Zoysiagrass (Zoysia spp. Willd.) is a perennial warm-season turfgrass widely used in lawns, sports fields, and urban landscapes across the southern United States and East Asia (Patton et al. 2017). Species in this genus are allotetraploids (2n = 4× = 40) with compact genomes (~300–400 Mb) and exhibit extensive variation in tolerance to multiple abiotic stresses, making them important targets for breeding and stress biology research (Tanaka et al. 2016). Zoysia japonica Steud. and Z. matrella (L.) Merr. are the primary species used in breeding programs (Chandra et al. 2017). However, high heterozygosity resulting from outcrossing has limited genome assembly quality, often yielding haplotype-collapsed references that obscure allelic and regulatory variation and restrict the development of phased markers for genetic mapping and breeding. Here, we report a near gap-free, fully chromosome-phased genome assembly of the Z. japonica cultivar ‘Palisades’, a widely used coarse-textured zoysiagrass in the southern United States. First, a fully phased de novo contig assembly was generated using PacBio HiFi reads and Hi-C data (Table S1) with hifiasm (Cheng et al. 2021), and subsequently scaffolded to the chromosome-scale using the allele-aware HapHiC (Zeng et al. 2024). Chromosome numbers were assigned based on alignment with the genetic map (Wang et al. 2015). The high quality of the assembly was supported by clear diagonal Hi-C contact matrices (Figure S1) and strong concordance with the genetic map (Figure S2). The final assembly comprised 20 chromosome pairs, totaling 317.2 Mb for haplotype 1 (hap-1) and 317.0 Mb for haplotype 2 (hap-2) (Table S2, Figure 1A). All chromosomes were gapless, except for five chromosomes in hap-1 and one in hap-2 (Table S2). Telomeric repeats were detected at both ends of all chromosomes (Figure S3), indicating telomere-to-telomere continuity in the genome. Gene space assessment by BUSCO identified 98.3% of conserved Poales orthologs genes, and assembly evaluation by Merqury estimated 99% completeness with a consensus quality of 76.5 (Table S3), confirming the high accuracy of the assembly. Because the progenitors of Z. japonica remain unidentified, we attempted subgenome assignment using k-mer profiling by SubPhaser (Jia et al. 2022) and phylogenetic approaches; however, both methods were inconclusive, indicating the requirement for genomic information from at least one diploid progenitor. Hap-1 and hap-2 showed overall collinearity with the previously published haplotype-collapsed Z. japonica cultivar ‘Compadre’ assembly (Figure S4) (Shen et al. 2025). However, each Palisades haplotype exhibited a distinct set of structural sequence differences relative to Compadre (Table S4), consistent intraspecific variation. Furthermore, comparison of raw HiFi read alignments at regions where structural variants (SVs) were identified between hap-1 and hap-2 in Palisades revealed chimeric alignments in the corresponding regions of the Compadre assembly (Figure S5). These patterns indicate mosaic allelic representation in the haplotype-collapsed assembly and underscore the necessity of phased genome assemblies. Consistent with previous studies (Tanaka et al. 2016; Wang et al. 2015), both haplotypes also exhibited strong collinearity with rice (Oryza sativa IRGSP-1.0) and sorghum (Sorghum bicolor NCBIv3) (Figure S6), highlighting the utility of Z. japonica for cross-species comparative genomics and translational research. Annotation of repetitive sequences accounted for approximately 50% of each haplotype (158 Mb in hap-1 and 156 Mb in hap-2) (Table S5). Gene prediction, integrating short- and long-read transcriptome data (Table S6) and protein homology, after our rigorous manual curation (Figure S7), identified 40 106 and 40 223 protein-coding genes in hap-1 and hap-2, respectively (Table S7), with 96% BUSCO completeness. Comparison of hap-1 and hap-2 using SyRI (Goel et al. 2019) revealed extensive collinearity (~250 Mb per haplotype) and substantial haplotype-specific sequences (142 Mb in hap-1 and 141 Mb in hap-2) (Figure 1A). Notably, we identified approximately 2.9 million SNPs and numerous structural variants (SVs), including 188 877 insertions, 196 125 deletions, 46 inversions, and 52 translocations (Table S8, Figure 1A). Most SVs were 50 bp) are shown in dark grey. Figure S6: Collinearity analysis between Sorghum bicolor and Oryza sativa. (A) Sorghum bicolor. (B) Oryza sativa. Figure S7: Representative cases of gene annotation model curation. Blue boxes indicate curated models, and yellow boxes represent models prior to curation. MAS-seq alignments are shown below the models. (A) An incorrectly split gene model. (B) An incorrectly merged gene model. (C) A gene model without supporting RNA evidence. Figure S8: Validation of SV markers. PCR products were resolved on a 2% agarose gel. PCR gel image of randomly selected five variants between hap-1 and hap-2 with ladder. Figure S9: Genome-wide heatmap of RNA expression profiles in haplotype 1 and haplotype 2. From top to bottom, rows correspond to rhizome, stolon, and root. Table S1: Summary of raw sequencing data. Table S2: Comparison of assembly statistics between hap-1 and hap-2 and previous assemblies. Table S3: Assembly completeness and consensus quality assessment. Table S4: Structural variants identified in comparison with the previously published genome. Table S5: Summary of repeat annotation. Table S6: Summary of transcriptome sequencing data used in gene annotation. Table S7: Gene annotation results. Table S8: Structural variants counts and size distributions between haplotype 1 and haplotype 2. Table S9: Information of the haplotype specific primers. Table S10: Raw data resources utilised in the gene annotation. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.
Lee et al. (Mon,) studied this question.