Haplotype-resolved genome of Schizothorax nukiangensis sheds light on polyploidy and altitudinal adaptation in freshwater fishes Authors: Chong Wang, Xue-Die Gu, Guo-Liang Lin, Shu-Wei Liu, Guo-Zheng Liao, Yu-Hui Du, Bang-Yang Sun, Ting-Chu Wei, Cheng-Zhi Ding, Ya-Ping Zhang, Yan Li, Xiao-Yong Chen, Jing Luo Published: 2026 ABSTRACT Polyploidy results in genome redundancy and drives ecological diversification in freshwater fishes, yet the genomic architecture and adaptive significance of recent polyploidization in high-altitude ecosystems remain poorly understood. Herein, after determining genome size and karyotype, we tackle the long-standing challenge of resolving highly similar homologous chromosomes in autopolyploid vertebrates. By integrating long-read sequencing, Hi-C scaffolding, and haplotype-specific phasing, we assemble a haplotype-resolved genome of the autotetraploid Schizothorax nukiangensis, comprising 100 chromosomes grouped into 25 homologous sets, with a genome size of 3.59 Gb per cell. Evidence consistently supports its autotetraploid origin, with a lineage-specific polyploidization event dated to ~0.68 million years ago. Comparative genomics reveals substantial expansions of gene families in innate immunity, redox regulation, and transposon activity, suggesting genome remodeling might have contributed to adaptation in hypoxic, pathogen-rich, and fast-flowing mountain rivers. Population genomics across three elevational zones reveal subtle yet significant structure and identify candidate genes under selection related to skeletal development, immune defense, and ciliary function, consistent with adaptation to the torrents or rapids in fast-moving rivers characteristic of steep mountain ranges. Transcriptomic profiling further shows altitude-associated upregulation of genes related to immune responses and oxidative stress mitigation, which is consistent with other reported schizothoracines, highlighting convergent molecular strategies for altitude adaptation. Collectively, these demonstrate that recent autotetraploidization, coupled with gene family innovation and homeolog-specific expression divergence, has facilitated the ecological success of S. nukiangensis in extreme environments of the Salween River. This work provides important insights into the genomic imprints underlying adaptation in high-altitude freshwater vertebrates. INTRODUCTION Whole-genome duplication (WGD) is a key driver of genomic variation, rapidly increasing genome size and resulting in gene redundancy and greater genetic diversity (Otto, 2007; Sémon & Wolfe, 2007; Vlček et al., 2025). WGD can also trigger genomic shock, causing gene loss, chromosomal rearrangements, and alterations in gene expression regulation, thereby contributing to the emergence of novel phenotypic traits (Almeida-Silva & Van De Peer, 2023; Hämälä et al., 2024; Yu et al., 2024). Many studies have shown that WGD events often coincide with major climatic and geological transitions, indicating that WGD plays a crucial role in adaptation and species diversification (Cai et al., 2019; Carruthers et al., 2024; David, 2022; Van De Peer et al., 2021). Although WGD is more prevalent in plants, hundreds of allopolyploid and autopolyploid species have also been documented in animals, particularly in freshwater fish (Leggatt & Iwama, 2003; Lien et al., 2016; Mezzasalma et al., 2023; Zhou & Gui, 2017). Snow carp (Cyprinidae: Schizothoracinae), also called naked carps, one of the youngest known polyploid lineages, are distributed across the Qinghai-Xizang Plateau and adjacent regions and represent one of the few freshwater fish taxa adapted to extreme plateau environments (Du et al., 2024). Their diversification is primarily associated with recent and recurrent polyploidization, supporting the view that polyploidy may facilitate adaptation, particularly under global climate change (Li et al., 2024c). Moreover, fossil evidence indicates that diversification in snow carp corresponds closely to the uplift history of the Qinghai-Xizang Plateau, with fossil distribution patterns aligning with current spatial distributions, reflecting long-term biological adaptation to plateau uplift (Deng et al., 2020; Wang & Chang, 2010). All known snow carp are polyploid, with chromosome numbers ranging from 2n=66 to 2n=470, further emphasizing the important role of polyploidization in their evolutionary history (Dai & Han, 2018; Li et al., 2024c; Yu & Yu, 1990). The Hengduan Mountains, located at the southeastern margin of the Qinghai-Xizang Plateau, arose from complex regional tectonic processes and form a critical part of the southeastern extension of the Himalayas (Wang et al., 2022). Their steep altitudinal gradients and rugged topography have profoundly influenced the Salween River Basin, creating a series of microhabitats along its course. Acting as a climatic barrier, the Hengduan Mountains intercept warm, moist air from the Indian Ocean, resulting in dramatic variations in temperature, precipitation, and humidity over short distances. These microclimatic heterogeneities have facilitated the coexistence of tropical, subtropical, temperate, and alpine ecosystems within a narrow geographic range, fostering exceptional biodiversity and high levels of endemism among both flora and fauna (Chaplin, 2005; Li et al., 2024b; Liu et al., 2020; Zhou & Li, 2024). Equally important, the dynamic aquatic environment of the Salween River imposes strong selective pressures on aquatic species. The river exhibits a pronounced longitudinal gradient in hydrological conditions, including fast-flowing rapids, deep pools, and braided channels (Yang et al., 2021). The ecological environment of the Salween River varies greatly along its elevation gradient. At higher altitudes, the water flows rapidly, temperatures are lower, and food resources are limited, creating harsher survival conditions for aquatic organisms. In contrast, lower-altitude sections are characterized by slower water, warmer temperatures, and abundant food resources. Seasonal fluctuations in discharge, driven by monsoonal rains and glacial meltwater, introduce substantial variability in flow velocity, sediment load, and turbidity (Yao et al., 2012). Furthermore, dissolved oxygen concentrations decrease with increasing altitude and fluctuating temperatures, posing significant physiological challenges for aquatic organisms in high-elevation reaches. Collectively, these complex hydrological and physicochemical factors generate a mosaic of aquatic habitats that, in conjunction with the region’s topography and climate, shape the adaptive landscape of native freshwater species (Kawecki & Ebert, 2004; Zhou & Li, 2024). From a biogeographical perspective, the combined effects of climatic barriers, rugged terrain, and dynamic hydrology have promoted population fragmentation, genetic divergence, and local adaptation to heterogeneous environmental conditions (Wang & Bradburd, 2014). Endemic freshwater fishes of the Salween River, including Schizothorax nukiangensis, a representative snow carp in this river, has shown specialized morphological, physiological, and behavioral traits to cope with strong currents, low oxygen availability, and fluctuating hydrological regimes. Among these, S. nukiangensis exhibits a distinctive shovel-like lower jaw that enables firm scratch algae on rocks in turbulent streams, facilitating efficient feeding (Chen, 2013). Distributed across elevations ranging from 612 to nearly 3890 m above sea level (asl), S. nukiangensis exhibits remarkable ecological plasticity, thriving in diverse habitats along the Salween River (Ding et al., 2019; Xu et al., 2023). This extensive altitudinal habitats within a single river system raises important questions about intraspecific variation: have geographically structured populations emerged along different elevations, and how differentiated are these populations in terms of genetic composition and adaptive traits? However, the genomic basis of this broad altitudinal and ecological adaptability, as well as the evolutionary implications of its recent autotetraploidization, remains largely unexplored. Resolving homologous chromosome sets in recent autopolyploid vertebrates poses significant technical challenges due to high sequence similarity between homeologs and pervasive genome redundancy. Previous studies on polyploid fish genomes have often resulted in collapsed assemblies or unresolved haplotypes, obscuring evolutionary and functional insights. In this work, we present the first haplotype-resolved, chromosome-level assembly of S. nukiangensis. By leveraging high-fidelity long-read sequencing, Hi-C scaffolding, and homeolog-specific phasing, we successfully partitioned the genome into 100 chromosomes corresponding to 25 homologous chromosome sets. By integrating genome resequencing of populations inhabiting distinct altitudinal gradients, we examined patterns of genetic diversity, population structure, and historical demography. Additionally, we identified candidate genes and genomic regions potentially involved in adaptation to the extreme and heterogeneous environments of the Hengduan Mountains, and the dynamic aquatic habitats of the Salween River. Our findings offer new insights into the genetic mechanisms underlying high-altitude and hydrological resilience in freshwater fishes and advance our understanding of how geological, climatic, and aquatic processes jointly shape biodiversity in mountainous river systems. MATERIALS AND METHODS Tissue sampling and sequencing for genome assembly of Schizothorax nukiangensis Samples of S. nukiangensis (sample ID is YPQ230120 for de novo genome assembly, more details in Supplementary Table S1) were collected from the Salween River Basin in Xizang and Yunnan Province, China (Supplementary Table S1). Muscle tissues were flash-frozen in liquid nitrogen for PacBio HiFi (CCS) long-read sequencing, DNBSEQ-T7 short-read sequencing, and Hi-C chromatin conformation capture sequencing. A total of 128.35 Gb of PacBio HiFi reads and 112.88 Gb (~71.31×) of paired-end short reads (150 bp) were generated for the de novo genome assembly of S. nukiangensis. Hi-C library preparation, including DpnII restriction enzyme digestion, was followed by sequencing on an DNBSEQ-T7 platform, yielding 192.26 Gb (~106.81×) of clean paired-end reads (Supplementary Table S2). To support genome annotation, both full-length transcriptome and short-read RNA sequencing were performed using pooled RNA samples from multiple tissues, including heart, liver, gill, kidney, swim bladder, muscle, skin, fin, brain, eye, spleen, and intestine. For full-length transcriptome sequencing, a total of 1.04 Gb of Iso-Seq circular consensus sequencing (CCS) data were generated on the PacBio platform. Additionally, short-read RNA-seq libraries were constructed and sequenced on the Illumina platform, generating 15.63 Gb of paired-end reads (150 bp) (Supplementary Table S2). These transcriptome data were integrated into genome annotation to improve gene model prediction. Population sampling and sequencing For population studies, a total of 37 wild S. nukiangensis individuals were collected from various locations along the Salween River, spanning altitudes from 683 m asl to 1987 m asl (Supplementary Table S1) and sequenced. Genomic DNA was extracted from muscle tissue, and whole-genome sequencing (WGS) was conducted on the DNBSEQ-T7 platform, producing 150-bp paired-end reads. The average sequencing depth of the samples was approximately 30×. Detailed information on sample collection, including sampling sites, altitude, and sequencing depth, is listed in Supplementary Table S1. Genome size estimation and assembly of Schizothorax nukiangensis The genome size of S. nukiangensis was estimated by flow cytometry using fresh peripheral blood, with chicken (Gallus gallus domesticus, 1C=1.25 pg) as an internal reference. Nuclei suspensions from fish and chicken blood were mixed, stained with propidium iodide, and analyzed on a BD FACSymphony A1 flow cytometer (BD, USA). Genome size was calculated based on the fluorescence intensity ratio between the sample and the reference, with measurements performed in triplicate. To further estimate genome size, heterozygosity, and repeat content with sequences, Jellyfish v.2.3.1 (parameters: -m 31 -s 1G -t 4 -C) (Marçais & Kingsford, 2011) and GenomeScope v.2.0 (Ranallo-Benavidez et al., 2020) were used to generate the k-mer frequency distribution. The genome was primarily assembled from 128.35 Gb of CCS reads and 192.26 Gb of clean Hi-C reads using hifiasm v.0.24.0-r703 with the parameters -l 0 --n-hap 4 (Cheng et al., 2021, 2022). Chromosome assignment was performed using Hi-C technology. High-quality paired-end reads were aligned to contigs using BWA v.0.7.17-r1188 (Li & Durbin, 2009) to calculate contact frequency. Following the YaHS v.1.2.2 pipeline with default parameters (Zhou et al., 2022), the process began with an initial round of error correction, followed by the generation of an interaction matrix using directional brackets in conjunction with Juicer, which enabled subsequent inspection and manual correction via the Juicebox assembly tool (Durand et al., 2016). Multiple strategies were applied to evaluate genome assembly quality. The phasing accuracy of haplotypes was assessed using Merqury v.1.4.1 (Rhie et al., 2020) with default parameters. Genome completeness was evaluated using BUSCO v.5.8.0 with the actinopterygii_odb10 (Tegenfeldt et al., 2025). In addition, both HiFi and Hi-C reads were realigned to the assembly to calculate the mapping ratio and assess the reliability of the genome assembly. Annotation of repeat elements and protein-coding regions A combination of homology-based and de novo approaches was used to identify repeat elements. Known transposable elements (TEs) were identified using RepeatMasker v.4.1.8 (parameters: -xsmall, -e rmblast, -nolow, -no_is, -norna) (https://www.repeatmasker.org/) (Tarailo-Graovac & Chen, 2009) with the Repbase TE library(Jurka et al., 2005). Additionally, a de novo repeat library was constructed using RepeatModeler v.2.0.5 (Flynn et al., 2020). The two libraries were then merged and used to screen the genome for repetitive sequences with RepeatMasker. Protein-coding genes were predicted using the BRAKER3 v.3.0.6 pipeline with default parameters (Gabriel et al., 2024), which integrates ab initio gene prediction via GeneMark-ETP v.4.71 (Brůna et al., 2024), TSEBRA v.1.1.2.3 (Gabriel et al., 2021), and AUGUSTUS v.3.5.0 (Stanke et al., 2008) with RNA-seq evidence to improve annotation accuracy. RNA-seq alignments provided extrinsic hints to guide gene structure prediction, while GeneMark-ETP v.4.71 enabled model training based on transcript evidence. For homology-based predictions, GeMoMa v.1.9 (Keilwagen et al., 2018) was used to compare the genome of S. nukiangensis with 10 related species, including Culter alburnus (GCA_040182925.1), Danio rerio (GRCz11 release-112, https://ftp.ensembl.org/pub/release-112/fasta/danio_rerio/dna, Gasterosteus aculeatus (GCA_016920845.1), Megalobrama amblycephala (GCA_018812025.1), Oryzias latipes (GCA_002234675.1), Cyprinus carpio (GCA_018340385.1), Carassius auratus (Luo et al., 2020) Ctenopharyngodon idella (GCA_019924925.1), Oreochromis niloticus (GCA_001858045.3), and Poropuntius huangchuchieni (Chen et al., 2021). For RNA-seq data, both de novo and reference-guided transcriptome assemblies were performed. RNA-seq reads were aligned to the genome using HISAT2 v.2.2.1 with default parameters. (Kim et al., 2019), and genome-guided transcript assembly was carried out with StringTie v.1.3.3b (Shumate et al., 2022). TransDecoder v.5.7.1 was employed to identify candidate coding regions by predicting open reading frames (ORFs), providing protein-coding evidence to support gene annotation. In parallel, Trinity v.2.8.5 (Grabherr et al., 2011) was used for de novo transcriptome assembly. To improve gene structure annotation, a comprehensive transcriptome database was constructed using both RNA-seq and Iso-Seq assemblies following the Program to Assemble Spliced Alignments (PASA v.2.4.1) (Haas et al., 2008) pipeline with default parameters. Transcripts were aligned to the reference genome using GMAP v.2021.08.25 and BLAT v.36, and PASA v.2.4.1 subsequently constructed high-quality gene models, annotated alternative splice isoforms, and refined UTR and exon-intron structures. EVidenceModeler v.2.1.0 (Haas et al., 2008) was used to integrate all predicted gene sets into a nonredundant, comprehensive, and high-confidence gene annotation set. Finally, the PASA pipeline was used to update the EVidenceModeler output by adding UTR annotations and alternative splicing models. The resulting protein-coding genes were functionally annotated using the eggNOG-mapper (Cantalapiedra et al., 2021) comprehensive database. Phylogenetic reconstruction of haplotype chromosomes based on single-copy orthologs Phylogenetic relationships among the 25 chromosome sets of each haplotype were inferred using single-copy orthologous genes. First, protein-coding sequences from S. nukiangensis and comparison species (D. rerio, Ca. auratus, and Cy. carpio) were clustered using OrthoFinder v.2.5.4 (Emms & Kelly, 2019) to identify single-copy orthologous gene families. The resulting orthologues were then partitioned into 25 groups based on their chromosome of origin. For each group, coding sequences were aligned using MUSCLE v.3.8.31 (Edgar, 2022). Maximum-likelihood (ML) phylogenies were reconstructed with RAxML v.8.2.12 (Stamatakis, 2014), under the JTT+I+G+F substitution model and with 1000 bootstrap replicates to assess node support. The resulting ML trees were exported in Newick format and interactively visualized using FigTree v.1.4.4. Genome comparison and synteny analyses To investigate relationships of S. nukiangensis to other cyprinids, we conducted a comparative genomics analysis using the published genomes of 10 species including Cirrhinus molitorella (Tu et al., 2025), D. rerio, Ct. idella, Cu. alburnus, M. amblycephala, P. huangchuchieni, Sinocyclocheilus grahami (Yin et al., 2021), Oxygymnocypris stewartii (Liu et al., 2019), Ca. auratus, and Cy. carpio. Orthologous gene families were identified using OrthoFinder based on protein sequences. Single-copy orthogroups shared across all species were retained for phylogenetic inference. Protein sequences from these orthogroups were aligned using MUSCLE v.3.8.31, and the resulting alignments were stringently trimmed using Gblocks v.0.91b (Talavera & Castresana, 2007) to remove ambiguously aligned and poorly conserved regions. Phylogenetic tree reconstruction was performed using the ML method implemented in RAxML, with automatic selection of the best-fit substitution model and 1000 ultrafast bootstrap replicates to assess node support. Divergence times were estimated using a Bayesian relaxed molecular clock implemented in the MCMCtree module of PAML v.4.9j (Yang, 2007), calibrated with fossil records and divergence time priors obtained from the TimeTree database (Kumar et al., 2017). Gene family expansion and contraction analyses were carried out using café (Mendes et al., 2021), based on a birth-death process model mapped onto the phylogeny. Significantly expanded or contracted gene families (false discovery rate (FDR)<0.05) were subjected to Gene Ontology (GO) and KEGG pathway enrichment analysis using Fisher’s exact test with multiple testing correction. Finally, genome synteny between S. nukiangensis and other species was analyzed using the jcvi v.1.5.4 (Tang et al., 2024) toolkit to identify conserved collinear blocks across genomes. Whole-genome duplication (WGD) evidence We investigated and compared WGD events between the S. nukiangensis genome and 8 Cyprinidae species (D. rerio, Ct. idella, Cu. alburnus, M. amblycephala, P. huangchuchieni, O. stewartii, Ca. auratus, and Cy. carpio). Paralogous gene pairs were identified using BLAST-based methods, and syntenic paralogs were detected using WGDI (Sun et al., 2022). The number of synonymous substitutions per synonymous site (Ks) was calculated for gene pairs using the WGDI pipeline. To estimate the timing of WGD events, we applied a synonymous substitution rate of 3.51×10-9 mutations per site per year, as previously established for common carp (David et al., 2003). Single nucleotide polymorphism (SNP) calling For variant calling, paired-end reads from each accession were mapped to the four haplotype genomes using BWA v.0.7.17-r1188. After filtering out low-quality alignments, SNP calling was conducted using the HaplotypeCaller module of the Genome Analysis Toolkit (GATK) (McKenna et al., 2010). High-confidence single nucleotide polymorphisms (SNPs) were obtained using VCFtools v.0.1.13 (Danecek et al., 2011) with the following parameters: --max-missing 0.8, --maf 0.05, --minDP 10, --maxDP 500, --minQ 30, --minGQ 30, --min-alleles 2, --max-alleles 2 (Lu et al., 2024). SNP annotation was performed based on genomic location and predicted coding effects, using the S. nukiangensis genome annotation as a reference, with SnpEff v.4.1g (Cingolani et al., 2012). Population genetics analyses Phylogenetic analysis was conducted using a SNP dataset derived from the S. nukiangensis genome with IQ-TREE v.2.2.2.6 (Minh et al., 2020). A ML phylogenetic tree was reconstructed using 1 000 ultrafast bootstrap replicates. Model selection was carried out with ModelFinder, implemented within IQ-TREE, which evaluated 242 candidate nucleotide substitution models. The best-fitting model, TVM+F+ASC+R2, was selected based on the bayesian information criterion (BIC) and subsequently applied for tree inference. Population structure analysis (K=2–5) were performed using PLINK v.1.90 (Chang et al., 2015) and TBtools v.2.482 (Chen et al., 2023) respectively, based on the same SNP dataset. Demographic history To investigate the demographic history of S. nukiangensis, the effective population size (Ne) over the past 10 million years was estimated using the pairwise sequentially markovian coalescent (PSMC) model (Li & Durbin, 2011). The parameters were set as follows: -N25 -t15 -r5 -p "4+25*2+4+6", with a generation time (g) of 6 years (Ma, 2022; Mu et al., 2025). And Mutation rate was calculated through a linear regression of the relationship between log-genome size and log-mutation rate. The best fitting model for mutation rate μ in bp was μ=2.59×10−10×g0.584 for genome size g in Mb (genome size=3585.88 Mb), giving an estimate of 3.08×10-8 per bp per generation (Crellen et al., 2016; Lynch, 2010). Selection analyses To detect signals of selective sweeps between the southern and northern groups, FST values were calculated using VCFtools with a sliding window size of 50 kb and a step size of 10 kb. XP-CLR (Chen et al., 2010) scores were also computed using the same window and step sizes. Genes overlapping with the identified sweep regions were extracted and considered candidate selective sweep genes. Quantitative real-time PCR (qRT-PCR) Analysis Total RNA was extracted from tissues using TRIzol reagent (Invitrogen, USA) and reverse-transcribed into cDNA using the PrimeScript RT reagent kit (Takara, Japan) following the manufacturer’s instructions. Quantitative real-time PCR was performed on a CFX96 Real-Time PCR System (Bio-Rad, USA) using SYBR Green Master Mix (Takara, Japan). The relative expression levels of target genes were calculated using the 2–ΔΔCt method, with ef1a as the internal reference. RNA sequencing and differential expression analyses Total RNA was extracted from intestinal tissues of S. nukiangensis individuals collected from three distinct altitudinal zones, with each elevation group represented by three biological replicates (n=9 in total) (Supplementary Table S3). We grouped the samples based on the individuals' sampling altitudes (Supplementary Table S3), with the following specific groupings: Group 1: Samples from an altitude of 1068 m asl; Group 2: Samples from an altitude of 718 m asl; Group 3: Samples from an altitude of 1255 m asl. All transcriptome data were used for haplotype-specific expression bias analysis. For the haplotype-specific expression bias analysis, we identified 1:1:1:1 homologous gene pairs across hapA, hapB, hapC, and hapD using OrthoFinder v.2.5.4. Group 2 and Group 3 samples were used for differential gene expression analysis. RNA libraries were constructed and sequenced on the Illumina platform. Clean reads were aligned to the reference genome using HISAT2 with default parameters. Gene expression levels were quantified for each sample using featureCounts v.2.0.3 (Liao et al., 2014), and differential expression analysis was performed using DESeq2 v.1.50.2 (Love et al., 2014). Genes with a fold change≥2 and a FDR<0.05 were considered significantly differentially expressed. RESULTS High-Quality haplotype-resolved genome assembly and annotation of autotetraploid Schizothorax nukiangensis We determined the genome size and karyotype of S. nukiangensis for the first time. Flow cytometry estimated its nuclear DNA content at 2.15 pg (Genome size≈1.80 Gb), using Chicken as a reference (Supplementary Figure S1A; Supplementary Table S4). Karyotype analysis showed 100 chromosomes (Supplementary Figure S1B), a count often associated with tetraploidy in cyprinids. K-mer analysis conducted with Jellyfish confirmed the autotetraploid nature of S. nukiangensis (Figure 1A; Supplementary Figure S1C). De novo assembly using hifiasm produced 853 scaffolds totaling 3585.89 Mb, with an N50 of 35.11 Mb. Integration of Hi-C data anchored 3463.55 Mb (96.59% of the assembly) onto 100 chromosomes, resulting in a haplotype-resolved genome assembly (Figure 1B; Supplementary Figure S2; Supplementary Table S5). These 100 chromosomes were organized into 25 homologous groups, each comprising four homologous chromosomes corresponding to haplotypes A, B, C, and D (Figure 1C; Supplementary Table S5). Genome assembly and annotation of Schizothorax nukiangensis A: K-mer frequency distribution (k=31) showing peaks at depths of 32, 65, and 130, consistent with autotetraploid genome characteristics. B: Hi-C contact heatmap of the assembled chromosomes, illustrating chromosomal interactions and validating assembly continuity and phasing accuracy. C: Circos plot depicting the genomic landscape of Schizothorax nukiangensis: I. 100 chromosomes organized into 25 homologous groups; II. GC content; III. repeat density; IV. gene density; V. syntenic relationships between chromosomes visualized by curved links. D: K-mer multiplicity spectrum comparing distinct k-mers absent from the assembly (black) and present with different copy numbers: 1 (red), 2 (blue), 3 (green), 4 (purple), and ≥4 times (orange) in the assembly relative to raw HiFi reads. Multiple validation approaches demonstrated the high accuracy and reliability of the S. nukiangensis genome assembly. K-mer spectra analysis with Merqury clearly confirmed the tetraploid nature and accurate phasing of the assembled haplotypes (Figure 1D; Supplementary Figure S3). Base-level quality assessment revealed an exceptionally low estimated error rate of 3.13×10-7 and a consensus quality value (QV) of 65.05, indicating excellent agreement between the HiFi reads and the assembly (Supplementary Table S6; Supplementary Figure S4). Hi-C contact maps exhibited strong diagonal signals, supporting correct phasing, ordering, and orientation of chromosomes (Figure 1B). Mapping of clean reads yielded high alignment rates, exceeding 99% for both PacBio HiFi and Hi-C reads, and over 95% for RNA-seq reads (Supplementary Table S7). BUSCO analysis demonstrated 99.76% completeness of the vertebrata_odb10 gene set across haplotypes (Supplementary Table S8; Supplementary Figure S5). Repetitive elements constitute more than half of the S. nukiangensis genome, with comparable proportions across all four haplotypes (Supplementary Table S5). Comprehensive repeat annotation revealed that repetitive sequences account for 52.84%, 51.33%, 50.75%, and 50.94% of haplotypes A, B, C, and D, respectively, resulting in an overall repeat content of approximately 51.96% (Supplementary Table S5). Transposable elements (TEs) predominated in the genome, accounting for 34.89% of the assembly, with retrotransposons contributing 13.32% of this total (Supplementary Table S9). This high repeat proportion is consistent with the large genome size and autotetraploid nature of S. nukiangensis. Divergence analyses of TEs showed that DNA transposons, LINEs, long terminal repeats (LTRs), and short interspersed nuclear elements (SINEs) have all substantially contributed to genome expansion (Supplementary Figure S6). Among these, DNA transposons and LINEs exhibited prominent divergence peaks at ~2%–5%, indicating recent bursts of TE amplification. Gene annotation revealed a high number of protein-coding genes per haplotype and near-complete coverage of the vertebrata_odb10 gene set. Specifically, 25956, 25096, 25878, and 26193 protein-coding genes were identified in haplotypes A, B, C, and D, respectively, collectively covering 98.48% of vertebrate core genes (Supplementary Table S8; Figure S5). Functional annotation of these genes was performed using the comprehensive eggNOG-mapper database. In total, 39869, 38137, 39883, and 40139 transcripts were predicted for in haplotypes A, B, C, and D, respectively, corresponding to an average of 1.52–1.54 splice isoforms per gene (Figure 1C; Supplementary Table S5). Evidence for autotetraploid origin of Schizothorax nukiangensis Among Cyprinidae species, polyploidy is common (Li et al., 2024c), with origins spanning both autopolyploid and allopolyploid events. Notably, well-studied allotetraploid species such as Ca. auratus (Luo et al., 2020) and Cy. carpio (Xu et al., 2019a) exhibit pronounced subgenomic divergence, enabling clear partitioning into distinct subgenomes. To elucidate the polyploid origin of S. nukiangensis, we conducted k-mer frequency analysis, collinearity assessment, and phylogenetic reconstruction. The 31-mer frequency plot displayed three distinct peaks at depths of approximately 32×, 65×, and 128× (Figure 1A), a pattern consistent with an autotetraploid genome. Analysis of nucleotide heterozygosity revealed a high proportion of homozygous (aaaa, 97.95%) and aaab-type k-mers, with 1.19% aaab and 0.57% aabb heterozygous k-mers (Supplementary Table S10). This distribution matches the expected pattern for autotetraploids, in which aaab-type heterozygosity exceeds aabb, further supporting the autotetraploid nature of S. nukiangensis. Synteny analysis revealed extensive conservation of collinear blocks among the four haplotypes, as demonstrated by both the synteny dot plot and the collinearity alignment (Figures 2A, B). This high degree of genomic homology strongly supports the hypothesis of autopolyploidy. Evidence supporting the autotetraploid origin of Schizothorax nukiangensis A: Syntenic relationships among the four haplotypes and representative cyprinid genomes. Comparison of syntenic blocks among Ca. auratus (Caru_M and Caru_P), D. rerio (Drer), M. amblycephala (Mamb), and the four haplotypes (hapA-hapD) of Schizothorax nukiangensis. B: Dot plot illustrating the syntenic blocks among the four haplotypes of Schizothorax nukiangensis and the reference genome of D. rerio. C: Phylogenetic tree of four Schizothorax nukiangensis haplotypes (hapA-hapD) and eight related cyprinid species, constructed from genome-wide orthologs; yellow blocks show Schizothorax nukiangensis and the sister clade of Schizothorax nukiangensis, orange blocks show Cy. carpio, Ca. auratus, Si. grahami M subgenome clade and blue blocks show Cy. carpio, Ca. auratus, Si. grahami P subgenome clade, grey blocks show Culter alburnus, D. rerio, M. amblycephala, respectively; red block represents the data generated in this work (hapA, hapB, hapC, and hapD represent the 4 haplotypes, respectively), and the green block represents downloaded species data. D: Upset plot illustrating the distribution of orthogroups shared across haplotypes, including unique and shared orthologs. To further validate this inference, we constructed a ML phylogenetic tree using genome-wide orthologs from eight closely related cyprinid genomes along with the four haplotypes of S. nukiangensis. The resulting tree showed that the four haplotypes clustered tightly together with high bootstrap support (Figure 2C), consistent with an autotetraploid origin. Additional individual phylogenetic trees constructed for each of the 25 chromosomes exhibited congruent topologies, reinforcing this conclusion (Supplementary Figure S7). In contrast, allotetraploid species such as Cy. carpio and Ca. auratus display distinct subgenomic topologies, with duplicated haplotypes partitioned into separate clades corresponding to different progenitor lineages, which is not observed in S. nukiangensis (Figure 2C; Supplementary Figure S7). Furthermore, orthogroup analysis among the four haplotypes revealed extensive genetic similarity and a high degree of shared gene content. A total of 24466 orthogroups were identified, of which 17460 were shared by all four haplotypes, 4403 by three haplotypes, 2386 by two haplotypes. Only 188 orthogroups were unique to a single haplotype (hapA: 65; hapB: 45; hapC: 43; hapD: 35) (Figure 2D). This high level of gene content uniformity provides additional evidence for an autotetraploid origin via WGD within a single lineage. Phylogeny and adaptive genome evolution of autotetraploid Schizothorax nukiangensis An ML phylogenetic tree was reconstructed using 698 single-copy orthologous genes from S. nukiangensis and ten other cyprinids: Cu. alburnus, D. rerio, Ct. idella, Ci. molitorella, M. amblycephala, Cy. carpio, Ca. auratus, O. stewartii, P. huangchuchieni, and Si. grahami (Figure 3A). All nodes in the resulting phylogeny were supported by strong bootstrap values (100%). Divergence time estimation using the Bayesian MCMCtree method indicated that S. nukiangensis diverged from the Barbinae lineage approximately 7.58 million years ago (Ma), with a subsequent split from O. stewartii around 5.33 Ma. These divergence events coincide with major uplift phases of the Qinghai-Xizang Plateau (Royden et al., 2008), suggesting a potential role for geological processes in shaping the evolutionary trajectory of S. nukiangensis (Li et al., 2024c). Phylogenetic relationships, comparative genomics, and evolutionary analyses of Schizothorax nukiangensis A: Phylogenetic tree with divergence times and gene family expansions (blue) and contractions (red) for 11 cyprinid species, including Cu. alburnus, D. rerio, Ct. idella, Ci. molitorella, M. amblycephala, Cy. carpio, Ca. auratus, O. stewartii, P. huangchuchieni, Si. grahami. B: Venn diagram of orthologous gene families of Cy. carpio, Ca. auratus, Si. grahami, Schizothorax nukiangensis, M. amblycephala, O. stewartii and P. huangchuchieni. C: Kernel density plots of Ks distributions for paralogous gene pairs within Schizothorax nukiangensis haplotypes. D: Ks distributions of orthologous gene pairs among Schizothorax nukiangensis and related species; the upper panel shows distributions consistent with Cyprinidae-wide WGD, while the lower panel highlights recent speciation between Schizothorax nukiangensis, P. huangchuchieni, and O. stewartii. In total, 1127 gene families were expanded, collectively gaining 4515 genes, while 1470 families were contracted, resulting in the loss of 1641 genes. This net expansion is reflected by a positive average expansion index (0.1257), suggesting substantial gene family innovation following WGD events (Figure 3A). GO and KEGG enrichment analyses of the 702 significantly expanded gene families highlighted biological processes central to genome plasticity and innate immunity, including DNA recombination (p<0.01), transposon activity (p<0.01), and signaling pathways such as NF-kappa B (p<0.01) and NOD-like receptor signaling (p<0.01) (Supplementary Figure S8). These findings suggest that TE activation and immune-related gene family expansions may have played a role in facilitating rapid adaptation to the pathogen-rich, high-altitude river environments inhabited by S. nukiangensis (Chuong et al., 2017; Oliver & Greene, 2009) (Supplementary Table S11; Supplementary Table S12). Orthologous gene family clustering across seven representative cyprinid species revealed a shared core of 9 119 gene families conserved across all species, reflecting essential biological functions maintained throughout Cyprinidae evolution (Figure 3B). Notably, S. nukiangensis possessed 441 unique gene families absent from the other genomes analyzed, representing lineage-specific genetic innovations potentially underlying ecological specialization. These gene families were enriched in functions related to transposon activity, nucleic acid metabolism, and developmental regulation, which may influence lineage-specific traits. While speculative, such features could be associated with physiological or morphological plasticity relevant to dynamic riverine habitats (Supplementary Table S13). Additionally, we identified a set of positively selected genes associated with high-altitude adaptation. These genes are primarily involved in cell cycle regulation, oxygen sensing, signal transduction, and stress response. Nitric oxide synthase (NOS) and natriuretic peptide receptor genes likely regulate nitric oxide production and vascular dilation under hypoxic conditions. Phospholipase D2 and TRP channels play key roles in calcium homeostasis and cellular stress responses. Cell division cycle genes and CDK2 contribute to cell proliferation and tissue repair, whereas EGLN1 regulates the HIF-1 pathway under low-oxygen stress. Collectively, these genes facilitate adaptation to high-altitude environments by enhancing oxygen utilization, promoting cellular repair, and improving stress tolerance (Supplementary Table S14). Meanwhile, WGD analysis was conducted using WGDI, which identified paralogous gene pairs among haplotype assemblies. Kernel density estimation of synonymous substitution rates (Ks) for these paralogous pairs revealed a prominent peak at Ks≈0.0018–0.0066, indicative of a recent large-scale duplication event likely corresponding to the tetraploidization of S. nukiangensis. Based on the Ks distribution, this WGD event is estimated to have occurred approximately 0.68 Ma, suggesting a relatively recent timeframe for the origin of autotetraploidy in S. nukiangensis (Figure 3C; Supplementary Figure S9). Analysis of gene pairs from nine representative cyprinid genomes revealed distinct patterns of WGD history across specie. In several species, a single major Ks peak corresponded to the well-characterized third round of WGD (3R-WGD), a duplication event conserved across all teleost. In contrast, other cyprinid species displayed bimodal Ks distributions, reflecting both this ancestral event and an additional, lineage-specific fourth round of WGD (4R-WGD) (Figure 3D). In S. nukiangensis, a clear bimodal Ks distribution was observed, with a prominent peak at low Ks values (≈0.0034), consistent with a recent duplication event, and a broader secondary peak at higher Ks values (≈1.9035–2.1065), reflecting the ancient 3R-WGD shared by all teleost (Figure 3D, upper panel). Additionally, Ks distributions for collinear orthologs between S. nukiangensis and P. huangchuchieni or O. stewartii also revealed bimodal patterns. Primary peaks occurred at Ks≈0.1153 (Snuk_vs_Phua) and Ks≈0.0913 (Snuk_vs_Oste), corresponding to more recent speciation events, while broader secondary peaks at Ks≈1.9133 and ≈1.9854, respectively (Figure 3D, lower panel), reflect deeper evolutionary relationships. The similarity of these secondary peaks suggests a shared ancient WGD event in the common ancestor of these species. Collectively, these findings indicate that S. nukiangensis has undergone both recent lineage-specific genome duplication and retains genomic signatures of an ancient WGD shared with P. huangchuchieni and O. stewartii. These results illuminate the complex evolutionary history of schizothoracine fishes and underscore the pivotal role of polyploidization and subsequent diversification in shaping cyprinid genomic architecture. Population structure and demographic history of autotetraploid Schizothorax nukiangensis along an altitudinal gradient To investigate the population genetic structure of S. nukiangensis along the Salween River, we sampled 37 individuals across a pronounced elevational gradient (632–1987 m asl) in The Hengduan Mountains (Figure 4A; Supplementary Table S1). The sampling sites represented habitats with distinct vertical stratification, reflecting the species’ adaptation to heterogeneous, high-altitude riverine environments. Whole-genome resequencing of each individual was performed at an average depth of ~33.54×, generating approximately 60.37 Gb of clean data per sample (Supplementary Table S1). Following read alignment to the S. nukiangensis reference genome, we identified 6553219 high-confidence SNPs. Based on these genome-wide SNPs, we constructed a ML unrooted phylogenetic tree, which revealed three major genetic clusters, designated as populations H, M, and L (popH, popM, and popL) (Figure 4B; Supplementary Table S1). PopH was predominantly distributed at high elevations (1793–1987 m asl), popM at mid-elevations (1190–1286 m asl), and popL at low elevations (632–683 m asl). Nonetheless, several individuals from each group were found outside their representative elevational ranges, indicating overlapping distributions and potential gene flow among populations. Population structure analysis based on genome-wide admixture modeling further revealed complex genetic patterns across elevations (Figure 4C). Although the lowest cross-validation (CV) error was observed at K=1, suggesting overall weak genome-wide differentiation, higher K-values (K=2–5) uncovered pronounced clustering patterns that aligned with the phylogenetic results. These patterns indicate the presence of cryptic population structure, likely shaped by ecological or historical factors rather than strict altitudinal isolation. Sampling locations, phylogenetic relationships, and population structure of Schizothorax nukiangensis along an altitudinal gradient of the Salween River. A: Sampling sites of 37 Schizothorax nukiangensis individuals along the Salween River in the region of The Hengduan Mountains, southwestern China. Elevations of sampling sites range from 632 to 1987 m asl, spanning habitats with pronounced vertical stratification. B: ML unrooted phylogenetic tree constructed from genome-wide SNPs, revealing three major genetic population (H, M, L) with individuals from different elevations intermixed within clusters. C: Population structure analysis inferred from genome-wide SNPs using admixture modeling at K=2 to K=5, showing varying degrees of clustering among individuals. PSMC analysis of individual genomes from eight sampling localities revealed highly consistent historical demographic trajectories across populations (Supplementary Figure S10). Estimates of effective population size (Ne) indicated a substantial decline beginning approximately 1 Ma, culminating in pronounced bottlenecks during the late Pleistocene (~104–105 years ago), coinciding with major glaciation events on the Qinghai-Xizang Plateau. Following these bottlenecks, all populations exhibited clear signals of demographic recovery, with Ne gradually increasing toward the present. These results suggest that S. nukiangensis experienced synchronous population contractions and subsequent expansions in response to Pleistocene climate oscillations, which have shaped the species’ present-day genetic diversity and distribution along the Salween River. Population genomic analyses reveal candidate regions of altitudinal adaptation of autotetraploid Schizothorax nukiangensis FST comparison analysis revealed consistent and strong differentiation signals on chromosome chr03A between popH (high altitude) and the mid- and low-altitude populations (popM and popL). Genomic regions exceeding the top 1% FST threshold (red dashed line) were designated as candidate selective sweeps (Figure 5A; Supplementary Figures S11A, B). By intersecting the top 1% FST peaks from both comparisons on chromosome chr03A, we identified a robust set of candidate genes predominantly enriched in transcriptional regulation (Supplementary Table S15). GO enrichment analysis of these candidate genes revealed significant molecular functions related to DNA binding and transcription factor activity. In particular, several key functions related to transcription regulation were identified, including nucleic acid binding transcription factor activity (GO:0001071), sequence-specific DNA binding (GO:0003700), and RNA polymerase II core promoter proximal region sequence-specific DNA binding (GO:0000978). These findings indicate that the genes within this region are involved in essential regulatory processes, such as the control of transcription and the binding of specific DNA sequences, which may contribute to adaptive responses to high-altitude environments. Additional functions like scavenger receptor binding (GO:0005124) and N-formyl peptide receptor activity (GO:0004982) further suggest roles in immune response and environmental stress adaptation. Genomic signatures of selection and candidate genes in Schizothorax nukiangensis populations A: Manhattan plot of genome-wide FST values between high-altitude (popH) and low-altitude (popL) populations. Red dashed line indicates top 1% FST threshold. B: Manhattan plot of XP-CLR scores for the same population comparison; top 1% threshold marked with a red dashed line. C: Scatterplot of FST versus log(πL/πH), highlighting candidate regions (red) exceeding top 1% thresholds in both metrics. Marginal histograms show the distribution of each variable across the genome. D: Line plots of XP-CLR, FST, and log(π-ratio) in candidate regions on Chr05B, respectively, showing co-localized peaks of selection signals. E: Line plots of XP-CLR, FST, and log(π-ratio) in candidate regions on Chr23B, respectively, showing co-localized peaks of selection signals. F: Genotype heatmap of candidate gene ift70, illustrating distribution of heterozygous, homozygous, and homozygous mutant genotypes across individuals from popH, popM, and popL. Parallel Cross Population Composite Likelihood Ratio (XP-CLR) analyses comparing popH with both popM and popL consistently revealed strong signals of selection on chromosome chr04A. Genomic regions exceeding the top 1% threshold (red dashed line) were designated as candidate selective sweeps (Figure 5B; Figures S11C, D). Overlapping XP-CLR peaks revealed several candidate genes on chr04A, including genes associated with critical cellular functions such as protein interaction, receptor signaling, and gene regulation (Supplementary Table S16). Notably, genes such as transmembrane protein 131-like and endothelin receptor B (ETB) are implicated in cellular signaling pathways and may play roles in stress adaptation, as transmembrane proteins and receptors are often involved in transmitting signals in response to environmental changes. Tripartite motif-containing protein 2-like and mitogen-activated protein kinase kinase kinase kinase 4-like are involved in protein regulation and immune response pathways, where tripartite motif proteins are known to regulate various cellular processes, and MAPK signaling plays a critical role in stress responses. Genes like zinc finger protein basonuclin-2-like and zinc finger protein 501-like indicate regulatory roles in gene expression, where zinc finger proteins are key in regulating transcription and protein synthesis. These findings suggest that the genes on chr04A are involved in a broad range of adaptive processes, including immune modulation, cellular signaling, and stress response mechanisms, all potentially contributing to adaptation in high-altitude environments. To further refine the signals of local adaptation and identify robust candidate genes, we integrated three genomic metrics, FST, XP-CLR, and the nucleotide diversity ratio (π-ratio, log(πL/πH)), by comparing popH and popL. Scatterplots of FST versus π-ratio (highlighted in red) delineated genomic regions exceeding the top 1% thresholds for both metrics (Figure 5C), and intersecting these with XP-CLR signals further strengthened evidence for selection. This integrative approach identified 11 robust candidate genes likely involved in altitudinal adaptation (Figure 5B; Supplementary Table S17), including sdk1a (cell adhesion), fgf23 and fgf6b (skeletal and muscle development), and mx (immune defense). Notably, fgf23 and fgf6b are associated with mineral metabolism and muscle development, respectively. These functions may underlie morphological or physiological traits that facilitate local adaptation to high-altitude river environments, such as strong currents and fluctuating environmental pressures. For example, enhanced muscle growth mediated by fgf6b could support locomotor performance in turbulent waters, while fgf23-related mineral regulation may influence skeletal maintenance under variable resource availability. These adaptations likely contribute to the ecological success of S. nukiangensis in its extreme habitat. Detailed regional analyses of two representative candidate regions on Chr05B and Chr23B revealed coincident peaks in XP-CLR, FST, and log(π-ratio) profiles (Figures 5D, E), reinforcing the likelihood of strong local selection at these loci. Furthermore, a genotype heatmap for the candidate gene ift70 (Figure 5F) showed distinct allele frequency patterns among popH, popM, and popL. Notably, popH exhibited a high frequency of homozygous mutations, suggestive of altitude-associated selective pressure. Interestingly, FST and XP-CLR analyses, each performed as pairwise comparisons between popH and both popM and popL, highlighted divergent regions on different chromosomes: intersections of top FST peaks were enriched on Chr03A, while XP-CLR intersections were concentrated on Chr04A. This non-overlapping distribution suggests that each metric captures distinct signatures of selection, underscoring the complementary value of combining multiple approaches to detect selective sweeps. Final candidate genes were identified by integrating FST, XP-CLR, and log(πL/πH) metrics, specifically focusing on the comparison between popH and popL. This integrative analysis yielded a focused and robust set of loci potentially involved in altitudinal adaptation in S. nukiangensis, including several significantly differentiated genes such as fgf6b, fgf23, mx, and ift70. Homeolog expression bias and altitude-associated transcriptional responses in the autotetraploid Schizothorax nukiangensis To investigate expression bias among the four haplotypes of S. nukiangensis, we quantified transcript abundance for 15 063 homeologous gene groups present in a complete 1:1:1:1 correspondence across hapA, hapB, hapC, and hapD (Figure 6A; Supplementary Table S18). Expression bias was defined as a ≥2-fold increase (log2(fold change)≥1) in expression of a single haplotype compared to the other three. Analysis revealed that expression dominance was relatively rare: only 315 homeolog groups (2.1%) displayed significant bias toward one haplotype, whereas the vast majority (97.9%) showed no detectable bias, indicating largely balanced expression among homeologs (Figure 6B). GO enrichment analysis of these biased genes revealed significant overrepresentation of functions related to mitochondrial activity, oxidative phosphorylation, and ribosomal structure. Specifically, enriched terms included respiratory chain (GO:0070469), mitochondrial inner membrane (GO:0005743), and ATP metabolic process (GO:0046034), indicating a functional preference toward energy production and protein synthesis. Notably, several components of the electron transport chain, including NADH dehydrogenase complex and respiratory chain complex I, were also significantly enriched. These biased genes may reflect subgenome-specific optimization of energy metabolism and translational control, potentially contributing to the physiological demands of surviving in cold, oxygen-variable, high-altitude, and high-flow river environments such as the Salween River (Supplementary Table S19). Homeolog expression bias and altitude-associated transcriptional responses in Schizothorax nukiangensis 多倍化造成基因组冗余,并推动了淡水鱼类生态多样性的形成与扩展。然而,高海拔生态系统中近期多倍化的裂腹鱼类基因组结构及其适应性意义仍知之甚少。本研究确定了四倍体怒江裂腹鱼(Schizothorax nukiangensis)基因组大小和染色体数目,组装了四倍体怒江裂腹鱼的单倍型基因组,组装总基因组大小为3.59 Gb,挂载到了100条染色体,分为25个同源染色体组。kmer评估、共线性分析以及系统发育树等多个证据一致支持怒江裂腹鱼的同源多倍化起源,并推断其谱系特异性多倍化事件发生在约0.68百万年前。在此基础上,第一,我们比较基因组学揭示了先天免疫、氧化还原调控和转座子活性相关基因家族在怒江裂腹鱼中显著扩张,表明基因组重塑可能促进了怒江裂腹鱼在恶劣的生态环境中适应。第二,群体基因组学分析揭示了怒江裂腹鱼三个海拔群体的细微群体结构,并鉴定出高海拔群体与骨骼发育、免疫防御和纤毛功能相关的环境适应候选基因,这与怒江裂腹对湍急河流的适应相一致。第三,转录组分析进一步显示,与免疫反应和氧化应激缓解相关的基因表达随海拔升高而上调,这与其他已报道的裂腹鱼一致,凸显了趋同的分子策略在海拔适应中的应用。综上所述,本研究组装了怒江裂腹鱼的单倍型基因组,揭示了其近期同源多倍化事件,在此基础上,发现因同源多倍化基因家族的扩张和同源基因特异性表达分化可能促进了怒江裂腹鱼对高原急溪等极端生态环境的适应。本研究为高海拔急溪和气候多变的淡水鱼类适应性基因组特征研究提供了重要见解。 A: Violin plots showing overall gene expression levels (log2(TPM+1)) across four haplotypes (hapA, hapB, hapC, hapD) in 15063 homeolog groups. B: Heatmap of 315 homeolog groups with significant expression bias (log2(fold-change)≥1) towards a single haplotype; color intensity indicates relative expression levels. C: Volcano plot of differential expression analysis in intestinal transcriptomes comparing high- (1255 m asl) and low-altitude (718 m asl) groups; red dots denote upregulated, blue dots downregulated genes (P-adjusted<0.05, | log2(fold change) |≥1). D: GO enrichment analysis of upregulated genes from high-altitude samples, highlighting terms related to immune response, leukocyte activation, and oxidative stress adaptation. To explore transcriptional mechanisms underlying local adaptation along the altitudinal gradient, we performed differential expression analysis of intestinal transcriptomes from two elevation groups: group 1 (1068 m asl) and group 2 (718 m asl) (Figure 6C). We identified 481 significantly upregulated and 60 downregulated genes (| log2(fold change) |≥1, P-adjusted<0.05) in the high-altitude group relative to the low-altitude group. GO enrichment analysis of upregulated genes revealed significant overrepresentation of biological processes related to innate and adaptive immunity, including response to bacterial molecules, leukocyte activation, and immune signaling pathways, as well as oxidative stress responses such as hydrogen peroxide catabolism (Figure 6D; Supplementary Table S20). These enrichments suggest that S. nukiangensis individuals inhabiting higher altitudes enhance immune function and oxidative stress defenses to cope with increased environmental stressors, such as hypoxia, temperature fluctuations, and pathogen pressures, representative of fast-flowing mountain streams. DISCUSSION Polyploidy might be a major driving force in teleost diversification; yet understanding recent polyploidization events in high-altitude species remains challenging due to genome complexity and high redundancy. Within the family Cyprinidae, both autopolyploidy and allopolyploidy have occurred repeatedly, contributing significantly to species richness and ecological diversity (Glasauer & Neuhauss, 2014; Leggatt & Iwama, 2003). However, despite the importance of autopolyploidy, the genomes have remained poorly characterized due to technical difficulties in resolving highly similar homologous sequences. Genome assembly of polyploid species has been particularly challenging; early efforts often resulted in collapsed assemblies or artificial merging of homologous regions, especially in autopolyploids where sequence similarity among homeologs is high. Here, we present the first high-quality, haplotype-resolved genome assembly of the autotetraploid S. nukiangensis (Figures 1B, C), providing a foundational resource for investigating polyploid genome evolution in schizothoracine fishes. The exceptional continuity, completeness, and phasing accuracy of our assembly demonstrate the feasibility of generating high-quality, haplotype-resolved genomes for complex polyploid species (Figure 1D). This assembly not only enriches genomic resources for Cyprinid but also establishes a robust foundation for future studies on polyploidy in teleost fishes. Polyploidy has emerged as a key driver of diversification in fishes inhabiting extreme environments, with recent studies highlighting its repeated occurrence among schizothoracine fishes of the Qinghai-Xizang Plateau (Li et al., 2024c; Liu et al., 2019; Xiao et al., 2020). Our finding of an autotetraploid origin for S. nukiangensis is supported by multiple lines of evidence (Figure 2; Supplementary Figure S1C). The genome size of S. nukiangensis is 1.80 Gb (Supplementary Figure S1A; Supplementary Table S4), approximately twice that of diploid cyprinids, consistent with other known tetraploid species such as Cy. carpio (Xu et al., 2019a) and Ca. auratus (Luo et al., 2020). Its chromosome number (2n=100) is also comparable to that of other autotetraploid cyprinids (Figure 1B; Supplementary Table S21). K-mer analysis shows an autotetraploid pattern, with four copies of most loci (Figure 1D; Supplementary Figure S1C, S3). The high proportion of homozygous k-mers and the low heterozygosity further support an autotetraploid origin (Supplementary Table S10). Haplotype analysis demonstrates that the four assembled haplotypes of S. nukiangensis have similar genome sizes and gene numbers and cluster closely in phylogenetic trees (Figure 2C; Supplementary Figure S7; Supplementary Table S5), which is consistent with an autotetraploid origin. In contrast, allotetraploid species such as Cy. carpio and Ca. auratus display distinct subgenomic topologies, with subgenomes grouping into different phylogenetic clades. These results, supported by chromosome-level phylogenetic trees, confirm that S. nukiangensis is an autotetraploid species, highlighting the importance of polyploidization in its evolutionary history. The autotetraploidy of S. nukiangensis may have arisen through multiple mechanisms. One possibility is the formation of unreduced gametes caused by meiotic errors, producing gametes with double the normal chromosome number, which subsequently fused to form a tetraploid organism (Leggatt & Iwama, 2003). Alternatively, spontaneous genome duplication may have occurred due to mitotic errors. Environmental factors may also have facilitated polyploidization (Otto, 2007). Cooling events on the Qinghai–Xizang Plateau could have forced populations into geothermal refugia, such as hot springs. Temperature shocks during reproduction in these environments may have induced genome duplication in fertilized eggs, promoting polyploid formation (Gu et al., 2024; Yan et al., 2022). Our Ks-based analyses of S. nukiangensis revealed clear signatures of both ancient and recent WGD events (Figure 3C; Supplementary Figure S9). A prominent Ks peak at very low divergence (~0.0018–0.0066) indicates a recent autotetraploidization event estimated at ~0.68 Ma (Figure 3C), while broader secondary peaks align with the ancient 3R-WGD shared by all teleosts (Leggatt & Iwama, 2003). Similar patterns were reported for Schizothorax o’connori, in which Xiao et al. identified a Ks peak at ~0.0086 corresponding to a recent WGD~1.23 Ma (Xiao et al., 2020). These independent but temporally proximate polyploidization events occurred during Pleistocene uplift phases of the Qinghai-Xizang Plateau, a period marked by dramatic climatic and geological changes that imposed strong selective pressures (Wang et al., 2016; Zhang et al., 2018). Recent studies on snow carps across the plateau further revealed multiple independent and recurrent polyploidization events within schizothoracine fishes, each associated with population divergence and ecological adaptation (Li et al., 2024c; Wang et al., 2016). Climate change on the Qinghai-Xizang Plateau, particularly cooling and glacial expansion during the Pleistocene, profoundly influenced species distributions, population dynamics, and adaptive evolution. As temperatures declined and glaciers advanced, formerly warm and humid environments became colder and drier, causing habitat contraction and population isolation, which promoted speciation and increased biodiversity (Hewitt, 2000). The tectonic uplift of the plateau, which began in the Miocene and intensified during the Pliocene (Pan et al., 2012), further amplified these climatic shifts by altering hydrological systems and fragmenting habitats. These conditions may have facilitated autotetraploidization in species such as S. nukiangensis (Jia et al., 2020). PSMC analysis reveals a population bottleneck in S. nukiangensis during the Late Pleistocene (Supplementary Figure S10), consistent with climate-driven fluctuations in population size (Maron et al., 2015; Zhou et al., 2023). This bottleneck likely resulted from repeated cycles of glacial advance and uplift-induced habitat instability. Over the past 1000 years, the continued decline in effective population size corresponds with both climatic variability and intensified human disturbance. Human activities, including hydropower development, habitat destruction, and urban expansion, have further reduced suitable habitats and increased adaptive pressure on the species (Li et al., 2024a). Expanded gene families and positively selected genes reveal convergent molecular strategies for high-altitude adaptation. In S. nukiangensis, gene family expansions were significantly enriched in immunity, nucleic acid metabolism, transposition, cell repair, and metabolic regulation—paralleling patterns observed in other high-altitude schizothoracines like Schizopygopsis (Supplementary Figure S8; Supplementary Table S11; Supplementary Table S14) (Zhou et al., 2020, 2023, 2025). Meanwhile, we found that the unique gene family is involved in several key biological processes, particularly transposition, phospholipid metabolism, cell differentiation, and metabolic regulation. These functions may influence cell activity, genome stability, and immune regulation (Supplementary Table S13). The recent expansion of immune-related genes and bursts of TEs suggest a potential link between genome instability and rapid adaptation, consistent with the hypothesis that WGD provides genomic raw material under environmental stress (Leggatt & Iwama, 2003). However, compared with other schizothoracine species, the genome of S. nukiangensis shows more pronounced changes in transposon activity and genomic remodeling. These changes are likely linked to its more recent autotetraploidization event and the resulting higher genome instability. Such instability may have generated greater adaptive variation, facilitating the rapid evolution of S. nukiangensis in the highly dynamic high-altitude environment. Positive selection analyses further identified genes involved in neural development, neurotransmission, and immune regulation, echoing signatures from S. o’connori (Gao et al., 2024) and Schizopygopsis (Zhou et al., 2023). These convergent genomic patterns point to a shared adaptive strategy involving nervous system modulation, energy homeostasis, and immune enhancement to cope with the hypoxia, cold, and pathogen challenges of high-altitude river environments. Although S. nukiangensis is continuously distributed along the Salween River, population genomic analyses reveal clear genetic structuring into high-, mid-, and low-altitude population (Figure 4). The FST values reveal varying degrees of genetic differentiation along the altitudinal gradient (Supplementary Table S22). While modest in magnitude, the FST values still reflect some genetic differentiation between the high-altitude population and the medium- and low-altitude groups, likely reflecting adaptive responses of S. nukiangensis to harsher high-altitude conditions such as reduced oxygen availability and greater temperature fluctuations. In contrast, the relatively low FST value between the medium- and low-altitude populations suggests weaker genetic differentiation, potentially due to higher gene flow or similar environmental pressures between these populations. Together, these patterns demonstrate that altitude and associated environmental factors have played an important role in shaping the genetic structure of S. nukiangensis and driving local adaptation. Although the weak population structure observed in the admixture analyses (with the lowest CV error at K=1) may reduce the resolution of selection scans, the results are still informative for identifying loci under environmental selection, particularly in high-altitude populations. Additionally, the phylogenetic tree constructed from SNP data separates the populations into three major clusters, further supporting the presence of population structure across high-, middle-, and low-altitude groups. This structure is likely shaped by ecological barriers and life-history traits. Its demersal, adhesive-spawning strategy likely limits juvenile dispersal, promoting local differentiation despite geographic continuity (Chen et al., 2015). Occasional mismatches between sampling location and genetic clusters may result from downstream displacement during seasonal floods. This fine-scale structure aligns with previous findings in schizothoracines such as S. o’connori (Gao et al., 2024) and Schizopygopsis younghusbandi (Zhou et al., 2025), where elevational gradients and river fragmentation shaped divergence (Gao et al., 2024; Zhou et al., 2023). Genome-wide selective sweep analyses identified candidate genes likely involved in high-altitude adaptation in S. nukiangensis, including immune responses (e.g., sdk1a, lgals2a, mx), cellular transport (e.g., ift70), and metabolic regulation (e.g., eef2kmt, alg1). Additionally, fgf6b and fgf23 are associated with muscle growth regulation and phosphate metabolism, respectively (Figures 5A–C; Supplementary Table S17). These genes may support survival in hypoxic, fast-flowing mountain streams. Direct recurrence of the exact same loci in other Schizothoracinae is rare (likely due to lineage-specific demography and heterogeneous selection targets). However, several of our candidates fall into pathways repeatedly implicated in Qinghai-Xizang Plateau fish adaptation. Similar signatures in energy metabolism and DNA repair pathways have been reported in S. o’connori and Schizopygopsis younghusbandi, suggesting convergent molecular adaptations among schizothoracines (Xu et al., 2023; Yu et al., 2017; Zhou et al., 2023). Notably, mx gene has been characterized in Schizothorax richardsonii, supporting recurrent selection on immune modules in snow carp (Saxena et al., 2018). In teleost, fgf6b is strongly expressed in muscle tissues and promotes muscle fiber growth and differentiation (Hu et al., 2023; Xu et al., 2019b; Yang et al., 2024). Given the high mechanical and energetic demands of swimming in turbulent, oxygen-poor plateau rivers, positive selection on fgf6b might have contributed to improve muscle performance and endurance, optimizing movement and propulsion in turbulent. Differentiation at loci such as ift70 aligns with altitude-related expression patterns previously reported in S. nukiangensis (Yu et al., 2017). Furthermore, metabolomic differences between high- and low-altitude populations (Xu et al., 2023), particularly in lipid metabolism and oxidative stress, correspond to the functional roles of these candidate genes. We further validated the expression of several candidate genes using qRT-PCR (Supplementary Table S23). With the exception of lgals2a, all other genes showed higher expression levels in the high-altitude population compared with the low-altitude population. Notably, lrp2a and fgf6b exhibited significantly elevated expression in the high-altitude group (Supplementary Figure S12). Together, these findings highlight consistent local adaptation at key functional loci, despite overall weak population structure—underscoring the molecular basis of ecological specialization in high-altitude riverine environments. Our analysis of expression bias among the four haplotypes of S. nukiangensis revealed largely balanced expression, with only 2.1% of homeologous gene groups showing significant bias toward a single haplotype. The WGD event in S. nukiangensis occurred relatively recently (0.68 Ma), much later than in other schizothoracine species such as Gymnocypris przewalskii (2.55 Ma) (Tian et al., 2022) and S. o’connori (1.23 Ma) (Xiao et al., 2020). In those species, subgenome differentiation and functional specialization have occurred following WGD. For example, in G. przewalskii, gene expression analyses revealed subfunctionalization and neofunctionalization of duplicated genes, whereas in S. o’connori, bursts of transposable elements contributed to re-diploidization. In contrast, S. nukiangensis may not yet have undergone extensive subgenome divergence, which could explain the lower homeolog expression bias observed in this species. This weak bias may also indicate a higher degree of functional redundancy among homeologs, allowing one copy to compensate for another and maintain balanced gene expression. Additionally, regulatory mechanisms may stabilize gene dosage and prevent harmful over- or under-expression, particularly for genes involved in essential processes such as development and stress response. Collectively, these patterns suggest that S. nukiangensis is in an early post-WGD stage and is maintaining genomic stability through balanced homeolog expression. Differential expression analysis between high- and low-altitude populations identified multiple immune- and oxidative stress–related genes that were upregulated at higher elevations. This suggests that high-altitude individuals experience greater environmental stress, including hypoxia and temperature fluctuations, and have adapted by enhancing immune responses and oxidative stress defenses. These results reveal the transcriptional mechanisms contributing to local adaptation across the altitudinal gradient and emphasize the role of immune and metabolic processes in surviving the extreme conditions of fast-flowing, high-altitude rivers such as the Salween. Recent autotetraploidy in S. nukiangensis—coincident with late-Pleistocene climatic and geomorphic change—provides the genomic foundation for adaptation to high-altitude rivers. Within this background, TE dynamics and genomic innovation, together with expansions in immunity, metabolism, cellular transport and repair gene families, point to increased genomic flexibility under environmental stress. PSMC indicates bottleneck–expansion cycles consistent with glacial oscillations. Candidate genes associated with high-altitude adaptation are consistent with the physiological demands of cold, fast-flowing, and food-limited habitats. Notably, minimal homeolog expression bias suggests dosage balance and regulatory stability after a recent WGD, potentially buffering essential pathways while permitting targeted divergence at ecologically relevant loci. While these multi-layer patterns are mutually reinforcing, they remain largely correlative. Genome-wide structure is weak, and selection inferences can be sensitive to filtering and linkage, functional validation through gene editing, expression assays, and physiological experiments will be essential to confirm their adaptive roles. Future integrative studies combining genomics with experimental approaches will further clarify the molecular mechanisms underlying these adaptations. CONCLUSION At the core of this work is the high-quality, haplotype-resolved genome assembly of S. nukiangensis, the first of its kind for an autotetraploid schizothoracine species. This genomic resource provides a foundation for future investigations into polyploidy and high-altitude adaptation. Our results confirm that S. nukiangensis is an autotetraploid, with tetraploidization occurring approximately 0.68 Ma. This recent polyploidy event, coupled with active transposable elements, has driven structural variation and genomic innovation, likely facilitating adaptation to the extreme conditions of the Qinghai-Xizang Plateau. Gene family expansions in immune and energy-related pathways support the hypothesis that polyploidy enhances ecological plasticity. Moreover, population genomic analyses along elevational gradients reveal subtle genetic structure shaped by both historical climate fluctuations and local environmental pressures. In conclusion, S. nukiangensis illustrates how recent polyploidy can fuel evolutionary innovation and local adaptation in high-altitude environments, offering key insights into genome evolution in polyploid vertebrates. SUPPLEMENTARY DATA Supplementary data to this article can be found online.