Chromosome-level genome assembly of Manglietia pachyphylla Authors: Chuqiao Tang, Jiamei Yang, Shuai Yuan, Yuling Li Published: 2026 ABSTRACT Manglietia pachyphylla, an endangered evergreen tree within the Magnoliaceae family, is renowned for its exceptional ornamental value in landscape horticulture. Despite its classification as a Category II nationally protected plant species in China, the genetic basis of its adaptive traits and conservation priorities remains poorly understood. To address this, we present the first chromosome-scale genome assembly of M. pachyphylla utilizing an integrated approach combining PacBio HiFi long-read and Hi-C chromosome conformation capture sequencing technologies. The assembled genome spans 2.15 Gb (contig N50 = 43.57 Mb), exhibiting a heterozygosity rate of 0.78% and repeat content of 78.64%, predominantly comprising long terminal repeat (LTR) retrotransposons (52.86%). Hi-C scaffolding anchored 99.57% of the assembly to 19 pseudochromosomes, achieving a BUSCO completeness score of 96.4%. Annotation revealed 42,505 putative protein-coding genes, with 84.46% of predicted genes were functionally annotated. Phylogenomic analysis positioned M. pachyphylla and Oyama sieboldii clustered together in a well-supported group. This high-contiguity genome assembly enables future investigations into adaptive evolution, functional genomics, and evidence-based conservation strategies for this endangered species. Background & Summary Manglietia pachyphylla (Fig. 1), an evergreen arborescent species endemic to Guangdong Province, China, occupies a specialized ecological niche in subtropical evergreen broad-leaved forests at 800 m elevation1,2. This taxon exhibits significant horticultural potential due to its distinctive phenotypic characteristics, an architecturally attractive form, coriaceous foliar morphology, and fragrant floral displays, rendering it a valuable candidate for ornamental cultivation3. However, M. pachyphylla is currently facing critical conservation challenges, including fragmented populations, constrained distributional range (<100 km2), limited seedling recruitment, and escalating anthropogenic pressures. M. pachyphylla is classified as Vulnerable (VU) in the China Biodiversity Red List, as well as designated as a Category II nationally protected plant in China4–7.Fig. 1The morphology of M. pachyphylla. The morphology of M. pachyphylla. Despite increasing genomic efforts within Magnoliaceae, only a few high-quality genome assemblies within Magnoliaceae have been published (such as Liriodendron chinense8, Yulania biondii9, Woonyoungia septentrionalis10, Pachylarnax sinica11, Oyama sieboldii12). The genomic resources for the genus Manglietia remain extremely limited, and no chromosome-scale genome has previously been available for this lineage, The establishment of this chromosomal framework provides three critical advances: (1) enables comparative phylogenomic analyses within Magnoliaceae through integration with existing chromosome level assemblies; (2) facilitates identification of genomic signatures underlying ecological specialization and reproductive limitations; (3) creates molecular tools for assessing genetic diversity in wild populations. The genomic resources developed here will facilitate targeted conservation interventions including habitat corridor design, ex situ conservation genomics, and molecular-assisted breeding programs, which are all urgently needed to preserve this threatened genus in the face of rapid environmental change. Methods and Results Sample preparation and sequencing Samples (fresh leaves, buds, fruits, and seeds) were collected from Nankunshan Provincial Nature Reserve, Guangdong Province and stored at −80 °C. All samples were sent to BGI Genomics for High-fidelity long-read (HiFi) sequencing, Hi-C sequencing and transcriptome sequencing. For HiFi sequencing with Circular Consensus Sequencing (CCS), the high-fidelity long read library was constructed with the SMRTbell Express Template Prep Kit and sequenced on the PacBio Revio platform. The concentration and fragment size of the library were assessed using Qubit and Agilent 2100. Subsequently, the library was bound to sequencing primers and polymerase using the PacBio Binding Kit and purified with AMPure PB beads to ensure sequencing quality. Finally, sequencing generated a total of 95.57 Gb of high-quality HiFi reads, with an N50 read length of 19.05 Kb and an average read length of 18.79 Kb (Table 1).Table 1Summary of sequencing data for Manglietia pachyphylla.Library typePlatformData size (Gb)Coverage(X)Read length (bp)HiFi readsPacBio Revio95.5741.619049 (N50)Hi-C readsDNBSEQ273.87119.08150RNA-seqDNBSEQ36.75—150 Table 1Summary of sequencing data for Manglietia pachyphylla.Library typePlatformData size (Gb)Coverage(X)Read length (bp)HiFi readsPacBio Revio95.5741.619049 (N50)Hi-C readsDNBSEQ273.87119.08150RNA-seqDNBSEQ36.75—150 Summary of sequencing data for Manglietia pachyphylla. Hi-C library construction was performed on the DNBSEQ platform (PE150 sequencing mode). Sequencing data underwent rigorous quality control and filtering, removing contaminants, adapter sequences, and low-quality reads. Consequently, we generated a total of 912,898,976 clean reads, with a total base yield of 273.87 Gb, where Q20 reached 99.34%, Q30 reached 97.56%, and the GC content was 41.62% (Table 1). To predict genes correctly, four independent 150 bp paired-end RNA-seq libraries were constructed from young leaves, buds, fruits, and seeds, respectively, and sequenced on the DNBSEQ platform. Trimming the adapters and low-quality raw sequence reads using Fastp v0.23.413 yielded approximately 36.75 Gb of clean data (Table 1). A total of 5.45, 5.79, 6.02, and 5.53 Gb of clean data were generated from leaves, buds, fruits, and seeds respectively (Table 2).Table 2RNA sequencing statistics.SampleSequencing platformClean dataTotal number of reads(bp)Total number of bases(bp)RNA leafDNBSEQ19,053,6045,716,081,200RNA stemDNBSEQ20,222,6546,066,796,200RNA seedDNBSEQ21,044,6136,313,383,900RNA fruitDNBSEQ19,346,1445,803,843,200 Table 2RNA sequencing statistics.SampleSequencing platformClean dataTotal number of reads(bp)Total number of bases(bp)RNA leafDNBSEQ19,053,6045,716,081,200RNA stemDNBSEQ20,222,6546,066,796,200RNA seedDNBSEQ21,044,6136,313,383,900RNA fruitDNBSEQ19,346,1445,803,843,200 RNA sequencing statistics. De novo genome assembly The PacBio HiFi reads were de novo assembled by using Hifiasm v.0.19.814. The short reads were mapped to the draft assembled genome by using Minimap2 v 2.28-r120915. Subsequently, The polishing genome assembly was conducted by using Racon v1.5.011 with default parameters for three rounds. Then, to enhance the accuracy of the assembly, Purge_dups v 1.2.516 was employed to align the PacBio HiFi reads with the preliminary assembled genome and calculate the boundary point of base coverage and identify haplotigs and overlaps to classify and remove redundant contigs caused by heterozygosity. To improve the quality of genome assembly, based on the Hi-C reads, the contigs were scaffolded to the pseudochromosome. Adaptor sequences and low-quality Hi-C reads were filtered using SOAPnuke17, and the resulting clean Hi-C reads were used to cluster, order and orient contigs onto the chromosomes. First, contigs were indexed with BWA v0.7.1818. The Hi-C reads aligned to the contig sequences, resulted in a BAM file. HapHic v1.0.619 generated sacffolds using Hi-C derived BAM files. Subsequently, visualization and manual adjustments were made based on the intensity of chromosomal interactions using the Juicebox v2.13.0720. The Hi-C interaction heatmap validates the precision of the assembly, confirming that M. pachyphylla has a diploid chromosome number 2n = 38, and the total of 2161.84 Mb (99.86%) of the contigs were anchored to the 19 chromosomes (Table 3). The integrity and high quality of the chromosome-level assembly were supported by Hi-C interaction heatmaps and genome-wide synteny and chromosomal organization analyses (Figs. 2, 3).Table 3Summary of the 19 pseudochromosomes.IDNo.of contigsLength(bp)GC content(%)Chr013179,412,85640.34Chr024162,278,24940.41Chr037155,766,94140.3Chr046141,089,37840.45Chr057137,036,39140.41Chr064125,313,14340.36Chr074112,335,83340.4Chr087109,783,42440.13Chr099109,046,32440.24Chr103106,602,72440.38Chr115105,429,25640.28Chr123103,706,79339.89Chr138100,473,39640.24Chr14398,656,35640.25Chr15493,122,00440.36Chr16791,899,31040.84Chr17489,990,45940.24Chr18789,195,63940.3Chr19887,171,58140.21unplaced503,040,864Fig. 2Hi-C contacts Heatmap illustrating chromosomal interactions in the M. pachyphylla genome.Fig. 3Genome characteristics of M. pachyphylla. The track from outer to inner circle represents the nineteen chromosomes(chr01-chr19), LTR density, Gene density, GC content and syntenic gene blocks within the genome indicated by connecting lines. Table 3Summary of the 19 pseudochromosomes.IDNo.of contigsLength(bp)GC content(%)Chr013179,412,85640.34Chr024162,278,24940.41Chr037155,766,94140.3Chr046141,089,37840.45Chr057137,036,39140.41Chr064125,313,14340.36Chr074112,335,83340.4Chr087109,783,42440.13Chr099109,046,32440.24Chr103106,602,72440.38Chr115105,429,25640.28Chr123103,706,79339.89Chr138100,473,39640.24Chr14398,656,35640.25Chr15493,122,00440.36Chr16791,899,31040.84Chr17489,990,45940.24Chr18789,195,63940.3Chr19887,171,58140.21unplaced503,040,864 Summary of the 19 pseudochromosomes. Hi-C contacts Heatmap illustrating chromosomal interactions in the M. pachyphylla genome. Genome characteristics of M. pachyphylla. The track from outer to inner circle represents the nineteen chromosomes(chr01-chr19), LTR density, Gene density, GC content and syntenic gene blocks within the genome indicated by connecting lines. The genome size of M. pachyphylla was assembled to be 2099.37 Mb, with a GC content of 40.33%. The contig and scaffold N50 values were 43.57 Mb and 109.78 Mb respectively (Table 4).Table 4Statistics of M.pachyphylla genome assembly.Assembly featureAssembly size (Mb)2099.37Scaffold N50 (Mb)109.78Contig N50 (Mb)43.57Anchor ratio (%)99.86%GC content40.33%BUSCO (%)96.50% Table 4Statistics of M.pachyphylla genome assembly.Assembly featureAssembly size (Mb)2099.37Scaffold N50 (Mb)109.78Contig N50 (Mb)43.57Anchor ratio (%)99.86%GC content40.33%BUSCO (%)96.50% Statistics of M.pachyphylla genome assembly. Assessing assembly genome data The completeness of the assembly genome was analyzed by BUSCO v5.7.121 with the embryophyte_odb10 orthologous database. The results showed that 96.50% of complete genes were assembled in the genome, including 91.50% single-copy and 5.00% duplicated genes, and the remaining were identified as missing (1.90%) and fragmented (1.60%). These results confirms the high quality of genome assembly (Table 5). To further assess the assembly quality in repeat-rich regions, LTR_retriever v3.0.422 estimated a raw LTR Assembly Index (LAI) of 17.62 and an adjusted LAI of 12.04, placing it within the reference-quality range for plant genomes. Merqury23 based on PacBio HiFi reads. estimated a consensus quality value (QV) of 39.09, indicating high base-level accuracy.Table 5Result of BUSCO assessment of M.pachyphylla.TypeNumberPercentageComplete BUSCOs224496.50%Complete and single-copy BUSCOs212891.50%Complete and duplicate BUSCOs1165.00%Fragmented BUSCOs441.90%Missing BUSCOs381.60%Total BUSCO groups searched2326 Table 5Result of BUSCO assessment of M.pachyphylla.TypeNumberPercentageComplete BUSCOs224496.50%Complete and single-copy BUSCOs212891.50%Complete and duplicate BUSCOs1165.00%Fragmented BUSCOs441.90%Missing BUSCOs381.60%Total BUSCO groups searched2326 Result of BUSCO assessment of M.pachyphylla. Repeat annotation To annotate repetitive elements in the M. pachyphylla genome, we employed RepeatModeler v2.0.224 and RepeatMasker v4.1.525. The repeat sequence database of M. pachyphylla was generated using RepeatModeler v2.0.224 with the LTR search program (-LTRStruct) and merged with RepBase-2018102626 database. RepeatMasker used the databases to identify and mask repeat sequences. DeepTE pipelines (source code revision dated 2020-12-25)27 was employed to identify transposable elements (TEs) based on RepeatMasker annotations. Unclassified TEs were processed incorporating structural domain information for classification. The newly classified sequences were then merged with RepeatMasker-identified TEs to generate a comprehensive TE library. As a result, a total of 77.10% genome was identified as repeat elements. Retroelements occupy 59.46% of the genome with 52.86% LTR elements. BEL/Pao elements account for 0.90%, Ty1/Copia elements account for 12.51%, Gypsy/DIRS1 elements account for 39.29%, and Retroviral elements account for 0.57%. DNA transposons elements contribute 13.30%. Total interspersed repeats represent 77.10% of the genome (Table 6).Table 6Annotation of repeat elements in M. pachyphylla.Repeat typeNumber of elementsLength (bp)Percentage of sequenceRetroelements1927802130889091359.46% SINEs:143247387975311.76% Penelope:293412820400.06% LINEs:1506201064288374.83% CRE/SLACS000.00% L2/CR1/Rex1247830017960.14% R1/LOA/Jockey8371722620.01% R2/R4/NeSL12642374180.01% RTE/Bov-B47959157935410.72% L1/CIN488082872238203.96% LTR elements:1633935116366454552.86% BEL/Pao177219008760.09% Ty1/Copia27465327529948312.51% Gypsy/DIRS1126965186496001739.29% Retroviral53777124543010.57% DNA transposons113058729277055413.30% hobo-Activator255866596033662.71% Tc1-IS630-Pogo255695588827022.67% En-Spm000.00% MULE-MuDR50268215614530.98% PiggyBac000.00% Tourist/Harbinger79071145160880.66% Other000.00%Rolling-circles82501332183111.51%Unclassified:462604942202934.28%Total interspersed repeats:169716380077.10%Small RNA:144521396056331.80%Satellites:25907295260.03%Simple repeats:25798580.00%Low complexity:000.00% Table 6Annotation of repeat elements in M. pachyphylla.Repeat typeNumber of elementsLength (bp)Percentage of sequenceRetroelements1927802130889091359.46% SINEs:143247387975311.76% Penelope:293412820400.06% LINEs:1506201064288374.83% CRE/SLACS000.00% L2/CR1/Rex1247830017960.14% R1/LOA/Jockey8371722620.01% R2/R4/NeSL12642374180.01% RTE/Bov-B47959157935410.72% L1/CIN488082872238203.96% LTR elements:1633935116366454552.86% BEL/Pao177219008760.09% Ty1/Copia27465327529948312.51% Gypsy/DIRS1126965186496001739.29% Retroviral53777124543010.57% DNA transposons113058729277055413.30% hobo-Activator255866596033662.71% Tc1-IS630-Pogo255695588827022.67% En-Spm000.00% MULE-MuDR50268215614530.98% PiggyBac000.00% Tourist/Harbinger79071145160880.66% Other000.00%Rolling-circles82501332183111.51%Unclassified:462604942202934.28%Total interspersed repeats:169716380077.10%Small RNA:144521396056331.80%Satellites:25907295260.03%Simple repeats:25798580.00%Low complexity:000.00% Annotation of repeat elements in M. pachyphylla. Gene prediction and functional annotation The repeat-masked genome was used for annotation. Protein-coding gene prediction for the M. pachyphylla genome was conducted based on three approaches, including ab initio prediction, homology-based methods, and transcriptome data. For ab initio prediction, Augustus v3.5.028 and GeneMark-ES v4.3329 were used. GeneMark-ES was applied to predictions. Homology-based annotation was performed utilizing GeMoMa v1.6.430 with genome data from closely related species and model species, i.e., Y. biondii9, W. septentrionalis10, Oryza sativa31 and Arabidopsis thaliana32. RNA-seq reads were assembled by Trinity v2.14.033, and annotated by PASA v2.5.334. Predictions from all three approaches were integrated using EVidenceModeler(EVM) v1.1.135. Finally, a total of 42,505 protein-coding genes were identified from the repeat-masked genome of M. pachyphylla. The protein-coding genes had an average gene length of 9,951 bp, an average coding DNA sequence(CDS) of 214 bp, an average exon number per gene of 4.50 (Table 7).Table 7Summary of predicted protein-coding genes in M.pachyphylla genome assembly.MethodsGene setGene numberCDS numberAverage length of gene(bp)Average number of exonAverage length of CDS(bp)Ab initio annotationAugustus33,694170,01714461.55.04221.58GeneMark-ES40,079163,3265760.584.07232.27Homolgous annotationY. biondii28,807125,1067251.254.34225.08W. septentrionalis28,095126,3158329.854.50239.32O. sativa17,43995,67910827.25.49237.42A. thaliana18,53799,27010,4985.36234.45EVM42,505191,0749550.844.50213.91 Table 7Summary of predicted protein-coding genes in M.pachyphylla genome assembly.MethodsGene setGene numberCDS numberAverage length of gene(bp)Average number of exonAverage length of CDS(bp)Ab initio annotationAugustus33,694170,01714461.55.04221.58GeneMark-ES40,079163,3265760.584.07232.27Homolgous annotationY. biondii28,807125,1067251.254.34225.08W. septentrionalis28,095126,3158329.854.50239.32O. sativa17,43995,67910827.25.49237.42A. thaliana18,53799,27010,4985.36234.45EVM42,505191,0749550.844.50213.91 Summary of predicted protein-coding genes in M.pachyphylla genome assembly. To assess whether the predicted gene number is reasonable, we compared the gene count of M. pachyphylla with two phylogenetically close Magnoliaceae species. Y. biondii9 contains 41,181 genes, while W. septentrionalis10 contains 35,927 genes, both comparable to the 42,505 genes identified in M. pachyphylla. This comparison supports the overall plausibility of the predicted gene set (Table 8).Table 8predicted genes in M.pachyphylla genome.SpeciesGene numberY. biondii41,181W. septentrionalis35,927M. pachyphylla42,505 Table 8predicted genes in M.pachyphylla genome.SpeciesGene numberY. biondii41,181W. septentrionalis35,927M. pachyphylla42,505 predicted genes in M.pachyphylla genome. Functional annotation of the genome utilized five databases (KEGG, GO, UniProt, eggNOG-mapper, InterProScan). We performed genome annotation using the DIAMOND36, derived from the Kyoto Encyclopedia of Genes and Genomes37 (KEGG, https://www.kegg.jp/), Gene Ontology38 (GO, http://geneontology.org/) and UniProt39 (https://www.uniprot.org/) databases. In addition, protein-coding genes were supplemented using eggNOG-mapper40 and InterProScan41. In total, 35583 protein-coding genes were annotated out of 42,505 predicted genes (84.46%) in the M. pachyphylla genome (Table 9).Table 9Functional annotation of predicted genes in M.pachyphylla genome.Functional annotationGO29864 (70.88%)KEGG29864 (70.88%)Uniprot29864 (70.88%)InterPro34477 (81.83%)eggNOG28131 (66.77%)Total35583 (84.46%) Table 9Functional annotation of predicted genes in M.pachyphylla genome.Functional annotationGO29864 (70.88%)KEGG29864 (70.88%)Uniprot29864 (70.88%)InterPro34477 (81.83%)eggNOG28131 (66.77%)Total35583 (84.46%) Functional annotation of predicted genes in M.pachyphylla genome. Data Records The Hi-C sequencing data, PacBio HiFi sequencing data, transcriptome sequencing data have been deposited in the NCBI Sequence Read Archive under accession SRP4541594242. Data Overview Genome-wide synteny analysis Based on the chromosome level of M. pachyphylla, we identified the syntenic blocks within the M. pachyphylla genome using JCVI’s MCScan pipeline43. Additionally, syntenic relationships between the M. pachyphylla genome and the published genomes of Y. biondii9 and O. sativa12 were analyzed using the same pipeline, with a minimum c-score of 0.99. Genome-wide syntenic blocks were visualized using JCVI with the parameter -minspan = 30 (Fig. 4).Fig. 4Genome-wide synteny among three genome assemblies in Magnoliaceae. Genome-wide synteny among three genome assemblies in Magnoliaceae. Phylogenetic analysis A phylogenetic tree was constructed based on 428 single-copy orthologous genes identified using Orthofinder v3.0.144 and RAxML v8.2.1245 to elucidate the evolutionary relationships among M. pachyphylla and four other Magnoliaceae species, with Amborella trichopoda46, O. sativa31, A. thaliana32 and Annona cherimola47 designated as outgroups. Sequence alignment was performed with Muscle v3.8.3148, conversed regions were preserved using Gblocks v0.91b49, and the phylogenetic tree was constructed with RAxML45 using the JTT + I + G + F model via the maximum likelihood method. Each branch node is showed with bootstrap support values ≥ 85, indicating the high statistical reliability of the phylogenetic tree structure (Fig. 5).Fig. 5Maximum likelihood analysis based on orthologous genes for five species from Magnoliaceae family. The numbers at the node indicate the bootstrap support values (BS). Maximum likelihood analysis based on orthologous genes for five species from Magnoliaceae family. The numbers at the node indicate the bootstrap support values (BS). The phylogenetic tree reveals the evolutionary relationships among the species in the Magnoliaceae family. Y. biondii9, P. sinica11 and W. septentrionalis10 formed a monophyletic clade (BS = 100), while O. sieboldii12 and M. pachyphylla showed the closest affinity (BS = 100). These results enhance understanding of the genetic diversity and phylogenetic relationships within Magnoliaceae, providing critical insights for conservation. Technical Validation By using ~43.05x PacBio HiFi reads and ~123.8x reads, The chromosome-scale assembled genome of M. pachyphylla measures 2.22 Gb. The genome quality was evaluated based on the following results: (1) The genome assembly achieved a contig N50 of 116.15 Mb and a scaffold N50 of 107.21 Mb. (2) The completeness of the genome was evaluated by BUSCO v5.7.121 with the embryophyta_odb10 reference gene set. The assembly exhibited 96.50% BUSCO complete genes, comprising 91.50% single-copy, 5.00% duplicated genes, 1.60% fragmented and 1.90% missing. (3) PacBio HiFi reads were aligned to the assembly using BWA v0.7.1818, achieving a mapping rate of 99.94%. (4) The assembly obtained the LAI value of 12.04. (5) A k-mer–based evaluation using Merqury23 estimated a consensus QV of 39.09. (6) The completeness of the predicted protein-coding gene set exhibited 71.3% complete BUSCOs, including 68.7% single-copy and 2.6% duplicated.