight genes (atpF, pafI, ndhA, ndhB, petB, petD, rpl16, rpl2, and rps16) contained one intron and they were cis-spliced. The rps12 gene was a trans-spliced gene, with the 5′-end located in the LSC and the 3′-end located in an IR region.

Comparative genome analysis within Chesneyeae

Collinearity in gene placement among the Chesneyeae revealed large-scale inversions and translocations (Fig. 3). Among the tribe, Tibetia showed strong similarity with Caragana, indicating that genome have remained stable during the evolution. Meanwhile, Chesneya had ~15 kb genus-specific inversion from the trnN to ndhB (E). In contrast, Gueldenstaedtia revealed several large-scale inversions and translocations: trnK–psbK (A, ~50 kb), rps15–trnL (C, ~10 kb), and ycf2 (D) inversions in all speceis; and accD–rpl16 (B, ~28 kb) and trnL–rps15 (G, ~6 kb) inversions in G. verna and G. guangxiensis; D, C, F (ycf1), and G blocks translocated in all species. The RSCU was calculated on the concatenate of 68 core protein-coding (CDS) genes of the Chesneyeae species. These CDSs contained 19,431–19,648 codons in most species, whereas G. monophylla contained 18,369 codons due to the partial loss of the ycf1 gene. Among the amino acids, leucine (10.75%) was the most frequently detected, whereas cysteine (1.18%) was the least ubiquitous. All species exhibited similar patterns in 64 codons, with high RSCU values for AGA (arginine), TTA (leucine), and GCT (alanine) (Fig. 4). Codons with A or T in the third position had a strong codon bias. Among the three stop codons, TGA was the most commonly detected. Based on hierarchical clustering of RSCU pattern, Chesneya, Gueldenstaedtia, and Tibetia species formed separate clusters. Within the Gueldenstaedtia, the two accessions of G. verna were clustered together which was closely related to G. guangxiensis. While G. monophylla displayed a distinct position within the genus. Overall, only minor variation in codon usage bias was observed among species, indicating strong conservation of plastid codon preferences within Chesneyeae.

The Ka/Ks analysis revealed substantial variation in selective pressure among protein-coding genes across the three Gueldenstaedtia species, with most genes showing purifying selection (Fig. 5). Several genes exhibited a Ka/Ks ratio of 0, including atpH, ndhJ, petL, petN, psaI, psbE, psbF, psbH, psbK, psbM, rpl20, rps12, and rps18, reflecting the absence of detected nonsynonymous substitutions and indicating strong conservation at the protein level. However, few genes showed the strongest signal of positive selection (Ka/Ks > 1) between G. monophylla and others, particularly ndhI, petB, rps11, and clpP (Table 1). As expected, synonymous and nonsynonymous differences of positively selected genes were highly pronounced between G. monophylla and others.

Phylogenetic analysis and divergence time estimation

To identify the phylogenetic position of G. monophylla, 68 core protein-coding genes were extracted from the chloroplast genome of Fabaceae species and aligned for phylogenetic construction. The final dataset comprised 55,339 bp, of which 42,556 bp (76.9%) were constant and 3,043 bp (5.5%) were parsimony-informative. Three Cicer species were used as outgroups. The phylogenetic analysis recovered all sampled species of Chesneyeae as a strongly supported monophyletic clade (BI=1, BS=100). Within the tribe, Gueldenstaedtia and Tibetia placed as a sister groups that together were sister to Chesneya (BI=1, BS=100). Within Gueldenstaedtia, G. monophylla formed a well-supported lineage sister to the clade comprising G. guangxiensis and G. verna (BI=1, BS=100). The close relationship between G. guangxiensis and G. verna was strongly supported, whereas G. monophylla occupied a distinct phylogenetic position as the earliest-diverging extant lineage within the genus.

Divergence time estimation indicated that the major lineages of the IRLC originated during the Eocene and Oligocene (Fig. 6). The split between the Cicerinae lineage (Cicer) and the remaining taxa was estimated at approximately 44.05 MYA. Subsequently, Astragaleae diverged from the lineage containing Chesneyeae, Caraganeae, and Hedysareae around 39.70 MYA. The separation of Hedysareae from the Caraganeae-Chesneyeae clade was estimated at 35.07 MYA, suggesting an Oligocene origin for these tribes. The divergence between the tribes Caraganeae and Chesneyeae occurred approximately 33.37 MYA. The genus Chesneya diverged from the lineage comprising Gueldenstaedtia and Tibetia around 27.51 MYA, whereas Gueldenstaedtia and Tibetia separated at approximately 21.09 MYA. Within Gueldenstaedtia, G. monophylla diverged from the common ancestor of G. guangxiensis and G. verna around 7.21 MYA. The divergence between G. guangxiensis and G. verna was estimated at approximately 0.65 MYA. G. monophylla diverged earlier than other species of the genus and may preserve ancestral genomic characteristics.

Discussion

Chloroplast genome feature of Gueldenstaedtia

The complete chloroplast genome size, structure, and gene content are highly conserved in angiosperms (Dobrogojski et al. 2020). However, rare genome characteristics, such as gene loss, inversion, and IR loss, have been reported in the IRLC legumes of Fabaceae (Moghaddam et al. 2022, 2023; Seol et al. 2025). Gueldenstaedtia is one of the most structurally rearranged chloroplast genomes, characterized by multiple inversions, translocations, and partial gene loss (Son and Choi 2022). In this study, the chloroplast genome of G. monophylla exhibited distinct structural variation compared with previously reported species within the tribe Chesneyeae.

Gene and intron losses are the most distinctive evolutionary features of the IRLC legumes (Lei et al. 2016; Jo et al. 2022; Cauz-Santos 2025), including rps16, rpl22, infA, accD, and ycf4, as well as the loss of introns from clpP, rpoC1, and atpF genes. These events have been documented across multiple IRLC legumes, including Astragalus L., Oxytropis DC., Vicia L., Onobrychis Mill., and others (Jo et al. 2022; Moghaddam et al. 2022, 2023; Son and Choi 2022; Cui et al. 2024; Liu et al. 2024, Li et al. 2025; Seol et al. 2025; Munkhtulga et al. 2026). Among these changes, the loss of rps16 and rpl22 appears to be particularly widespread and may be associated with functional transfer of these genes to the nuclear genome (Lei et al. 2016). Furthermore, the recurrent loss of the intron of rps12 and clpP has been recognized as a common feature of many IRLC and is considered an important evolutionary marker within the group (Jansen et al. 2008). Such as IRLC legumes, Gueldenstaedtia have completely lost rpl22 and rps16 genes, as well as the introns of clpP and rps12 (Son and Choi 2022; Byun et al. 2025). These gene and intron loss were appeared within the tribe, including Chesneya (Nie et al. 2021) and Tibetia (Guo and Wariss 2021).

The chloroplast genome size of tribe Chesneyeae showed a decreasing trend, from 128–127 kb in Chesneya species to 124–123 kb in Tibetia and approximately 122 kb in Gueldenstaedtia (G. guangxiensis and G. verna), whereas G. monophylla has the smallest genome size at 118,980 bp. This gradual reduction in chloroplast genome size across Chesneyeae suggests lineage-specific genome contraction during the evolutionary diversification of the tribe. The ycf1 gene of G. monophylla has a relatively smaller size (1,521 bp) compared with all examined species (5,094–5,466 bp). The ycf1 gene is among the largest and most rapidly evolving plastid genes and frequently serves as a hotspot for structural variation in angiosperms (Dong et al. 2015). The association between rearrangement breakpoints and partial gene loss suggests that large-scale inversion events may have contributed directly to chloroplast genome reorganization in Gueldenstaedtia.

Evolutionary distinctiveness of Gueldenstaedtia monophylla

Phylogenetic analyses strongly support the position of Gueldenstaedtia within tribe Chesneyeae and confirm a sister relationship between Gueldenstaedtia and Tibetia, consistent with previous molecular studies based on ITS and plastid markers (Zhang et al. 2015; Xie et al. 2016; Zhao et al. 2021; Son and Choi 2022; Byun et al. 2025). However, the inclusion of G. monophylla for the first time provides a more complete understanding of evolutionary relationships within the genus. Within Gueldenstaedtia, G. monophylla occupies the earliest-diverging position and separated from the common ancestor of G. guangxiensis and G. verna approximately 7.21 MYA (Fig. 6). This divergence is noticeable longer than the recent split between G. guangxiensis and G. verna (0.65 MYA), suggesting independent evolutionary history for G. monophylla.

Gueldenstaedtia chloroplast genome has large-scale inversions and translocation within tribe Chesneyeae (Fig. 3). After the ancestor of Gueldenstaedtia diverged from Chesneya and Tibetia, structural changes occurred within its chloroplasts. These included ~50 kb inversion of trnK–psbK (A), ~7 kb translocation of ycf1–rps15 (F-G), and ~10 kb inversion and translocation of rps15–trnL (C) and ycf2 (D). The ancestor of Gueldenstaedtia diverged into two lines. G. verna and G. guangxiensis belonged to the same line and first evolutionary change of this line was represented by the accD–rpl16 (B) and trnL–rps15 (G) inversions, which were not detected in the chloroplast genome of G. monophylla. It may suggest an ancestral condition retained within G. monophylla, whereas the alternative arrangement likely originated later during the evolution of the genus.

In addition, codon usage patterns within tribe Chesneyeae were relatively consistent in contrast to the structural rearrangements. Similar codon preferences were observed across all taxa, with a strong bias toward A/T-ending codons, due to the AT-rich feature of chloroplast genomes (Dobrogojski et al. 2020). However, significant differences in codon usage were revealed between the species, and RSCU-based clustering closely mirrored phylogenetic positions. In particular, G. verna and G. guangxiensis formed a closely related group, whereas G. monophylla formed a distinct position. These species-specific differences in codon usage and nucleotide variation provide additional support for the phylogenetic relationships among Gueldenstaedtia species.

Selective pressure analysis revealed positive selection in G. monophylla relative to other species, particularly in ndhI, petB, rps11, and clpP, genes involved in essential metabolic processes. Positively selected genes can play a role in adaptive evolution (Yang and Nielsen 2000). When plants grow in harsh environments, specific sites of photosystem-related genes tend to exhibit positive selection signals, potentially pointing to adaptations for improved photosynthetic efficiency and stress tolerance (Gao et al. 2019; Geng et al. 2026). G. monophylla is adapted to xeric, rocky-alkaline habitats, in contrast to G. verna and G. guangxiensis, which grow in meadows and on shrubby hillslopes (Zhu 2004; Stupnikova 2018; Baasanmunkh et al. 2022). Four positively selected genes in G. monophylla have a functional role that plausibly connects to this ecological contrast. ndhI encodes a subunit of the NADH dehydrogenase, which drives cyclic electron flow around photosystem I and contributes to photoprotection under high light, drought, and temperature stress (Sirpio et al. 2009). Higher Ka/Ks ratio in ndhI is frequently reported in plants from arid or high-altitude habitats (Chen et al. 2022; Sabater 2021). petB is a core subunit of the cytochrome b6/f complex, central to the photosynthetic electron transport chain (Stoppel et al. 2011). Similarly, adaptive substitutions here could reflect selection to maintain photosynthetic efficiency under fluctuating light and temperature (Ghotbi and Soorni 2026; Liu et al. 2023; Zhang et al. 2024). Ribosomal protein rps11 has been linked to adjusting protein synthesis rates under abiotic stress (Robles and Quesada 2022). In parallel, clpP encodes a subunit of the ATP-dependent protease, responsible for degrading damaged/misfolded chloroplast proteins (Shikanai et al. 2001). This is one of the more commonly reported stress adaptation plastid genes, since efficient protein turnover helps sustain photosynthesis under oxidative and thermal stress (Erixon and Oxelman 2008). The functional profile of these genes suggests that positive selection in G. monophylla may be linked to physiological adaptation to its harsher habitat. Future studies incorporating experimental validation, such as physiological or transcriptomic analyses under stress conditions, will be needed to confirm this link.

The congruence among phylogenetic relationships, chloroplast genome rearrangements, codon usage clustering, and selective pressure analysis provides independent lines of evidence supporting the evolutionary distinctiveness of G. monophylla. As the earliest lineage within the genus, G. monophylla may retain ancestral genomic characteristics and therefore represents a key taxon for understanding the origin and diversification of Gueldenstaedtia. Future studies incorporating the remaining species (G. henryi) and nuclear genomic data will help clarify the origin of the observed rearrangements and the evolutionary history of Gueldenstaedtia.

Conclusion

This study presents the first complete chloroplast genome of Gueldenstaedtia monophylla and places it in a comparative framework with related genera of tribe Chesneyeae. The genome retains the typical gene content of IRLC legumes but shows a distinct, smaller architecture than other species within the genus, driven largely by partial ycf1 loss. The evidence from phylogenetic, structural, codon-usage, and selective pressure analysis converges on the same conclusion: G. monophylla is the earliest-diverging lineage within the genus and retains genomic features that are ancestral relative to G. verna and G. guangxiensis. These combined lines of evidence indicate that G. monophylla occupies a distinct evolutionary position within Gueldenstaedtia and offers a useful reference point for reconstructing the ancestral plastome architecture of the genus. Future work incorporating G. henryi and nuclear genomic data will help clarify when and how the structural rearrangements identified here arose, and will further resolve the evolutionary history of Gueldenstaedtia within tribe Chesneyeae.

Data availability

The complete chloroplast genomes of Gueldenstaedtia monophylla was submitted to GenBank under accession numbers PZ504763.

Acknowledgments

This work was supported by the National University of Mongolia (Grant No. P2025-5052), the Korea National Arboretum (Grant No. KNA1-2-42, 22-2) and the Korea Basic Science Institute (National Facilities and Equipment Center) grant funded by the Ministry of Education (Grant No. 2023R1A6C101B022).