Geographic genetic divergence and ecological viral communities shape the virome of Chilo suppressalis Authors: Jia-Bao Lu, Ying Tian, Yi-Yuan Li, Yu-Hua Qi, Fang He, Peng-Peng Ren, Hai-Jian Huang, Ji-Chong Zhuo, Qing-Ling Hu, Gang Lu, Qian-Zhuo Mao, Jian-Ping Chen, Jun-Min Li, Chuan-Xi Zhang Published: 2026 ABSTRACT AbstractBackgroundInsect-specific and insect-borne viruses (particularly RNA viruses) may pose a serious threat to global rice production. However, the effects of geographic genetic divergence and ecological viral communities in shaping insect viromes remain poorly understood. The striped stem borer, Chilo suppressalis—a notorious chewing pest widely distributed across China—serves as an ideal model to study RNA virus evolution and cross-species transmission dynamics in rice ecosystems.ResultsMeta-transcriptomic sequencing analysis of 48 C. suppressalis samples collected nationwide revealed four major geographic populations and identified 20 high-abundance viruses (15 novel). These include the widespread core virus Hangzhou sesamia inferens peribunyavirus 1 and Chilo suppressalis sedoreo-like virus 1, as well as several climate-adapted viruses that may represent extreme-environment specialists. Diversity analyses revealed that the geographic genetic divergence of C. suppressalis influences viral species composition but not the overall virome structure. Parallel sequencing of 5 Cotesia chilonis (endoparasitoid) samples revealed 8 high-abundance viruses (5 novel). Significant variations in distribution patterns, viral loads, and vsiRNA profiles within C. suppressalis–C. chilonis parasitic system suggested host-driven adaptive evolution of these viruses. Co-occurrence network analysis demonstrated that sympatric species (such as Sesamia inferens and rice planthoppers) and their host plant, rice, significantly influence C. suppressalis virome composition, highlighting cross-kingdom viral transmission dynamics.ConclusionsThis study elucidates the synergistic roles of geographical divergence and ecological viral communities in driving virome diversity in C. suppressalis. Our findings advance the understanding of virus-host coevolution in agricultural ecosystems and provide a framework for developing integrated strategies targeting both insect pests and their viral pathogens in rice cultivation. Video AbstractSupplementary InformationThe online version contains supplementary material available at 10.1186/s40168-026-02407-8.Keywords: Striped stem borer, Cotesia chilonis, Meta-transcriptomic sequencing, Virus, SRNA, Geographic population, Ecological network Background Viruses represent the most abundant microbial entities on earth and have been recognized as critical biological pathogens capable of causing diseases in humans and other animal/plant hosts since the nineteenth century [1–4]. Over the recent years, the advent of metagenomic next-generation sequencing (mNGS) has facilitated the identification of numerous asymptomatic and covert RNA viruses [5, 6]. A substantial number of novel RNA viruses have been identified in arthropods, which can be divided into insect-specific viruses (ISVs) and arthropod-borne viruses (arboviruses) [7, 8]. ISVs infect insects exclusively and are unable to replicate in vertebrates or vertebrate cells [7]. Phylogenetic investigations suggest that ISVs may represent ancestral forms of arthropod viruses [9, 10], with some closely related to plant viruses [11]. Comprehensive characterization of the virome and in-depth investigation of the basic biology of ISVs in diverse geographically distributed insect species may provide a promising approach to elucidate the ecological and evolutionary context of various ISVs and their hosts, as well as the complex interaction mechanisms between ISVs and host/host associated microorganisms [12–15]. Notably, ISVs host range and transmission dynamics are intricately modulated by multifaceted environmental and biological factors, including seasonal variations [12, 16], temperature fluctuations [17], host-associated microorganisms [18] and parasites influences [19]. Accumulating in vivo and in vitro experiments have demonstrated that ISVs can regulate host susceptibility to arboviruses, vector competence, and host fitness, highlighting their potential as biocontrol agents [7, 18, 20–22]. Arboviruses can infect both insects and vertebrates/plants, besides, arboviruses are the main microbial pathogens causing epidemic diseases, exemplified by dengue viruses in human [23] and southern rice black-streaked dwarf virus (SRBSDV) in rice [24]. In the past, considerable attention was focused on viruses in hematophagous insects of medical importance, including mosquitoes [25], sandflies, and ticks [26], as well as the economically important insect bees [27]. In recent years, research has progressively expanded to investigate RNA viruses in other insect vector of agricultural importance [28, 29], like whitefly [30], aphids [31], planthoppers [32], and grasshoppers [33], which may cause devastating threats to important food and commercial crops via arboviruses. As one of the world’s most vital food crops, rice faces persistent threats from viral diseases, primarily associated with fungal pathogens and rice-feeding insect vectors [34]. Notably, many of the most damaging rice viruses are RNA viruses transmitted by insects, such as SRBSDV [24] and rice stripe virus (RSV) [35]. Recent global transcriptomic surveys of rice-associated viruses have uncovered a substantial number of novel viral taxa, many of which are hypothesized to originate from rice-feeding insects [36]. Thus, rice-feeding insects serve not only as potential spillover reservoirs but also as critical bridging vectors for viral transmission. Consequently, in-depth investigations into the RNA viromes of key rice-feeding insects hold significant value, as they can not only help enrich the virus database, but also elucidate virus-insect relationships and their interaction mechanisms. To date, most research has concentrated on sap-sucking insects such as planthoppers [32] and leafhoppers [37], whereas studies on RNA viruses in chewing insects remain notably scarce. Up to now, a few studies have demonstrated that evolutionary and geographically proximate species tend to share the same pathogens [38], and cross-infection capabilities of RNA viruses among various hosts have been verified [29, 30, 39]. However, how geographic genetic divergence and ecological viral communities shape an insect virome remain largely unknown. The striped stem borer, Chilo suppressalis (Walker) (Lepidoptera, Crambidae), is one of the most notorious chewing insect pests that significantly impacts rice production around the world, including Asian, Oceania, and Europe [40–43]. Notably, in China, a wide geographical distribution of C. suppressalis was found, ranging from the cold northeastern region to the hot southern region, mainly divided by the Yangtze River as a boundary [44]. As a polyphagous [45] and non-migratory insects [46], significant climatic and topographic differences indicate the existence of different ecological phenotypes of C. suppressalis. Based on the above research, C. suppressalis, as a representative species with a well-defined geographical distribution of populations, is expected to serve as a good model for studying the evolution and cross-talk of RNA viruses among geographically distributed populations. Besides, Cotesia chilonis (Hymenoptera: Braconidae), a parasitoid wasp (endoparasitoid) of C. suppressalis, has been used as a natural enemy for the biological control of this pest [47]. The comparative analyses of RNA virome in the C. suppressalis–C. chilonis parasitoid system will help elucidate the potential host-driven adaptive evolution of viruses, allowing for a virus perspective to better understand the intrinsic mechanisms of pest outbreaks and natural enemy-mediated pest control. To further assess the influence of ecological viral communities in shaping the composition and dynamics of C. suppressalis virome, this study conducted a comparative analysis involving rice and the two sympatric species, Sesamia inferens (Lepidoptera, Noctuidae) and Laodelphax striatellus (Hemiptera, Delphacidae). S. inferens is another major rice stem-boring pest that shares a highly similar ecological niche with C. suppressalis, whereas L. striatellus is a key vector of rice viruses by feeding on the phloem sap of rice plants [32]. Together, rice, C. suppressalis, S. inferens, and L. striatellus form an interconnected ecological network. However, it remains unclear whether this network facilitates virus transmission through direct or indirect interactions. Notably, among these four species, only C. suppressalis serves as the natural host for the parasitoid wasp C. chilonis. Overall, this study provides a foundation for investigating how geographic divergency and ecological viral communities collectively shape the virome of C. suppressalis, offering insights into virus evolution, host adaptation, and potential biocontrol applications in agricultural systems. Results Geographic diversity in C. suppressalis To elucidate the factors shaping the virome of C. suppressalis, we conducted a large-scale meta-transcriptomic study encompassing diverse geographic and ecological contexts across China. Our sampling strategy was designed to capture the influence of both geographic genetic divergence and ecological virus communities on the composition of the C. suppressalis virome. In total, we collected 48 samples, including 40 whole insect samples, 6 dissected tissue samples, and 2 larvae samples fed on rice seedlings or water bamboo (Zizania latifolia). These samples were obtained from 23 cities in 14 provinces spanning 7 major climate zones [14, 48] between 2022 and 2024 (Fig. 1a). This broad environmental gradient ensured robust representation of virome variation under different ecological conditions in China. To further explore potential virus-host interactions and cross-species transmission, we isolated and sequenced the parasitic wasp C. chilonis from 5 host C. suppressalis samples. Additionally, we acquired 3 samples of S. inferens, a closely related moth species occupying overlapping ecological niches with C. suppressalis, to assess virome divergence in sympatric species. A total of > 530 GB of raw sequencing data was generated (Tables S1–S3 for assembly statistics). After De novo assembly and rigorous bioinformatic analysis as described in methods, 30 contigs classified at the species level via VITAP, 51 homolog-based viral contigs, and 86 contigs encoding RNA-dependent RNA polymerase (RdRP) or conserved RdRP motifs were screened out, respectively, ensuring high-confidence viral detection. Sampling locations and population assignment of Chilo suppressalis in China. a. Sampling locations of C. suppressalis and its parasitoid wasp Cotesia chilonis. 7 major climate zones were marked with different color in the map of China, with each insect symbol representing a sampling location; b. Assignment of C. suppressalis geographic populations based on mitochondrial genome data using IQ-Tree. 1000 ultrafast bootstraps were used with Diatraea saccharalis as the outgroup. Notes with bootstrap values > 60% are marked with solid blue circles A population genetic analysis with 37 larval C. suppressalis samples collected from diverse geographic regions across China using mitochondrial genome data via phylogenetic construction (Fig. 1b). Result revealed that 30 samples could be classified into four major geographic populations (groups 1–4), exhibiting distinct distribution patterns (Table S4). Notably, Group 1 (12 samples) was predominantly composed of eastern samples (75%), while group 2 displayed the most complex distribution, encompassing samples from eastern (3 samples), central (2 samples), southern (1 sample), northwestern (2 samples), and southwestern (1 sample) regions, suggesting potential gene flow or historically human-mediated dispersal events. Group 3 consisted of eastern (3) and southwestern (2) samples, whereas group 4 included eastern (2), southwestern (1), and northern (1) samples. Additionally, the remaining seven samples (5 samples in eastern, 2 samples in northeast) formed independent branches without clear clustering, possibly indicating localized adaptive evolution or recent invasive populations. These results demonstrate that C. suppressalis in China exhibits eastern-centered genetic diversity, with eastern samples distributed across all four major geographic populations, implying that this region may serve as a dispersal hub or historical migration passage. The presence of northwestern and southwestern samples in multiple populations suggests potential long-distance migration via agricultural trade. Furthermore, cross-regional clustering (like in group 2) may be linked to climate-adaptive gene flow or human-mediated dispersal activities. In conclusion, this analysis provides critical insights into the population genetic structure and biogeographic history of C. suppressalis in China. RNA virome identification in C. suppressalis Following criteria annotation standards described in methods, we identified 20 candidate RNA viruses with high coverage (> 20X) across all the C. suppressalis libraries. Among these, 5 were previously reported, while 15 were newly identified (9 potential ISVs and 6 arboviruses) (Tables S5 and S6). The known viruses included: Saccharomyces 20S RNA narnavirus (S20SRNV, Narnaviridae), a positive-sense single-stranded RNA (+ ssRNA) virus originally identified in yeast [49]; Hangzhou sesamia inferens carmotetravirus 1 (HZSICV1, Carmotetraviridae), a + ssRNA virus that primarily infects lepidopteran insects, such as the rice borer S. inferens (its original host); Hangzhou sesamia inferens Solinvi-like virus 1 (HZSISolV1, Solinviviridae), a + ssRNA virus detected in C. suppressalis, S. inferens, and Monolepta lauta (Coleoptera); Hangzhou phasmavirus 1 (HZPV1, Phasmaviridae), a negative-sense single-stranded RNA (-ssRNA) virus found in Cnaphalocrocis medinalis, another major lepidopteran rice pest; and Hangzhou sesamia inferens peribunyavirus 1 (HZSIPbV1, Peribunyaviridae), a -ssRNA virus present at high abundance in both C. suppressalis and S. inferens. Notably, HZSICV1, HZSISolV1, HZPV1, and HZSIPbV1 were all isolated from Fuyang District, Hangzhou City, Zhejiang Province, China [29]. Phylogeny and genomic structures of the 15 novel viruses, together with related reference viruses, are revealed in (Fig. 2 and Figs. S1–S6). A total of 6 novel + ssRNA viruses were identified in C. suppressalis. 5 of these belonged to the order Picornavirales, suggesting that picorna-like viruses may constitute the core + ssRNA virome in C. suppressalis, which is consistent with findings in another rice pest, L. striatellus [32]. These included 2 locarnaviruses (Marnaviridae), 2 novel dicistro-like viruses (Dicistroviridae), and 1 iflavirus (Iflaviridae) (Fig. 2a–c; see Fig. S1 for detailed phylogenetic trees). Besides, another + ssRNA virus in the order Martellivirales was identified as a novel kita-like virus (Kitaviridae) (Fig. 2d; see Fig. S2 for detailed phylogenetic trees). 7 novel -ssRNA viruses were also discovered. Three were classified within the order Mononegavirales. Phylogenetic analysis with established families from the International Committee on Taxonomy of Viruses (ICTV) database placed them in Rhabdoviridae, Artoviridae, and Xinmoviridae (Fig. 2e–g; see Fig. S3 for detailed phylogenetic trees). 1 quaranjavirus and 1 yingvirus, belonging to the family Orthomyxoviridae and Qinviridae, respectively, were identified (Fig. 2h, i; see Fig. S4 for detailed phylogenetic trees); Furthermore, 1 mobuvirus and an unclassified phenui-like virus (Phenuiviridae) were found (Fig. S5); Finally, we detected an unclassified partiti-like double-stranded RNA (dsRNA) virus (Partitiviridae) and another unclassified sedoreo-like dsRNA virus (Sedoreoviridae) (Fig. S6). Among these, viruses identified as potential arboviruses included 2 from Marnaviridae, 1 from Kitaviridae, 1 from Partitiviridae, 1 from Sedoreoviridae, and 1 from Orthomyxoviridae; the remainder were likely ISVs. Notably, through the analysis of abundance correlations among contigs and homology assessment, we identified additional genomic segments, excluding those encoding the RdRP or conserved RdRP motifs, derived from viruses belonging to the families Qinviridae and Phenuiviridae (Fig. 2i and Fig. S5). Detailed descriptions for each of the above discovered novel viruses are provided in (Supplementary file S1). Besides, 4 viruses [Sanya narnavirus 2 (SYNNV2), Cotesia chilonis Solemo-like virus 1 (CCSlV1), Cotesia chilonis mobuvirus 1 (CCMoV1), and Cotesia chilonis beny-like virus 1 (CCBlV1)] which were initially identified in the parasitic wasp C. chilonis and subsequently detected in C. suppressalis, were also included in the subsequent analyses. Description of these viruses are provided in the “Comparative analysis of viral diversity between C. suppressalis and C. chilonis” section of the results. Phylogeny and genomic structures of newly identified viruses in Chilo suppressalis. Phylogeny (left) and genomic structures (right) of viruses from the families Marnaviridae (a), Dicistroviridae (b), Iflaviridae (c), Kitaviridae (d), Rhabdoviridae (e), Artoviridae (f), Xinmoviridae (g), Orthomyxoviridae (h), and Qinviridae (i). Phylogenetic trees were constructed using the maximum likelihood method based on conserved viral RdRP sequences containing domains A, B, and C. Newly identified viruses and their corresponding taxonomic classification at the genus level are highlighted in red. All trees are midpoint-rooted. Nodes with bootstrap values ≥ 60% are denoted by solid blue circles, with larger circles representing higher values. Branches belonging to the same genus have been collapsed and are labeled with the genus name, and the detailed phylogenetic trees are provided in supplementary figures (Figs. S1–S6). Except for panel (d), all the other panels contain representative reference virus genomes (labeled in red) that are phylogenetically closely related to the newly identified viruses. GenBank accession numbers are provided in parentheses after the names of reference viruses, with domain names annotated at the figure bottom. Domain abbreviation: ORF (Open Reading Frame); RdRP (RNA-dependent RNA polymerase); RNA_helicase (RNA_helicase super family); rhv_like (Picornavirus capsid protein domain_like); Dicistro_VP4 (a family of minor capsid proteins). The library dataset corresponding to the identified novel viruses is listed in Table S6 Composition and diversity analyses of viromes in C. suppressalis Viromes analysis across different geographic populations and climatic zones of C. suppressalis revealed a distribution pattern characterized by the coexistence of ubiquitous and population/region-specific viruses. Given the complexity of environmental factors, viral contamination in samples is often unavoidable. Viral infections typically increase viral RNA abundance in organisms, making RNA abundance a useful indicator for assessing viral infection or contamination [36]. Through viral sequence alignment and transcripts per kilobase millions (TPMs) calculation, we restricted our analysis to viruses with TPMs > 10. For known viruses: HZSIPbV1 was the most ubiquitously present with high abundance across all geographic populations (groups 1–4 and 7 independent populations), and this virus was also detected in the arid Xinjiang Uygur Autonomous Region (Cs-l (1) XJ 2024 and Cs-l (2) XJ 2024) and the cold northeast regions (Cs-l SY 2023 and Cs-l CC 2023), suggesting exceptional host adaptability in different climate or cross-climatic transmission advantages. Meanwhile, S20SRNV was exclusively detected in Changsha sample (Cs-l CS 2024, in Group 2, Central China), suggesting that localized environmental conditions may shape unique viral community. For novel viruses, a similar trend of broad distribution and specificity was observed. Chilo suppressalis sedoreo-like virus 1 (CSSelV1) was widely distributed with high abundance in 8 populations, except for Cs-l (3) NBjs 2024, Cs-l SX 2023, and Cs-l SY 2023 (Fig. 3a–c). CSSelV1 was detected in multiple samples from seven climatic zones, implying exceptional host adaptability (Fig. 3a, e). Notably, several viruses were only found in individual samples, including Chilo suppressalis peropuvirus 1 (CSPPV1) in Cs-l WH 2023 (Group 1), Chilo suppressalis quaranjavirus 1 (CSQRV1) in Cs-l (3) NBjs 2024, and CCBlV1 in Cs-l CC 2023 (Fig. 3c), highlighting the potential for environment to structure geographically restricted viral populations. The eastern region harbored the highest number of region-specific viruses (n = 6) (Fig. 3e), which may be associated with the high genetic diversity of C. suppressalis populations in this area. The southern region ranked second (n = 3) (Fig. 3e), where viral populations likely evolved adaptations to elevated temperatures. Notably, Chilo suppressalis locarnavirus 1 and 2 (CSLNV1 and CSLNV2) were exclusively detected in the arid Xinjiang Uygur Autonomous Region (northwest China) (Fig. 3e), suggesting these viruses may have developed specialized adaptations to extreme aridity and high-temperature conditions. The marked difference between the East and Northwest regions provides critical insights into host-virus coevolution and environmental adaptation. In contrast, only two and one region-specific viruses were identified in the Central and Northeast zones (Fig. 3d, e), respectively, potentially reflecting more frequent cross-regional viral exchange or host migration events in these two zones. The viral abundance data are shown in (Tables S7 and S8). Composition and diversity analyses of RNA viruses in Chilo suppressalis. a. Heatmap of viral abundance (transcripts per kilobase million, TPM) for known and novel viruses across geographical populations of C. suppressalis. Virus name abbreviations are provided below the heatmap. The TPM dataset is listed in Tables S7 and S8. b-e. Comparative analysis of known and novel viral composition across geographical populations (b, c) and climatic zones (d, e). For panels (b) and (c), geographical populations and sample abbreviations are as follows: G1–G4 represent Group 1 to Group 4; NB: Cs-l(3) NBjs 2024; HZ: Cs-l HZ 2023; SX: Cs-l SX 2023; XM: Cs-l XM 2023; CC: Cs-l CC 2023; FZ: Cs-l(1) FZ 2023; SY: Cs-l SY 2023; For panels (d) and (e), climatic zone abbreviations are as follows: CC: Central China; NC: Northern China; NEC: Northeast China; NWC: Northwest China; SC: Southern China; SWC: Southwest China; EC: Eastern China. f–m. Alpha diversity analysis across geographical populations (f–i) and climatic zones (j–m), evaluated using Richness (f, j), Chao1 (g, k), Shannon (h, l), and Simpson (i, m) indices. Significant difference (p < 0.05) is marked with different lowercases To investigate the effects of geographic population, climate zone, and collection time on the viromes of C. suppressalis, we systematically conducted alpha and beta diversity analyses. Based on genetic differentiation, geographic populations classified as Group 1/2 and Group 3/4 showed significant differences in richness indices (Richness and Chao1) (p < 0.05), whereas no significant differences were observed in terms of Shannon and Simpson diversity indices (Fig. 3f–i). Furthermore, beta diversity analyses (PCoA and NMDS) indicated no significant separation in virome composition among groups (Fig. S7a, b). These findings suggest that although groups 1 and 2 harbored a richer variety of viruses, the overall virome structure remained highly conserved. Although the southwestern region exhibited a lower trend in Richness, no significant differences were detected in the Chao1, Shannon, or Simpson indices (Fig. 3j–m). Beta diversity analyses (PCoA and NMDS) also revealed no clear separation in virome structure among different climate zones (Fig. S7c, d), indicating that despite the presence of region-specific viruses in some areas, climate zones did not significantly alter the overall virome composition and structure. Additionally, similar to the pattern observed for climate zones, the collection time also showed no significant impact on the virome (Fig. S7e–j). The above results indicate that the geographic genetic divergence of C. suppressalis may influence the viral species composition. Dynamics of viral expression across development, tissues, and host plants in C. suppressalis To elucidate the expression characteristics of viruses across different developmental stages of C. suppressalis and their relationship with host plants, this study systematically analyzed the distribution patterns of viruses in larvae, pupae, adults, and eggs using a laboratory-reared C. suppressalis population maintained at Hunan Agricultural University (sample name: Cs-e(ad) CS 2024, Cs-l(ad) CS 2024, Cs-p(ad) CS 2024, and Cs-a(ad) CS 2024). Additionally, two major host plants of C. suppressalis, rice and water bamboo, were used to feed newly hatched larvae to investigate the influence of host plants on viral infection (sample name: Cs-l(wb-feeding) CS 2024 and Cs-l(rs-feeding) CS 2024). The results revealed that only three known viruses (S20SRNV, HZSISolV1, and HZSIPbV1) were detected in this laboratory population (Fig. 4a), suggesting that long-term laboratory rearing may lead to a simplification of virus species. All three viruses exhibited peak expression during the larval stage but displayed significant differences: S20SRNV was nearly undetectable in other developmental stages, HZSISolV1 maintained low-level expression in pupae and adults, and HZSIPbV1 showed low expression in pupae, adults, and eggs. For newly hatched larvae reared on rice or water bamboo, no S20SRNV was detected in larvae, consistent with its absence in eggs, suggesting potential horizontal transmission. HZSISolV1, despite extremely low abundance in eggs (TPM < 1), showed significantly higher expression in larvae. However, HZSIPbV1 was detectable under both host conditions but exhibited weak signal in eggs. In summary, for this laboratory population, the host plant species had a relatively limited impact on the expression level of the three viruses, whereas the intrinsic transmission characteristics of the viruses appeared to be the key determinant of their differential expression patterns. Expression analysis of viruses in Chilo suppressalis. Heatmaps showing the abundance (transcripts per kilobase million, TPM) of known (a) and novel (b) viruses across different development stages, host contexts, and tissues using laboratory-reared C. suppressalis (maintained at Hunan Agricultural University) or wild-type C. suppressalis (collected from water bamboo fields in Ningbo City). Virus name abbreviations are provided in the right of each heatmap. Wb, water bamboo. Tra, trachea. Int, integument. FB, fat body. MT, Malpighian tube. SG, silk gland. The database is listed in Table S9 Then, we systematically analyzed the viral expression profiles across six tissues in wild-type C. suppressalis collected from water bamboo fields in Ningbo City (sample name: Cs-l(trachea) NB 2024, Cs-l(integument) NB 2024, Cs-l(fat body) NB 2024, Cs-l(gut) NB 2024, Cs-l(Malpighian tube) NB 2024, and Cs-l(silk gland) NB 2024). A total of seven viruses were identified, including 3 known viruses [HZSISolV1, HZSIPbV1, and SYNNV2] and 4 newly discovered viruses [Chilo suppressalis phenui-like virus 1 (CSPhelV1), Chilo suppressalis gudgevirus 1 (CSGDV1), CCSlV1, and CCMoV1]. The results showed that these viruses exhibited distinct tissue expression patterns. For known viruses: HZSISolV1 showed significantly high expression in the integument and gut; HZSIPbV1 was primarily enriched in the trachea and fat body; SYNNV2 displayed strict tissue specificity, with high expression exclusively in the integument (Fig. 4a). For newly identified viruses, all four novel viruses exhibited peak expression in the integument but differed in tissue distribution: CSPhelV1 had strict tissue specificity, with scarce expression in other tissues; CSGDV1 maintained low-level expression in the trachea, fat body, and Malpighian tubules; CCSlV1 was widely distributed across all tissues except the silk gland, though expression levels varied; CCMoV1 was detected at low levels in the trachea and fat body (Fig. 4b). These findings suggest that differential tissue expression models among viruses may be closely associated with their infection routes and host interaction mechanisms. The corresponding data are shown in (Table S9). RNA virome discovered in C. chilonis As a parasitoid wasp of C. suppressalis, we further characterized the RNA virome in five C. chilonis libraries derived from C. suppressalis hosts. A total of 8 RNA viruses were identified (Tables S5 and S6). Among these, 3 viruses were previously reported: 2 + ssRNA botourmiaviruses, Laodelphax striatellus botourmia-like virus 1 and 2 (LSBolV1 and LSBolV2), were first detected in the rice pest L. striatellus [32]; 1 + ssRNA virus, SYNNV2, was originally isolated from S. inferens collected in a rice field in Sanya City, Hainan Province, China [29]. The remaining 5 were newly discovered: 1 putative ISV belonging to the genus Mobuvirus (family Phenuiviridae) was identified. For this virus, an additional genomic segment was characterized in addition to the segment encoding the RdRP (Fig. S5); 2 + ssRNA magouliviruses (Botourmiaviridae) and 1 + ssRNA Solemo-like virus (Solemoviridae), which were speculated to be arboviruses; 1 + ssRNA beny-like virus (Benyviridae) (Fig. S8). Detailed descriptions for the above viruses are provided in Supplementary file S2. Comparative analysis of viral diversity between C. suppressalis and C. chilonis C. suppressalis and its parasitoid C. chilonis have developed intricate molecular interaction mechanisms through long-term coevolution [50]. RNA viruses can not only regulate virus-parasitoids interactions [51], but emerging evidence suggests that cross-species viral transmission may also contribute to an evolutionary advantage of divergent parasitism strategy [52]. These findings underscore the importance of studying viral diversity from a whole perspective that encompasses the entire parasitic system. In this study, we further analyzed shared viral components (TPMs > 10) between C. suppressalis and C. chilonis samples. Among previously reported viruses, both HZSISolV1 and SYNNV2 were detected in C. suppressalis and C. chilonis samples, yet their distribution patterns and host specificity exhibited significant differences (Tables S5 and S7). HZSISolV1 was widely distributed in C. suppressalis (23 samples) (Fig. S9a) with high abundance, whereas it was detected at an extremely low level in only one C. chilonis sample (Cc-l NB 2024) from the Ningbo population. Notably, the paired C. suppressalis host sample (Cs-l NB 2024) also exhibited a high HZSISolV1 abundance (Fig. S9b), suggesting that the low viral signal in the C. chilonis likely originated from host contamination during dissection or feeding rather than real viral infection. In contrast, SYNNV2 displayed a multi-sample distribution in both hosts across multiple locations (4 C. suppressalis samples and 3 C. chilonis samples) (Fig. S9a) and was detected at an exceptionally high abundance in one C. chilonis sample (Cc-a WZyq 2023). Its corresponding host sample (Cs-l WZyq 2023) also showed relatively high viral abundance (Fig. S9b). Among the novel viruses identified, 6 (CSPhelV1, CSGDV1, CSPPV1, CCSlV1, CCMoV1, and CCBlV1) were detected in both C. suppressalis and C. chilonis (Tables S6 and S8). Notably, CSPPV1 was restricted to C. suppressalis from Wuhan (Cs-l WH 2023) and C. chilonis from Ningbo (Cc-l NB 2024), exhibiting low abundance in both species (Fig. S9b). The spatially distinct detection patterns between host and parasitoid may eliminate cross-species transmission possibility. CSPhelV1 showed consistently low abundance in Wenzhou samples (Cc-a WZyq 2023/Cs-l WZyq 2023), with incompleteness of the viral sequence (Fig. S9b). CCMoV1 and CCBlV1 were significantly enriched in C. chilonis compared to its paired host C. suppressalis (Fig. S9b), suggesting parasitoid-specific viral adaptation or replication advantages. CCSlV1 was highly abundant in C. chilonis from Ningbo (Cc-l NB 2024) but rare in its paired C. suppressalis host (Cs-l NB 2024); intriguingly, it was also detected in C. suppressalis from Wenzhou (Cs-l WZyq 2023) but rare in its paired C. chilonis from Wenzhou (Cc-a WZyq 2023) (Fig. S9b). This contradiction may imply environmental acquisition of this virus for C. suppressalis, but not C. chilonis-dependent transmission. CSGDV1 was widespread in C. suppressalis (5 samples) and C. chilonis (3 samples) (Fig. S9c), yet revealed high abundance in C. chilonis while low abundance in corresponding host C. suppressalis from Ningbo (Cc-l NB 2024/Cs-l NB 2024) and Wenzhou samples (Cc-a WZyq 2023/Cs-l WZyq 2023) (Fig. S9b), highlighting its parasitoid-predominant amplification. High-confidence genomes (identity and coverage > 95%) were assembled for the above 8 viruses in various samples (Table S10). Notably, the viral sequence of CSPhelV1 is incomplete, with only 27–30% sequence coverage aligning with its homologues found initially in the Cs-l (3) NBjs 2024 library. The C. chilonis COX1 gene sequence (0.0032%–0.1433%) can be detected in the above C. suppressalis samples (Table S4), which indicated potential parasitoid-derived contamination. Overall, cross-species viral transmission cannot be inferred with these novel viruses directly. However, it is speculated that CCMoV1, CCBlV1, CCSlV1, and CSGDV1 may exhibit stronger adaptability to C. chilonis. Additionally, CSGDV1 was widely detected in C. suppressalis across multiple locations, suggesting that it may have a broader host range. Analysis of virus-derived small RNAs (sRNAs) sRNA-mediated RNAi (siRNA) pathways is one of the most important antiviral pathways in insects by producing virus-derived small interfering RNAs (vsiRNAs) to restrict virus replication [53, 54]. To characterize siRNA-mediated antiviral immunity in C. suppressalis and its parasitoid wasp C. chilonis, we conducted sRNA sequencing in both species. In C. suppressalis, four highly abundant viruses [(HZSICV1, CSSelV1, Chilo suppressalis partiti-like virus 1 (CSPalV1), and Chilo suppressalis mobuvirus 1 (CSMoV1)] generated vsiRNAs predominantly 20-nt in length, with nearly equal contributions from sense and antisense strands of viral genomic RNAs (Fig. 5a–d). Notably, HZSIPbV1 produced not only 20-nt vsiRNAs but also a peak of 28-nt sRNAs, displaying a size distribution reminiscent of virus-derived piwi-interacting RNAs (piRNAs) [55]. To investigate whether these 28-nt sRNAs are involved in the piRNA pathway, their Ping-pong cycle and sequence signatures were explored. The results revealed a hallmark 10-nt overlap between 24–29 nt sRNAs from opposite strands. Furthermore, these sRNAs exhibited a uridine (U) bias at the first position of antisense reads and an adenosine (A) bias at the tenth positions of sense reads (Fig. 5e). Collectively, these findings revealed the presence of HZSIPbV1-derived piRNAs in C. suppressalis. Additionally, HZSISolV1 showed a strong sense-strand bias and a wider vsiRNA range (16–34 nt) (Fig. 5f), implying an unconventional biogenesis mechanism. Profiles of virus-derived sRNAs. a-f. Profiles of sRNAs from Chilo suppressalis. g. Profile of sRNAs from Cotesia chilonis, h–k. Comparative sRNA profiles from C. suppressalis (left panels) and C. chilonis (right panels). Virus name abbreviations are provided on the top of each panel. For (a-k), the upper subpanels reveal the size distribution and 5′ base preference of sRNAs. The lower subpanels show the coverage of 20- or 22-nt sRNAs along the corresponding viral genomes, with reads mapping to the sense and antisense strands colored blue and brown, respectively. For (e), the lower right subpanel displays the coverage of 24–29 nt sRNAs. The corresponding upper inset figures depict characteristics of these reads: a histogram showing the normalized frequency of read overlap sizes aligned to the sense and antisense strands (left), and sequence logos for the two strands (right) For one known virus (SYNNV2) and four novel viruses (CCBlV1, CCMoV1, CCSlV1 and CSGDV1) that were detected in both C. suppressalis and C. chilonis, we performed comparative sRNA analyses in both species. Results showed that CCBlV1-derived vsiRNA (peaking at 22 nt) were exclusively found in C. chilonis (Fig. 5g). CCMoV1, CCSlV1, CSGDV1, and SYNNV2 exhibited parasitoid-dominated vsiRNA patterns, in both hosts, producing vsiRNAs with the 22-nt dominant length and nearly equal sense/antisense strand representation from viral genomic RNA (Fig. 5h–k). Furthermore, CCMoV1, CCSlV1, and CSGDV1 showed significantly higher vsiRNA abundances in C. chilonis than in C. suppressalis. These divergences suggest that despite their intimate ecological relationship, C. suppressalis and C. chilonis have evolved independent antiviral strategies. Shared virome in ecological viral communities In order to investigate the influence of surrounding ecological viral communities on the virome of C. suppressalis, we further examined potential virus sharing among insect populations occupying similar ecological niches by conducting virome analysis on three S. inferens samples collected from rice fields in Ningbo City. A total of 9 RNA viruses were identified, comprising 3 known and 6 novel viruses (Tables S3, S5, and S6). The known RNA viruses included 2 -ssRNA viruses: Sanya sesamia inferens phasmavirus 1 (SSIPV1, belonging to Phasmaviridae) and Hangzhou peribunyavirus 1 (HZPbV1, belonging to Peribunyaviridae). Additionally, SYNNV2, which was detected in both C. suppressalis and C. chilonis in this study, was also found in S. inferens. Both SSIPV1 and HZPbV1 were previously identified in S. inferens and the yellow rice borer, Scirpophaga incertulas, in Sanya City, Hainan Province, China [29]. HZPbV1 was originally discovered in Frankliniella intonsa (Thysanoptera) from Fuyang District, Hangzhou City, Zhejiang Province, China (Accession: SAMN14734864). For novel viruses, phylogenetic analysis and genomic structural characteristics were shown in (Figs. S5, S6, S10). 1 newly discovered -ssRNA virus [Sesamia inferens phenui-like virus 1 (SIPhelV1)] was found in both C. suppressalis and S. inferens (Table S6). Phylogenetic analysis revealed that SIPhelV1 grouped with the CSPhelV1 (Fig. S5), a newly discovered virus in C. suppressalis in our study, and both viruses cluster with other ISVs, further supporting their classification as ISVs. Besides, 5 novel viruses [Sesamia inferens tombus-like virus 1 (SITlV1), Sesamia inferens mobuvirus 1 (SIMoV1), Sesamia inferens partiti-like virus 1 (SIPalV1), Sesamia inferens cypovirus 1 (SICpV1), and Sesamia inferens sedoreo-like virus 1 (SISelV1)] were only identified in S. inferens but not in C. suppressalis. In addition to the segment encoding the RdRp or its conserved motifs, a second genomic segment was identified for both SIMoV1 and SICpV1 (Figs. S5, S6). Detailed descriptions for the novel viruses are provided in (Supplementary file S3). High-confidence genomes (identity and coverage > 94%) were assembled for SSIPV1, SYNNV2, and SIPhelV1 in various samples (Table S11). We further analyzed vsiRNAs in S. inferens infected with 5 distinct viruses (SISelV1, SICpV1, SIPalV1, SIMoV1, and SIPhelV1). Strikingly, all infections exhibited a dominant 20-nt vsiRNA peak in both strands (Fig. S11), like the vsiRNA profiles observed in C. suppressalis. In order to further investigate the potential influence of environmental virus communities on the virome of C. suppressalis, we also conducted a screening of low-abundance (< 20X coverage) viruses associated with rice pests and rice plants in C. suppressalis and its parasitic wasp C. chilonis (Table S12). The detected viruses included: Nilaparvata lugens reovirus (Spinareoviridae) [56]; Oryza sativa alphaendornavirus (Endornaviridae), which is abundant in rice plant [36]; a picorna-like virus (Riptortus pedestris virus-1), which was found in the pest R. pedestris [57]; Hangzhou hydrellia griseola dicistrovirus 1 (Dicistroviridae), initially identified in the leaf-mining pest Hydrellia griseola; another Dicistroviridae member, Hangzhou dicistrovirus 2, identified in Paederus fuscipes, a natural predator of rice pests [58]. Besides, three fungal viruses were also identified: Saccharomyces cerevisiae virus L-BC (La), Saccharomyces cerevisiae virus L-A, and Cordyceps javanica negative-strand RNA virus 1 (CjNRSV1). CjNRSV1 infects entomopathogenic fungus Cordyceps javanica isolated from the rice pest L. striatellus [59]. The low abundance of these viruses may indicate multiple environmental sources in the same habitat or possible cross-species transmission events. Further field tracking and laboratory analyses are needed to clarify the dynamics of these viruses. Discussion In this study, we systematically described the composition and potential evolutionary mechanisms of the C. suppressalis virome from both macroecological and species interactions perspectives. Mitochondrial genome analysis revealed four mainly distinct geographic populations of C. suppressalis in China, with an eastern-centered genetic diversity hub (Fig. 1b). This pattern is similar to the findings from a recent genome resequencing study of the C. suppressalis, which similarly reported higher genetic diversity in central populations containing central-eastern individuals [60]. Notably, the eastern region not only contributed samples to all four predominant genetic groups (groups 1–4) but also harbored the highest number of region-specific viruses (n = 6) (Fig. 3e). Group 2 exhibits the broadest geographical spread across five regions, with 7 specific viruses (Fig. 3c). This distribution pattern implies that human-mediated transport and regional host diversity may collectively promote viral diversification, like in honeybee [14], warranting enhanced surveillance for emerging viral threats. 20 RNA viruses (5 known and 15 novel) with high coverage (> 20X) were identified in C. suppressalis samples. Among the 5 known viruses, the narnavirus S20SRNV belongs to the family Narnaviridae, members of which have been identified in a wide range of eukaryotic hosts, including fungi, plants, and arthropods [61, 62]. Many narnaviruses are known as mycoviruses and are commonly detected in invertebrates associated with rice ecosystems [29]. The other 4 viruses (HZSICV1, HZSISolV1, HZPV1, and HZSIPbV1) were previously reported from various rice-associated invertebrate pests collected from the same geographic region (Fuyang District, Hangzhou City, Zhejiang Province, China) [29]. For newly identified viruses, there were 9 potential ISVs [Chilo suppressalis dicistro-like virus 1 and 2 (CSDlV1 and 2), Chilo suppressalis iflavirus 1 (CSIfV1), CSMoV1, CSPhelV1, Chilo suppressalis yingvirus 1 (CSYV1), Chilo suppressalis alphahymrhavirus 1 (CSAHV1), CSPPV1, CSGDV1] and six arboviruses [CSLNV1, CSLNV2, CSPalV1, Chilo suppressalis kita-like virus 1 (CSKlV1), CSSelV1, CSQRV1]. Among all the 20 RNA viruses, two viruses demonstrated exceptional host adaptability: HZSIPbV1 and CSSelV1 were detected across all climatic regions (Fig. 3d, e), suggesting evolutionary mechanisms for climatic flexibility. It is reported that many insect species possess stable core viruses, indicating a long-term co-evolution between the viruses and their hosts. These viruses may have originated from the environment and established a stable infection relationship. For example, species-specific core viruses have been identified in the mosquitos Aedes aegypti and Culex quinquefasciatus [63], along with cryptic-species-specific core viruses in the important agricultural pest, the whitefly Bemisia tabaci [30]. Particularly noteworthy is CSSelV1, a member belonging to the family Sedoreoviridae. Several Sedoreoviridae members possess significant pathogenic potential, such as rice dwarf virus, which is primarily transmitted by green rice leafhoppers, Nephotettix cincticeps [64]. Notably, Southern-specific viruses (CSDlV1, CSDlV2, and CSIfV1) may reflect thermal adaptation, while Xinjiang's arid-adapted CSLNV1/2 and Changchun's cold-adapted CCBlV1 exemplify extreme-environment specialization (Fig. 3e). Furthermore, based on alpha and beta diversity analyses, the genetic differences among C. suppressalis geographical populations influenced the viral species composition but not the virome structure (Fig. 3f–i). Combined with viral abundance characteristics, we speculate that the virome of C. suppressalis is primarily dominated by a stable core set of viruses (such as HZSIPbV1, HZSICV1, HZSISolV1, and CSSelV1). Although region-specific viruses were detected in certain climatic environments, their low abundance might limit the impact on the overall virome structure and did not significantly alter the fundamental composition of the core viral community. These above findings collectively demonstrate how viral diversity in C. suppressalis is shaped by the host population structure and climatic selection pressures-a framework that could inform predictive models of agricultural virus emergence. By integrating host geographic population genetic structure and virome distribution patterns, we can develop environment-host-virus interaction models. These models not only provide a scientific basis for predicting viral epidemic trends across different ecoregions, but also offer novel perspectives for analyzing the historical impact of human agricultural activities on disease transmission. As early as 1968, researchers have already explored the viral diversity, transmission patterns, and interactions with other microorganisms in Lepidoptera by collecting insects from different geographical locations. This approach provided a theoretical foundation for epidemic prediction and forecasting [65]. Another successful application of this model involves honeybees, a crucial agricultural pollinator species, facilitating enhanced disease management strategies and promoting bee conservation and beekeeping industry development [12–14]. Furthermore, population structure analysis based on viral communities also has been successfully applied. For example, an epidemiological modeling of ISVs was utilized to elucidate the population structure and dynamics of mosquitoes [15]. A few studies have demonstrated that many viruses play crucial regulatory roles in parasitoid wasps by establishing symbiotic relationships through convergent evolution, thereby facilitating improved colonization within host species. For example: pioneering research utilizing viruses from divergent Nasonia parasitoid wasp species has firstly demonstrated the phylosymbiotic relationship between insect viromes and their respective hosts [66]. A poxvirus (DIEPV) in Diachasmimorpha longicaudata wasps exhibits functions similar to known viral elements that benefit parasitism, displaying high virulence within its fly host Anastrepha suspensa [67]; PpNSRV-1 in Pteromalus puparum, the closest homolog to CSPPV1 identified in our study, can manipulate the secondary sex ratio by reducing female offspring numbers while increasing wasp longevity, thereby enhancing parasitism success [68]. In contrast, DlNaLV, a virus specific to D. longicaudata, exclusively infects female wasps and may be lethal to males [69]. Another virus, the negevirus ZcNLV, can infect both D. longicaudata and its host fly Zeugodacus cucurbitae [69]. Similar to these established findings, our study reveals a complex host-parasitoid-virus interaction network involving multiple viruses co-detected in C. suppressalis and its parasitoid C. chilonis, which exhibit distinct dynamics through differential distribution patterns, viral abundances, and RNAi-mediated immune responses. Notably, SYNNV2 emerges as a strong candidate for cross-species transmission, showing widespread distribution and high viral/vsiRNA abundance in both species (Fig. 5k and Fig. S9a, b). CSGDV1 presents another potential case of dual-host infection, exhibiting parasitoid-biased amplification, as evidenced by its widespread detection alongside significantly higher replication in C. chilonis (elevated viral and vsiRNA abundance) compared to its host C. suppressalis (Fig. 5j and Fig. S9b, c). CCMoV1 and CCBlV1 display ecological dominance in C. chilonis, characterized by their high viral/vsiRNA abundance (Fig. 5g, h and Fig. S9b). Remarkably, all the above viruses exhibit 22-nt dominant vsiRNA profiles in both C. suppressalis and C. chilonis, a characteristic signature of C. chilonis identified in our study. This finding strongly suggests adaptive evolution of these viruses within C. chilonis, while viral replication in C. suppressalis may be influenced by RNAi mechanisms derived from the parasitoid. During oviposition, C. chilonis injects various factors (including venom and viruses) into C. suppressalis to manipulate host development and immunity [50, 70]. A previous study has demonstrated that parasitic insect-derived miRNAs can modulate host development [71], whether similar siRNA-mediated regulation exists between parasitic wasps and their hosts remains an important question for future investigation. Furthermore, whether C. chilonis manipulate their parasitic relationship with the C. suppressalis by transmitting beneficial viruses and utilizing their own viral replication mechanisms requires further experimental validation. Distinct vsiRNA profiles were identified in C. suppressalis and C. chilonis, suggesting they may employ different RNAi-mediated antiviral mechanisms (Fig. 5). This finding further implies that viruses may have evolved host-specific adaptive strategies or counter-defense mechanisms, driving an evolutionary ‘arms race’ between viruses and their hosts [72]. piRNA-like antiviral response is usually associated with dipteran insects (e.g., mosquitoes and Drosophila) [54, 55]. Interestingly, we also detected HZSIPbV1-derived piRNAs, in C. suppressalis, which displayed canonical piRNA signatures observed in mosquitoes [73] (Fig. 5e). A previous study has demonstrated that Bombyx mori latent virus infection in B. mori ovaries generates both 20-nt siRNAs and 26–30 nt piRNAs, which cooperatively regulate the balance between viral replication and cell growth [74]. Whether a similar piRNA-mediated antiviral pathway exists in C. suppressalis needs further study. Furthermore, dsRNA dicing activity in B. mori was confirmed to be mediated by Dicer-2, which specifically cleaves at the 20-nt site [75]. Similarly, Dicer-2 has been shown to produce vsiRNAs in the lepidopteran Helicoverpa armigera, yielding abundant 20-nt fragments [76]. Given the prevalence of 20-nt-dominant vsiRNAs in C. suppressalis and S. inferens (Fig. 5a–e and Fig. S11), we speculate that this may represent a conserved feature among Lepidoptera species. Additionally, the 22-nt vsiRNA profile may be a typical feature in parasitoid wasps, as observed in D. longicaudata [69] and the results in this study (Fig. 5g–k). It should be noted that RNAi pathways facilitate the establishment of persistent viral infections [77]. Therefore, vsiRNA characteristics should be carefully analyzed when studying host-switching events of viruses. Overall, a deeper investigation into the origin, transmission, and intricate dynamics of RNA viruses within parasitoid-host-virus systems is essential for promoting pest biological control strategies in agricultural ecosystems. SSIPV1 and SIPhelV1 were detected in both C. suppressalis and S. inferens, while SYNNV2 was found to be shared among C. suppressalis, C. chilonis, and S. inferens (Tables S5 and S6). The high sequence identities of these viruses (> 94% for SIPhelV1, > 99% for SSIPV1 and SYNNV2) across different host species indicate their potential for cross-species transmission among ecologically similar insects (Table S11). Notably, the SIPhelV1 robust grouped with CSPhelV1 in phylogenetic analysis (Fig. S5), although they share no significant nucleotide identity. This divergence may suggest divergent evolution from a common ancestor followed by host-specific adaptation, or convergent evolution driven by ecological niche pressures in sympatric host species. The rice ecosystem is characterized by its rich biodiversity. Rice serves as a common host for diverse organisms, facilitating extensive viral exchange events within the ecosystem. These interactions propose significant challenges for viral disease prevention in rice production [36, 78]. Studies have revealed a complex, bidirectional virus exchange network involving rice, rice-feeding insects, and surrounding plants [36]. For example, the rice tiller inhibition virus (RTIV), a newly discovered RNA virus, can be transmitted via aphids from wild rice to cultivated rice [79]. As one of the most notorious boring pests of rice, C. suppressalis may act as a potential reservoir or intermediate host for crop viruses. Therefore, comparative analysis of viruses associated with other biological factors is crucial for a deeper understanding of the viral composition in C. suppressalis. To date, systematic research on rice pest-associated viruses has primarily focused on sucking pests such as aphids [31], planthoppers, and leafhoppers [32, 80]. This study presents the first systematic virome analysis of C. suppressalis in China, expanding our understanding of viral diversity in chewing insects. The identification of numerous viruses within the C. suppressalis–C. chilonis parasitic system that were previously reported in rice or other rice-associated insects, such as LSBolV1 and LSBolV2 in L. striatellus [32] (Tables S5 and S12), highlights that C. suppressalis is a key component of the rice ecosystem. It may play an essential bridging role in viral transmission through direct physical contact or indirect pathways mediated by feeding, thereby facilitating cross-species viral exchange within the agricultural environment. Through systematic comparison of the compositional characteristics among five sample groups: rice viral families/rice-associated viral families (n = 57) [36], viral families in the important rice virus vector L. striatellus (n = 14) [32], viral families in S. inferens (n = 7), C. suppressalis (n = 24), and C. chilonis (n = 8) that were identified in this study, we revealed a distribution pattern of virome at family level in rice ecosystems (Fig. 6). Results demonstrated that the viral families Phenuiviridae and Narnaviridae were consistently present across all five groups, indicating their broad host adaptability and epidemic potential within rice ecosystems. Conversely, the 33 rice-specific viral families not detected in the other four species suggest potential origins from other biological sources such as alternative insect vectors, surrounding plants, or soil microorganisms etc. (Fig. 6a). Rice shares 19, 10, and 6 viral families with the three rice-feeding pests, C. suppressalis, S. inferens, and L. striatellus, respectively. Notably, C. suppressalis exhibits the highest level of viral sharing, though there is currently no clear evidence of rice virus transmission by C. suppressalis, suggesting its potential role as a key reservoir for rice-associated viruses. Strikingly, three viral families, Spinareoviridae, Partitiviridae, and Phenuiviridae, were detected across all four species (rice and the three insect pests), providing potential evidence for cross-species viral transmission in rice-insect interactions (Fig. 6b). Two potential transmission routes may be involved: (i) direct insect-to-insect contact within the parasitic system, and (ii) indirect transmission mediated by rice as a shared host, such as the oral-fecal transmission of a Picorna-like virus (Himetobi P virus) among planthoppers, with rice acting as a transient reservoir [81, 82]. Collectively, these findings not only confirm the complexity of virus-host interactions in rice ecosystems, but also provide novel insights for understanding viral transmission networks within agricultural ecosystems. Comparative analysis of different virus family in co-occurring species: Chilo suppressalis, Cotesia chilonis, Sesamia inferens, Laodelphax striatellus, and rice. a. Venn diagram illustrating virus family overlap across species. b. Schematic representation of virus-mediated interactions between species. Solid lines indicate direct connections; dashed lines represent indirect linkages. The colors black, blue, and red indicate that viruses from the corresponding family were identified with high abundance (> 20 × coverage), low abundance (< 20 × coverage), or both, respectively Conclusion This study reveals that both geographic genetic divergence and ecological viral communities collectively shape the composition and dynamics of the C. suppressalis virome. Virome analysis of C. suppressalis across China identified four major geographic populations (Groups 1–4) and unveiled 20 confident viral species (15 novel), featuring ubiquitous core virus (notably HZSIPbV1 and CsSelV1) and climate-adapted viruses. Parallel examination of C. chilonis detected 8 confident viruses (6 novel). Differential viral distribution patterns, loads, and vsiRNA profiles in the C. suppressalis-C. chilonis parasitic system indicated host-driven viral adaptation. Co-occurrence network analysis further showed that sympatric species (including S. inferens and L. striatellus) and rice may collectively influence C. suppressalis virome. Moreover, the abundant meta-transcriptomic and sRNA data obtained for C. suppressalis, C. chilonis, and S. inferens will provide valuable resources for research on viral epidemiology and contribute to the development of preventive strategies against viral diseases and pest outbreaks. Methods Sample collection Based on distinct climatic zones, we subdivided China into seven major regions according to Li et al. [14]: North China, Central China, East China, South China, Northwest China, Southwest China, and Northeast China. From 2022 to 2024, we collected 40 C. suppressalis samples across 23 cities in 14 provinces, ensuring nationwide representativeness (Fig. 1a, Table S1). The majority of the samples (n = 22) were collected from Eastern China: 14 larval samples from rice fields at 8 geographical sites in Zhejiang Province; 1 larval sample collected in Ningbo Yongjiang water bamboo field; 5 larval samples from rice fields in Jiangxi, Jiangsu and Fujian provinces, respectively; 2 larval samples from laboratory-reared populations maintained on an artificial diet from Fujian provinces. The second largest group of samples (n = 7) were collected from Central China: 3 larval samples from rice fields in Hubei, Henan, and Hunan provinces, respectively; 4 samples from laboratory cultures fed on artificial diet covering all development stages (egg, larva, pupa, adult) were provided by Hunan Agricultural University; The other samples (n = 11) were distributed in the other five geographical and climatic regions: 1 laboratory-reared larva (from Beijing, reared on artificial diet) in North China; in Northeast China, Northwest China, and South China, 2 rice field-collected larval samples per region; 4 rice field-collected larval samples in Southwest China. Additionally, we obtained 5 parasitic wasps (C. chilonis) larva or adult samples from corresponding C. suppressalis hosts (4 from rice fields in Hubei, Zhejiang, and Jilin provinces, respectively; 1 from Ningbo Yongjiang water bamboo field) (Fig. 1a, Table S2). Furthermore, for comparative analysis, 3 S. inferens larval samples were also collected from two rice fields in Ningbo City for sequencing (Table S3). All samples were subjected to meta-transcriptomic sequencing, and a total of 48 libraries were generated. RNA-seq library construction and dataset reassembly Each sample of the C. suppressalis consisted of 1–9 larvae, or 2–3 egg masses, or 1–3 adults. The C. chilonis samples contained either 5–10 adults or 15 larvae. For S. inferens, each sample comprised 2 or 3 larvae. Total RNA was extracted from each sample using the Trizol RNA Extraction Kit following the manufacturer's protocol. RNA quality and concentration were assessed using a Nanodrop 2000 spectrometer (Thermo Fisher Scientific, Waltham MA, USA). Eukaryotic ribosomal RNA was removed from the total RNA, and the remaining RNA was fragmented into approximately 250–300 bp fragments. Double-stranded cDNA was synthesized through reverse transcription. The cDNA underwent end repair, A-tailing, adapter ligation, size selection, and PCR amplification to construct meta-transcriptomic libraries. The library was checked with Qubit and real-time PCR for quantification. A total of 48 qualified libraries (40 from C. suppressalis, 5 from C. chilonis, and 3 from S. inferens) were subjected to paired-end (150 bp) sequencing on the Illumina NovaSeq X Plus platform (Illumina, USA) by Novogene (Tianjin, China). Raw sequencing data were processed using Fastp (v0.23.0) [83] and SOAPnuke (v2.1.8) with “-q 0.2” parameter [84] to remove adapter sequences and low-quality reads, yielding clean reads. Potential host-derived contamination was removed by aligning clean reads to respective host genomes (C. suppressalis, C. chilonis, and S. inferens) using BWA software (v0.7.17) with default parameter [85]. De novo assembly was performed using SPAdes (v4.0.0) [86] with the spades.py –rnaviral parameter. RNA viral identification This study employed bioinformatic prediction for systematic RNA virus identification, the analytical workflow consisted of six key steps: (i). Primary viral sequence screening. Putative RNA viral contigs were initially identified using VirSorter2 (v2.2.4) [87] with the parameter –include-groups RNA, utilizing its built-in ViPhOGs database for RNA virus detection. Additionally, contigs containing predicted hallmark genes encoding viral RNA-dependent RNA polymerase (RdRP) [88] domains were also screened. Open reading frames were predicted using Prodigal (v2.6.3) [89] with the “-p meta” parameter, followed by hmmscan (v3.3.2) alignment of predicted protein sequences against Hidden Markov Models (HMMs) from the NeoRdRp (2.1) and RVDB-prot (29.0) databases [90, 91]. (ii). Species-level viral clustering. Predicted RNA viral contigs were clustered into species-level viral operational taxonomic units (vOTUs) using CheckV's ANI_calc.py (v0.8.1), with Average Nucleotide Identity [ANI] ≥ 95% and coverage ≥ 85%. This step effectively reduced redundancy while improving the accuracy of downstream analyses. (iii). The Viral Taxonomic Assignment Pipeline (VITAP1.5) was employed [92] to screen contigs from vOTUs that could be assigned to the species level. This tool maintains synchronization with the latest ICTV reference database, incorporates an alignment-based graph algorithm, and provides confidence scores for each taxonomic assignment. (iv). Homology-based viral identification. vOTUs were dual-aligned with parallel searches against the National Center for Biotechnology Information (NCBI) viral RefSeq database (download date: 2024.09.12) and the NCBI non-redundant (NR) protein database (download date: 2024.08.20) using diamond blastx (v2.1.9) [93]. Only contigs that were consistently identified as viral in both alignments were retained. (v). To identify potential segmented fragments in viral genomes, we developed an analytical pipeline as follows: pairwise abundance correlations for all viral contigs across all samples were calculated using Spearman's rank correlation coefficient (ρ), and contig groups showing strong positive correlations (ρ > 0.8 and p < 0.05) were selected. Contigs homologous to known segmented viruses in the ICTV database were designated as “anchors”. Unannotated contigs that showed high abundance correlation with these anchor contigs were considered candidate novel segments. Finally, to confirm their identities, a targeted BLASTp (v2.16.0) search against a viral protein database extracted from the ICTV database (MSL40.v1.2025.03.07) was performed using the protein sequences encoded by these candidate contigs, with an E-value cutoff of < 1e-5. (vi) False-positive filtration. Stringent filtering criteria were applied: non-viral homologs were further excluded via blastx (v2.1.9) against NCBI NR database (e-value ≤ 1e-5, min-score ≥ 50), and potential endogenous viral elements were removed through blastn on the NCBI website with host genomes (coverage threshold ≥ 10%). Finally, a high-confident reference dataset of RNA viral contigs was constructed. Viral genome annotation and phylogenetic analysis Notably, during the process of virus identification, annotations of virus homologues can be obtained. While, for sequences with multiple homologous, predicted taxonomic assignment was determined using the lowest common ancestor (LCA) algorithm, with the most specific shared taxonomic level selected to maximize annotation precision. To ensure both comprehensive and reliable annotation of newly identified viral contigs, the RdRP was used for phylogenetic construction, together with referred RdRP protein sequences that loaded from ICTV. For contigs lacking initial RdRP hits underwent secondary validation via palm_annot to detect conserved RdRP motifs (domains A/B/C), ensuring sensitive detection through structural domain profiling [94], following the approach described by Zhu et al. [36]. The phylogenetic analysis was constructed based on maximum likelihood method with conserved viral RdRP sequences containing domains A/B/C to help refine their classification. Genomic structures analysis was conducted by Conserved Domain Search Service (2025.08) (https://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi). Dynamics of viral expression across development, tissues, and host plants in C. suppressalis To investigate the temporal expression dynamics of viruses in C. suppressalis, developmental samples (eggs, larvae, pupae, and adults) obtained from Hunan Agricultural University (Changsha, Hunan Province) were used. Additionally, newly hatched larvae from these cultures were reared on either rice seedlings or water bamboo to assess the influence of host plants on viral components and abundance. In order to examine spatial expression patterns, tissue samples (trachea, integument, fat body, gut, Malpighian tubules, and silk glands) were dissected from water bamboo field-collected larvae in Ningbo Yongjiang water bamboo field. Dissections were performed under a stereomicroscope (Olympus SZX7, Tokyo, Japan) using sharp forceps (Ideal-Tek, Switzerland) in 1 × phosphate-buffered saline. Total RNA was extracted from all samples using the TRIzol RNA Extraction Kit following the manufacturer’s protocol and subjected to meta-transcriptomic sequencing, according to the above RNA-seq methods. Viral abundances were reflected using TPMs. Clean reads were mapped to viral contigs using Bowtie2 (v2.4.5) with default parameters to calculate TPMs. For other viruses identified in this study, TPMs were similarly computed to evaluate their abundances. Mitochondrial genome-based phylogenetic analysis of C. suppressalis Trimmomatic version 0.39 [95] was used to remove adaptor contaminations and low-quality reads with the parameters: “ILLUMINACLIP:Illumina.fa:2:30:10:8:TRUE LEADING:3 TRAILING:3 SLIDINGWINDOW:4:15 MINLEN:150”. To assemble and annotate the mitochondrial genomes, we assembled the clean reads using MitoZ version 3.6 [96] in a Singularity (version 3.10.0) container. The annotated protein coding sequences were extracted and aligned based on codon sequences [97]. We then concatenated the coding sequences and infer the phylogenetic relationships among populations using IQ-Tree version 2.3.6 [98]. 1000 ultrafast bootstraps were used with Diatraea saccharalis as the outgroup. Abundance estimation of C. suppressalis and C. chilonis using COX1 gene To estimate the abundance of C. suppressalis and C. chilonis, we first extracted the COX1 gene from both species as suggested in the study of honeybee viruses [14]. We then mapped clean reads to the COX1 sequences using Bowtie2 (v2.4.5) [99] with default parameters. Lastly, we used SAMtools version 1.21 [100] to summarize the total number of reads and the number of reads mapped to the COX1 sequences of C. suppressalis and C. chilonis. Alpha and Beta diversity analyses To assess within-sample species diversity, we calculated alpha diversity indices. Specifically, the Richness, Chao1, Shannon, and Simpson indices were computed using the R package vegan (v2.7–1). To examine differences, overall significance was assessed with the Kruskal–Wallis non-parametric test. For indices showing significant differences, pairwise comparisons were performed using Dunn’s test, with p-values adjusted by the Benjamini–Hochberg method to control the false discovery rate. Differences were considered statistically significant at p-value < 0.05. To evaluate differences in virome composition between samples, a Bray–Curtis dissimilarity matrix was constructed based on viral abundance data using the R package vegan (v2.7–1). Permutational multivariate analysis of variance (PERMANOVA) was then applied, using the adonis2 function (vegan), with 999 permutations. To visualize differences in community structure, principal coordinate analysis (PCoA) was performed based on the Bray–Curtis dissimilarity matrix using the ape package (v5.8.1). Non-metric multidimensional scaling (NMDS) was also conducted using the metaMDS function in the same package for dimensionality reduction and visualization. The reliability of the NMDS ordination was assessed using the Stress value. sRNA analysis sRNA sequencing and analysis were performed on 6 C. suppressalis samples, 3 C. chilonis samples, and 2 S. inferens samples. A total of 11 cDNA libraries were constructed using the NEB Next® Multiplex Small RNA Library Prep Set for Illumina®(NEB E7300L, New England Biolabs) and sequenced on an NovaSeq 6000 platform (Novogene, Tianjin, China) (Table S13). Then, the raw sRNA reads were processed as follows: adapter trimming and quality filtering were performed using fastp (v0.23.0) [83] to remove low-quality sequences; clean reads (15–35 nt) were extracted using the seqkit toolkit (v2.8.2) [101]. These reads were then aligned to host genomes (genomes of C. suppressalis, C. chilonis, and S. inferens) using Bowtie1 (v1.3.1) [102] with “-v 0” parameter (allowing 0 mismatch) to remove host-derived reads. The unmapped reads were subsequently aligned to viral genomes using Bowtie1 (v1.3.1) with “-v 1” parameter (allowing 1 mismatch). Bioinformatic analyses, containing overall mapping statistics (read length distribution, 5’ terminal base preference, and strand bias) for reads of 15–35 nt, were performed using custom Python scripts. Data visualization was conducted using R (v4.3.1) and ggplot2 (v3.5.2). To characterize canonical vsiRNAs, predominant reads of 20 nt or 22 nt were filtered and mapped to viral genomes to reveal their strand-specific distribution using the ggplot2 (v3.5.1). For piRNA analysis, reads ranging from 24- to 29-nt were independently filtered and examined. First, the Ping-pong cycle was explored by analyzing the 5'-terminal distances between sense and antisense reads. The frequency distribution of these 5'−5' distances was then compared to a theoretical random distribution model and normalized using Z-score analysis [73, 103]. Result was visualized using ggplot2 (v3.5.1). Additionally, to investigate the canonical 1U/10A sequence signatures, the 24–29 nt reads were aligned by their 5’ ends, and the nucleotide frequency of the first ten positions was plotted separately for sense and antisense strands using the ggseqlogo 0.2 R package [104]. Moreover, the 24–29 nt reads were also mapped to the viral genome to reveal their strand-specific distribution using the ggplot2 (v3.5.1). Supplementary Information Supplementary Material 1. Fig. S1 Detailed phylogenetic analysis of novel RNA viruses from Chilo suppressalis within the order Picornavirales. Fig. S2 Detailed phylogenetic analysis of novel RNA virus from Chilo suppressalis within the order Martellivirales. Fig. S3 Detailed phylogenetic analysis of novel RNA viruses from Chilo suppressalis within the order Mononegavirales. Fig. S4 Detailed phylogenetic analysis of novel RNA viruses from Chilo suppressalis within the family Orthomyxoviridae and Qinviridae. Fig. S5 Detailed phylogenetic analysis and genomic structures of novel RNA viruses from Chilo suppressalis, Cotesia chilonis, and Sesamia inferens within the family Phenuiviridae. Fig. S6 Detailed phylogenetic analysis and genomic structures of novel RNA viruses from Chilo suppressalis and Sesamia inferens within the family Partitiviridae, Spinareoviridae, and Sedoreoviridae. Fig. S7 Alpha and Beta diversity analysis of virome in Chilo suppressalis. Fig. S8 Detailed phylogenetic analysis and genomic structures of novel RNA viruses from Cotesia chilonis within the family Botourmiaviridae, Solemoviridae, and Benyviridae. Fig. S9 Comparative analysis of known and novel viruses in Chilo suppressalis and Cotesia chilonis. Fig. S10 Detailed phylogenetic analysis and genomic structure of novel RNA virus from Sesamia inferens within the family Tombusviridae. Fig. S11 Profiles of virus-derived small interfering RNAs (vsiRNAs) from Sesamia inferens. Supplementary Material 2. File S1: Descriptions of novel viruses identified in the striped stem borer, Chilo suppressalis. File S2: Descriptions of novel viruses identified in the parasitoid wasp, Cotesia chilonis. File S3: Descriptions of novel viruses identified in Sesamia inferens. Supplementary Material 3. Table S1. Chilo suppressalis datasets used in this study derived from field investigation and lab cultures. Table S2. Cotesia chilonis datasets used in this study derived from field investigation. Table S3. Sesamia inferens datasets used in this study derived from field investigation. Table S4. Assignment of Chilo suppressalis geographical population and COX1 gene depth of C. suppressalis and Cotesia chilonis. Table S5. Known viruses identified across Chilo suppressalis,Cotesia chilonis, and Sesamia inferens samples. Table S6. Novel viruses identified across Chilo suppressalis, Cotesia chilonis, andSesamia inferens samples. Table S7. Distribution and abundance (Transcripts Per Kilobase Millions) of known viruses across Chilo suppressalis, Cotesia chilonis, and Sesamia inferens samples Table S8. Distribution and abundance (Transcripts Per Kilobase Millions) of novel viruses across Chilo suppressalis, Cotesia chilonis, and Sesamia inferens samples. Table S9. Spatio-temporal expression analysis of viruses in Chilo suppressalis. Table S10. Nucleotide identities and coverages of viral sequences shared in Chilo suppressalis/Cotesia chilonis libraries. Table S11. Nucleotide identities and coverages of viral sequences shared in Chilo suppressalis and Sesamia inferens libraries. Table S12. Known viruses identified (<20 coverage) across Chilo suppressalis and Cotesia chilonis samples. Table S13. Viruses and corresponding small RNA libraries. Acknowledgements We thank for Doc Lan-Zhi Han, You-Zhi Li, Bao-Qing Zhang, Yu Xi, Xiao-Meng Zhang, Sheng-Li Jing, Jin-Gang Xu, Xi Zhang, Ying Du, Ya-Ya, Yu, An-Di, Zhu for helps in sampling. Authors’ contributions LJB and ZCX conceived and designed research. LJB performed the experiments and Data analysis. TY, LYY, QYH, HF, RPP, HHJ, ZJC, HQL, LG, MQZ, and LJM help to perform partial experiments or Data analysis, LJB wrote the manuscript. ZCX, CJP reviewed the manuscript. All authors approved the final manuscript. Funding The study was supported by the National Natural Science Foundation of China (U21A20225, U23A6006). Data availability The meta-transcriptomic and sRNA datasets supporting the conclusions of this article are available in the Genome Sequence Archive in the National Genomics Data Center, China National Center for Bioinformation/Beijing Institute of Genomics, Chinese Academy of Sciences (GSA: CRA029234, GSA:CRA034937, and GSA: CRA031347) that are publicly accessible at https://ngdc.cncb.ac.cn/gsa. All newly discovered virus sequences from C. suppressalis, C. chilonis, and S. inferens in this study have been deposited into CNSA (https://db.cngb.org/cnsa/) with accession number CNP0007910. Declarations Ethics approval and consent to participate Not applicable. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes The original article has been updated: VideoID of the Video Abstract has been updated. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Change history 5/25/2026 The original article has been updated: VideoID of the Video Abstract has been updated. References Associated Data Supplementary Materials Supplementary Material 1. Fig. S1 Detailed phylogenetic analysis of novel RNA viruses from Chilo suppressalis within the order Picornavirales. Fig. S2 Detailed phylogenetic analysis of novel RNA virus from Chilo suppressalis within the order Martellivirales. Fig. S3 Detailed phylogenetic analysis of novel RNA viruses from Chilo suppressalis within the order Mononegavirales. Fig. S4 Detailed phylogenetic analysis of novel RNA viruses from Chilo suppressalis within the family Orthomyxoviridae and Qinviridae. Fig. S5 Detailed phylogenetic analysis and genomic structures of novel RNA viruses from Chilo suppressalis, Cotesia chilonis, and Sesamia inferens within the family Phenuiviridae. Fig. S6 Detailed phylogenetic analysis and genomic structures of novel RNA viruses from Chilo suppressalis and Sesamia inferens within the family Partitiviridae, Spinareoviridae, and Sedoreoviridae. Fig. S7 Alpha and Beta diversity analysis of virome in Chilo suppressalis. Fig. S8 Detailed phylogenetic analysis and genomic structures of novel RNA viruses from Cotesia chilonis within the family Botourmiaviridae, Solemoviridae, and Benyviridae. Fig. S9 Comparative analysis of known and novel viruses in Chilo suppressalis and Cotesia chilonis. Fig. S10 Detailed phylogenetic analysis and genomic structure of novel RNA virus from Sesamia inferens within the family Tombusviridae. Fig. S11 Profiles of virus-derived small interfering RNAs (vsiRNAs) from Sesamia inferens. Supplementary Material 2. File S1: Descriptions of novel viruses identified in the striped stem borer, Chilo suppressalis. File S2: Descriptions of novel viruses identified in the parasitoid wasp, Cotesia chilonis. File S3: Descriptions of novel viruses identified in Sesamia inferens. Supplementary Material 3. Table S1. Chilo suppressalis datasets used in this study derived from field investigation and lab cultures. Table S2. Cotesia chilonis datasets used in this study derived from field investigation. Table S3. Sesamia inferens datasets used in this study derived from field investigation. Table S4. Assignment of Chilo suppressalis geographical population and COX1 gene depth of C. suppressalis and Cotesia chilonis. Table S5. Known viruses identified across Chilo suppressalis,Cotesia chilonis, and Sesamia inferens samples. Table S6. Novel viruses identified across Chilo suppressalis, Cotesia chilonis, andSesamia inferens samples. Table S7. Distribution and abundance (Transcripts Per Kilobase Millions) of known viruses across Chilo suppressalis, Cotesia chilonis, and Sesamia inferens samples Table S8. Distribution and abundance (Transcripts Per Kilobase Millions) of novel viruses across Chilo suppressalis, Cotesia chilonis, and Sesamia inferens samples. Table S9. Spatio-temporal expression analysis of viruses in Chilo suppressalis. Table S10. Nucleotide identities and coverages of viral sequences shared in Chilo suppressalis/Cotesia chilonis libraries. Table S11. Nucleotide identities and coverages of viral sequences shared in Chilo suppressalis and Sesamia inferens libraries. Table S12. Known viruses identified (<20 coverage) across Chilo suppressalis and Cotesia chilonis samples. Table S13. Viruses and corresponding small RNA libraries. Data Availability Statement The meta-transcriptomic and sRNA datasets supporting the conclusions of this article are available in the Genome Sequence Archive in the National Genomics Data Center, China National Center for Bioinformation/Beijing Institute of Genomics, Chinese Academy of Sciences (GSA: CRA029234, GSA:CRA034937, and GSA: CRA031347) that are publicly accessible at https://ngdc.cncb.ac.cn/gsa. All newly discovered virus sequences from C. suppressalis, C. chilonis, and S. inferens in this study have been deposited into CNSA (https://db.cngb.org/cnsa/) with accession number CNP0007910.