Characterization of Ovarian Cell Types and Molecular Signatures in Mauremys reevesii Using Single‐Cell Sequencing Authors: Kaiqi Yuan, Xin Niu, Yue Yuan, Jiale Zhou, Meiling Hong, Li Ding Published: 2026 ABSTRACT ABSTRACTThe ovary orchestrates female reproduction, where cellular composition and intercellular communication dictate reproductive cycles. Despite advances in chelonian reproductive biology, the cellular composition and regulatory mechanisms of the ovary remain unclear, thereby restricting insights into reproductive ecology and conservation strategies. Here, we performed single‐cell RNA sequencing (scRNA‐seq) to systematically characterize the molecular signatures of ovarian cell types in the Chinese pond turtle ( Mauremys reevesii ). By integrating differential gene expression analysis and functional enrichment, we constructed a high‐resolution ovarian cell atlas and identified major cell populations, including granulosa cells and oocytes. Specifically, we identified Wnt family member 4 (WNT4) and Wnt family member 6 (WNT6) as specific markers for granulosa cells and Ubiquitin C‐terminal hydrolase L1 (UCHL1) for oocytes. Furthermore, the specific expression of these granulosa cell markers was validated using dual‐color in situ hybridization. Collectively, this study elucidates the ovarian biology of M. reevesii at the cellular and molecular levels, providing a vital theoretical foundation for understanding chelonian reproductive ecology and offering important insights into reproductive ecology in turtles. Introduction The ovary (e.g., in mammals) orchestrates the maintenance of the female phenotype, the production of developmentally competent oocytes, and the support of pregnancy through the synthesis and secretion of sex hormones [1, 2, 3]. Underpinning these complex functions is a highly heterogeneous cellular landscape, comprising germ cells (oocytes) and diverse somatic lineages such as granulosa and endothelial cells [4, 5, 6]. The intricate interplay among these cell types forms the basis of ovarian growth, development, and homeostasis [7]. However, previous studies have largely relied on bulk‐tissue analyses [8, 9], which often lack the resolution to dissect cellular heterogeneity. For example, Wang et al. [10] revealed that cell‐type‐specific differences in redox regulation during ovarian aging are obscured in bulk studies, thereby hindering the distinction of molecular damage pathways between granulosa cells and oocytes. Moreover, while follicle development and ovarian function hinge on the paracrine crosstalk between oocytes and granulosa cells, traditional bulk approaches are unable to resolve these dynamic signaling networks, as evidenced by Tan et al. [11]. The ovary is central to reptilian reproduction and population sustainability. Within this class, turtles occupy a unique evolutionary niche and perform critical ecological functions [12]. To date, research on turtle reproduction has predominantly centered on temperature‐dependent sex determination (TSD) and developmental strategies [13, 14]. For instance, environmental temperature dictates sex determination in both sea turtles and the red‐eared slider ( Trachemys scripta elegans ) [15]. At the molecular level, Guo [16] demonstrated that activating estrogen receptor 1 (Esr1) alone is sufficient to trigger female gonadal differentiation in T. s. elegans, identifying Esr1 as a key mediator of early ovarian differentiation. However, a systematic characterization of the cellular landscape and intercellular interaction networks within the turtle ovary remains absent. This lack of resolution impedes our understanding of the fine‐scale cellular mechanisms governing reproduction and constrains the development of precision conservation strategies informed by cell–cell communication insights. The Chinese Three‐keeled Pond Turtle ( Mauremys reevesii ) is a widespread testudine species in East Asia, characterized by its ecological adaptability. In this study, we employed M. reevesii as a model for investigating reptilian reproduction. We performed scRNA‐seq to construct a comprehensive cellular atlas of the M. reevesii ovary. By deconvoluting the cellular heterogeneity, we provide direct cellular evidence elucidating the mechanisms of ovarian physiology and reproductive regulation. Materials and Methods Sample Collection Healthy female Chinese Three‐keeled Pond Turtles were obtained from Hongwang Turtle Farm (Haikou, Hainan, China). Prior to the experiment, the turtles were acclimatized in laboratory conditions for 2 weeks. During this period, water quality was strictly maintained (pH 7.44–7.79; dissolved oxygen 8–8.5 mg/L) at a constant temperature of 25°C. The photoperiod was set to a 12‐h light/12‐h dark cycle. Ovarian Tissue Preparation A randomly selected 5‐year‐old Chinese pond turtle ( Mauremys reevesii ) weighing 525.7 g was dissected, and its ovaries were removed. After removal, the ovary was immediately rinsed with sterile 1× phosphate‐buffered saline (PBS) to remove surface blood and interstitial fluids. The tissue was then transferred to a 10 mL sterile cryotube, flash‐frozen in liquid nitrogen, and processed within 72 h. Nuclei Isolation and Suspension Preparation Nuclei were isolated by using the commercial kit (52009‐10, Bohao). Briefly, the frozen tissue was placed in lysis buffer (52009‐10, Bohao) and homogenized for 30 s using a tissue homogenizer (F6/F10, Jingxin), followed by incubation on ice for 5 min. The homogenate was filtered through a 40‐μm cell strainer (BD 352340) and centrifuged at 500 × g for 5 min at 4°C. The supernatant was discarded, and the pellet was resuspended in washing buffer (52009‐10, Bohao) and stored temporarily on ice. To purify the nuclei, a density gradient was established using Percoll buffers (PB1/PB2). The suspension was centrifuged at 3000 × g for 12 min at 4°C (acceleration/deceleration set to 5). The debris layer was removed, and 200 μL of the nuclei layer was collected. The nuclei were resuspended in a large volume of washing buffer and centrifuged at 500 × g for 5 min at 4°C. The supernatant was discarded, and the nuclei pellet was resuspended in an appropriate volume of PBS. Nuclei concentration and integrity were assessed by mixing 10 μL of the suspension with 10 μL of AO/PI stain. Counts were performed using a Countstar automated cell counter (S2, Countstar), and nuclear morphology was verified under a fluorescence microscope (DMi1, Leica). Library Construction and Sequencing Single‐nucleus RNA sequencing (snRNA‐seq) libraries were constructed using the 10× Genomics Chromium system (10× Genomics, USA) following the manufacturer's protocol. The nuclei suspension underwent quality control (viability 0%–5%, confirming pure nuclei) and counting, after which it was washed and resuspended to a concentration of 2000–8000 nuclei/μL. Approximately 5000 nuclei were loaded onto the Chromium Controller to generate Gel Beads‐in‐emulsion (GEMs). GEMs were harvested and subjected to reverse transcription in a thermal cycler (S1000, Bio‐Rad, USA). Post‐RT, the emulsion was broken, and cDNA was purified using magnetic beads and amplified. The quality and quantity of the amplified cDNA were assessed prior to library construction. The final libraries were sequenced on an Illumina NovaSeq X Plus platform (Illumina, USA) in PE150 mode, aiming for a sequencing depth of approximately 20 000 reads per nucleus. Data Processing and Quality Control Raw sequencing data were processed using the Cell Ranger pipeline (v7.1.0, 10× Genomics). Reads were aligned to the Mauremys reevesii reference genome using STAR to generate a feature‐barcode matrix. Quality control metrics, including the number of high‐quality nuclei, gene counts, and genome mapping rates, were evaluated. Further downstream analysis was performed using the Seurat R package [17]. Low‐quality nuclei were filtered based on the following criteria: (1) number of detected genes (nFeature_RNA), (2) unique molecular identifier (UMI) counts (nCount_RNA), and (3) percentage of mitochondrial gene expression. Outliers were identified and removed using the interquartile range (IQR) method [18]. Dimensionality Reduction, Clustering, and Annotation Following normalization and identification of highly variable genes (HVGs), Principal Component Analysis (PCA) was performed. The top 50 Principal Components (PCs) were selected based on the ElbowPlot. A Shared Nearest Neighbor (SNN) graph was constructed using FindNeighbors, and clusters were identified using FindClusters with a resolution of 0.8. The resulting clusters were visualized using Uniform Manifold Approximation and Projection (UMAP). Cell types were annotated by combining automated and manual strategies. Cluster‐specific marker genes were identified using FindAllMarkers. Annotation was performed using the SCINA algorithm and manually verified against the CellMarker database and existing literature. Functional enrichment analysis (GO and KEGG) of differentially expressed genes (DEGs) among different cell clusters was conducted using clusterProfiler. Pseudotime trajectory analysis was performed using Monocle 2. Cell–cell communication networks were inferred using CellPhoneDB (v3), with significant ligand‐receptor pairs (expression > 10%, p < 0.05) visualized via dot plots. Histological Analysis (H&E Staining) Ovarian tissues were rinsed with 1× PBS (Servicebio), fixed in universal fixative (Servicebio), and embedded in paraffin (JB‐P5, Junjie) at 62°C. Sections (3 μm thick) were cut using a microtome (RM2016, Leica) and baked at 65°C for 16 h. Sections were deparaffinized in environmental dewaxing solutions I/II (G1128, Servicebio; 20 min each) and rehydrated through a graded ethanol series. Staining was performed using a Hematoxylin and Eosin (H&E) kit (G1101, Servicebio) according to the manufacturer's instructions (Hematoxylin for 3–5 min, Eosin for 15 s). Sections were dehydrated through graded ethanol and n‐butanol, cleared in xylene, and mounted with neutral balsam. Images were acquired using a Nikon Eclipse 100 microscope. Double Fluorescence In Situ Hybridization (FISH) Ovaries were fixed in in situ hybridization fixative (G1113, Servicebio) for ≥ 12 h. Follicular regions were excised, dehydrated through a graded ethanol series, cleared in xylene, and embedded in paraffin. Sections (4 μm) were baked at 62°C for 2 h, deparaffinized, and rehydrated. To minimize background and enhance permeability, sections were treated with absolute ethanol, 85% ethanol, 75% ethanol, and DEPC‐treated water, followed by microwave antigen retrieval in citrate buffer (pH 6.0) and digestion with Proteinase K (G1234, Servicebio) at 40°C for 5 min. Pre‐hybridization was performed at 37°C for 1 h. Sections were then incubated with specific probes (50 nM) at 40°C overnight. Post‐hybridization washing was carried out using a gradient of SSC buffers (2×, 1×, 0.5×) at 37°C. Signal amplification was achieved using branched probes (40°C, 45 min) followed by incubation with Cy3‐ and 488‐labeled signal probes (1:400) at 42°C for 3 h. Nuclei were counterstained with DAPI (G1012, Servicebio) for 8 min, and sections were mounted with an anti‐fade medium (G1401, Servicebio). Images were captured using a Nikon upright fluorescence microscope. Results Cell Cluster and Marker Gene Selection Following stringent quality control procedures, we obtained a final dataset of 4105 high‐quality single nuclei, in which a total of 5688 genes were detected. For dimensionality reduction, the top 50 Principal Components (PCs) were determined based on the Elbow Plot visualization to facilitate robust downstream clustering and analysis (Figure S1). Through unsupervised clustering, the M. reevesii ovarian cellular landscape was resolved into 19 distinct clusters (Figure 1A). To characterize these populations, we analyzed the expression profiles of differentially expressed genes (DEGs). As shown in the dot plot (Figure 1B), each cluster possessed a robust set of specific marker genes with high expression levels and broad coverage within the cluster. For example, Cluster 1 was defined by the significant upregulation of Solute carrier family 7 member 11 (SLC7A11), TIAM Rac1 associated GEF 1 (TIAM1), and Melanocyte inducing transcription factor (MITF). The distinct transcriptional signatures of these clusters were further corroborated by visualizing 12 representative DEGs (e.g., SLC7A11, Glutamate receptor interacting protein 1) on the UMAP projection, which revealed highly restricted expression patterns specific to their respective clusters (Figure 1C). Moreover, a heatmap displaying the top three variable features for each cluster confirmed their exclusive specificity across the 19 populations (Figure 1D), validating the robustness of the clustering resolution. Transcriptomic landscape and marker gene characterization of the Mauremys reevesii ovary. (A) UMAP visualization of the 19 identified cell clusters. Each point represents a single nucleus, colored by cluster identity. (B) Dot plot depicting the expression of representative marker genes across clusters. Dot size reflects the proportion of expressing cells, and color intensity (green to red) indicates average scaled expression levels. (C) Feature plots visualizing the spatial distribution and specificity of key marker genes on the UMAP embedding. Expression levels range from light gray (low/absent) to dark blue (high). (D) Heatmap showing the expression dynamics of the top three differentially expressed genes (DEGs) per cluster. The color scale indicates relative expression levels from purple (low) to yellow (high). Annotation of Major Cell Lineages We successfully resolved the 19 cell clusters into six major lineages based on canonical marker expression: oocytes (2 clusters, 12.96%), granulosa cells (3 clusters, 30.74%), endothelial cells (3 clusters, 11.01%), immune cells (2 clusters, 19.22%), fibroblasts (3 clusters, 9.99%), and uncharacterized populations (6 clusters, 16.8%) (Figure 2A). The robustness of these annotations was corroborated by heatmap analysis, where the top 10 DEGs per cell type demonstrated exclusive specificity (Figure 2B). To spatially map these identities, we overlayed the expression of 12 key marker genes—including Wnt family member 4 (WNT4), Wnt family member 6 (WNT6), Platelet and endothelial cell adhesion molecule 1 (PECAM1), Von willebrand factor (VWF), Cadherin5(CDH5), Ubiquitin C‐terminal hydrolase L1 (UCHL1), Nuclear receptor subfamily 2 group F member 2 (NR2F2), Collagen type I alpha 1 chain (COL1A1), Decorin (DCN), Protein tyrosine phosphatase receptor type C (PTPRC), and the LOC120373389/90 loci—onto the UMAP embedding (Figure 2C). These markers exhibited distinct spatial enrichment within their respective clusters, aligning with the transcriptomic signatures observed in the heatmap. Furthermore, comparative expression analysis (Figure 2D) established a panel of robust molecular markers for each lineage: UCHL1, LOC120373389, and LOC120373390 for oocytes; WNT4 and WNT6 for granulosa cells; PECAM1, VWF, and CDH5 for endothelial cells; NR2F2, COL1A1, and DCN for fibroblasts; and PTPRC for immune cells. Identification and molecular characterization of ovarian cell lineages. (A) UMAP visualization of the six major cell types resolved in the Mauremys reevesii ovary. Each point represents a single nucleus, colored according to cell type identity. (B) Heatmap depicting the transcriptional profiles of the top 10 differentially expressed genes (DEGs) per cell type. The color scale indicates relative expression from purple (low) to yellow (high). (C) Feature plots illustrating the spatial specificity of canonical marker genes on the UMAP embedding. Color intensity corresponds to expression level (light gray: Low/absent; dark blue: High). (D) Violin plots demonstrating the specific expression of marker genes across cell types. The y‐axis denotes log‐normalized expression scores. Functional Characterization via GO and KEGG Enrichment To elucidate the functional distinctiveness of the identified cell lineages, we performed Gene Ontology (GO) enrichment analysis. By selecting the top three biological processes with the lowest FDR for each lineage, we highlighted 18 representative functional terms across the six major cell types (Figure 3A). Importantly, these functional annotations recapitulated the canonical roles of each cell identity. For instance, DEGs highly expressed in endothelial cells were significantly associated with angiogenesis, small GTPase‐mediated signal transduction, and transmembrane receptor protein tyrosine kinase signaling. In parallel, granulosa cells exhibited enrichment in metabolic and biosynthetic processes, including organic cation transport and the biosynthesis of proteoglycans and glycoproteins. These functional profiles were further corroborated by KEGG pathway analysis; notably, endothelial cells showed distinct enrichment in the Notch signaling pathway, a hallmark of vascular development and homeostasis (Figure 3B). Functional characterization: (A) Heatmap of top‐enriched Gene Ontology (GO) Biological Processes across cell types. Color intensity (yellow to green) denotes the enrichment score/significance of each term. (B) Heatmap depicting key signaling pathways identified via KEGG enrichment analysis, with the color scale (yellow to green) indicating enrichment levels. Temporal Dynamics and Intercellular Communication Networks To delineate transcriptional dynamics, we defined the pseudotime for each cell based on its projection along the differentiation trajectory. Heatmap analysis (Figure S2) revealed that representative genes exhibit dynamic, progressive expression patterns along the pseudotime axis, with peak expression occurring at distinct stages. These temporal profiles underscore the stage‐specific functions of these genes in ovarian development. As shown in Figure S2, Peroxiredoxin 1 (PRDX1) exhibited relatively high expression at the early stage of the developmental trajectory, followed by a gradual decline as pseudotime progressed. We next reconstructed the cell–cell communication network to explore the “social” interactions within the ovary. As shown in Figure 4A, oocytes act as key regulators of immune cells, granulosa cells, and fibroblasts, while integrating signals from granulosa cells, fibroblasts, and autocrine loops. Notably, granulosa cells function as a dominant signaling hub. They exert strong regulatory effects on immune cells, oocytes, and fibroblasts—most prominently on the immune population—and engage in bidirectional crosstalk with oocytes and endothelial cells. Analysis of interaction strength (Figure 4B) confirmed the role of granulosa cells as a major outgoing signaling source. They secrete a repertoire of ligands, including Amyloid Beta Precursor Protein (APP), Neurexin 3 (NRXN3), Semaphorin 3C (SEMA3C), Semaphorin 4D (SEMA4D), Protein Tyrosine Phosphatase Receptor Type D (PTPRD), and Collagen Type IV Alpha (COL4A) family members, effectively targeting receptor‐bearing immune cells and oocytes. Conversely, granulosa cells are equipped with diverse receptors (e.g., LRFN5, PLXNB2, and ADGRL2) to sense signals from oocytes, endothelial cells, and autocrine inputs, Such as among the predicted interactions, oocyte‐expressed PTPRD, PTPRF, and PTPRS (Protein Tyrosine Phosphatase Receptor Type D/F/S) were paired with the granulosa cell receptor Leucine‐rich repeat and fibronectin type III domain‐containing 5 (LRFN5). This connectivity suggests that granulosa cells orchestrate ovarian function through precise feedback mechanisms and upstream regulation. Significant ligand‐receptor pairs driving these interactions are further visualized in Figure S3. Intercellular signaling networks. (A) Network visualization of the inferred intercellular communication landscape. Directed edges represent significant ligand–receptor interactions, capturing both autocrine and paracrine signaling events. (B) Heatmap summarizing the interaction strength between cell lineages. The color gradient (blue to red, with the scale capped at 100 for visualization; values above this threshold are shown in the deepest red) reflects the frequency of significant ligand‐receptor pairs detected between sender and receiver cells. Functional Heterogeneity and Developmental Trajectories of Granulosa Cells Granulosa cells (GCs) represented a major cellular component, accounting for 30.74% of the ovarian population. Beyond the canonical markers WNT4 and WNT6, GCs were distinguished by the specific upregulation of Regulating Synaptic Membrane Exocytosis 2 (RIMS2) (Figure 5A). To dissect their functional heterogeneity, we performed sub‐clustering analysis, which resolved the GCs into three distinct subpopulations: subclusters 0, 1, and 2 (Figure 5B). Examination of subtype‐specific transcriptomes revealed unique molecular signatures for each group (Figure 5C). Subcluster 0 was defined by the specific expression of EPH Receptor A5 (EPHA5); subcluster 1 was characterized by Glutamate Metabotropic Receptor 7 (GRM7); and subcluster 2 was marked by the enrichment of RUNX Family Transcription Factor 2 (RUNX2). These distinct molecular profiles imply that these subclusters correspond to GCs residing in discrete functional states or developmental phases. Functional heterogeneity and differentiation kinetics of granulosa cells. (A) Violin plots characterizing the lineage identity of granulosa cells via the specific expression of marker genes. (B) UMAP visualization of the fine‐scale structure within the granulosa lineage, revealing three subtypes. (C) Violin plots demonstrating the molecular signatures defining each granulosa cell subtype. (D) Pseudotime trend plots depicting the transcriptional kinetics of key marker genes across the differentiation continuum. To reconstruct the differentiation continuum, we performed pseudotime trajectory analysis (Figure S4). This analysis elucidated the transcriptional kinetics of key regulators: notably, RUNX2 exhibited a “high‐to‐low” expression pattern (early upregulation followed by downregulation), whereas EPHA5 displayed an inverse “low‐to‐high” trend. The synchronized expression dynamics of these signature genes along the pseudotime axis (Figure 5D) suggest a coordinated regulatory program driving GC differentiation. Furthermore, mapping the three subtypes onto the pseudotime manifold (Figure S5) revealed their distinct temporal occupancy, thereby delineating their specific developmental statuses along the lineage trajectory. As can be seen from the figure, Pseudotime analysis arranged the three granulosa cell subtypes along a continuous trajectory. Cluster 2 and Cluster 0 cells were predominantly located at opposite ends of the trajectory, while Cluster 1 cells were distributed throughout the intermediate region, suggesting that Cluster 1 may represent a transitional cellular state between the other two subtypes (Figure S5). Histological and Molecular Validation of Granulosa Cell Markers To substantiate the fidelity of the scRNA‐seq‐derived cell annotations, we performed histological and molecular validation. Histological examination (H&E staining) of the ovarian cortex confirmed the anatomical localization of granulosa cells, which were observed enveloping oocytes within primordial follicles, consistent with the canonical follicular architecture (Figure 6A). We further verified the spatial specificity of key markers using RNA in situ hybridization (RNA‐ISH). As shown in Figure 6B, distinct mRNA hybridization signals for WNT4 and WNT6 were exclusively enriched in the granulosa cells of primordial follicles. Notably, these signals were predominantly localized to the cytoplasm, consistent with the expected subcellular distribution of mRNA transcripts. Collectively, these in situ data corroborate the robustness of WNT4 and WNT6 as specific molecular signatures for granulosa cells. Histological characterization and in situ validation of granulosa cell markers. (A) Representative H&E‐stained sections of the Mauremys reevesii ovary, highlighting the typical morphological features of primordial follicles and the arrangement of granulosa cells. (B) Double fluorescence in situ hybridization confirming the spatial specificity of marker genes. WNT4 (red) and WNT6 (green) transcripts are specifically localized within the granulosa cell layer surrounding the oocytes. Nuclei are visualized with DAPI (blue). Discussion The ovary serves as the central organ of the reptilian reproductive system, playing a pivotal role in regulating reproductive capacity [19, 20]. However, systematic research on the cellular composition and intercellular interaction networks of the turtle ovary remains lacking. This gap severely limits our in‐depth understanding of reproductive regulation mechanisms and restricts the development of conservation strategies based on a cellular perspective. In this study, we utilized single‐cell RNA sequencing (scRNA‐seq) technology to analyze the ovary of the Chinese Three‐keeled Pond Turtle ( Mauremys reevesii ). We identified oocytes, granulosa cells, endothelial cells, immune cells, and fibroblasts. Furthermore, we found that the marker genes for oocytes are UCHL1, LOC120373389, and LOC120373390, for granulosa cells are WNT4 and WNT6, for endothelial cells are PECAM1, VWF, and CDH5, for immune cells is PTPRC, and for fibroblasts are NR2F2, COL1A1, and DCN. The results fill a critical gap in turtle research and provide a key data foundation for understanding the cellular architecture and functional mechanisms of the turtle ovary. Oocytes and granulosa cells (GCs) constitute the fundamental functional unit of the follicle, maintaining intimate crosstalk to orchestrate ovarian function [21]. In our study, LRFN5 was identified as a putative receptor in granulosa cells, which may interact with oocyte‐derived PTPRD, PTPRS, and PTPRF. Previous studies have demonstrated that LRFN5 (SALM5) can directly bind to members of the LAR‐RPTP family, including PTPRD, PTPRS, and PTPRF, thereby mediating intercellular adhesion and signal transduction [22]. We hypothesize that although the functional role of this signaling axis in ovarian development has not yet been reported, its presence within the oocyte–granulosa cell communication network suggests a potential involvement in maintaining intercellular interactions during folliculogenesis. In mammals (e.g., humans, monkeys, and mice), the genes Growth Differentiation Factor 9 (GDF9), Factor in the Germline Alpha (FIGLA), and Zona Pellucida Glycoprotein 3 (ZP3) are widely recognized as canonical oocyte‐specific markers [23, 24]. Intriguingly, we observed a distinct molecular signature in M. reevesii oocytes: these mammalian markers were conspicuously absent. Instead, UCHL1 and two uncharacterized loci (LOC120373389 and LOC120373390) emerged as highly specific oocyte signatures. The identification of UCHL1 parallels findings in human ovaries [25], suggesting a deep evolutionary conservation of this gene's role in germ cell identity despite the divergence in other markers. Regarding GCs, which underpin oocyte maturation, we identified WNT4 and WNT6 as robust, specific markers. This observation recapitulates findings by Gong et al. [25], solidifying the status of WNT4 and WNT6 as reliable indicators of granulosa lineage. We further corroborated this specificity via RNA in situ hybridization, which precisely mapped WNT4 and WNT6 transcripts to the granulosa layer enveloping the oocytes. Beyond lineage identity, our analysis unveiled significant heterogeneity within the GC population. The identification of three distinct subtypes with unique transcriptional profiles points to divergent functional states. This heterogeneity echoes recent insights from goat prolificacy studies by Zhang et al. [26], where distinct GC subpopulations were found to differentially modulate apoptosis, anabolism, and hormonal responsiveness to fine‐tune reproductive output. Despite the phylogenetic distance between reptiles and mammals, the convergence of GC functional specialization suggests that maintaining heterogeneous granulosa subpopulations represents an evolutionarily conserved strategy across vertebrates. The RUNX gene family is a pivotal regulator of mammalian reproduction [27]. Here, we observed specific upregulation of RUNX2 in the granulosa cells of M. reevesii . This finding recapitulates functional studies in mammals: for instance, Gao [27] reported that RUNX2 in goat granulosa cells is hormonally responsive and drives progesterone synthesis and cell proliferation, while Park [28] highlighted its role in transcriptional regulation within rat luteinized granulosa cells. Furthermore, previous studies have demonstrated that granulosa cell‐specific deletion of RUNX2 and Cbfb in mice does not impair mature follicle development, but severely disrupts ovulation and luteinization, suggesting that RUNX2 plays a critical regulatory role in the fate transition of periovulatory granulosa cells [29]. This suggests that the RUNX2‐mediated regulatory axis in granulosa cells is evolutionarily conserved. Endothelial cells are indispensable for follicle maturation and the vascular dynamics of ovarian remodeling [30, 31]. Comprising approximately 11% of the total population, the endothelial cluster was defined by the robust expression of VWF and PECAM1. The specificity of these markers aligns with extensive evidence from mammalian models, including primates [10, 32] and yaks [21, 33]. The consistency of these molecular signatures across diverse taxa underscores the evolutionary conservation of ovarian vascular markers and validates the fidelity of our endothelial annotation in the turtle ovary. Furthermore, immune cells and fibroblasts act as key architects of ovarian homeostasis and development [34, 35]. We found that immune cells constitute ~19.22% of the ovarian cell landscape, a proportion comparable to previous reports by Wu et al. [34]. These cells were distinguished by high expression of PTPRC (CD45), an essential modulator of adaptive immunity [36]. In parallel, the fibroblast population was marked by the specific expression of NR2F2, DCN, and notably, COL1A1. Ovarian fibroblasts are key components of the ovarian stroma and the primary cells responsible for extracellular matrix (ECM) production and remodeling. By secreting collagen, fibronectin, and other matrix‐associated molecules, they support the maintenance and dynamic remodeling of ovarian tissue structure [37]. The ovarian stroma and ECM microenvironment are essential for follicular maturation, with dynamic ECM remodeling supporting granulosa cell proliferation, follicular expansion, ovulation, and luteinization [38]. COL1A1 is a key structural component of the ECM and serves as one of the marker genes of fibroblasts [39]. In the ovary, COL1A1 is widely expressed in the ovarian stroma and theca‐associated cells, where it participates in follicular development, ECM remodeling, corpus luteum formation, and ovarian fibrosis [38]. In bovine ovaries, COL1A1 is highly expressed in the theca layer and stroma, supporting follicular structure, angiogenesis, and theca differentiation [40]. Given that COL1A1 serves as a canonical fibroblast marker in avian species [41], our identification of COL1A1 in turtles bridges the phylogenetic gap between birds and mam mals, confirming it as a robust and conserved fibroblast marker in reptiles. Notwithstanding the comprehensive cellular landscape presented here, certain limitations warrant consideration. First, as an initial foundational study, our analysis was conducted using ovarian samples from a single individual, which represents a limitation in assessing the broader applicability and representativeness of the identified ovarian cell populations. Nevertheless, this study provides a preliminary framework for understanding ovarian cellular composition and molecular characteristics in this species. Future studies with expanded sample sizes will be conducted to further validate the stability and generalizability of these cell populations and to elucidate the roles of ontogenetic processes and environmental plasticity in shaping variations in ovarian cellular composition. Additionally, the annotation of rare or highly specialized cell types was constrained by the limited annotation of reptilian genomes in current reference databases. Due to the current lack of relevant literature for reference, this study was unable to highlight the evolutionary differences in ovarian cell lineages and marker genes between reptiles and mammals. In future research, we will further investigate these evolutionary differences between the two groups. Crucially, however, the persistence of these uncharacterized populations hints at the existence of species‐specific cell types that diverge from evolutionarily conserved lineages. Far from being mere artifacts, these “unknown” clusters represent promising frontiers for uncovering novel biological mechanisms and offer fertile ground for future functional exploration in reptilian reproduction. Conclusion In summary, we have comprehensively mapped the cellular landscape of the adult Mauremys reevesii ovary at single‐cell resolution, defining six distinct lineages: oocytes, granulosa cells, immune cells, endothelial cells, fibroblasts, and uncharacterized populations. By leveraging cross‐species comparative analysis, we validated canonical markers and established WNT4 and WNT6 as robust, specific molecular signatures for turtle granulosa cells. Moreover, our reconstruction of the intercellular communication network unveiled the intricate signaling architectures governing the ovarian microenvironment. Collectively, these insights not only elucidate the cellular and molecular biology of the turtle ovary but also serve as a foundational resource for advancing our understanding of reproductive ecology and informing evidence‐based conservation strategies for chelonian species. Author Contributions L.D. and K.Y.: conceptualization. L.D. and K.Y.: methodology. K.Y., X.N. and Y.Y.: validation. K.Y., J.Z. and Y.Y.: formal analysis. K.Y. and X.N.: investigation. K.Y.: writing – original draft preparation. L.D. and X.N.: writing – review and editing. K.Y.: visualization. L.D. and M.H.: supervision. L.D.: project administration. L.D.: funding acquisition. Funding This work was supported by the National Natural Science Foundation of China (Nos. 32571774 and 32271577). Ethics Statement All animal welfare concerns and experimental protocols received approval from the Animal Research Ethics Committee of Hainan Ecological Environment Education Center, Hainan Normal University (Approval No. HNECEE 2023‐005). Conflicts of Interest The authors declare no conflicts of interest. Supporting information Figure S1: The elbow plot shows standard deviations of the top 50 principle components (PCs) in principle component analysis (PCA). Figure S2: Pseudotime‐ordered heat map demonstrating of pseudotime order of the selected marker genes for all the cell types. The expression level of each gene from low to high is indicated by a color gradient from blue to red. Figure S3: Putative communications between granulosa cells and the other cell types through ligand–receptor interactions. Figure S4: Pseudotime‐ordered heat map demonstrating of pseudotime order of the selected marker genes for granulosa cells subtypes. The expression level of each gene from low to high is indicated by a color gradient from blue to red. Figure S5: Scatterplot showing the differential trajectories of four granulosa cells subtypes with pseudotime scale by Monocle.