Complete chloroplast genomes of four Triumfetta species obtained from herbarium specimens: comparative analysis and identification of a synapomorphic inversion Authors: Rushan Yan, Abdullah, Abdul Sammad, Hui Li, Mingzhuo Fu, Ying Cui, Parviz Heidari, Xiaoxuan Tian Published: 2026 ABSTRACT The genus Triumfetta L. (Malvaceae) comprises about 177 species, several of which are of medicinal importance. However, genomic resources for this genus remain limited, with only ten chloroplast (cp.) genomes published to date. Here, we report the sequencing and de novo assembly of complete cp. genomes for four Triumfetta species (Triumfetta annua L., Triumfetta cordifolia A.Rich., Triumfetta lappula L., and Triumfetta rhomboidea Jacq.) using the Illumina NovaSeq 6000 platform. Our objectives were to characterize cp. genome features, conduct comparative analyses with previously reported species, identify polymorphic loci, and reconstruct a preliminary phylogeny. The cp. genomes exhibited the typical quadripartite structure, ranging from 159,245 to 160,604 bp, and contained 112 unique genes (78 protein-coding, 30 tRNA, and 4 rRNA). Comparative analysis revealed a six-gene synapomorphic inversion (trnC, petN, psbM, trnD, trnY, and trnE) in the large single-copy (LSC) region. Codon usage analysis revealed a strong bias toward A/T-ending codons (RSCU > 1). Leucine was the most abundant amino acid, while cysteine was the least abundant. Simple sequence repeat analysis identified 63–81 repeats, with a predominance of mononucleotide A/T repeats. Nucleotide substitution analysis revealed a transition bias across all regions (Ts/Tv > 1), most pronounced in the LSC (Ts/Tv = 1.29–1.35). Seven highly polymorphic loci were identified as potential DNA barcoding candidates, including two intergenic spacers (trnL-ccsA, ndhD-psaC), and five coding regions (clpP, rpl22, ycf1, rpl23, and ndhI). This study provides genomic resources for Triumfetta taxonomy, highlighting key structural variations and identifying molecular markers with potential applications in taxonomic research.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-026-57295-7. Introduction The chloroplast (cp.) genome exhibits a conserved quadripartite structure, consisting of a large single-copy (LSC) region, a small single-copy (SSC) region, and a pair of inverted repeat regions (IRa and IRb)1,2. It typically encodes 110–130 genes, including protein-coding sequences (CDS), transfer RNAs (tRNAs), and ribosomal RNAs (rRNAs)2. By virtue of their moderate polymorphism and predominantly uniparental inheritance, cp. genomes serve as valuable resources for phylogenetics, DNA barcoding, and conservation genetics2–6 and have been proposed as super-barcodes for plant identification7. Numerous mutational events, including substitutions, insertions-deletions (indels) events, and inversions have been reported in cp. genome2,3,5. The family Malvaceae Juss. represents a diverse and economically important angiosperm lineage comprising approximately 244 genera and 4,225 species distributed worldwide, with the greatest diversity concentrated in tropical and subtropical regions8–10. The genus Triumfetta L. (Grewioideae, Malvaceae) is pantropical and comprises approximately 177 species10–12 distributed across Africa, Asia, the Americas, and Australasia10,11. Taxonomic delimitation within Triumfetta has relied predominantly on morphological characters13–15, which are susceptible to environmental influences and convergent evolution, leaving interspecific relationships poorly resolved. Although ITS-based phylogenies have been attempted for Grewioideae, they include relatively few Triumfetta representatives16. The first systematic cp. genomic study of the genus was recently conducted by Yan et al.17, who assembled and compared cp. genomes of ten species, characterized genome structure, identified positive selection in six genes (psbK, rpl20, rpl23, rps16, rps7, and ycf2), and reconstructed a preliminary phylogeny revealing two biogeographically structured clades. However, that study was geographically restricted, with sampling dominated by Australian taxa and no representation from the Americas — the most species-rich region within the distribution of the genus. Furthermore, the LSC inversion involving six genes (trnC, petN, psbM, trnD, trnY, and trnE) reported in T. tomentosa has not been evaluated across a broader taxonomic sample, leaving it uncertain whether it is species-specific, clade-specific, or more widespread within the genus. Species within Triumfetta are also noteworthy for their diverse pharmacological activities, including antimicrobial, antidiabetic, anti-inflammatory, and antitumour properties18,19, making accurate species delimitation a priority for ethnobotanical and pharmaceutical applications. Prior cp. genomic studies in related Malvaceae genera have demonstrated that expanded geographic and taxonomic sampling substantially refines phylogenetic resolution and barcode performance20–25. In this study, we sequenced and de novo assembled cp. genomes for four Triumfetta species: Triumfetta rhomboidea Jacq., Triumfetta lappula L., Triumfetta cordifolia A.Rich., and Triumfetta annua L. from diverse geographical regions, representing their native ranges in the Americas, Africa, and Asia (Table 1). Integrating these genomes with the ten previously published Triumfetta cp. genomes17, we pursued four objectives: (i) to characterize the structural features and genomic properties of the newly sequenced cp. genomes; (ii) to reconstruct an expanded phylogeny of Triumfetta with broader geographic coverage, incorporating the first American representatives; (iii) to assess the taxonomic scope of the LSC inversion previously reported in T. tomentosa17 across a more comprehensive species set; and (iv) to identify and validate polymorphic loci as candidate DNA barcodes for taxonomic and phylogenetic research within the genus. Table 1Clade classification, native ranges, growth forms, and primary biomes of Triumfetta species.Species nameCladeInversion presentNative rangeGrowth formPrimary biomeTriumfetta annua L.Clade IbNoTropical & S. Africa, Madagascar, Indian Subcontinent to S. China and Peninsula MalaysiaSubshrubDry tropical Triumfetta appendiculata F.Muell.Clade IINoAustraliaSubshrub or shrubDry tropical Triumfetta cana BlumeClade IbNoHimalaya to S. China and PeninsulaSubshrub or shrubDry tropical Triumfetta chaetocarpa F.Muell.Clade IINoAustraliaSubshrub or shrubDesert or dry shrubland Triumfetta clementii (Domin) RyeClade IINoAustraliaSubshrubDesert or dry shrublandTriumfetta cordifolia A.Rich.Clade IaYesAfricaShrubWet tropical Triumfetta japonica MakinoClade IbNoKorea, S. Central & S. Japan to PhilippinesSubshrub or shrubSubtropical Triumfetta johnstonii F.Muell.Clade IINoAustraliaSubshrub or shrubDesert or dry shrublandTriumfetta lappula L.Clade IbNoMexico to Tropical AmericaSubshrubWet tropical Triumfetta leptacantha F.Muell.Clade IINoAustraliaSubshrub or shrubDesert or dry shrubland Triumfetta maconochieana HalfordClade IINoAustraliaSubshrub or shrubDesert or dry shrubland Triumfetta ramosa Sprague & Hutch.Clade IINoAustraliaSubshrub or shrubDesert or dry shrublandTriumfetta rhomboidea Jacq.Clade IaYesTropical & Subtropical Old WorldSubshrub or shrubWet tropical Triumfetta tomentosa BojerClade IaYesAfricaSubshrub or shrubDry tropical Table 1Clade classification, native ranges, growth forms, and primary biomes of Triumfetta species.Species nameCladeInversion presentNative rangeGrowth formPrimary biomeTriumfetta annua L.Clade IbNoTropical & S. Africa, Madagascar, Indian Subcontinent to S. China and Peninsula MalaysiaSubshrubDry tropical Triumfetta appendiculata F.Muell.Clade IINoAustraliaSubshrub or shrubDry tropical Triumfetta cana BlumeClade IbNoHimalaya to S. China and PeninsulaSubshrub or shrubDry tropical Triumfetta chaetocarpa F.Muell.Clade IINoAustraliaSubshrub or shrubDesert or dry shrubland Triumfetta clementii (Domin) RyeClade IINoAustraliaSubshrubDesert or dry shrublandTriumfetta cordifolia A.Rich.Clade IaYesAfricaShrubWet tropical Triumfetta japonica MakinoClade IbNoKorea, S. Central & S. Japan to PhilippinesSubshrub or shrubSubtropical Triumfetta johnstonii F.Muell.Clade IINoAustraliaSubshrub or shrubDesert or dry shrublandTriumfetta lappula L.Clade IbNoMexico to Tropical AmericaSubshrubWet tropical Triumfetta leptacantha F.Muell.Clade IINoAustraliaSubshrub or shrubDesert or dry shrubland Triumfetta maconochieana HalfordClade IINoAustraliaSubshrub or shrubDesert or dry shrubland Triumfetta ramosa Sprague & Hutch.Clade IINoAustraliaSubshrub or shrubDesert or dry shrublandTriumfetta rhomboidea Jacq.Clade IaYesTropical & Subtropical Old WorldSubshrub or shrubWet tropical Triumfetta tomentosa BojerClade IaYesAfricaSubshrub or shrubDry tropical Clade classification, native ranges, growth forms, and primary biomes of Triumfetta species. Materials and methods DNA extraction and sequencing Leaf material was obtained from herbarium specimens at the Finnish Museum of Natural History for four species: T. rhomboidea (C.196802), T. lappula (C.196729), T. cordifolia (C.196804), and T. annua (C.196717). Detailed specimen information can be accessed via the LUOMUS herbarium database (e.g., http://id.luomus.fi/C.196802; the voucher number in the URL can be replaced to access records for the other species). The collection dates of the samples are provided in Table 2. Genomic DNA was extracted from approximately 10 mg of dried leaf tissue per specimen using the TIANGEN Plant Genomic DNA Kit (TIANGEN Biotech, Beijing, China), with the following protocol modifications designed to accommodate the physicochemical challenges of aged herbarium material. First, the volume of GP1 lysis buffer was increased from the standard 0.7 mL to 1.5 mL. The elevated buffer volume serves a dual purpose: it ensures complete resuspension of mechanically recalcitrant, desiccated leaf tissue — which does not homogenize efficiently at standard buffer volumes — and dilutes the phenolic compounds, alkaloids, and other secondary metabolites that accumulate in plant cells over time and are known to co-purify with DNA from aged specimens, which inhibit downstream enzymatic reactions26,27. Second, the homogenate was divided equally into two Eppendorf tubes and processed as two parallel half-volume extractions. Table 2Collection details, total DNA yield, sequencing metrics—including coverage depth, whole genome sequencing (WGS) reads, and chloroplast reads—as well as GenBank and Sequence Read Archive (SRA) accession numbers for four Triumfetta species.SpeciesCollection dateTotal DNA (µg)SRA accession no.Data in GBCoverage depthWGS reads (millions)Chloroplast reads (millions)GenBank submission Triumfetta rhomboidea 1967-07-180.31066SRR354127556.3344718.190.47PX395925 Triumfetta lappula 16900.30694SRR354127568.0159923.030.62PX395924 Triumfetta cordifolia 1967-07-060.12830SRR354127574.1919912.030.17PX395923 Triumfetta annua 1972-03-230.27608SRR354127587.8946322.690.48PX395922 Table 2Collection details, total DNA yield, sequencing metrics—including coverage depth, whole genome sequencing (WGS) reads, and chloroplast reads—as well as GenBank and Sequence Read Archive (SRA) accession numbers for four Triumfetta species.SpeciesCollection dateTotal DNA (µg)SRA accession no.Data in GBCoverage depthWGS reads (millions)Chloroplast reads (millions)GenBank submission Triumfetta rhomboidea 1967-07-180.31066SRR354127556.3344718.190.47PX395925 Triumfetta lappula 16900.30694SRR354127568.0159923.030.62PX395924 Triumfetta cordifolia 1967-07-060.12830SRR354127574.1919912.030.17PX395923 Triumfetta annua 1972-03-230.27608SRR354127587.8946322.690.48PX395922 Collection details, total DNA yield, sequencing metrics—including coverage depth, whole genome sequencing (WGS) reads, and chloroplast reads—as well as GenBank and Sequence Read Archive (SRA) accession numbers for four Triumfetta species. Third, both tubes underwent an extended incubation at 65 °C for 60 min, rather than the recommended 20 min, to ensure thorough cell wall disruption and complete protein denaturation in the rigid, cross-linked cell walls of desiccated herbarium tissue. The extended thermal treatment also facilitated the partial degradation of thermolabile secondary metabolites — including alkaloids and polyphenolics — that would otherwise reduce DNA purity and co-elute with the DNA extract, as evidenced by elevated A230 absorbance ratios in short-incubation extractions from aged specimens26. Following incubation, the supernatants from both tubes were combined and loaded onto a single CB3 column in sequential aliquots, enabling elution of the combined DNA yield into a single final volume. Fourth, during elution, the silica membrane was incubated with 70 µL of elution buffer for 30 min at room temperature rather than the recommended 5 min. This extended incubation is justified by two considerations: (i) the kit manufacturer recommends a minimum elution volume of 50 µl; concentrating two parallel extractions into a single 70 µl eluate thus provides the highest attainable DNA concentration while remaining within the validated elution volume range, and (ii) the extended incubation time allows elution buffer to fully penetrate and re-hydrate the silica membrane, which improves the release efficiency of low-molecular-weight, fragmented DNA — the predominant form present in aged herbarium specimens26,27. All remaining extraction steps followed the manufacturer’s standard protocol without further modification. The quantity and integrity of the extracted DNA were evaluated using an Agilent 5400 Fragment Analyzer (Agilent Technologies, Santa Clara, CA, USA). DNA concentration and fragment size distribution were assessed for each specimen. As expected for herbarium material, DNA was partially degraded across all four specimens. The total DNA amounts of the four samples ranged from 0.128 to 0.311 µg (Table 2). A paired-end sequencing library with a 350-bp insert size was prepared for each sample, and sequencing (2 × 150 bp) was performed on an Illumina NovaSeq 6000 platform at Novogene Co., Ltd. (Tianjin, China), following the approach previously described28. Chloroplast genome assembly and annotation Raw reads were processed using fastp v1.0.129 for quality control. Paired-end reads were discarded if either read contained > 10% ambiguous bases (N) of its length, or if > 50% of bases in either read had a quality score (Q) ≤ 5. The cp. genomes were assembled de novo from the clean reads using GetOrganelle v1.7.7.130 with multiple k-mer sizes (21, 45, 65, 85, 105). To validate the assemblies and evaluate the depth distribution across genomic regions, the clean reads were mapped back to their respective draft cp. genomes using BWA-MEM v0.7.18 with default parameters. The resulting alignments were sorted and indexed with SAMtools v1.21, and per-base coverage depth was calculated using samtools depth under the default parameter. Overall, coverage depth was highly uniform across the whole genome for all four specimens, indicating the absence of major assembly artifacts or contaminating nuclear DNA. However, to rigorously exclude the incorporation of nuclear mitochondrial DNA segments or nuclear plastid DNA segments — which can co-assemble with organellar reads, particularly in fragmented herbarium DNA—we manually inspected any localized regions exhibiting relatively low coverage depths. We confirmed those regions further by performing pairwise alignments against reference Triumfetta cp. genomes using the MAFFT aligner implemented in Geneious Prime 2025. This curation step confirmed that the low-coverage regions exhibited syntenic gene boundaries and sequence homology consistent with the cp. genome, verifying that no nuclear or mitochondrial sequences were mistakenly incorporated into the final assemblies. Gene annotation was performed using GeSeq v2.0.331, and tRNA predictions were further validated using both ARAGORN v1.2.3832 and tRNAscan-SE v2.0.733. Start and stop codons for protein-coding genes were manually verified and corrected in Geneious Prime 2025 by comparing them with homologous genes from Corchorus capsularis L. (NC044467.1), T. cana (OP581196.1), and T. tomentosa (PV550975.1). Finally, the genomes were visualized using Chloroplot v134. Comparative chloroplast genomics among Triumfetta species Genome features, including complete genome length, LSCs, SSCs, IRs, their GC content, and the GC content of functional genes, were analyzed using CGAS v1.0.135 and cross-checked in Geneious Prime 2025. CGAS results were cross-validated against manual calculations in Geneious Prime 2025, and no discrepancies were found. The arrangement of genes for potential inversion events was examined using collinear blocks using the Mauve progressive alignment algorithm36. The boundary regions of the IRs and single copies were visualized with CPJSdraw to explore IR contraction and expansion patterns37. To characterize sequence-level variation within the inverted region, the six genes encompassed by the inversion (trnC-GCA, petN, psbM, trnD-GUC, trnY-GUA, and trnE-UUC) were individually aligned across all 14 Triumfetta species using MAFFT v7.52538 to identify substitutions and indels. For the two protein-coding genes (petN and psbM), nucleotide sequences were translated in silico using the first reading frame to assess whether single-nucleotide polymorphisms (SNPs) identified among clades resulted in amino acid changes. For the four tRNA genes, sequence conservation was evaluated at the nucleotide level. Analysis of amino acid frequency, codon usage, and microsatellites Amino acid frequency and codon usage bias were evaluated by calculating relative synonymous codon usage (RSCU) using CGAS35. Simple sequence repeats (SSRs) were identified using the MISA-web server39(https://webblast.ipk-gatersleben.de/misa/) and CGAS. The minimum thresholds for repeat units were: ten for mononucleotides, five for dinucleotides, four for trinucleotides, and three for tetranucleotides, pentanucleotides, and hexanucleotides. Both tools yielded identical SSR sets; CGAS output was used for downstream reporting because it provides positional annotation relative to genomic regions (LSC, SSC, IR) and gene boundaries. Analysis of substitution, identification of polymorphic loci, and reconstruction of phylogeny Nucleotide substitution analysis was conducted on pairwise alignments of the LSC, SSC, and IR regions extracted from Triumfetta species. C. capsularis was selected as the reference species due to its close phylogenetic relationship with Triumfetta17. Transitions (Ts), transversions (Tv), and their ratios (Ts/Tv) were calculated using CGAS. Additionally, we performed Ts/Tv analysis on 78 protein-coding genes. For this, we extracted all coding sequences of each species, concatenated them, and aligned using MAFFT v7.525. Nucleotide diversity (π) across different cp. genome regions, including CDS, introns, and intergenic spacers (IGS), was evaluated using CPStools40 and CGAS. Both approaches produced largely consistent results; however, CGAS additionally identified a novel intergenic spacer generated by a genomic inversion. The shared 78 coding sequences were extracted from each Triumfetta species, from representative species of related Grewioideae genera (Microcos, Colona, Grewia, and Corchorus), and from Aquilaria sinensis (Lour.) Spreng. (NC029243), which served as the outgroup. These sequences were aligned with MAFFT v7.52538 and then concatenated. The maximum likelihood phylogenetic tree was reconstructed using this alignment in IQ-TREE v3.0.141. The best-fit substitution model (TVM + F+R3) was selected using ModelFinder42 according to the Bayesian Information Criterion (BIC). Branch support was assessed using 10,000 ultrafast bootstrap43 replicates (with the –bnni option) and 10,000 SH-aLRT tests as described previously5. The final tree was visualized using the Interactive Tree of Life platform44. Results Characterization of the chloroplast genome of Triumfetta species High-quality sequencing data for four Triumfetta species were generated on the Illumina NovaSeq 6000 platform, yielding 4.19–8.01 GB (12.03–23.03 million reads). De novo assembly produced complete cp. genomes in circular form, with average coverage depths ranging from 199× for T. cordifolia to 599× for T. lappula (Table 2, Fig. S1). The Triumfetta cp. genomes exhibited typical quadripartite structures, with total lengths ranging from 159,245 bp (T. lappula) to 160,604 bp (T. tomentosa). These structures consisted of an LSC region ranging from 88,069 bp (T. lappula) to 89,386 bp (T. tomentosa), an SSC region ranging from 20,142 bp (T. appendiculata) to 20,266 bp (T. tomentosa), and each IR region ranging from 25,440 bp (T. maconochieana) to 25,498 bp (T. appendiculata) (Fig. 1; Table 3). Fig. 1Circular map of the Triumfetta chloroplast genome. Genes transcribed clockwise are shown inside the circle, while those transcribed counterclockwise are positioned outside. Genes are color-coded according to their functional categories. The inner gray circle indicates the GC content. Insets (A) and (B) illustrate two distinct structural types observed within the genus. Circular map of the Triumfetta chloroplast genome. Genes transcribed clockwise are shown inside the circle, while those transcribed counterclockwise are positioned outside. Genes are color-coded according to their functional categories. The inner gray circle indicates the GC content. Insets (A) and (B) illustrate two distinct structural types observed within the genus. Table 3Comparison of chloroplast genomes among Triumfetta species.SpeciesNucleotide length (bp)Number of genesGC content (%)TotalLSCSSCIRUniqueProteintRNArRNADuplicateTotalLSCSSCIRCDSrRNAtRNAAll gene Triumfetta annua 160,14088,96020,21625,482112783041737.135.031.043.138.255.652.939.7 Triumfetta appendiculata 160,23789,09920,14225,498112783041737.135.031.343.138.255.752.939.7 Triumfetta cana 160,19489,00620,23825,475112783041737.135.131.143.138.155.652.939.6 Triumfetta chaetocarpa 160,20189,13620,15125,457112783041737.135.031.343.138.255.752.939.7 Triumfetta clementii 160,11489,05120,16125,451112783041737.134.931.143.138.155.752.939.7 Triumfetta cordifolia 160,59389,37820,26325,476112783041737.135.131.143.038.255.652.939.7 Triumfetta japonica 160,19189,00820,22525,479112783041737.135.131.243.138.255.652.939.7 Triumfetta johnstonii 160,16589,11520,15625,447112783041737.235.131.343.138.255.752.939.7 Triumfetta lappula 159,24588,06920,21225,482112783041737.235.231.243.138.255.652.839.7 Triumfetta leptacantha 160,33489,25120,16925,457112783041737.135.031.343.138.255.752.939.7 Triumfetta maconochieana 160,07589,03420,16125,440112783041737.135.031.243.138.255.752.639.7 Triumfetta ramosa 160,24189,17920,15625,453112783041737.135.031.343.138.255.752.939.7 Triumfetta rhomboidea 160,53589,35420,23125,475112783041737.135.131.143.138.255.652.939.7 Triumfetta tomentosa 160,60489,38620,26625,476112783041737.135.131.043.138.255.652.939.7 Table 3Comparison of chloroplast genomes among Triumfetta species.SpeciesNucleotide length (bp)Number of genesGC content (%)TotalLSCSSCIRUniqueProteintRNArRNADuplicateTotalLSCSSCIRCDSrRNAtRNAAll gene Triumfetta annua 160,14088,96020,21625,482112783041737.135.031.043.138.255.652.939.7 Triumfetta appendiculata 160,23789,09920,14225,498112783041737.135.031.343.138.255.752.939.7 Triumfetta cana 160,19489,00620,23825,475112783041737.135.131.143.138.155.652.939.6 Triumfetta chaetocarpa 160,20189,13620,15125,457112783041737.135.031.343.138.255.752.939.7 Triumfetta clementii 160,11489,05120,16125,451112783041737.134.931.143.138.155.752.939.7 Triumfetta cordifolia 160,59389,37820,26325,476112783041737.135.131.143.038.255.652.939.7 Triumfetta japonica 160,19189,00820,22525,479112783041737.135.131.243.138.255.652.939.7 Triumfetta johnstonii 160,16589,11520,15625,447112783041737.235.131.343.138.255.752.939.7 Triumfetta lappula 159,24588,06920,21225,482112783041737.235.231.243.138.255.652.839.7 Triumfetta leptacantha 160,33489,25120,16925,457112783041737.135.031.343.138.255.752.939.7 Triumfetta maconochieana 160,07589,03420,16125,440112783041737.135.031.243.138.255.752.639.7 Triumfetta ramosa 160,24189,17920,15625,453112783041737.135.031.343.138.255.752.939.7 Triumfetta rhomboidea 160,53589,35420,23125,475112783041737.135.131.143.138.255.652.939.7 Triumfetta tomentosa 160,60489,38620,26625,476112783041737.135.131.043.138.255.652.939.7 Comparison of chloroplast genomes among Triumfetta species. Analysis of GC content revealed a conserved profile across Triumfetta cp. genomes (37.1–37.2%), with notable regional variation: IR regions (43.0–43.1%) > LSC (34.9–35.2%) > SSC (31.0–31.3%). Gene content and intron distribution were highly conserved across Triumfetta species. Each cp. genome contained 112 unique genes, comprising 78 CDS, 30 tRNA genes, and 4 rRNA genes (Table 3). The IR regions duplicated 16 genes, including 5 protein-coding genes (rpl2, rpl23, rps7, ycf2, ndhB), 7 tRNA genes (trnA-UGC, trnI-CAU, trnI-GAU, trnL-CAA, trnN-GUU, trnV-GAC, trnR-ACG), and 4 rRNA genes (rrn16, rrn23, rrn4.5, rrn5). Among the 18 intron-containing genes, 12 were protein-coding (rpoC1, rpl16, rps12, rpl2, clpP, atpF, petB, ndhA, rps16, ndhB, petD, ycf3) and 6 were tRNA genes (trnG-UCC, trnK-UUU, trnV-UAC, trnL-UAA, trnA-UGC, trnI-GAU), with intron positions conserved across all species (Table 4). Table 4Chloroplast gene content and functional classification in Triumfetta species.Category for genesGroup of genesName of genesTotalSelf-replicationLarge subunit of ribosomerpl14, rpl16*, rpl2*a, rpl20, rpl22, rpl23a, rpl32, rpl33, rpl3611Small subunit of ribosomerps11, rps12**, rps14, rps15, rps16*, rps18, rps19, rps2, rps3, rps4, rps7a, rps813DNA dependent RNA polymeraserpoA, rpoB, rpoC1*, rpoC24rRNA genesrrn16a, rrn23a, rrn4.5a, rrn5a8tRNA genestrnA-UGC*a, trnC-GCA, trnD-GUC, trnE-UUC, trnF-GAA, trnG-GCC, trnG-UCC*, trnH-GUG, trnI-CAUa, trnI-GAU*a, trnK-UUU*, trnL-CAAa, trnL-UAA*, trnL-UAG, trnM-CAU, trnN-GUUa, trnP-UGG, trnQ-UUG, trnR-ACGa, trnR-UCU, trnS-GCU, trnS-GGA, trnS-UGA, trnT-GGU, trnT-UGU, trnV-GACa, trnV-UAC*, trnW-CCA, trnY-GUA, trnfM-CAU37PhotosynthesisPhotosystem IpsaA, psaB, psaC, psaI, psaJ5Photosystem IIpsbA, psbB, psbC, psbD, psbE, psbF, psbH, psbI, psbJ, psbK, psbL, psbM, psbN (pbf1), psbT, psbZ (lhbA)15NADPH dehydrogenasendhA*, ndhB*a, ndhC, ndhD, ndhE, ndhF, ndhG, ndhH, ndhI, ndhJ, ndhK12Cytochrome b/f complexpetA, petB*, petD*, petG, petL, petN6Large subunit of Rubisco rbcL 1Photosynthesis assembly genesycf3 (pafI)**, ycf4 (pafII)2Other genesProtease clpP** 1Maturase matK 1Envelop membrane protein cemA 1Subunit of Acetyl-CoA-carboxylase accD 1C-type cytochrome synthesis gene ccsA 1Conserved open reading framesycf1, ycf2a3Pseudogenes infA Ψ 1Subunits of ATP synthaseatpA, atpB, atpE, atpF*, atpH, atpI6Total number of genes129* and ** indicate genes containing one and two introns, respectively; a and Ψ indicate duplicated genes in inverted repeat (IR) regions and pseudogenes, respectively. The rps12 gene is a trans-spliced gene and is not marked as duplicated despite appearing in multiple locations. Table 4Chloroplast gene content and functional classification in Triumfetta species.Category for genesGroup of genesName of genesTotalSelf-replicationLarge subunit of ribosomerpl14, rpl16*, rpl2*a, rpl20, rpl22, rpl23a, rpl32, rpl33, rpl3611Small subunit of ribosomerps11, rps12**, rps14, rps15, rps16*, rps18, rps19, rps2, rps3, rps4, rps7a, rps813DNA dependent RNA polymeraserpoA, rpoB, rpoC1*, rpoC24rRNA genesrrn16a, rrn23a, rrn4.5a, rrn5a8tRNA genestrnA-UGC*a, trnC-GCA, trnD-GUC, trnE-UUC, trnF-GAA, trnG-GCC, trnG-UCC*, trnH-GUG, trnI-CAUa, trnI-GAU*a, trnK-UUU*, trnL-CAAa, trnL-UAA*, trnL-UAG, trnM-CAU, trnN-GUUa, trnP-UGG, trnQ-UUG, trnR-ACGa, trnR-UCU, trnS-GCU, trnS-GGA, trnS-UGA, trnT-GGU, trnT-UGU, trnV-GACa, trnV-UAC*, trnW-CCA, trnY-GUA, trnfM-CAU37PhotosynthesisPhotosystem IpsaA, psaB, psaC, psaI, psaJ5Photosystem IIpsbA, psbB, psbC, psbD, psbE, psbF, psbH, psbI, psbJ, psbK, psbL, psbM, psbN (pbf1), psbT, psbZ (lhbA)15NADPH dehydrogenasendhA*, ndhB*a, ndhC, ndhD, ndhE, ndhF, ndhG, ndhH, ndhI, ndhJ, ndhK12Cytochrome b/f complexpetA, petB*, petD*, petG, petL, petN6Large subunit of Rubisco rbcL 1Photosynthesis assembly genesycf3 (pafI)**, ycf4 (pafII)2Other genesProtease clpP** 1Maturase matK 1Envelop membrane protein cemA 1Subunit of Acetyl-CoA-carboxylase accD 1C-type cytochrome synthesis gene ccsA 1Conserved open reading framesycf1, ycf2a3Pseudogenes infA Ψ 1Subunits of ATP synthaseatpA, atpB, atpE, atpF*, atpH, atpI6Total number of genes129* and ** indicate genes containing one and two introns, respectively; a and Ψ indicate duplicated genes in inverted repeat (IR) regions and pseudogenes, respectively. The rps12 gene is a trans-spliced gene and is not marked as duplicated despite appearing in multiple locations. Chloroplast gene content and functional classification in Triumfetta species. * and ** indicate genes containing one and two introns, respectively; a and Ψ indicate duplicated genes in inverted repeat (IR) regions and pseudogenes, respectively. The rps12 gene is a trans-spliced gene and is not marked as duplicated despite appearing in multiple locations. Among the gene categories, rRNA genes exhibited the highest GC content (55.6–55.7%), followed by tRNA genes (52.6–52.9%) and CDS (38.1–38.2%) (Table 3). However, Progressive Mauve alignment revealed a distinct inversion event within the LSC region (from ~ 28,200 bp to ~ 32,220 bp). This inversion, previously documented in only one Triumfetta species (T. tomentosa), was confirmed in that species and newly identified in two additional species (T. cordifolia, T. rhomboidea) in the present study. The rearranged segment length was about 4,000 bp and encompassed six consecutive genes: trnC, petN, psbM, trnD, trnY, and trnE (Fig. 2). This inversion was shared by T. cordifolia, T. rhomboidea, and T. tomentosa, and was absent from all other species examined; its evolutionary significance is addressed in the Discussion (Sect. 4). Fig. 2Collinearity analysis of Triumfetta species based on MAUVE alignment. Red boxes denote inversion regions spanning six genes (trnC, petN, psbM, trnD, trnY, trnE), corresponding to positions approximately 28,200–32,220 bp in the non-inverted reference orientation. This inversion event distinguishes Clade Ia from all other Triumfetta lineages (Clade Ib and Clade II). Collinearity analysis of Triumfetta species based on MAUVE alignment. Red boxes denote inversion regions spanning six genes (trnC, petN, psbM, trnD, trnY, trnE), corresponding to positions approximately 28,200–32,220 bp in the non-inverted reference orientation. This inversion event distinguishes Clade Ia from all other Triumfetta lineages (Clade Ib and Clade II). Inverted repeat boundaries analysis Analysis of IR boundaries revealed broadly conserved patterns across Triumfetta species at the four junctions: JLB (LSC/IRb), JSB (IRb/SSC), JSA (SSC/IRa), and JLA (IRa/LSC), with only minor variations in IR expansion and contraction (Fig. 3). In all 14 species, the rps19 gene was located at the JLB junction, starting in the IRb region (with 8 bp in the IRb region) and extending toward the LSC (with 271 bp in the LSC). The rpl2 gene was positioned in the IRb and IRa regions, with distances from the JLB and JLA junctions ranging from 74 bp to 165 bp. The trnN gene was found in both IR regions, at distances of 333–344 bp from the JSA and JSB boundaries. The ndhF gene began in the SSC region and extended into the IRb region from 32 bp to 43 bp. The ycf1 gene was located near the JSA boundary in the SSC region, at a distance of 1–11 bp from the JSA boundary across species. Finally, the trnH gene was located at the JLA junction (Fig. 3). Fig. 3Comparison of inverted repeat (IR) expansion and contraction among Triumfetta species. JLB, JSB, JSA, and JLA correspond to the junctions between LSC/IRb, IRb/SSC, SSC/IRa, and IRa/LSC, respectively. Species newly assembled in this study are highlighted with red boxes. Comparison of inverted repeat (IR) expansion and contraction among Triumfetta species. JLB, JSB, JSA, and JLA correspond to the junctions between LSC/IRb, IRb/SSC, SSC/IRa, and IRa/LSC, respectively. Species newly assembled in this study are highlighted with red boxes. Amino acid frequency, codon usage, and microsatellites Leucine was the most abundant amino acid, followed by isoleucine, whereas cysteine was the least abundant (Fig. 4A, Table S1). RSCU analysis revealed 30 codons with RSCU values greater than 1. Of these, 13 codons ended in A, 16 in T, and 1 in G (Table S2). No differences were observed in codon preferences among the four species. RSCU analysis revealed a strong preference for A- or T-ending codons (RSCU > 1), while codons ending with C or G were used less frequently (RSCU < 1) (Fig. 4B). This pattern reflects the AT-rich composition of the cp. genome. Fig. 4Amino acid frequency and codon usage in the chloroplast genomes of Triumfetta species. (A) Distribution of amino acid frequency, where the x-axis represents amino acid types and the y-axis indicates their relative frequency. (B) Relative synonymous codon usage (RSCU), showing amino acids along the x-axis and bar heights representing RSCU values across species; codons are indicated within each bar. Amino acid frequency and codon usage in the chloroplast genomes of Triumfetta species. (A) Distribution of amino acid frequency, where the x-axis represents amino acid types and the y-axis indicates their relative frequency. (B) Relative synonymous codon usage (RSCU), showing amino acids along the x-axis and bar heights representing RSCU values across species; codons are indicated within each bar. The number of identified SSRs ranged from 63 in T. cordifolia to 81 in T. annua, with their distribution primarily concentrated in the LSC (41–59) region, followed by the SSC (12–15) and IR (4–10) regions (Fig. 5A; Table S3). SSR analysis identified a predominance of A/T-rich mononucleotide repeats, followed by di- and trinucleotide motifs (Fig. 5B). In addition, we identified hexanucleotide-type SSRs in both T. annua and T. lappula. For detailed information on all SSR types, including their start and end positions in the genome, see Table S3. Fig. 5Analysis of simple sequence repeats (SSRs) in the chloroplast genomes of Triumfetta. (A) Number of SSRs identified in the LSC, SSC, and IR regions. (B) Distribution and frequency of different SSR types. Analysis of simple sequence repeats (SSRs) in the chloroplast genomes of Triumfetta. (A) Number of SSRs identified in the LSC, SSC, and IR regions. (B) Distribution and frequency of different SSR types. Nucleotide substitution A decreasing nucleotide substitution frequency was observed across regions: the LSC region had the highest count (3,650–3,944), followed by the SSC (1,128–1,255) and then by the IRs (294–408) (Fig. 6A). A transition bias (Ts/Tv > 1) was observed across all regions, being most pronounced in the LSC (Ts/Tv: 1.29–1.35), followed by the SSC (Ts/Tv: 1.24–1.30) and then by the IRs (Ts/Tv: 1.06–1.38) (Fig. 6B). A/G and C/T transitions were the most frequent (Table S4). The Ts/Tv ratios based on protein-coding genes ranged from 1.544 to 1.682. This indicated a stronger transition bias in coding sequences. Fig. 6Comparative analysis of nucleotide substitutions among Triumfetta species. (A) Distribution of nucleotide substitutions across the LSC, SSC, and IR regions. (B) Transition-to-transversion (Ts/Tv) ratio among the three genomic regions. Comparative analysis of nucleotide substitutions among Triumfetta species. (A) Distribution of nucleotide substitutions across the LSC, SSC, and IR regions. (B) Transition-to-transversion (Ts/Tv) ratio among the three genomic regions. Nucleotide diversity and phylogenetic relationships Nucleotide diversity (π) analysis revealed that IGS regions exhibited the highest diversity (average π = 0.0161), followed by intronic regions (average π = 0.0100) and CDS (average π = 0.0075). Among CDS, clpP, rpl22, ycf1, rpl23, and ndhI showed the greatest variation, with π values ranging from 0.03706 to 0.01557. Within IGS regions, five polymorphic loci were identified, including rpl33-rps18, trnH-psbA, ccsA-ndhD, trnL-ccsA, and ndhD-psaC with π values ranging from 0.05596 to 0.04108. These regions displayed high levels of polymorphism but also generated substantial missing data (2.6–38.4%) due to indel events (Table 5). Notably, trnH-psbA showed the highest proportion of missing data (38.4%), followed by rpl33-rps18 (38.1%) and ccsA-ndhD (26.7%). All polymorphic regions are summarized in Table S5 and illustrated in Fig. 7. Table 5Polymorphic loci identified by comparative chloroplast genome analysis of Triumfetta species.Serial numberRegionNucleotide diversityNumber of substitutionsNumber of indelsLength (excluding indels)Alignment lengthMissing data (%)CDS region 1 clpP 0.0370668305916214.8 2 rpl22 0.021432604234230 3 ycf1 0.0180132896550255981.7 4 rpl23 0.017391262732792.1 5 ndhI 0.015573735015040.6IGS region 6 rpl33-rps18 0.055967528045473438.1 7 trnH-psbA 0.044613919931951838.4 8 ccsA-ndhD 0.042723410729340026.7 9 trnL-ccsA 0.042211441071113.6 10 ndhD-psaC 0.041081741451492.6 Table 5Polymorphic loci identified by comparative chloroplast genome analysis of Triumfetta species.Serial numberRegionNucleotide diversityNumber of substitutionsNumber of indelsLength (excluding indels)Alignment lengthMissing data (%)CDS region 1 clpP 0.0370668305916214.8 2 rpl22 0.021432604234230 3 ycf1 0.0180132896550255981.7 4 rpl23 0.017391262732792.1 5 ndhI 0.015573735015040.6IGS region 6 rpl33-rps18 0.055967528045473438.1 7 trnH-psbA 0.044613919931951838.4 8 ccsA-ndhD 0.042723410729340026.7 9 trnL-ccsA 0.042211441071113.6 10 ndhD-psaC 0.041081741451492.6 Polymorphic loci identified by comparative chloroplast genome analysis of Triumfetta species. Fig. 7Nucleotide diversity of Triumfetta chloroplast genome regions. (A) Nucleotide diversity of coding regions, including protein-coding, tRNA, and rRNA genes. (B) Nucleotide diversity of non-coding regions, comprising intergenic spacers and introns. Nucleotide diversity of Triumfetta chloroplast genome regions. (A) Nucleotide diversity of coding regions, including protein-coding, tRNA, and rRNA genes. (B) Nucleotide diversity of non-coding regions, comprising intergenic spacers and introns. Sequence alignment of the six inversion-region genes across all 14 Triumfetta species revealed distinct patterns of conservation and polymorphism. The two protein-coding genes, petN and psbM, each harbored one synonymous SNP in Clade Ia relative to all other species; translation of the aligned nucleotide sequences confirmed that these substitutions did not alter the encoded amino acid sequences (Figs. S2 and S3). Among the four tRNA genes, trnC-GCA exhibited three nucleotide differences in Clade Ia (Fig. S4), while trnD-GUC, trnY-GUA, and trnE-UUC were invariant across all 14 species. Phylogenetic reconstruction resolved Triumfetta into two strongly supported clades (100% bootstrap) (Fig. 8). Clade I was further subdivided into Clades Ia and Ib, each with 100% bootstrap support. Clade Ia comprised three Old World species (T. cordifolia, T. rhomboidea, T. tomentosa), whereas Clade Ib included four species, one from the New World (T. lappula) and three from the Old World (T. annua, T. cana, T. japonica). In contrast, Clade II consisted exclusively of Australian and New World species. Fig. 8Maximum-likelihood phylogenetic tree of Triumfetta species. Bootstrap support values below 100% are indicated. Species newly sequenced in this study are shown in boldface. Maximum-likelihood phylogenetic tree of Triumfetta species. Bootstrap support values below 100% are indicated. Species newly sequenced in this study are shown in boldface. Structural variation provided additional phylogenetic support. A large LSC inversion was detected in Clade Ia (T. cordifolia, T. rhomboidea, and T. tomentosa) but was absent in the remaining 11 species. This shared inversion among these three Old World taxa represents a synapomorphic character that reinforces their close relationship. Discussion Herbarium specimens are invaluable sources of taxonomic material, yet their utility for genomic studies is constrained by extensive DNA degradation and low yields inherent to aged, desiccated tissue45,46. In the present study, the modified extraction protocol yielded 0.12–0.31 µg of genomic DNA per specimen, sufficient for whole genome sequencing (WGS) library preparation. This success is attributable to the protocol modifications detailed in the Methods, particularly the increased lysis buffer volume, extended incubation, and dual-tube extraction strategy, which collectively improved DNA recovery from recalcitrant herbarium tissue. Indeed, genome skimming has been shown to succeed with as little as 500 pg of starting DNA from specimens47, and complete plastomes have been recovered from herbarium specimens46. Crucially, chloroplasts are present at 100–1,000 copies per cell, far exceeding the single nuclear genome copy, so that even low total DNA yields from degraded specimens contain a disproportionately high fraction of plastid reads in WGS data45,48. This copy-number advantage enabled high-coverage cp. genome assembly from all four herbarium specimens, demonstrating that herbarium collections remain a viable and largely untapped resource for plastome-scale phylogenomic studies in taxonomically under sampled genera such as Triumfetta49. The cp. genomes of the four newly sequenced Triumfetta species conform to the conserved quadripartite architecture universally observed in angiosperms, with genome dimensions, gene inventories, and GC profiles consistent with those reported for Triumfetta and other genera of Malvaceae17,20,23,25,50,51. The elevated GC content of the IR regions relative to the LSC and SSC is a universal feature of land plant cp. genomes, attributed to the presence of rRNA genes in the IR 2. This pattern has been reported across all examined Malvaceae, including Tilia20, Hibiscus51,52, Firmiana50. The most significant finding of the present study is the discovery that the six-gene LSC inversion, previously documented as a feature exclusive to T. tomentosa17, is in fact shared by three phylogenetically coherent Old World species: T. cordifolia, T. rhomboidea, and T. tomentosa. The shared presence of this rearrangement among species forming a monophyletic lineage (Clade Ia; bootstrap = 100%) demonstrates unambiguously that the inversion is synapomorphic; that is, it arose once in the common ancestor of Clade Ia and was vertically inherited by all three descendant species. The inverted segment comprises six functionally heterogeneous genes: four tRNA genes (trnC-GCA, trnD-GUC, trnY-GUA, and trnE-UUC) and two protein-coding genes (petN and psbM). The evolutionary persistence of this inversion across species occupying geographically disparate Old World habitats implies either functional neutrality at the transcriptional level or compensatory cis-regulatory changes within the inversion-bearing lineage that restored coordinated gene expression. No comparable inversion involving this six-gene block has been reported in any other Malvaceae genus examined to date, including the morphologically and phylogenetically proximate Grewioideae genera Corchorus, Grewia, and Colona20,23,24,53,54; this confirms that the rearrangement is strictly lineage-specific and did not predate the origin of Triumfetta. Sequence alignment of the six inverted genes revealed that petN and psbM carry Clade Ia-specific synonymous SNPs (Figs. S1 and S2), confirming that the inversion has not disrupted the reading frames or coding capacity of these genes. These clade-specific substitutions also indicate that nucleotide mutations have continued to accumulate in the inverted segment after the inversion event. Among the tRNA genes, only trnC-GCA showed a single-nucleotide difference in Clade Ia (Fig. S3), further supporting post-inversion mutation accumulation within this lineage. The functional integrity of all six genes thus appears uncompromised by the inversion event, consistent with the hypothesis of transcriptional neutrality discussed above. Large-scale structural inversions are rare events in cp. genome evolution55, and this pattern parallels the clade-defining synapomorphies reported in Styracaceae (Ericales)56 and in the columbine genus Aquilegia (Ranunculaceae)57, where structural characters resolved inter-clade relationships that nucleotide substitution data had left ambiguous. Because the inversion is absent from Clades Ib and II, its origin can be temporally constrained to after the divergence of Clade Ia from Clade Ib within the Old World crown of the genus, providing a non-homoplastic genomic character that anchors the deepest split within the African-centred subclade of Triumfetta. Additional sampling of unsequenced Triumfetta species can shed further light on the precise boundaries and phylogenetic distribution of this inversion. The IR boundary organization was broadly conserved across all 14 Triumfetta cp. genomes, with no gene transfer events detected at the LSC/IR or SSC/IR junctions and only minor length variation at the boundary-proximal genes rps19, rpl2, trnN, ndhF, ycf1, and trnH (Fig. 3), consistent with the structural stability reported for Firmiana50 and Hibiscus52 within Malvaceae. By contrast, IR expansion in Abelmoschus esculentus produced duplication of rps19, rpl22, rps3, and a pseudogenized rpl16, while IR contraction in Durio zibethinus returned rpl2 and rpl23 to single-copy status52. The absence of comparable events in Triumfetta, combined with near-identical IR boundary patterns across both major clades, indicates that IR-mediated expansion and contraction have been negligible within the genus. The clade-specific IR variation reported at fine phylogenetic scales in Berberis (Berberidaceae)6 and within populations of Medicago minima58 was likewise not observed, underscoring the genus-level stability of the Triumfetta cp. genome architecture. Leucine and isoleucine were the most abundant amino acids, and cysteine was the least abundant across all 14 Triumfetta cp. genomes, a conserved property of angiosperm cp. proteomes reflecting the combined influence of codon usage, gene function, and amino acid physicochemical properties2,17,28,52,59. RSCU analysis revealed a pronounced preference for A/T-ending codons across all species, consistent with findings across Malvaceae25,50,52, with no detectable difference between species carrying the six-gene LSC inversion and those lacking it, as petN and psbM contribute negligibly to the total coding pool of 78 genes. This bias reflects the well-established AT mutational pressure at synonymous (wobble) positions in angiosperm cp. genomes, where the cp. genome replication and repair machinery generates a systematic bias toward AT nucleotides, rather than active selection for translational efficiency60. Preferred codons in Triumfetta are consistently decoded by the cp-encoded tRNA complement (~ 30 genes), supporting co-evolutionary coupling of AT mutational pressure and tRNA gene content61,62. Simple sequence repeat analysis identified 63–81 SSRs per genome, with mononucleotide A/T repeats predominating, consistent with the AT-compositional bias documented across Malvaceae17,20,23,25,50,51 and other land plant cp. genomes. Mononucleotide and dinucleotide repeats, which evolve rapidly through replication slippage, are the most informative markers for population-genetic analyses63–65, whereas longer repeat units (tetranucleotide to hexanucleotide) are better suited for variety discrimination and phylogenetic marker development66. The notably higher SSR count in T. annua most likely reflects proportionally longer or more AT-rich non-coding spacers in its LSC region, as AT-rich templates are more susceptible to replication slippage, generating larger SSR inventories as a by-product of AT-pressure rather than selection61,67. Hexanucleotide SSRs were restricted to T. annua and T. lappula, which are phylogenetically non-sister within Clade Ib, most parsimoniously supporting two independent gain events via de novo tandem duplication of shorter repeat units. Transitions to transversions ratios vary considerably across angiosperm cp. genomes; values at or below unity have been reported in some lineages50, while ratios exceeding unity, as observed here, have been documented in others68. The transition bias in Triumfetta may be attributable to the preferential generation of transitions through spontaneous deamination of cytosine to uracil (C→T) and adenine to hypoxanthine (A→G) in cpDNA69, consistent with the predominance of C/T and A/G transitions observed across all compartments. This mutational bias is compounded by purifying selection against transversions, which are on average more disruptive, causing non-conservative amino acid substitutions in protein-coding genes and greater destabilization of secondary structures in rRNA and tRNA genes70,71. The markedly higher Ts/Tv in protein-coding genes (1.544–1.682) compared to the overall compartmental values further supports this interpretation, as functional constraint on coding sequences selectively removes transversions more efficiently than transitions. This bias represents an emergent property of cp. genome biology under specific combinations of mutational chemistry, functional constraint, and repair enzyme activity rather than a lineage-specific anomaly72. Nucleotide diversity analysis confirmed that intergenic spacer (IGS) regions are the most polymorphic compartment of the Triumfetta cp. genome (average π = 0.0161), followed by intronic regions (π = 0.0100) and coding sequences (π = 0.0075), consistent with the expected gradient from reduced functional constraint in non-coding sequences to strong purifying selection in CDS, a pattern documented across Malvaceae21,50,52,73 and other angiosperm families. Seven loci stood out as particularly informative: five CDS regions (clpP, rpl22, ycf1, rpl23, and ndhI) and two IGS (trnL-ccsA, ndhD-psaC). Not all candidates are equally applicable across research contexts. The high missing data rates for trnH-psbA (38.4%) and rpl33-rps18 (38.1%), attributable to extensive indel events that cause alignment gaps, are particularly problematic for herbarium-derived DNA, which is already fragmented and therefore prone to amplification failure in longer, indel-rich regions. For such specimens, trnL-ccsA and ndhD-psaC (low missing data; high π) and the five CDS loci are preferable primary candidates. Compared with the universal barcodes matK and rbcL, which suffer well-documented limitations in amplification reproducibility and phylogenetic resolution at shallow taxonomic depths7, loci such as clpP and rpl22 exhibited notably higher nucleotide diversity and no missing data, and represent more reliable markers for intrageneric discrimination in Triumfetta. The taxon-sampling dependence of locus performance, demonstrated in Erigeron74 and Hibiscus51, where the optimal locus set differed substantially between two-species and broad-taxon analyses, underscores the need for validation of these candidates across a wider Triumfetta species panel before their adoption as standard barcodes. Phylogenetic analysis resolved Triumfetta into two strongly supported major clades (bootstrap = 100%), with Clade I further divided into Clades Ia (bootstrap = 100%) and Ib (bootstrap = 100%), and with Corchorus as the confirmed sister genus, consistent with prior ITS-based analyses16,17. Structural variation provided independent, non-homoplastic corroboration of this topology: the six-gene synapomorphic inversion was present in all three Clade Ia species (T. cordifolia, T. rhomboidea, T. tomentosa) and absent in all 11 remaining species across Clades Ib and II (Table 1). This concordance between the sequence-based tree and the structural synapomorphy (two independent data sources) substantially reinforces the monophyly of Clade Ia and eliminates the possibility that the inversion’s phylogenetic signal results from homoplasy. The geographic structure of the two major clades illuminates the biogeographic history of the genus. Clade I is composed exclusively of Old World species: Clade Ia comprises African (T. cordifolia, Central Africa; T. tomentosa, pantropical Old World) and broadly distributed (T. rhomboidea, pan-tropical Old World) taxa, while Clade Ib adds Asian representatives (T. annua, T. cana, T. japonica) alongside the geographically anomalous New World species T. lappula. Clade II is composed predominantly of Australian and New World taxa. This pronounced Old World versus New World/Australasian dichotomy is consistent with a biogeographic scenario in which the ancestral Triumfetta lineage may have originated in Africa and subsequently diversified into Asia, with one or more independent long-distance dispersal events potentially seeding the New World and Australian lineages of Clade II; however, formal ancestral area reconstruction and divergence time estimation would be required to test this hypothesis. Africa as a possible center of diversity for the genus is further suggested by the position of the earliest-diverging lineages within Clade Ia. The placement of T. lappula (Americas) within Clade Ib rather than Clade II suggests that New World colonization by Triumfetta may have occurred on at least two independent occasions, although this inference requires corroboration through molecular dating and expanded taxon sampling. Taken together, these findings demonstrate that integration of structural cp. genome variation with nucleotide sequence data provides a more robust reconstruction of the evolutionary and biogeographic history of pantropical plant genera such as Triumfetta. Conclusion This study presents four newly sequenced Triumfetta cp. genomes and reveals a conserved quadripartite structure with a synapomorphic six-gene inversion (trnC, petN, psbM, trnD, trnY, trnE) in three species (T. cordifolia, T. rhomboidea, T. tomentosa). We identified seven highly polymorphic loci (trnL-ccsA, ndhD-psaC, clpP, rpl22, ycf1, rpl23, ndhI) suitable for DNA barcoding applications. Phylogenetic analysis resolved two major clades with strong biogeographic structure, supporting the phylogenetic utility of structural rearrangements. These genomic resources will facilitate taxonomic resolution, molecular marker development, and conservation genetics in Triumfetta. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1: Figure S1-S4. Supplementary Material 1: Figure S1-S4. Supplementary Material 2: Table S1. Supplementary Material 2: Table S1. Supplementary Material 3: Table S2. Supplementary Material 3: Table S2. Supplementary Material 4: Table S3. Supplementary Material 4: Table S3. Supplementary Material 5: Table S4. Supplementary Material 5: Table S4. Supplementary Material 6: Table S5. Supplementary Material 6: Table S5.