Comparative chloroplast genomics of 22 species of Sida (Malvaceae) reveals structural conservation, adaptive evolution, and preliminary phylogenetic structure Authors: Rushan Yan, Abdul Sammad, Sayed Afzal Shah, Atinga Vallada, Xiaoxuan Tian, Abdullah Published: 2026 ABSTRACT BackgroundThe genus Sida L. (Malvoideae, Malvaceae) comprises 275 species with a near-cosmopolitan distribution. Persistent taxonomic uncertainties, driven by morphological complexity and low resolution of traditional molecular markers, hinder systematic understanding of the genus. Chloroplast (cp) genomes provide genome-scale data capable of resolving these taxonomic uncertainties, but genomic resources for Sida remain limited. Here, we substantially expand cp genome sampling, perform comparative and selection analyses, identify candidate loci for species discrimination and phylogenetic inference, and conduct phylogenomic inference to clarify infrageneric relationships.ResultsWe sequenced and de novo assembled the complete cp genomes of Sida cordata (Burm.f.) Borss.Waalk. and Sida rhombifolia L., and assembled cp genomes of 18 additional species de novo from publicly available raw sequencing reads. Together with two previously reported cp genomes, 22 cp genomes were subjected to comparative and phylogenetic analyses. Genome sizes ranged from 159,706 to 160,513 base pairs (bp), comprising a large single-copy region (88,771–89,551 bp), inverted repeats (IRa/IRb; 25,288–25,640 bp), and a small single-copy region (19,863–20,294 bp). Each genome contained 113 unique genes (79 protein-coding, 30 tRNAs, 4 rRNAs). Comparative analyses revealed conserved gene order, guanine-cytosine content, codon usage, and simple sequence repeat composition, as well as a consistent transition bias in substitution patterns across all Sida species. Gene-level selection analyses identified episodic positive selection in psaB, psbB, rbcL, and rps8, while site-level analyses detected positive selection at specific codons in rbcL and ycf1. Ten highly polymorphic loci with elevated nucleotide diversity and low missing data were identified as candidates for species identification and phylogenetic inference, including psbT-psbN, ndhD-psaC, psaC-ndhE, rps15-ycf1, and psaI. Phylogenetic analysis of the 22 species resolved two strongly supported clades: a monophyletic endemic Australian lineage and a broadly distributed Asian/pantropical lineage.ConclusionsOur findings reveal conserved cp genome structure across 22 Sida cp genomes, identify polymorphic loci as potential barcodes, and resolve two clades (an endemic Australian lineage and a widespread Asian/pantropical lineage), demonstrating the value of whole-cp genome data. However, due to sampling limitations and sole reliance on the maternally inherited cp genome, we recommend that future studies refine biogeographic inferences using denser taxon sampling and nuclear genomic data.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12870-026-09208-z. Introduction The genus Sida L. (Malvoideae, Malvaceae), commonly known as fanpetals, comprises approximately 275 species of annual, biennial, and perennial herbs with a near-cosmopolitan distribution across tropical and subtropical regions [1]. These species exhibit remarkable ecological versatility, occupying habitats ranging from dry savannas to disturbed weedy landscapes [1, 2]. Beyond their ecological success, several species possess significant pharmacological value; for example, Sida cordata Burm. f. and Sida rhombifolia L. are widely used in Indian Ayurveda to treat respiratory and neural disorders [3]. Early taxonomic studies subdivided Sida into multiple sections based on morphological characters. In 1957, Clement established eight sections in his foundational monograph, which were later revised and expanded [4–8]. Currently, 17 sections are recognized [9]. However, molecular phylogenetic analyses have revealed that many morphologically defined sections are non-monophyletic, and Sida itself exhibits paraphyletic relationships with other genera in the tribe [10, 11]. Species identification within Sida has primarily relied on the ITS region and universal chloroplast (cp) markers, which provide insufficient resolution [12, 13]. This taxonomic complexity, combined with the limited resolving power of universal markers, underscores the need for expanded genomic resources and robust loci for DNA barcoding and phylogenetic reconstruction. The cp genome typically exhibits a conserved quadripartite structure, comprising a large single-copy (LSC) region, a small single-copy (SSC) region, and two inverted repeats (IRs: IRa and IRb) [14, 15]. Its structural conservation, generally maternal inheritance, lack of meiotic recombination, and moderate mutation rate make it a valuable marker for phylogenetic reconstruction, population genetics, and DNA barcoding [15–18]. Although the cp genome is highly conserved in structure and gene content, it accumulates evolutionary changes over time, including nucleotide substitutions, insertions and deletions (indels), inversions, and IR boundary shifts, which can alter gene order, content, and genome architecture [15, 19–21]. Many of these substitutions and indels are associated with repetitive sequences within the genome [20]. In this study, the complete cp genomes of Sida cordata and Sida rhombifolia were sequenced and de novo assembled, and cp genomes of 18 additional Sida species were assembled from publicly available sequencing data on the National Center for Biotechnology Information (NCBI). Two previously published cp genomes, S. szechuensis (NC051877) and S. fallax (NC080995), were also incorporated, yielding a comparative dataset of 22 Sida cp genomes, the most comprehensive sampling of the genus to date and the maximum coverage achievable from currently available genomic resources. Importantly, this dataset spans both major phylogenetic lineages within the genus: (1) an arid-adapted Australian lineage comprising S. arsiniata, S. cardiophylla, S. echinocarpa, S. fibulifera, S. trichopoda, and several undescribed accessions from the Pilbara, Hamersley Range, and Barlee Range of Western Australia; and (2) a broadly distributed Asian and pan-tropical clade comprising S. rhombifolia (tropical and subtropical Old World), S. spinosa (pantropical), S. cordata (Socotra and tropical Asia), S. szechuensis (southern China to northern Vietnam), and S. fallax (Pacific islands). This combined dataset thus captures deep phylogenetic divergence across ecologically contrasting environments although it represents only ~ 8% of the genus taxonomically. Detailed information on species names, native distributions, and data sources is provided in Table S1. The objectives of this study were to: (i) characterize cp genome structure and variation through comparative genomic analyses; (ii) identify highly polymorphic loci with elevated nucleotide diversity suitable for species identification and DNA barcoding; and (iii) reconstruct preliminary phylogenetic relationships among the 22 Sida species and evaluate phylogeographic patterns inferred from cp genome data. Materials and methods Sample collection, DNA extraction, and sequencing Fresh leaves of Sida cordata and Sida rhombifolia were collected from Quaid-i-Azam University, Islamabad, Pakistan (33°44′54.7″N, 73°08′04.3″E) and taxonomically identified by Dr Sayed Afzal Shah. Voucher specimens were deposited in the herbarium of the National University of Medical Sciences under accession numbers NUMS00065 (S. cordata) and NUMS00066 (S. rhombifolia). No permits were required from local or national authorities for plant collection or subsequent research use. Leaves were immediately dried with silica gel to preserve DNA integrity. Total genomic DNA was extracted from 15 mg of dried tissue using the Plant Genomic DNA Kit (TIANGEN Biotech, Beijing, China), following the manufacturer’s instructions. DNA integrity was verified at Annoroad Genomics (Beijing, China) by electrophoresis on a 1% agarose gel run at 120 V for 45 min. DNA quality and concentration were assessed using a NanoDrop spectrophotometer and further quantified with a Qubit fluorometer. Sequencing libraries with an insert size of 350 bp were constructed from high-quality DNA, and paired-end sequencing (2 × 150 bp) was performed on the Illumina NovaSeq 6000 platform. Chloroplast genome assembly and annotation Raw sequencing reads were quality-filtered using fastp v1.0.1 [22]. Paired-end reads were discarded if either read contained more than 10% ambiguous bases (N) or if more than 50% of bases had a quality score ≤ 5. Clean reads were used for de novo cp genome assembly with GetOrganelle v1.7.7.1 [23], employing multiple k-mer sizes (21, 45, 65, 85, and 105). To expand the comparative dataset, raw sequencing data for 18 additional Sida species were downloaded from NCBI and de novo assembled using the described approach. Annotations of all twenty genomes were performed using GeSeq v2.0.3 [24]. Transfer RNA genes were further predicted and verified using both ARAGORN v1.2.38 [25] and tRNAscan-SE v2.0.7 [26]. Start and stop codons of protein-coding genes were manually curated by comparing them with homologous genes from S. szechuensis (NC051877) and S. fallax (NC080995) in Geneious Prime 2025. To validate assembly accuracy, cleaned reads of each species were mapped back to the assembled genomes to assess coverage depth and detect potential assembly errors in Geneious Prime 2025. A circular genome map was generated using Chloroplot v1 [27]. Submission files for NCBI submission were prepared from GenBank files using a custom Python script (Script 1), which is now integrated into Chloroplast Genome Analysis Suite (CGAS v1.0.1) as module 4 [28]. Comparative chloroplast genomics of Sida species For comparative analyses, the complete cp genomes of S. szechuensis and S. fallax retrieved from NCBI were combined with the 20 cp genomes assembled in this study. Genome characteristics, including total genome length, sizes of the LSC, SSC, and IR regions, gene content, GC content, and intron number and length, were calculated using the CGAS v1.0.1 [28] and subsequently manually verified in Geneious Prime 2025. Genome structural variation and potential inversions were examined using collinear block analysis with the Mauve progressive aligner [29]. Inverted repeat boundary variation was further examined by visualizing junctions between IR and single-copy regions using CPJSdraw v0.0.1 [30]. Gene maps of cis- and trans-splicing intron–containing genes were illustrated using CPGview v1 [31], while intron features across all species—including intron number, start and end positions, and intron length—were analyzed using CGAS v1.0.1. Codon usage, amino acid frequency, microsatellite analysis, and substitution analysis Relative synonymous codon usage (RSCU), amino acid frequencies, simple sequence repeats (SSRs), and nucleotide substitutions were analyzed using CGAS v1.0.1. SSRs were identified based on the following minimum repeat thresholds: 10 for mononucleotide repeats, 5 for dinucleotide repeats, 4 for trinucleotide repeats, and 3 for tetra-, penta-, and hexanucleotide repeats. To validate SSR detection, CGAS-derived results were cross-checked against MISA-web [32], and consistent outputs were obtained. In addition to SSR identification, CGAS provides detailed positional information, including the distribution of SSRs across coding sequences (CDS), intronic regions, and intergenic spacers (IGS), as well as their localization within the LSC, SSC, and IR regions. Given the concordance between CGAS and MISA-web results, and the integrated annotation framework of CGAS, only CGAS-based SSR results were presented in this study. The LSC, SSC, and IR regions of all Sida cp genomes were extracted and aligned pairwise against S. cordata to assess nucleotide substitution patterns. S. cordata was selected as the reference genome because preliminary phylogenetic analysis indicated it occupied a relatively basal position within the sampled taxa. Transition and transversion substitutions, as well as their ratios, were calculated using CGAS. Analysis of adaptive evolution on protein-coding genes Selective pressure on protein-coding genes was evaluated using HyPhy v2.5.7 [33]. Codon-based alignments for each protein-coding gene were generated using the MACSE v2 [34] aligner across all 22 Sida cp genomes. For each gene, a corresponding phylogenetic tree was reconstructed using FastTree [35] under the GTR + G substitution model and used as input for subsequent selection analyses. Gene-wide episodic positive selection was assessed using the branch-site unrestricted statistical test for episodic diversification (BUSTED) [36], whereas site-level episodic and pervasive selection were evaluated using the Mixed Effects Model of Evolution (MEME) [37] and the Fast Unconstrained Bayesian Approximation (FUBAR) [38], respectively. GenBank files parsing, gene extraction and alignment, execution of selection tests, application of multiple-testing corrections, and generation of result tables in Excel format were performed using a custom Python script (Script 2). Analysis of polymorphic loci and phylogenetic inference Nucleotide diversity (π) across coding and non-coding regions of Sida cp genomes was calculated using CGAS v1.0.1 [28] to identify suitable polymorphic loci. For phylogenetic inference, 78 shared protein-coding genes of Sida species, along with the outgroup Malva phoenicea (NC063266), were extracted from GenBank files using a custom Python script 3 implemented in the CGAS [28] pipeline (based on Biopython). Each gene was aligned separately using MAFFT v7.525 [39] with the --auto strategy. The aligned gene sequences were then concatenated into a single supermatrix, which was subsequently inspected and manually curated in Geneious Prime 2025 to remove ambiguously aligned regions and correct potential alignment inconsistencies. Maximum likelihood (ML) phylogenetic analysis was conducted using IQ-TREE v3.0.1 [40], with the best-fit nucleotide substitution model (TVM + F + I) selected using ModelFinder (implemented within IQ-TREE v3.0.1) under the Bayesian Information Criterion [41]. Branch support was evaluated using 10,000 ultrafast bootstrap replicates with -bnni optimization [42] and 10,000 SH-aLRT replicates. The same concatenated alignment was used for Bayesian inference (BI) in MrBayes v3.2.7 under the GTR + I + Γ model (nst = 6, rates = invgamma) [43]. Two independent runs, each with four Markov chains, were performed for 10 million generations, with sampling every 1,000 generations. Convergence was assessed based on the average standard deviation of split frequencies and potential scale reduction factor (PSRF) values, and the first 25% of sampled trees were discarded as burn-in. The resulting consensus trees were visualized and annotated using the Interactive Tree of Life (iTOL) platform [44]. We provided the results of ML phylogeny as data file 1 and MrBayes as data file 2. In addition, species distribution data for Sida were compiled from Plants of the World Online (POWO: https://powo.science.kew.org/) and the NCBI, and are provided in Table S1. Results Genome sequencing and de novo assembly of genomes Whole-genome sequencing of S. cordata and S. rhombifolia generated 6.89–7.85 GB of data (20.01–22.82 million paired-end reads). Of these, 0.34–0.46 million reads were cp-derived, resulting in high coverage depths of 321 × and 435 × for the de novo assembled cp genomes of the two species (Table 1). In addition, 18 Sida cp genomes were de novo assembled from publicly available SRA datasets, with data sizes ranging from 0.9 to 2.35 GB. These datasets yielded approximately 0.02–0.45 million cp-derived reads, resulting in coverage depths of 21 × to 356 × (Table 1), which provided sufficient data quality for subsequent comparative genomic analyses. Only four species exhibited coverage depths below 50 × (Table 1); however, all assemblies were validated by read remapping and showed consistent coverage across the genome.Table 1Accession numbers, quantity of raw data, and coverage depth of de novo assembled chloroplast genomesSpeciesSRA accession noData in GBCoverage depthWGS reads (millions)Chloroplast reads (millions)Sida arsiniata R.M.BarkerSRR86665971.80374.910.04Sida cardiophylla F.MuellSRR86665961.291353.540.17Sida clementii DominSRR86665951.23613.380.08Sida echinocarpa F.MuellSRR86664241.94455.310.06Sida ectogama W.R.Barker & R.M.BarkerSRR86665941.60674.360.08Sida fibulifera LindlSRR86666001.64954.480.12Sida platycalyx F.Muell. ex BenthSRR86664571.591284.360.16Sida rohlenae subsp. rohlenaeSRR86664591.823564.990.45Sida sp. Articulation belowSRR86664581.92535.260.06Sida sp. Barlee RangeSRR86664600.90212.410.02Sida sp. dark green fruitsSRR86664561.77604.850.07Sida sp. Hamersley RangeSRR86664231.76332.410.04Sida sp. PilbaraSRR86661992.35746.420.09Sida sp. Shovelanna HillSRR86664611.30653.580.08Sida sp. spiciform paniclesSRR86666011.22853.330.10Sida sp. verrucose glandsSRR86665921.751004.800.12Sida spinosa LSRR86665931.061712.910.21Sida trichopoda F.MuellSRR86667772.091935.700.24Sida rhombifolia LSRR368017606.8943520.010.46Sida cordata (Burm.f.) Borss.WaalkSRR368017617.8532122.820.34 Table 1Accession numbers, quantity of raw data, and coverage depth of de novo assembled chloroplast genomesSpeciesSRA accession noData in GBCoverage depthWGS reads (millions)Chloroplast reads (millions)Sida arsiniata R.M.BarkerSRR86665971.80374.910.04Sida cardiophylla F.MuellSRR86665961.291353.540.17Sida clementii DominSRR86665951.23613.380.08Sida echinocarpa F.MuellSRR86664241.94455.310.06Sida ectogama W.R.Barker & R.M.BarkerSRR86665941.60674.360.08Sida fibulifera LindlSRR86666001.64954.480.12Sida platycalyx F.Muell. ex BenthSRR86664571.591284.360.16Sida rohlenae subsp. rohlenaeSRR86664591.823564.990.45Sida sp. Articulation belowSRR86664581.92535.260.06Sida sp. Barlee RangeSRR86664600.90212.410.02Sida sp. dark green fruitsSRR86664561.77604.850.07Sida sp. Hamersley RangeSRR86664231.76332.410.04Sida sp. PilbaraSRR86661992.35746.420.09Sida sp. Shovelanna HillSRR86664611.30653.580.08Sida sp. spiciform paniclesSRR86666011.22853.330.10Sida sp. verrucose glandsSRR86665921.751004.800.12Sida spinosa LSRR86665931.061712.910.21Sida trichopoda F.MuellSRR86667772.091935.700.24Sida rhombifolia LSRR368017606.8943520.010.46Sida cordata (Burm.f.) Borss.WaalkSRR368017617.8532122.820.34 Accession numbers, quantity of raw data, and coverage depth of de novo assembled chloroplast genomes Comparative analysis of Sida chloroplast genome features Complete cp genomes were successfully assembled for 20 Sida species. Two additional genomes, S. szechuensis (NC051877) and S. fallax (NC080995), were retrieved from NCBI, resulting in a comparative dataset of 22 genomes. These genomes exhibited conserved genome structure, and the total genome length ranged from 159,706 to 160,513 bp (Table 2). The LSC, SSC, and IR regions were 88,771–89,551 bp, 19,863–20,294 bp, and 25,288–25,640 bp, respectively (Fig. 1). Overall GC content was highly conserved (36.83–36.97%), with limited regional variation: LSC (34.64–34.77%), SSC (31.29–31.56%), and IR (42.81–43.05%) (Table 2).Table 2Comparative analysis of chloroplast genome features in Sida speciesGenome Length (bp)GC (%)CompleteLSCSSCIRCompleteLSCSSCIRtRNArRNACDSSida arsiniata160,31088,93920,23525,56836.9234.6931.4442.9853.2555.4238.10Sida cardiophylla160,35388,90320,29225,57936.9234.6931.4642.9553.2555.4238.08Sida clementii160,35688,87120,25925,61336.9334.7431.4942.9053.2555.3938.10Sida cordata160,49989,16520,11825,60836.9634.7731.4242.9653.2455.5438.16Sida echinocarpa160,18688,92320,23325,51536.9334.6731.4443.0553.2555.4238.10Sida ectogama160,30088,99920,25325,52436.9134.6431.4743.0453.2555.4238.09Sida fallax160,05889,55119,86325,32236.8834.6831.5642.8553.1655.4538.14Sida fibulifera160,36488,88820,26025,60836.9434.7431.5042.9053.2555.4238.11Sida platycalyx160,42788,99820,26125,58436.9334.7131.4842.9453.2555.4238.08Sida rhombifolia159,70689,21819,87625,30636.9134.7331.5542.8553.1655.4538.12Sida rohlenae subsp rohlenae160,07489,37719,98725,35536.8334.6831.2942.8153.1355.4338.11Sida sp Articulation below160,31188,90220,26325,57336.9434.7231.4742.9653.2555.4238.09Sida sp Barlee Range160,29388,88820,25725,57436.9634.7631.4942.9653.2555.4238.09Sida sp dark green fruits160,22388,94920,24425,51536.9534.7031.5143.0553.2555.4238.10Sida sp Hamersley Range160,27388,77120,22125,64036.9734.7731.4842.9653.2555.4238.12Sida sp Pilbara160,51389,07820,29325,57136.9334.7131.4542.9753.2555.4238.10Sida sp Shovelanna Hill160,48989,17120,29425,51236.9234.6631.4443.0553.2555.4238.10Sida sp spiciform panicles160,26288,98320,25525,51236.9434.6931.4743.0453.2555.4238.08Sida sp verrucose glands160,33888,92220,26425,57636.9534.7431.4942.9653.2555.4238.10Sida spinosa159,78989,24819,90725,31736.8734.7031.4442.8253.2055.4238.11Sida szechuensis159,87889,42619,87625,28836.8834.6831.5142.8953.1655.4538.11Sida trichopoda160,21288,80220,27225,56936.9534.7531.4642.9553.2555.4238.11GC guanine-cytosine content, LSC large single-copy region, SSC small single-copy region, IR inverted repeat region, Complete complete chloroplast genome, tRNA transfer RNA, rRNA ribosomal RNA, CDS protein-coding geneFig. 1Circular chloroplast genome map of Sida. Genes transcribed clockwise are displayed inside the circle, and those transcribed counterclockwise are placed outside. Genes are color-coded according to their functional categories. The darker gray band in the inner circle indicates the GC content Table 2Comparative analysis of chloroplast genome features in Sida speciesGenome Length (bp)GC (%)CompleteLSCSSCIRCompleteLSCSSCIRtRNArRNACDSSida arsiniata160,31088,93920,23525,56836.9234.6931.4442.9853.2555.4238.10Sida cardiophylla160,35388,90320,29225,57936.9234.6931.4642.9553.2555.4238.08Sida clementii160,35688,87120,25925,61336.9334.7431.4942.9053.2555.3938.10Sida cordata160,49989,16520,11825,60836.9634.7731.4242.9653.2455.5438.16Sida echinocarpa160,18688,92320,23325,51536.9334.6731.4443.0553.2555.4238.10Sida ectogama160,30088,99920,25325,52436.9134.6431.4743.0453.2555.4238.09Sida fallax160,05889,55119,86325,32236.8834.6831.5642.8553.1655.4538.14Sida fibulifera160,36488,88820,26025,60836.9434.7431.5042.9053.2555.4238.11Sida platycalyx160,42788,99820,26125,58436.9334.7131.4842.9453.2555.4238.08Sida rhombifolia159,70689,21819,87625,30636.9134.7331.5542.8553.1655.4538.12Sida rohlenae subsp rohlenae160,07489,37719,98725,35536.8334.6831.2942.8153.1355.4338.11Sida sp Articulation below160,31188,90220,26325,57336.9434.7231.4742.9653.2555.4238.09Sida sp Barlee Range160,29388,88820,25725,57436.9634.7631.4942.9653.2555.4238.09Sida sp dark green fruits160,22388,94920,24425,51536.9534.7031.5143.0553.2555.4238.10Sida sp Hamersley Range160,27388,77120,22125,64036.9734.7731.4842.9653.2555.4238.12Sida sp Pilbara160,51389,07820,29325,57136.9334.7131.4542.9753.2555.4238.10Sida sp Shovelanna Hill160,48989,17120,29425,51236.9234.6631.4443.0553.2555.4238.10Sida sp spiciform panicles160,26288,98320,25525,51236.9434.6931.4743.0453.2555.4238.08Sida sp verrucose glands160,33888,92220,26425,57636.9534.7431.4942.9653.2555.4238.10Sida spinosa159,78989,24819,90725,31736.8734.7031.4442.8253.2055.4238.11Sida szechuensis159,87889,42619,87625,28836.8834.6831.5142.8953.1655.4538.11Sida trichopoda160,21288,80220,27225,56936.9534.7531.4642.9553.2555.4238.11GC guanine-cytosine content, LSC large single-copy region, SSC small single-copy region, IR inverted repeat region, Complete complete chloroplast genome, tRNA transfer RNA, rRNA ribosomal RNA, CDS protein-coding gene Comparative analysis of chloroplast genome features in Sida species GC guanine-cytosine content, LSC large single-copy region, SSC small single-copy region, IR inverted repeat region, Complete complete chloroplast genome, tRNA transfer RNA, rRNA ribosomal RNA, CDS protein-coding gene Circular chloroplast genome map of Sida. Genes transcribed clockwise are displayed inside the circle, and those transcribed counterclockwise are placed outside. Genes are color-coded according to their functional categories. The darker gray band in the inner circle indicates the GC content All Sida cp genomes contained 113 unique genes, including 79 protein-coding sequences (including the infA pseudogene), 30 tRNA genes, and 4 rRNA genes, and exhibited a highly conserved gene order (Table S2). Sixteen genes were duplicated in the IR regions, comprising 5 CDS, 7 tRNA genes, and all 4 rRNA genes, resulting in a total of 129 genes (Table S3). Eighteen genes contained introns, including 12 CDS and 6 tRNA genes, and these intron-containing genes were conserved across species. Among protein-coding genes, ycf3 and clpP each contained two introns, whereas the remaining nine genes harbored a single intron (Fig. 2A). Notably, rps12 was trans-spliced, with exon 1 located in the LSC region and exons 2 and 3 duplicated within the IR regions (Fig. 2B).Fig. 2Representation of protein coding genes that contain introns in the chloroplast genome. A Genes with cis-spliced introns. B The gene with a trans-spliced intron Representation of protein coding genes that contain introns in the chloroplast genome. A Genes with cis-spliced introns. B The gene with a trans-spliced intron Intron lengths varied among species. The rps12 intron was the shortest and most conserved (536 bp), whereas rpl16 exhibited the greatest length variation (952–1188 bp). The longest intron occurred in ndhA (1114–1171 bp). For clpP, intron I ranged from 648 to 662 bp and intron II from 885 to 936 bp, while ycf3 introns ranged from 764 to 785 bp (intron I) and 755 to 795 bp (intron II) (Table 3, Table S4).Table 3Length variation of chloroplast genes and introns across Sida speciesMinimum length (bp)Maximum length (bp)atpF766812clpP Intron I648662clpP Intron II885936ndhA11141171ndhB682683petB817830petD737760rpl2678687rpl169521188rpoC1764825rps12536536rps16870979ycf3 Intron I764785ycf3 Intron II755795trnA-UGC798798trnG-UCC781850trnI-GAU956969trnK-UUU25582596trnL-UAA556609trnV-UAC588596 Table 3Length variation of chloroplast genes and introns across Sida speciesMinimum length (bp)Maximum length (bp)atpF766812clpP Intron I648662clpP Intron II885936ndhA11141171ndhB682683petB817830petD737760rpl2678687rpl169521188rpoC1764825rps12536536rps16870979ycf3 Intron I764785ycf3 Intron II755795trnA-UGC798798trnG-UCC781850trnI-GAU956969trnK-UUU25582596trnL-UAA556609trnV-UAC588596 Length variation of chloroplast genes and introns across Sida species Multiple sequence alignment with Mauve revealed a high degree of collinearity, with no evidence of large-scale inversions (Figure S1). The consistent arrangement of locally collinear blocks (LCBs) across all species supported structural conservation. IR contraction and expansion showed minor variation at the four junction sites—LSC/IRb (JLB), IRb/SSC (JSB), SSC/IRa (JSA), and IRa/LSC (JLA)—across the 22 Sida species (Fig. 3). The rps19 gene was generally located at the JLB junction, initiating within the IRb region and extending toward the LSC. However, in seven species (Sida echinocarpa, Sida ectogama, Sida rohlenae subsp. rohlenae, Sida sp. Articulation below, Sida sp. dark green fruits, Sida sp. Shovelanna Hill, and Sida sp. spiciform panicles), rps19 was entirely located within the LSC region. This minor IR contraction is not associated with monophyly and appears to represent convergent evolution or homoplasy, having arisen independently in multiple lineages.Fig. 3Comparison of inverted repeat expansion and contraction among Sida species. Boundary junctions are labeled as JLB (LSC/IRb), JSB (IRb/SSC), JSA (SSC/IRa), and JLA (IRa/LSC) Comparison of inverted repeat expansion and contraction among Sida species. Boundary junctions are labeled as JLB (LSC/IRb), JSB (IRb/SSC), JSA (SSC/IRa), and JLA (IRa/LSC) The rpl2 gene was duplicated in the IRb and IRa regions, with distances from the JLB and JLA junctions ranging from 36 bp in Sida sp. Shovelanna Hill to 176 bp in Sida spinosa. The trnN gene was also duplicated in both IR regions and was positioned 348–386 bp away from the JSB and JSA junctions. The ndhF gene was located within the SSC region, with distances from the JSB junction ranging from 59 to 167 bp. The ycf1 gene spanned the IR–SSC boundary, initiating in the IR region (31–69 bp) and extending into the SSC. Finally, the trnH gene was located at the JLA junction. Relative synonymous codon usage, amino acid frequency, microsatellites, and substitution analysis Codon usage was analyzed across all 22 Sida cp genomes (Table S5). Relative synonymous codon usage (RSCU) revealed 30 codons with RSCU values > 1. Among these 30 preferred codons, 13 ended in A and 16 ended in T, while only 1 ended in G (Table S5), reflecting a general preference for codons ending in A or T over those ending in C or G. Codons with no synonymous alternatives, ATG (Met) and TGG (Trp), showed no usage bias. Amino acid frequency was highly conserved among species, with leucine being the most abundant (10.58–10.63%), followed by isoleucine (8.58–8.63%), and cysteine the least frequent (1.14–1.15%). Overall, the distribution of all 20 amino acids was remarkably similar across genomes (Table S6). Simple sequence repeats (SSRs) were detected in all species, ranging from 57 to 90 per genome (Table S7). The majority of SSRs were located in the LSC region (36–72), followed by the SSC (8–16) and IRs (2–8). Most repeats were found within IGS regions (40–59), followed by introns (6–21), and then CDS (11–14). Mononucleotide repeats were the most abundant (35–67), followed by dinucleotide (8–12) and tetranucleotide repeats (4–9). Despite variation in SSR counts, the distribution of repeat motif types and their A/T-rich nucleotide composition were highly conserved across species (Table S7). Nucleotide substitution analysis revealed that the LSC region harbored the highest number of substitutions (1,346–1,702), followed by the SSC (400–500) and IR regions (102–120) (Table S8). Transition-to-transversion (Ts/Tv) ratios indicated a bias toward transitions, with the highest ratio in the SSC (1.494–1.881), followed by IRs (1.400–1.757) and LSC (1.264–1.412). The most frequent substitutions were A → G and C → T, although their occurrence varied across species and genome regions (Table S8). Evaluation of adaptive evolution Gene-wide selection analysis using BUSTED detected four genes under episodic positive selection (psaB, psbB, rbcL, rps8) across the 22 Sida cp genomes (Table S9). Site-specific analyses revealed variable patterns depending on the method applied. MEME initially identified up to 15 codons under episodic positive selection across several genes. However, after applying a global false discovery rate (FDR) correction, only two codon sites remained significant (FDR < 0.05): one in rbcL and one in ycf1. (Table S10). FUBAR analysis detected additional codons under pervasive selection (posterior probability PP > 0.9) distributed across multiple genes, including accD, atpH, matK, ndhF, ndhK, rbcL, rpl16, rpl20, rpl22, rpl23, rpoC1, rpoC2, rps15, ycf1, and ycf2. Notably, some genes identified as positively selected by BUSTED, such as psaB and psbB, did not yield significant sites in MEME, whereas FUBAR captured additional low-signal codons. These results indicate differences in detection between gene-wide and site-specific methods, as well as between episodic and pervasive selection analyses, while highlighting rbcL and ycf1 as consistently supported targets of positive selection. Nucleotide polymorphism and phylogenetic analysis Nucleotide diversity (π) was calculated for CDS, tRNA, rRNA, intronic, and IGS regions based on comparative cp genome analysis using CGAS. The analysis showed that the IGS regions exhibited the highest average nucleotide diversity (π = 0.0137), followed by the intronic region (π = 0.0082), and then the CDS region (π = 0.0057). We identified 10 polymorphic regions suitable for phylogenetic inference at the Sida genus level, comprising 6 IGS regions, 3 protein-coding sequences, and 1 intronic region (Table 4). Regions with high levels of missing data (> 5%), such as trnH-psbA and rps19-rpl2, were excluded from the list of suitable polymorphic loci due to multiple indels (Table S11). Among these candidate regions, ycf1 was the longest, while psbT-psbN was the shortest. Detailed information on all polymorphic regions is provided in Table S12 and Fig. 4.Table 4Polymorphic loci identified by comparative plastome analysis of Sida speciesSerial numberRegionNucleotide diversityNumber of substitutionsNumber of indelsLength (excluding indels)Alignment lengthMissing data (%)1psbT-psbN0.027226166671.42ndhD-psaC0.0245541161251314.53psaC-ndhE0.0189812272472542.74rps15-ycf10.0176742483633712.15psaI0.0165198011411406ycf10.016184377105565557601.87atpI-rps20.0151921142122161.88ndhJ-ndhK0.015005841331372.99trnK-UUU intron0.011523133114253126454.310ndhI0.01109716174975143.3Fig. 4Nucleotide diversity (π) across different regions of the chloroplast genome. A Coding regions, including protein-coding genes, tRNA genes, and rRNA genes. B Non-coding regions, including intergenic spacers and introns. The ten most suitable polymorphic loci are highlighted with red circles Table 4Polymorphic loci identified by comparative plastome analysis of Sida speciesSerial numberRegionNucleotide diversityNumber of substitutionsNumber of indelsLength (excluding indels)Alignment lengthMissing data (%)1psbT-psbN0.027226166671.42ndhD-psaC0.0245541161251314.53psaC-ndhE0.0189812272472542.74rps15-ycf10.0176742483633712.15psaI0.0165198011411406ycf10.016184377105565557601.87atpI-rps20.0151921142122161.88ndhJ-ndhK0.015005841331372.99trnK-UUU intron0.011523133114253126454.310ndhI0.01109716174975143.3 Polymorphic loci identified by comparative plastome analysis of Sida species Nucleotide diversity (π) across different regions of the chloroplast genome. A Coding regions, including protein-coding genes, tRNA genes, and rRNA genes. B Non-coding regions, including intergenic spacers and introns. The ten most suitable polymorphic loci are highlighted with red circles The phylogenetic analysis resolved Malva phoenicea (NC063266) as the outgroup. All Sida accessions formed a monophyletic ingroup that was divided into two major clades, which were separated by a strongly supported node (BS = 100%) (Fig. 5). Clade I comprised all Australian Sida accessions and was recovered as a monophyletic group with strong support (BS = 100%). Internal relationships within this clade showed a mixture of strong and weak bootstrap support. At the base of Clade I, Sida sp. Hamersley Range occupied a strongly supported basal position (BS = 100%) and exhibited a relatively long branch length (0.0127), indicating substantial divergence from other Australian taxa. Sida cardiophylla was placed as a sister to the remaining taxa with strong support (BS = 87%). Sida sp. Articulation below formed a distinct lineage (BS = 98%) that was sister to a well-supported subclade. Within this subclade, Sida clementii and Sida fibulifera formed a sister pair with moderate support (BS = 59%). The Sida clementii–Sida fibulifera pair clustered with Sida sp. verrucose glands with strong support (BS = 100%). This assemblage was subsequently associated with Sida sp. dark green fruits (BS = 90%) and Sida platycalyx (BS = 63%). Sida sp. Barlee Range was placed as a distinct lineage with strong support (BS = 87%) and showed affinity to taxa from Sida sp. Shovelanna Hill and Sida sp. Pilbara, which formed a weakly supported pair (BS = 48%). The placement of this geographic assemblage within Clade I received strong support (BS = 95%). Sida trichopoda was resolved as a distinct lineage within Clade I with strong support (BS = 100%). A strongly supported sister relationship was observed between Sida arsiniata and Sida echinocarpa (BS = 100%), which together grouped with Sida ectogama with moderate support (BS = 72%). This assemblage showed a weaker association with Sida sp. spiciform panicles (BS = 66%). Overall, Clade I represented a single Australian lineage with variable bootstrap support across internal nodes, reflecting shallow divergence among several closely related taxa.Fig. 5Maximum-likelihood phylogenetic inference of 22 Sida species. All nodes with bootstrap support values ≤ 99% are shown. Species of Sida are divided into two clades (Clade I and Clade II), which are indicated by different colors Maximum-likelihood phylogenetic inference of 22 Sida species. All nodes with bootstrap support values ≤ 99% are shown. Species of Sida are divided into two clades (Clade I and Clade II), which are indicated by different colors Clade II consisted of Asian and broadly distributed Sida species and was recovered as a monophyletic group with strong support (BS = 100%). Sida cordata occupied a more divergent position within Clade II and formed a sister lineage to the remaining taxa with strong support (BS = 100%), consistent with its longer branch length (0.0448). Within the remaining Clade II taxa, Sida szechuensis (NC051877) and Sida rhombifolia formed a strongly supported sister pair (BS = 100%). This pair clustered with Sida fallax (NC080995) with strong support (BS = 100%). The resulting lineage was further associated with Sida spinosa with high support (BS = 99%). Sida rohlenae subsp. rohlenae grouped with these taxa with strong support (BS = 100%). Bayesian inference yielded a topology fully congruent with the ML tree, with the majority of nodes receiving posterior probabilities (PP) ≥ 0.95. Several relationships that showed moderate bootstrap support in ML analysis (e.g., BS = 55%, 59%, 66%, 72%, 87%) received stronger support under Bayesian inference (PP > 0.95) (Figure S2). Discussion Chloroplast genome features and comparative analysis of Sida species In this study, the cp genomes of 20 Sida species were de novo assembled and analyzed together with two publicly available cp genomes, yielding a comparative dataset of 22 cp genomes. Overall genome size, gene content, gene order, and GC content were highly conserved, consistent with cp genome structure reported for other Malvaceae lineages [19, 20, 45–47]. Whole-genome collinearity analyses detected no large-scale rearrangements or inversions, indicating pronounced structural stability across Sida [19, 45, 46, 48]. Such conservatism is characteristic of most angiosperm cp genomes and is commonly attributed to several molecular and biological features, including the organization of cp genome genes into operons, predominantly uniparental (usually maternal) inheritance, the presence of efficient DNA repair mechanisms, and the rarity of fusion events, all of which collectively constrain large-scale structural change [49]. Nevertheless, departures from this conserved architecture have been documented in multiple plant lineages. Extensive cp genome rearrangements, inversions, and gene losses have been reported in ferns, Cyperaceae, and Thymelaeaceae [15, 50, 51], illustrating that cp genome structure can be evolutionarily labile under certain genomic or ecological contexts. Even within Malvaceae, localized structural variation has been observed; for example, a small synapomorphic inversion involving six genes was recently reported in Triumfetta [52]. In contrast, no comparable rearrangements were detected among the Sida species examined here, underscoring the highly conserved nature of the Sida cp genome despite diversification across broad geographic and ecological ranges. Similar structural conservation has been documented in Hibiscus, Gossypium, and other genera, including in comprehensive comparisons across Malvaceae [19, 47, 48, 53–56]. However, although the 22 species compared here are biogeographically diverse, they represent only 8% of the genus; therefore, denser taxonomic sampling is recommended to comprehensively test for lineage-specific gene rearrangements or gene losses across the full diversity of Sida. IR boundary variation across the 22 Sida cp genomes was overall minor, consistent with the conserved IR dynamics reported for other Malvaceae genera such as Triumfetta, Hibiscus, Theobroma, and Firmiana [19, 46, 48, 53]. However, a noteworthy exception was observed at the JLB junction, where a localized contraction of the IRb boundary displaced the rps19 gene entirely into the LSC region in seven species: Sida echinocarpa, Sida ectogama, Sida rohlenae subsp. rohlenae, Sida sp. Articulation below, Sida sp. dark green fruits, Sida sp. Shovelanna Hill, and Sida sp. spiciform panicles. Importantly, these seven species are scattered across the phylogeny and do not form a monophyletic group, indicating that this boundary shift arose independently in multiple lineages rather than being inherited from a common ancestor. This pattern of phylogenetically scattered IR boundary variation has precedent in the literature. At broader taxonomic scales, IR contraction and expansion events have been shown to lack a consistent association with phylogenetic placement, as documented across Malvaceae and Araceae [47, 57–59]. By contrast, clade-specific IR dynamics have been reported at lower taxonomic levels in lineages such as Asclepiadoideae, Berberis, and Medicago, where IR shifts are concentrated within particular clades [16, 21, 60]. The present findings in Sida therefore align more closely with the former pattern: rather than defining a monophyletic lineage, the IRb contraction at JLB appears to have occurred in an independent and scattered manner across the genus. Given that the overall magnitude of IR size variation among all 22 Sida cp genomes is small (25,288–25,640 bp, a range of only ~ 352 bp), this localized boundary displacement should be interpreted as a subtle, lineage-specific structural adjustment rather than evidence of large-scale IR instability. Adaptive evolution and selection dynamics Comparative selection analyses of 78 protein-coding cp genes in Sida revealed heterogeneous patterns of adaptive evolution that varied depending on both the evolutionary scale examined and the analytical framework applied. Gene-wide tests using BUSTED [36] identified episodic positive selection in a limited subset of loci (psaB, psbB, rbcL, rps8), indicating that these genes experienced elevated nonsynonymous substitution rates along one or more branches of the Sida phylogeny. At the site level, MEME [37] initially detected numerous candidate codons under episodic positive selection; however, following global false FDR correction, only two sites—located in rbcL and ycf1—remained statistically supported (FDR < 0.05). In contrast, FUBAR [38] inferred a broader distribution of codons under pervasive selection (posterior probability PP > 0.9) across multiple genes. Together, these results demonstrate that adaptive signals in Sida cp genomes vary in strength (magnitude of selection coefficient), temporal mode (episodic vs. pervasive), and lineage distribution, rather than being uniformly distributed across genes or codon positions. Discrepancies among BUSTED, MEME, and FUBAR reflect the distinct biological questions and statistical frameworks of these approaches. MEME detects episodic diversifying selection at individual sites using a stringent likelihood ratio test with FDR correction [37], which explains why only a small number of codons remained significant despite a larger initial pool. BUSTED operates at the gene level and can detect positive selection even when only a small number of lineages or codon positions are affected [36]. Consequently, genes such as psaB and psbB—where episodic selection may be concentrated on a limited subset of branches—may produce a significant BUSTED signal without surviving MEME's site-level FDR correction. By contrast, FUBAR estimates selection parameters across all branches under a Bayesian random effects model [38, 61] and is more sensitive to weak but persistent deviations from neutrality. The combined application of these complementary methods therefore provides a more complete depiction of selection dynamics, capturing both lineage-restricted episodic events and pervasive, low-intensity selective pressures. From a functional perspective, the consistent support for positive selection in rbcL across gene-wide and site-specific analyses is notable and concordant with patterns reported across diverse angiosperm lineages. The rbcL gene encodes the large subunit of ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco), a central enzyme of the Calvin cycle, and has frequently been shown to evolve under positive selection in association with ecological diversification, photosynthetic efficiency, and environmental gradients [62, 63]. Signatures of positive selection in rbcL have been reported in numerous cp genome studies, including analyses of Erigeron [64], Monsteroideae [65], and Aroideae [66], reinforcing the role of rbcL as a recurrent target of adaptive evolution. Similarly, ycf1—one of the largest and most rapidly evolving cp genome genes—exhibited robust site-level support after FDR correction and has been widely proposed as both a powerful cp genome barcode and a locus prone to adaptive divergence [67, 68]. Site-specific positive selection in ycf1 has also been detected in Erigeron [64] and Pothoideae [66], suggesting that discrete adaptive signatures may accumulate at specific codon positions even when genes are largely conserved overall. The concordant detection of selection in these genes across analytical frameworks and taxonomic lineages suggests that they represent recurrent targets of adaptive or fine-tuning selection in cp genomes. In addition to rbcL, BUSTED identified episodic positive selection in psaB and psbB, which encode core reaction center proteins of Photosystem I and Photosystem II, respectively. These genes play essential roles in light harvesting and electron transfer, and signatures of episodic or relaxed selection in photosystem components have been reported in cp genome studies of lineages experiencing shifts in light environment, photosynthetic performance, or ecological niche [49, 63]. Episodic selection in psaB and psbB may therefore reflect lineage-specific optimization of photosystem function rather than sustained, pervasive adaptive change. BUSTED also detected episodic selection in rps8, a cp ribosomal protein gene, indicating that adaptive or compensatory changes may occasionally involve components of the cp genome translational machinery. Positive selection in rps8 has been reported in Monsteroideae [65] and may be associated with altered demands on protein synthesis, cp genome–nuclear co-evolution, or shifts in translational efficiency under changing physiological conditions [49, 69]. Beyond the genes highlighted by BUSTED and MEME, FUBAR inferred additional codons under pervasive selection in loci involved in photosynthetic electron transport (ndhF, ndhK), ATP synthesis (atpH), RNA splicing (matK), transcription (rpoC1, rpoC2), ribosomal structure and translation (rps15, rpl16, rpl20, rpl22, rpl23), fatty acid biosynthesis (accD), and hypothetical chloroplast reading frames (ycf2) in addition to rbcL and ycf1. Several of these genes, including accD, ndhF, ndhK, rpoC1, rpoC2, and ycf2, have been identified under positive selection in comparative cp genome analyses of Erigeron [64], Nicotiana [70], Monsteroideae [65], and Pothoideae [66]. For example, Mehmood et al. (2020) detected positive selection in rpoA, atpB, ndhD, and ndhF in Nicotiana species using FUBAR and MEME after FDR correction, suggesting that these genes may be associated with adaptation to specific environmental conditions such as coastal climates. Comparable low-level selection signals in these functional categories have been reported across cp genome-wide analyses and are generally interpreted as reflecting long-term fine-scale optimization or partial relaxation of functional constraints rather than strong adaptive sweeps [49, 69, 71]. Because FUBAR integrates information across the entire phylogeny, such signals should be interpreted cautiously and in conjunction with results from episodic selection tests. Overall, these findings indicate that adaptive evolution in Sida cp genomes is multifaceted, involving both episodic and pervasive components acting across multiple functional classes of genes. The integration of complementary selection tests, together with appropriate correction for multiple comparisons, provides a robust and nuanced framework for interpreting cp genome evolution. This approach highlights a small number of well-supported candidate genes for adaptive evolution while emphasizing the broader contribution of weak but persistent selective pressures in shaping cp genome diversity. Nucleotide polymorphism The identification of highly variable regions within the cp genome offers important resources for addressing long-standing taxonomic challenges in Sida. Traditional species identification in the genus has relied largely on the nuclear ribosomal ITS region or a limited number of short cp markers; however, these loci often provide insufficient phylogenetic signal for reliable species delimitation and resolution of interspecific relationships, particularly among closely related species [10–13, 72]. The limited discriminatory power of previously used molecular markers has therefore contributed to persistent taxonomic ambiguity within this morphologically complex genus. Consistent with general patterns of cp genome evolution, nucleotide diversity in Sida was markedly higher in IGS regions than in protein-coding sequences. Among coding genes, psaI and ycf1 exhibited comparatively elevated levels of nucleotide diversity, indicating relaxed functional constraint or lineage-specific divergence at these loci. Notably, ycf1 is one of the largest and most rapidly evolving cp genome genes and has been widely recommended as a core cp barcode for angiosperms due to its high discriminatory power [67]. Integrating phylogenetics and biogeography to decipher the evolutionary dynamics of Sida Clade I represents a cp-defined, predominantly Australian lineage in the present sampling. All taxa recovered in this clade, including Sida cardiophylla, Sida echinocarpa, and Sida fibulifera, are native to Australia, with many species restricted to Western Australia and northwestern Australia, such as Sida arsiniata, Sida clementii, and Sida ectogama [2, 4–6]. This clade also includes all undescribed Sida accessions sampled from localized regions such as Sida sp. Hamersley Range, Sida sp. Pilbara, and Sida sp. Barlee Range. Their consistent placement alongside established Western Australian endemics supports their inclusion within this cp-based Australian lineage in the present sampling, although formal taxonomic assignment will require morphological and multi-locus corroboration. Although several internal nodes within Clade I exhibited moderate to low bootstrap support (BS = 48%–72%), the overall monophyly of the Australian lineage was strongly supported (BS = 100%), as was its separation from Clade II (BS = 100%). Short internodes and weakly resolved relationships are consistent with recent diversification or limited cp genome divergence, incomplete lineage sorting, or low substitution rates, and similar patterns have been widely reported in Australian plant radiations, particularly in arid and semi-arid regions [73–75]. In Sida, this pattern likely reflects geographic isolation and limited cp genome divergence among recently differentiated species, as cp genomes often provide reduced resolution among closely related lineages [76]. More strongly supported sub-groupings within Clade I should be interpreted as cp-based working hypotheses representing early stages of lineage divergence rather than deeply established phylogenetic subdivisions. In contrast, Clade II represents a broadly distributed, trans-oceanic lineage within Sida. Species in this clade, including Sida spinosa (pantropical distribution), Sida rhombifolia (the tropical and subtropical Old World), and Sida cordata (tropical and subtropical Asia), exhibit wide native ranges consistent with long-distance dispersal and ecological generalism [4, 11]. Sida szechuensis, restricted to southern China and northern Vietnam, and Sida fallax, occurring primarily on Pacific islands, further characterize this lineage as exhibiting a predominantly Asian–Pacific distribution pattern within the sampled taxa. Notably, Sida rohlenae subsp. rohlenae represents the only Australian member of Clade II in the current sampling with a documented native distribution. Its phylogenetic nesting within an otherwise Asian–Pacific lineage is compatible with either a relatively recent dispersal into Australia from an Asian–Pacific source area or the retention of an ancestral Asian–Australian distribution that was subsequently lost in other Clade II taxa, rather than participation in the endemic Australian lineage represented by Clade I. Similar biogeographic patterns involving secondary dispersal into Australia from Asian or Pacific source areas have been inferred in multiple angiosperm groups [74, 77]. Given that only one Clade II accession is Australian and that cp genome data alone cannot discriminate between dispersal-from-Asia and range-retention scenarios, we present these two alternatives as equally supported working hypotheses rather than advancing a preferred biogeographic model. Although the phylogenetic patterns recovered here are well supported, the sampling comprised only 22 species, representing approximately 8% of the genus Sida (~ 275 species), and reliance on the maternally inherited cp genome [78] introduces specific limitations. Several processes may compromise cp-based phylogenetic inference in such lineages. Polyploidy and reticulate evolution have been widely documented across the Malveae alliance, including Sida and closely allied genera [10, 79]; notably, allopolyploidy is well established in Gossypium, where it has produced well-documented cytonuclear discordance [80, 81]. Hybridization and introgression can lead to cp genome capture, producing cp-based trees that do not fully reflect nuclear evolutionary history, as demonstrated in several Malvaceae lineages where cp and nuclear phylogenies show significant topological conflict [79, 82, 83]. Incomplete lineage sorting (ILS) is particularly pronounced in rapidly radiating lineages with short internodes—consistent with the weakly resolved substructure observed within Australian Clade I—and both cp capture and ILS are documented sources of cytonuclear phylogenetic discordance within Malvaceae [83–85]. Unexpected placements, such as the nesting of Sida rohlenae subsp. rohlenae within the otherwise Asian–Pacific Clade II, may also stem from these processes. Given these limitations and the known paraphyly of Sida with respect to allied genera [10, 11], the present cp phylogeny should be regarded as a preliminary framework rather than a definitive species tree. Integrating nuclear genomic data—via multi-locus or genome-wide approaches—together with broader taxon sampling will be essential to disentangle the confounding effects of hybridization, polyploidy, and ILS, and to refine the taxonomy and evolutionary history of Sida within tribe Malveae as these approaches have been found effective in other plant lineages [86–88]. Conclusion and future perspectives This study presents a comparative analysis of 22 Sida cp genomes, establishing a genomic framework for the genus. The cp genomes exhibit high structural conservation in size, gene content, and organization, with no significant rearrangements. Hypervariable regions, particularly within ycf1 and specific intergenic spacers, were identified as promising DNA barcodes for resolving taxonomic ambiguities. While purifying selection predominates, genes such as rbcL and ycf1 show signatures of episodic positive selection, suggesting lineage-specific adaptive evolution. Phylogenomic analysis resolved two strongly supported clades—an endemic Australian lineage and a widespread Asian/pan-tropical lineage—providing preliminary insights into the phylogeographic structure of the genus and demonstrating the value of whole-cp genome data for phylogenetic reconstruction. The current dataset encompasses only ~ 8% of the genus and relies solely on maternally inherited cp genome data, which cannot capture the full complexity of reticulate evolution, hybridization, and incomplete lineage sorting documented across Malveae. Future studies should prioritize denser taxonomic sampling alongside nuclear genomic data—via multi-locus or phylogenomic approaches—to test the monophyly and internal relationships of the clades recovered here, resolve weakly supported nodes, and clarify the evolutionary and biogeographic history of Sida within tribe Malveae. Supplementary Information Supplementary Material 1: Figure S1. Comparative analysis of chloroplast structure and gene organization among Sida species based on mauve alignment. Colored blocks indicate locally collinear regions, highlighting conserved genomic architecture across species. Figure S2. Bayesian phylogenetic relationships among Sida species inferred from plastome data, showing two major clades (Clade I and Clade II). Posterior probability values are indicated at the nodes. Supplementary Material 2: A GenBank-to-FASTA/TBL converter (Script 1), a gene selection detection pipeline (Script 2), and a phylogenetic analysis script (Script 3). Scripts 1 and 3 are now part of CGAS. Supplementary Material 3: Results of maximum likelihood phylogeny inference. Supplementary Material 4: Results of Bayesian phylogenetic inference. Supplementary Material 5: Table S1-S12. Supplementary Material 1: Figure S1. Comparative analysis of chloroplast structure and gene organization among Sida species based on mauve alignment. Colored blocks indicate locally collinear regions, highlighting conserved genomic architecture across species. Figure S2. Bayesian phylogenetic relationships among Sida species inferred from plastome data, showing two major clades (Clade I and Clade II). Posterior probability values are indicated at the nodes. Supplementary Material 2: A GenBank-to-FASTA/TBL converter (Script 1), a gene selection detection pipeline (Script 2), and a phylogenetic analysis script (Script 3). Scripts 1 and 3 are now part of CGAS. Supplementary Material 3: Results of maximum likelihood phylogeny inference. Supplementary Material 4: Results of Bayesian phylogenetic inference. Supplementary Material 5: Table S1-S12.