Novel 20th century rotavirus alphagastroenteritidis strains identified in formalin-fixed and ethanol-preserved museum specimens

Abstract

Archival specimens held in biorepositories (e.g. natural history collections) offer rare temporal snapshots of global biodiversity. These collections not only preserve species morphology and aspects of ecology, but increasingly provide access to historical molecular data, including insights into wildlife disease. As several pandemics have originated from animal viruses spilling over into the human population (i.e. SARS-CoV-2/COVID-19, 2009 H1N1 influenza, and HIV/AIDS), characterizing the diversity of viruses circulating in wildlife populations is essential for proactive pandemic preparedness. Yet, current surveillance remains biased toward contemporary viruses of economic importance. One solution to bridging spatiotemporal gaps in wildlife virus knowledge is retrospective screening of vouchered wildlife specimens. However, molecular analysis of specimens has been hindered by formalin fixation, which degrades and cross-links nucleic acids. Here, we demonstrate that formalin-fixed vouchered wildlife specimens retain both host and viral RNA fragments after being stored for up to 60 years. We recovered fragments of divergent strains of Rotavirus alphagastroenteritidis from two Australian species of order Chiroptera; Nyctophilus geoffroyi (lesser long-eared bat) and Rhinolophus megaphyllus (smaller horseshoe bat), representing the first characterization of R. alphagastroenteritidis (RVA) in Australian bats, and the oldest identification of the virus to date worldwide. Concurrently, we sequenced endogenous host RNA, providing a proof-of-concept for dual host–virus transcript recovery from vouchered specimens. This study highlights the role biorepositories can play in reconstructing unbiased historical viral landscapes from specimens, irrespective of the host disease status, and enabling spatiotemporal host–virus insight to advance both biodiversity science and global pandemic preparedness.

Introduction

Emerging infectious diseases, including zoonotic viruses, cause a substantial burden for human and animal health, with approximately one novel or reemerging threat detected annually (Bloom et al. 2017). Climate and land use changes (e.g. habitat destruction), resource scarcity (e.g. decreased food availability), and increased interactions between humans, wildlife, and domesticated animals have expanded the human–wildlife and wildlife–domestic animal interfaces, and are fueling the risk of zoonotic viral spillover (Woolhouse and Gaunt 2007; Bloom et al. 2017; Plowright et al. 2024).

To date, viral discovery and characterization efforts have been skewed toward pathogens that affect human health or agricultural productivity. As of 2022, 75% of animal virus entries into a major public repository for nucleotide sequences, the United States National Center for Biotechnology Information, were derived from humans or other animals with anthropogenic importance (i.e. disease vectors, economic value) (Harvey and Holmes 2022). This pronounced sampling bias is perpetuated in public genetic sequence repositories, limiting the effectiveness of viral discovery tools that rely upon this data and hindering identification of novel viruses across the broader viral ecosphere (Krishnamurthy and Wang 2017). Furthermore, the vast majority of wildlife viruses with zoonotic potential remain uncharacterized (Carroll et al. 2018) and are investigated only once there is sustained circulation in the human population (Vora et al. 2022). Practical pandemic preparedness requires broad-scale, unbiased, proactive surveillance to advance viral discovery in wildlife (Plowright et al. 2024).

Natural history collections may offer a powerful, yet underutilized, resource for retrospective disease surveillance (Colella et al. 2021). Historical specimens provide unparalleled temporal and taxonomic coverage of wildlife species. Traditionally, these specimens have long been used to describe new eukaryotic species, monitor long-term population dynamics, assess the ecological responses to environmental change (Suarez and Tsutsui 2004; Cook et al. 2014; Funk 2018; Meineke et al. 2019; Nachman et al. 2023), and provide insights into ecosystems before the onset of anthropogenic impact (Card et al. 2021; Davis and Knapp 2025). Collections value, both academically and to society, will continue to grow considering accelerating losses to biodiversity (Cowie et al. 2022; Kapun et al. 2025). The significance of specimens is established in research fields of biodiversity, systematics, and taxonomy (Kahrer and Zettel 2022; Enghoff and Akkari 2024; Schwentner and Hethke 2025), however, molecular genetics has broadened the multidisciplinary research in collections to include evolutionary biology, ecology, and environmental biology (Card et al. 2021 ; Raxworthy and Smith 2021; Fong et al. 2023). Linking data, such as genetic sequences, to a physical specimen enhances the scientific value of specimens and collections more broadly, a concept known as the “extended specimen.” This framework integrates data types including physical components (e.g. tissue samples), digital data (e.g, μCT-scans, genomic data), and contextual metadata (e.g. georeferenced collection coordinates, functional traits) (Webster 2017; Lendemer et al. 2020). Previous work has demonstrated how vouchered specimens have contributed the extended specimen principle through sequence data (Blaimer et al. 2016) and parasite linkage (Teixeira-Costa et al. 2023). Whole vouchered specimens make a particularly important contribution, as they have intact internal anatomy suitable for imaging (Gutiérrez et al. 2018), and for microbiome or parasite sequence analysis. Whole specimens collected prior to the 1980s were preserved in formalin, a chemical that crosslinks cellular molecules and was previously thought to prohibit molecular genetic analysis (Simmons 2014).

Recent biochemical and sequencing innovations have overturned the assumption that molecular data from formalin-fixed natural history specimens is inaccessible (Campos and Gilbert 2012; Hykin et al. 2015; Ruane and Austin 2017; McGuire et al. 2018; Derkarabetian et al. 2019; Zacho et al. 2021; Speer et al. 2022; Hahn et al. 2024 ; Holleley and Hahn 2025; Kapun et al. 2025, Straube et al. 2021, Totoiu et al. 2020). Studies now demonstrate that DNA and RNA can be recovered from clinical formalin-fixed paraffin-embedded (FFPE) samples (reviewed in Mathieson and Thomas 2019) and natural history collection formalin-fixed specimens, and used for microbial genomic and phylogenetic analysis (Table 1). For example, in samples with known or suspected pathology: the detection of the measles virus (Düx et al. 2020) and influenza A virus (Patrono et al. 2022) in formalin-fixed clinical specimens, and the detection of Lamprey reddening syndrome in FFPE samples of pouched lamprey (Geotria australis) (Miller et al. 2025). Due to the successful identification of several important pathogens in museum specimens from medical and natural history collections (Table 1), there have been proposed actions to integrate natural history collections and microbiological research to enhance pandemic preparedness (Colella et al. 2021; Thompson Cody et al. 2021). Linking pathogen data, including viral sequences, to a physical specimen enhances the scientific value of specimens under the extended specimen principle (Webster 2017 ; Lendemer et al. 2020). However, as discussed previously (Hykin et al. 2015; Speer et al. 2022; Holleley and Hahn 2025 ), collecting genomic data from natural history collection formalin-fixed or ethanol-preserved specimens present specific challenges, for example, long-term storage, a lack of standardized conditions (e.g. fixation conditions, documentation). Furthermore, compared to medical specimens (such as FFPE samples), historical natural history collection specimens have not been prepared with a focus on molecular preservation (Speer et al. 2022), and do not have a pathological association with a disease or infection. Despite these challenges, a range of microbes have been identified from natural history collection formalin-fixed and ethanol-preserved tissues with no established disease data using targeted enrichment approaches (PCR, hybridization capture). Collections have been utilized to screen a range of formalin-fixed and ethanol-preserved specimens, suspected to be the reservoir for a specific pathogen: for example, a range of 19th and 20th centuries echolocating bat species for white nose syndrome (Campana et al. 2017), screening 209 great ape species for 99 DNA viruses (Hämmerle et al. 2024), screening European Pipistrellus pipistrellus for bandaviruses (Gieraths et al. 2025), and Mycobacterium leprae and M. lepromatosis across armadillo species (Romero-Alvarez et al. 2023). However, to determine the total viral diversity present within a single host (e.g. the “virome”), including divergent and novel viruses, within natural history collection specimens requires unbiased viral discovery methodology, such as metatranscriptomics (Cobbin et al. 2021), an approach we present in our study. This technique can be applied with natural history collections to investigate the diversity and evolution of wildlife viromes in species and across populations, through space and time.

Table 1.

A summary of museums that have been involved in microbiological research using voucher specimens to identify pathogens from various components/tissues, using a range of technologies.

Years	Microbe characterized	Reservoir host (Scientific name, common name)	Tissue	Methodology	Collection	Citation
1994–1996	Sin nombre virus	Peromyscus maniculatus (Deer mice)	Whole blood	Hantavirus- antibody detection	Museum of Southwestern Biology	Mills et al. 1998
17,870 ± 230 years BP.	Mycobacterium tuberculosis	Bison cf. antiquus (Ancient bison)	Bone	DNA extraction and DNA sequencing	N/A	Rothschild et al. 2001
1957–2010	Cichlidogyrus spp.	Simochromis and Pseudosimochromis	Whole specimen mounted on a slide under a cover-slip and fixed using ammonium picrate glycerine	Parasitology (the practice of characterizing parasites)	Royal Museum for Central Africa, Muséum National d’Histoire Naturelle, Iziko South African Museum	Van Steenberge et al. 2015
1880–1971	Pseudogymnoascus destructans	Myotis bechsteinii (Bechstein's bat), Myotis brandtii (Siberian bat),Myotis chinensis (large myotis), Myotis daubentoniid (Daubenton’s bat), Myotis spp., Plecotus austriacus (Grey long-eared bat), Rhinolophus euryale (Mediterranean horseshoe bat), Rhinolophus ferrumequinum (Greater horseshoe bat), Eptesicus serotinus (Serotine bat), Pipistrellus pipistrellus (Common pipistrelle), Pipistrellus kuhlii (Kuhl’s pipistrelle), Miniopterus schreibersii (Common bent-wing bat), Plecotus auritus (Brown long-eared bat)	Dry skin and formalin- and ethanol-preserved wet specimens	Ancient DNA extraction and DNA sequencing from swabs	National Museum of Natural History	Campana et al. 2017
1899–1933	Monkeypox virus	Funisciurus spp. (African rope squirrels)	Skin	DNA extraction	American Museum of Natural History	Tiee et al. 2018
Sixth century BCE	Measles virus	Homo sapiens (Human)	Formalin-fixed lung	Nucleic acid extraction, cDNA synthesis, and DNA sequencing	Berlin Museum of Medical History at the Charité	Düx et al. 2020
1913–1918	Influenza A virus	Homo sapiens (Human)	Formalin-fixed lung	Nucleic acid extraction, cDNA synthesis, and DNA sequencing	Berlin Museum of Medical History at the Charité	Patrono et al. 2022
2011–2013	Lamprey reddening syndrome	Geotria australis (Pouched lamprey)	Several tissue types (skin, muscle, heart, kidney, liver, gut, and visceral fat) in formalin-fixed paraffin-embedded blocks	Total RNA extraction and RNAseq	Aotearoa New Zealand Ministry of Primary Industries Collections	Miller et al. 2025
1974–2017	Mycobacterium leprae and M. lepromatosis	Dasypus novemcinctus (nine-banded armadillo), Tolypeutes matacus (southern three-banded armadillo), Cabassous unicinctus (southern naked-tailed armadillo), Chaetophractus vellerosus (screaming hairy armadillo), Zaedypus pichiy (pichi), Chaetophractus villosus (big hairy armadillo), Cabassous tatouay (greater naked tailed armadillo), Chaetophractus sp. (hairy armadillos), Euphractus sexcinctus (six-banded armadillos), and Priodontes maximus (giant armadillo)	A variety of frozen, ethanol-preserved (95%, 90% and 70%), and DMSO preserved tissue, including muscle, spleen and liver samples.	DNA extraction, qPCR, DNA sequencing	Field Museum, Sam Noble Museum, Lousiana State University Museum of Natural Science, Florida Museum of Natural History, yale Peabody Museum, University of Alaska Museum of the North, Museum of Southwestern Biology, Texas Tech University Museum, Angelo State Natural History Collections, Museum of Vertebrate Zoology Berkeley	Romero-Alvarez et al. 2023
1838–2014	Hepatitis B virus, monkeypox	Pan sp., Gorilla sp., and Pongo sp.	Teeth, phalanges, and soft tissues	DNA extraction, DNA sequencing with viral baits, using aDNA best practices in modern laboratory	Senckenberg Forschungsinstitut, Naturmuseum Frankfurt, Naturhistorisches Museum Vienna, Přírodovědecké muzeum Prague, Senckenberg Naturhistorische Sammlungen Dresden, Museum für Naturkunde Berlin, Museum Koenig Bonn, Haus der Natur Salzburg, Staatliches Museum für Naturkunde Stuttgart	Hämmerle et al. 2024

Here, we applied RNA sequencing methodology to identify viral fragments from a set of formalin-fixed and ethanol-preserved Australian Chiroptera specimens from the Australian National Wildlife Collection (ANWC), including N. geoffroyi, Rhinolophus megaphyllus, Ozimops ridei, Scotorepens greyii, Hipposideros diadema, and Mops jobensis. We have reconstructed fragments of RVA from two specimens, representing divergent strains. These strains are the first RVA detections in Australian bats and include one of the oldest RVA detections worldwide. RVA is a leading cause of acute gastroenteritis, which carries a serious mortality risk in children (>120,000 pediatric deaths per annum), agricultural livestock and wild-living animals (Adams and Kraft 1963; Lanata et al. 2013; Mebus 1976; Tate et al. 2012; Rodger et al. 1975; GBD 2013 Mortality and Causes of Death Collaborators 2015). Our findings demonstrate that formalin-fixed natural history collection specimens can retain both viral and host RNA for over 60 years, contributing both host and virome data to the extended specimen. This highlights the potential of natural history collections to contribute to new avenues of museum-centered research, including historical, long-term viral surveillance data, and expanding our knowledge of viral diversity in wildlife populations. Advances in these fields will have impacts on future work in host–pathogen evolution, biodiversity, and One Health initiatives.

Results

Archival RNA of host and viral origin is retrievable from museum specimens

We successfully extracted RNA from seven vouchered Chiroptera specimens preserved in either formalin or ethanol (Table 2). Despite chemical fixation and specimen ages exceeding 60 years, tissue integrity was sufficient for specimen morphology and dissection (Fig. 1a and b) and to enable RNA recovery (Fig. 1c and d). RNA yields ranged from 2.9 ng/μL to 48.4 ng/μL (Fig. 1c; Table 2) and all samples produced libraries with suitable read counts and lengths for downstream analysis (Fig. 1d; Table S1). Post-trimming quality scores exceeded Q34, and no sequences were flagged as poor quality by FASTQC.

Table 2.

Details of Australian Chiroptera specimens screened for viral RNA.

Registration number	Species name	Location	Collection date	Preservation method	Library preparation methodology	RNA (yield ng/µL)
ANWC M05058	Nyctophilus geoffroyi (lesser long-eared bat)	NT	25-2-1964	Formalin-fixed and kept at room temperature	Total nucleic acid library preparation and targeted enrichment using Twist probes	3.1
ANWC M24834	Rhinolophus megaphyllus (smaller horseshoe bat)	QLD	22-5-1999	Ethanol-preserved and frozen at −20 °C after collection	Illumina TruSeqStranded total RNA prep (RiboZero Plus depletion)	452
ANWC M16448	Ozimops ridei or Mormopterus loriae (eastern free tailed bat)	QLD	23-2-1992	Ethanol-preserved and frozen at −20 °C after collection	Illumina TruSeqStranded total RNA prep (RiboZero Plus depletion)	780
ANWC M16474	Scotorepens greyii (Little broad-nosed bat	QLD	25-2-1992	Formalin-fixed and kept at room temperature	Total nucleic acid library preparation and targeted enrichment using Twist probes	48.4
ANWC M15981	Hipposideros diadema (Diadem leaf-nosed bat)	QLD	7-2-1966	Formalin-fixed and kept at room temperature	Total nucleic acid library preparation and targeted enrichment using Twist probes	4.9
ANWC M02438	Nyctophilus geoffroyi (lesser long eared bat)	NT	26-4-1968	Formalin-fixed and kept at room temperature	Total nucleic acid library preparation and targeted enrichment using Twist probes	16.1
ANWC M19055	Mops jobensis (Northern freetail bat	NT	30-6-1974	Formalin-fixed and kept at room temperature	Total nucleic acid library preparation and targeted enrichment using Twist probes	22.8

We screened seven specimens from the ANWC for viral RNA using liver tissue samples. For each specimen, we report the ANWC specimen registration number, species name (scientific and common), collection location (Australian state), collection date, preservation method, library preparation approach, and total RNA yield (ng/μL) as measured by Agilent 2100 Bioanalyzer. Bolded specimens yielded viral hits discussed in this study. NT, Northern Territory; QLD, Queensland.

Figure 1.

Specimen condition, RNA-seq output, and phylogenetic placement of novel historical strain. a) Anterior view of specimen M24834 (Rhinolophus megaphyllus), an adult male smaller horseshoe bat with a fleshy horseshoe-shaped nose leaf and grey-brown fur. Body mass at collection was 8.8 g. The nose leaf can be observed in close-up (A[i]). b) Anterior view of specimen M05058 (Nyctophilus geoffroyi), an adult female lesser long-eared bat with pale grey fur on the back and white fur on the underbelly. Forearm length at collection was 30.4 mm. The facial features can be observed in close-up (B[i]). c) Total RNA yield (ng/µL) recovered from liver tissue per specimen. d) Total RNA-seq read count recovered per specimen. e) Taxonomic origin of reads inferred using the NCMI nt database visualized for specimen M24834.

Across libraries, a high proportion of reads (mean 87%) were repetitive or consisted of short k-mer dimers (<30 bp) of unknown origin (sequencing duplication, microsatellites, promoter regions, repetitive elements), indicating that only a subset of reads represented identifiable endogenous host or microbial RNA (Figure S1, Table S2). Nevertheless, taxonomic profiling revealed in two specimens, ANWC M24834 and ANWC M16448 that most assignable reads originated from the host (Fig. 1e, Figure S2), with smaller fractions attributed to other eukaryotes, fungi, and viruses (Fig. 1e; Figure S2). In specimen M24834, viral reads were attributed to the families Sedoreoviridae (containing the Rotavirus genus) and Retroviridae. Three specimens, M16474, M19055, and M16648, had the majority of reads assigned to order Mammalia, with a small proportion assigned to Chiroptera, while specimen M02438 and M15981 had almost half of reads unassigned (Figure S2) and M05058 had 17% of reads map to Mammalia, 42% map to viral order Reovirales (containing Sedoreoviridae), and 26% of reads unassigned.

As closely related host species lacked annotated reference genomes, we mapped the Rhinolophus affinis genome (Supplementary Methods, Figure S12, Table S3) and to partial host gene sequences. For specimen M24834 (R. megaphyllus), we mapped reads to the 653 bp cytochrome oxidase subunit 1 (COI) gene fragment from the relative R. philippinensis, and found that 0.12% of reads uniquely mapped (Table S4). For specimen M16448, 0.03% of reads mapped to the 759 bp NADH gene fragment from O. ridei (Table S4). In three specimens (M16474, M19055, and M15981), we could not map any reads to a host gene.

We constructed virus-like contigs (e.g. de novo sequences with significant homology to viral genes) from sequencing libraries from three specimens (M24834, M05058, and M16448). In M24834 and M05058, most viral contigs matched RVA, while additional contigs from M24834 resembled bat gammaherpesvirus and bat fecal-associated retrovirus. In M16448, a subset of contigs showed similarity to R. ferrumequinum gammaretrovirus. From M24834 and M05058, we assembled 16 candidate RVA contigs, ranging from 129 bp to 2,508 bp in length (Table 3). The relative abundance of RVA-like reads in these libraries ranged from 0.00001% to 6.1% (Table S2).

Table 3.

Summary of viral genome segments recovered from two formalin-fixed bat museum specimens.

Strain	Segment and protein	Genotype	%GC	Sequence length (bp)
RVA/Nyctophilus geoffroyi-wt/AUS/M05058/1964/G3PX	Segment 2 (VP2), partial	C2*	39.6%	321
	Segment 4 (VP4), partial	P[3]*	37.2%	234
	Segment 6 (VP6), partial	I3	42.3%	657
	Segment 9 (VP7), partial	G3	37.8%	870#
	Segment 8 (NSP2), partial	n/a	45.0%	129
	Segment 11 (NSP5), partial	H6	39.6%	492
RVA/Rhinolophus megaphyllus-wt/AUS/M24384/1999/G3P[3]	Segment 1 (VP1), partial	R3	33.3%	3,267#
	Segment 2 (VP2), cds	C3	33.6%	2,664
	Segment 3 (VP3), cds	M3	30.9%	2.508
	Segment 4 (VP4), partial	P[3]	37.1%	1,896#
	Segment 5 (NSP1), partial	A9	33.3%	1,395
	Segment 6 (VP6), partial	I3	33.5%	564
	Segment 7 (NSP3), partial	T3	34.1%	954
	Segment 8 (NSP2), partial	N3	33.5%	477
	Segment 9 (VP7), cds	G3	35.2%	981
	Segment 11 (NSP5), partial	H6	36.3%	408

For each specimen, we list the RVA strain name, genomic segment with corresponding protein, assigned genotype, GC content (%GC), and contiguous sequence length (bp). Genotypes are based on established classification criteria for RVA. Complete coding sequences (cds) are indicated where recovered. Sequences with gaps—represented by stretches of ambiguous bases (‘Ns’)—are noted with a superscript hash (#). Genotypes marked with an asterisk (*) could not be confidently assigned. VP, viral protein; NSP, nonstructural protein.

Assembly of historical RVA genomes reveals specimen and method-specific variation

We recovered historical viral RNA from two specimens (formalin-fixed, M05058 and ethanol-preserved, M24834) collected in 1964 and 1999, respectively. Both yielded sequences from RVA, a double-stranded RNA virus of the family Sedoreoviridae, comprising 11 genomic segments that encode structural (VP1–VP6) and nonstructural proteins (NSP1–NSP6) (Estes et al. 2007). We detected RVA using two complementary methods in different specimens; total RNA-seq (M24834) and targeted enrichment (M05058) using the Twist Comprehensive Viral Research Panel, supplemented with custom probes targeting Gammaretrovirus (see Supplementary Methods, Table S6).

From specimen M24834 (R. megaphyllus), we recovered three coding sequences (cds) (VP2, VP3, and VP7) and partial sequences for VP1, VP4, VP6, NSP1, NSP2, NSP3, and NSP5 (Table 3). The VP1 and VP4 segments were assembled from multiple contigs, with ambiguous bases (“Ns”) inserted to span unsupported gaps. In specimen M05058 (N. geoffroyi), we reconstructed partial sequences for VP2, VP4, VP6, NSP2, NSP5, and VP7. All but VP7 were recovered as single contigs while VP7 was assembled from two contigs bridged with 257 ambiguous bases (Table 3).

We note that the RVA sequences obtained from specimen M05058 were generally shorter and more fragmented than those from specimen M24834. This discrepancy may reflect differences in viral load, specimen preservation quality, or biases introduced by the probe set, which was not specifically optimized for Sedoreoviridae (Supplementary Methods). Despite inclusion of retroviral probes in our enrichment panel, we did not detect Retroviridae sequences in any of the five probe-enriched libraries, only a small proportion of Retroviridae reads that did not assemble into large contigs were detected in the total RNA libraries. As with previous studies (Düx et al. 2020; Patrono et al. 2022), the RNA molecules from M05058 and M24834 appeared to be preserved with little damage (i.e. deamination) (Supplementary Results, Figure S3). We note that potentially RNA is impacted differently to other nucleic acids when exposed to formalin. Previous studies have examined DNA damage patterns that are associated with a range of sample storage options, including FFPE and ultralow temperature frozen tissue (Arreaza et al. 2016), formalin-fixed museum specimens (Hykin et al. 2015; Steiert et al. 2023), or DNA degradation resulting from flash freezing tissue samples (Salis et al. 2025), and there will likely be future work that examines RNA in a similar process.

Rotavirus A genotypes reveal reassortment and potential host connections

RVA exhibits high genetic diversity driven by two evolutionary mechanisms: genetic drift and genetic shift. Genetic drift arises through the accumulation of point mutations introduced by the virus's error-prone RNA-dependent RNA polymerase (Ramig 1997), although some mutations are retained through positive selection. In contrast, genetic shift occurs through reassortment of two or more distinct RVA strains during co-infection of a host cell, allowing gene segment exchange; a key mechanism driving viral diversification and cross-species transmission (Matthijnssens et al. 2008a, 2011b; Martella et al. 2010).

To characterize viral diversity, RVA was classified using a dual genotyping system based on the outer capsid proteins VP7 and VP4 genes. These define the G-genotype (VP7 is glycosylated) and P-genotype (VP4 is protease sensitive). A more comprehensive classification known as genotype constellation assigned genotypes to all 11 genome segments using a standard nomenclature: Gx – P[x] – Ix – Rx – Cx – Mx – Ax – Nx – Tx – Ex – Hx, representing VP7 – VP4 – VP6 – VP1 – VP2 – VP3 – NSP1 – NSP2 – NSP3 – NSP4 – NSP5 (Matthijnssens et al. 2011a).

We reconstructed partial or complete genotype constellations for both museum-derived strains. Historical strain RVA/Rhinolophus megaphyllus-wt/AUS/M24384/1999/G3P[3] demonstrated the constellation G3 – P[3] – I3 – R3 – C3 – M3 – A9 – N3 – T3 – Ex – H6 (Matthijnssens et al. 2011b), which we confirmed using the Rotavirus A Genotyping Tool, v1 (rivm.nl/mpf/typingtool/rotavirusa). Four segments (VP4, NSP2, NSP4, and NSP5) fell below the 500 bp threshold for confident classification and thus are considered tentative (Matthijnssens et al. 2011b). This Australian genotype constellation was identical to bat-derived strains from Bulgaria in 2008 (e.g. RVA/Bat-wt/BGR/BB89-15/2008/G3P[3]; RVA/Bat-wt/BGR/BR89-60/2008/G3P[3]).

In contrast, the historical strain RVA/Nyctophilus geoffroyi-wt/AUS/M05058/1964/G3PX exhibited a constellation of: G3 – P[3]* – I3* – Rx – C2* – Mx – Ax – Nx – Tx – Ex – H6, with asterisks indicating genotypes (VP4, VP6, NSP2) inferred from short sequences (Matthijnssens et al. 2011b). As such, the VP4 genotype remains unassigned. Segment 10 (NSP4) was not recovered, and three segments (VP2, VP4, and NSP5) were <500 bp and considered incomplete. This partial genotype constellation resembles early zoonotic strains reported in humans, including RVA/Human/ISR/Ro1845/1985/G3P[3] and RVA/Human/USA/HCR3A/1984/G3P[3], which were originally thought to be of canine or feline origin (Aboudy et al. 1988; Nakagomi et al. 1990; Li et al. 1993; Nakagomi and Nakagomi 2000; Tsugawa and Hoshino 2008).

We visualized recovered segments for both strains relative to a canonical segmented RVA genome (Fig. 2a) and identified their closest relatives using megablast nucleotide identity (Table S7).

Figure 2.

Genome characterization of two novel bat RVA strains, historic timeline of RVA, and phylogenetic placement of strain RVA/Rhinolophus megaphyllus-wt/AUS/M24384/1999/G3P[3] within VP4 phylogeny. a) Partial assemblies of RVA strains recovered from two archived bat specimens: [i] Nyctophilus geoffroyi (M05058, 1964) and [ii] Rhinolophus megaphyllus (M24834, 1999). Segments of the canonical RVA genome are shown as coloured boxes and annotated with segment and protein names. Recovered contigs are shown as red boxes aligned by segment, scaled to approximate length and placement. b) A historical timeline displaying major events in RVA discovery throughout the last 70 years. Left to right: 1964; the collection of specimen M05058, Nyctophilus geoffroyi, assumed to be infected with RVA, 1973; Bishop and colleagues discovered the particles that will be termed “rotavirus” under the electron microscope, seen in epithelial cells obtained from Australian patients, 1976; Rotaviruses were also identified in other mammals, including cows, pigs, mice and horses, 1980s; vaccine development for rotavirus begins, 1999; the collection of specimen M24834, Rhinolophus megaphyllus, assumed to be infected with RVA, 2013; first modern detection of RVA in bats by He and colleagues in China. c) Phylogenetic tree of RVA segment 4/VP4, containing a 450bp section from the partial sequence (three contigs joined by empty strings) from strain RVA/Rhinolophus megaphyllus-wt/AUS/M24384/1999/G3P[3], along with other partial or whole coding sequences of RVA VP4. genotype P[3] (n = 32). The novel historical strain has the respective branch highlighted in orange and an orange star next to the tip name. Bat-associated strains have a bat icon next to their tip name. Strains within the same clade as the novel historical strain have their associated animal host icon next to their tip name. To visualize the grouping of bat strains and the historical strain, the clade has been highlighted by a red box. Bootstrap replicates > 80% is visualized as filled-in circles at nodes. The scale bar represents nucleotide substitutions per site. For clarity, the phylogeny is rooted at the P[2] outgroup.

Twentieth century bat RVA strains point to long-standing zoonotic potential

Rotaviruses infect many species and RVA circulates widely across livestock, companion animals, and wildlife, forming a diverse reservoir with potential for interspecies transmission (Matthijnssens et al. 2008; Ghosh and Kobayashi 2014; Lagan et al. 2023). In humans, RVA lineages are thought to emerge through repeated zoonotic introductions and reassortment events, as evidenced by sporadic spillovers from bovine and porcine hosts (Chandler-Bostock et al. 2014; Desselberger 2014; Phan et al. 2016; Zeller et al. 2016; Sawant et al. 2020).

Bats are frequent hosts of RVA, harboring a wide diversity of genotypes. Some of these genotypes appear bat-specific, while others are shared with alternate mammalian hosts including humans, suggesting a history of cross-species transmission (Esona et al. 2010; He et al. 2013; Asano et al. 2016; Sasaki et al. 2016; Yinda et al. 2016; He et al. 2017; Simsek et al. 2021). Bat-associated RVA strains have been identified around the world in Saudi Arabia, Zambia, Brazil, China, Cameroon, Kenya, Bangladesh, and across Europe in multiple bat families including Pteropodidae, Rhinolophidae, Hipposideridae, Pyllostomidae, Vespertilionidae, Emballonuridae, and Miniopteridae (He et al. 2013; Dacheux et al. 2014; Xia et al. 2014; Asano et al. 2016; Sasaki et al. 2016; Yinda et al. 2016; He et al. 2017; Waruhiu et al. 2017; Mishra et al. 2019; Islam et al. 2020; Simsek et al. 2021).

Like zoonotic influenza viruses or other rotaviruses, bat RVA strains can become “humanized” through reassortment with co-circulating human strains (Matthijnssens et al. 2010; Li et al. 2016). For example, RVA/Human-wt/CHN/WZ101/2013/G3P[3] closely clustered with rodent-associated strains from the same geographic region, providing evidence of zoonotic spillover. The historical bat strains recovered in our study also cluster phylogenetically with animal-derived RVA strains, further supporting their animal origin.

In Australia, a putatively zoonotic strain RVA/Human-wt/AUS/RCH272/2012/G3P[14] was isolated from a 12-year-old child with gastroenteritis in Victoria in 2012. The individual lived in a household with domesticated cats and dogs, in close proximity to a gray-headed flying fox (Pteropus poliocephalus) colony. This strain exhibited a mosaic genome, with segments resembling international bat strains and lineages of bovine, canine, feline, human origins. However, no Australian bat RVA strains were available for comparison at the time (Fig. 2b) (Donato et al. 2014). While the novel strains discussed here do not cluster with RVA/Human-wt/AUS/RCH272/2012/G3P[14], knowledge of the unsampled diversity of RVA strains within Australian bat populations is critical for future genomic case definitions (particularly zoonoses) and public health interventions.

Expanding the known diversity of RVA in Australian bats

Bats are among the most species-rich orders of mammals, with over 1,400 species across 21 families (Burgin et al. 2018). Australian bats span both suborders of Chiroptera: Yangochiroptera (primarily echolocating bats) and Yinpterochiroptera (including the family Pteropodidae, or flying foxes). Bats host a broad array of viruses, many still undescribed (Tan et al. 2021), and several with zoonotic potential, including henipaviruses, lyssaviruses, filoviruses, and coronaviruses. As a result, bats have become a focal group for viral surveillance and virome characterization (Wu et al. 2016; Van Brussel and Holmes 2022  Wallau et al. 2023).

RVA is frequently detected in bat fecal samples, with over 20 bat-associated lineages now described globally (Esona et al. 2010; He et al. 2013; Xia et al. 2014; Asano et al. 2016; Sasaki et al. 2016; Yinda et al. 2016; He et al. 2017; Waruhiu et al. 2017; Mishra et al. 2019; Simsek et al. 2021). The sequences presented here represent the first RVA strains identified and characterized from Australian bats. Although previous studies speculated that Australian flying foxes (Pteropodidae) may act as RVA reservoirs (Donato et al. 2014), our data establish the first confirmed Australian bat-associated RVA strains from echolocating bats. Given the high bat diversity in Australia, it is likely that bat-associated RVA diversity in this region is far greater than currently recognized. These findings establish an initial baseline for RVA diversity in Australian bats and highlight the untapped potential of museum specimens for revealing previously undocumented viral lineages.

Phylogenetic relationships reveal global links and reassortment in historical bat RVA

To contextualize the evolutionary history of the recovered strain RVA/Rhinolophus megaphyllus-wt/AUS/M24384/1999/G3P[3], we conducted maximum likelihood phylogenetic analyzes for each segment using representative RVA sequences (models in Table S7). This strain clustered with zoonosis-associated clades in the VP1, VP6, VP7, NSP1, and NSP2 trees (Fig. 3, Figures S4 and S7–S9), suggesting broad host plasticity and potential for interspecies transmission. In contrast, segments VP4 and NSP3 clustered exclusively with other bat-associated strains (Fig. 2c, Figure S10), indicating more host-restricted evolutionary trajectories for these segments.

Figure 3.

Maximum likelihood phylogeny of VP7 (segment 9) sequences from genotype G3 RVA strains (n = 41). The novel historical strain RVA/Rhinolophus megaphyllus-wt/AUS/M24834/1999/G3/P[3] is highlighted in orange with a star. Tip icons indicate host origin (e.g., bat, human, mouse) and blue boxes denote the clade containing the historical strain. Bootstrap values > 80% are shown as filled circles and the scale bar represents nucleotide substitutions per site. The phylogeny is mid-point rooted using a G14 outgroup.

We identified a genotype constellation shared between the novel strain RVA/AUS/Rhinolophus megaphyllus-wt/M24834/1999/G3/P[3] and a subset of published sequences, which we refer to here as the “Au/zoo” constellation (Table 4). This constellation has been isolated from a broad array of hosts, including bats (Rhinolophus spp.), rodents (e.g. reed voles), horses, and humans, and has been detected multiple times over more than a decade (1999–2015) in China, Bulgaria, Argentina, and now Australia. Its recurrence in multiple hosts and geographic contexts suggests either long-term circulation in reservoir populations or repeated emergence through reassortment.

Table 4.

Genotype constellations of novel and related RVA strains across hosts.

Host

Strain

VP7

VP4

VP6

VP1

VP2

VP3

NSP1

NSP2

NSP3

NSP4

NSP5

Homo sapiens

RVA/Human-tc/USA/Wa/1974/G1P[8]

G1

P[8]

I1

R1

C1

M1

A1

N1

T1

E1

H1

Homo sapiens

RVA/Human-tc/USA/DS-1/1976/G2P[4]

G2

P[4]

I2

R2

C2

M2

A2

N2

T2

E2

H2

Homo sapiens

RVA/Human-wt/USA/3000640724/2015/G8P[14]

G8

P[14]

I2

R3

C2

M2

A3

N2

T6

E2

H3

Homo sapiens

RVA/Human-tc/JPN/K8/1977/G1P[9]

G1

P[9]

I1

R3

C3

M3

A1

N1

T3

E3

H3

Homo sapiens

RVA/Human-wt/AUS/RCH272/2012/G3P[14]

G3

P[14]

I2

R3

C3

M3

A9

N2

T6

E2

H3

Homo sapiens

RVA/Human-tc/KOR/CAU12-1/262/2014/G3P[9]

G3

P[9]

I3

R3

C3

M3

A3

N3

T1

E3

H6

Canis familiarus

RVA/Dog-wt/THA/CU23379/2019/G3P[3]

G3

P[3]

I3

R3

C3

M3

A9

N2

T3

E3

H6

Equus asinus

RVA/Donkey-wt/HB01/China/2021/G3P[12]

G3

P[12]

I2

R3

C3

M3

A9

N2

T3

E2

Homo sapiens

RVA/Human-tc/JPN/AU-1/1982/G3P[9]

G3

P[9]

I3

R3

C3

M3

A3

N3

T3

E3

H3

Homo sapiens

RVA/Human-tc/CHN/L621/2006 G3P[9]

G3

P[9]

I3

R3

C3

M3

A3

N3

T3

E3

H6

Homo sapiens

RVA/Human-wt/THA/CMH-S015-19/2019/G3P[10]

G3

P[10]

I8

R3

C3

M3

A9

N3

T3

E3

H6

Homo sapiens

RVA/Human-wt/THA/MS2015-1-0001/XXXX/G3P[10]

G3

P[10]

I8

R3

C3

M3

A9

N3

T3

E3

H6

Aselliscus stoliczkanus

RVA/Bat-tc/CHN/MYAS33/2013/G3P[10]

G3

P[10]

I8

R3

C3

M3

A9

N3

T3

E3

H6

Homo sapiens

RVA/Human-wt/THA/B4684/2018/G3P[10]

G3

P[10]

I3

R3

C3

M3

A9

N3

T3

E3

H6

Homo sapiens

RVA/Human-wt/THA/B5662/2018G3P103]

G3

P[10]

I3

R3

C3

M3

A9

N3

T3

E3

H6

Procyon lotor

RVA/Raccoon/USA/D2213150/2022/G3P[3]

G3

P[3]

I3

R3

C3

M3

A3

N3

T3

E3

H6

Au/zoo

Homo sapiens

RVA/Human-wt/CHN/WZ101/2013/G3P[3]

G3

P[3]

I3

R3

C3

M3

A9

N3

T3

E3

H6

Rhinolophus megaphyllus

RVA/Rhinolophus megaphyllus-wt/AUS/M24834/1999/G3P[3]

G3

P[3]

I3

R3

C3

M3

A9

N3

T3

H6

Rhinolophus blasii

RVA/Bat-wt/BGR/BB89-15/2008/G3P[3]

G3

P[3]

I3

R3

C3

M3

A9

N3

T3

E3

H6

Rhinolophus euryale

RVA/Bat-wt/BGR/BR89-60/2008/G3P[3]

G3

P[3]

I3

R3

C3

M3

A9

N3

T3

E3

H6

Hipposideros pomona

RVA/Bat-wt/CHN/LZHP2/2015/G3P[3]

G3

P[3]

I3

R3

C3

M3

A9

N3

T3

E3

H6

Alexandromys fortis

RVA/Mouse-wt/CHN/LQ6/2013/G3P[3]

G3

P[3]

I3

R3

C3

M3

A9

N3

T3

E3

H6

Equus ferus caballus

RVA/Horse-wt/ARG/E3198/2008G3P[3]

G3

P[3]

I3

R3

C3

M3

A9

N3

T3

E3

H6

Homo sapiens

RVA/Human-wt/CHN/M2-102/2014/G3P[3]

G3

P[3]

I3

R3

C3

M3

A9

N3

T3

E3

H6

Taphozous melanopogon

RVA/Bat-wt/CHN/BSTM70/XXXX/G3P[3]

G3

P[3]

I8

R3

C3

M3

A29

N3

T3

E3

H6

Rhinolophus hipposideros

RVA/Bat-wt/CHN/MSLH14/2011/G3P[3]

G3

P[3]

I8

R3

C3

M3

A9

N3

T3

E3

H6

Scotophilus kuhlii

RVA/Bat-wt/CHN/YSSK5/2015/G3P[3]

G3

P[3]

I8

R20

C2

M1

A9

N3

T3

E3

H6

Homo sapiens

RVA/Human-wt/RUS/Leningrad-568/1992/G3P[3]

G3

P[3]

I2

R2

C3

M3

A9

N2

E3

H6

Macaca mulatta (lab adapted)

RVA/Simian-tc/USA/RRV/1975/G3P[3]

G3

P[3]

I2

R2

C3

M3

A9

N2

T3

E3

H6

Nyctophilus geoffroyi

RVA/Nyctophilus geoffroyi-wt/AUS/M05058/1964/G3PX

G3*

P[3]*

I3*

C2*

H6

Rhinolophus simulator

RVA/Bat-wt/ZMB/LUS12-14/2012/G3P[3]

G3

P[3]

I3

R2

C2

M3

A9

N2

T3

E2

H3

Macaca nemestrina

RVA/Rhesus-tc/USA/TUCH/2002/G3P[24]

G3

P[24]

I9

R3

C3

M3

A9

N1

T3

E3

H6

We summarized the genotype constellations for RVA/Nyctophilus geoffroyi-wt/AUS/M05058/1964/G3PX, RVA/Rhinolophus megaphyllus-wt/AUS/M24384/1999/G3P[3] and other well-known RVA strains, and strains that share genotypes. For each strain, we indicate the host species (scientific name), strain designation, and assigned genotypes for all 11 RVA genome segments (VP7, VP4, VP6, VP1, VP2, VP3, NSP1, NSP2, NSP3, NSP4, and NSP5). Strains included in the “Au/zoo” genotype constellation are grouped within the gray-shaded box. Genotypes marked with an asterisk (*) are uncertain due to contig length or sequence identity thresholds.

Phylogenetic comparisons showed that the strain RVA/AUS/Rhinolophus megaphyllus-wt/M24834/1999/G3/P[3] clustered with two human strains, including RVA/Human-wt/CHN/WZ101/2013/G3P[3], a suspected zoonotic strain, and three related rodent strains from the same region (RVA/Rat-wt/CHN/LQ285/2013/G3P[3], RVA/Mouse-wt/CHN/LQ6/2013/G3P[3], and RVA/Mouse-wt/CHN/LQ321/2013/G3P[3]) (Li et al. 2016). This pattern was consistent across multiple gene trees, including VP7 (Fig. 3), VP6 (Figure S8), NSP1 (Figure S7), and NSP2 (Figure S9). A second human strain, RVA/Human-wt/CHN/M2-102/2014/G3P[3], suspected to have arisen from wildlife reassortment (Dong et al. 2016), also clustered with the historical strain in the VP4 (Fig. 2c), VP1 (Figure S4), NSP2, and NSP3 phylogenies.

Two previously reported bat strains RVA/Bat-wt/BGR/BB89-15/2008/G3P[3] and RVA/Bat-wt/BGR/BR89-60/2008/G3P[3], isolated from Bulgarian Rhinolophus spp. (Simsek et al. 2021), also clustered with the Australian strain in VP6 and VP7 phylogenies, strengthening evidence for a long-standing, geographically widespread lineage within Rhinolophus-associated RVA. An equine strain with evidence of multiple interspecies transmission and/or reassortment events, RVA/Horse-wt/ARG/E3198/G3P[3], also clustered with the Australian strain in the VP6 and VP3 phylogenies (Miño et al. 2013).

Due to short length (<300 bp) and high divergence of most segments (<80% nucleotide identity), we excluded strain RVA/Nyctophilus geoffroyi-wt/M05058/1964/G3PX from phylogenetic analysis except for VP2 and NSP5 (Matthijnssens et al. 2011a). Further investigation is required to fully characterize this genotype constellation.

Historical bat rotaviruses reveal deep evolutionary connections and surveillance gaps

We recovered two distinct partial historical RVA genomes from fragmented RNA in formalin-fixed museum specimens of Chiroptera. To our knowledge, these are the first viral genomes successfully reconstructed from such specimens and the first documented RVA strain from an Australian bat. Remarkably, one historical strain (RVA/Nyctophilus geoffroyi-wt/AUS/M05058/1964/G3PX) predates the discovery of human rotavirus in 1973 (Fig. 2b), placing it among the oldest RVA strains characterized to date. Given the absence of prior Australian bat RVA sequences, these genomes may represent endemic strains or ancestral forms of contemporary RVA strains. Both scenarios underscore a significant gap in our understanding of global RVA evolution and geographic spread.

We provide genomic coverage, read abundances (Fig. 2a, Table S2), genotypes (Table 4), and phylogenetic relationships (Figs. 3 and 2c, Figures S4–S11) for both strains. Viral recovery was not limited by total RNA yield (2.9 ng/μL for M24834 and 3.1 ng/μL for M05058), implicating viral load and specimen preservation quality as more influential factors. The near-complete recovery from M24384 was especially striking, given the specimen's age and preservation in formalin (Fig. 1a). Notably, we recovered substantial host RNA from libraries prepared without enrichment (M24834 and M16448; Fig. 1, Figure S2), consistent with expectations that probe-based enrichment introduces biases against endogenous transcripts. This underscores the value of total RNA sequencing for virus and host RNA recovery, particularly for studying reservoirs of zoonotic viruses (Lee et al. 2015). Furthermore, in the strain associated with specimen M05058, RVA/Nyctophilus geoffroyi-wt/AUS/M05058/1964/G3PX, viral recovery was likely limited by the use of probes predominantly designed to target Gammaretrovirus. Despite recovering only five segments from specimen M05058, some below confidence thresholds, these partial and cds segments extend the known temporal and taxonomic breadth of bat-associated RVA, representing the earliest molecular evidence of RVA in the genus Nyctophilus, which remains underrepresented in global surveillance.

The similarity of strain RVA/Rhinolophus megaphyllus-wt/M24384/1999/G3P[3] to 7 other strains across 10 segments, forming the distinct “Au/zoo” constellation (Table 4), demonstrating that this genotype constellation likely appeared multiple times, across more than one geographical location. The Au/zoo constellation included multiple bat strains and likely represented a broader bat-associated lineage (Li et al. 2016; He et al. 2017; Simsek et al. 2021). The presence of Au/zoo in Australian R. megaphyllus in the 1960s supports a long-standing Australian bat-associated strain. Moreover, the Au/zoo constellation included human, rodent, and equine strains, highlighting its ecological plasticity. Our findings reinforce earlier observations (Simsek et al. 2021) that bat-associated RVA strains frequently share genotypes with urban wildlife and human lineages, generating strains with zoonotic potential. Given the high frequency of reassortment along with broad host range of RVA strains, such interspecies spillover events are likely to recur (Matthijnssens et al. 2008b).

Understanding the zoonotic potential of rotaviruses of animal origin is critical for surveillance efforts. In Australia, RVA surveillance has primarily focused on livestock as reservoirs (e.g. dairy and beef herds, and piggeries), where zoonotic transmission to humans, particularly children, has been documented (Swiatek et al. 2010; Cowley et al. 2013; Genz et al. 2023). However, the role of bats in this transmission network remains poorly understood. Our data confirms that RVA circulates in Australian bats across the Northern Territory and Queensland, supporting calls for a One Health approach to RVA monitoring that integrates wildlife into public health frameworks.

Globally, bat RVA strains display high genetic diversity, with some strains exhibiting host specificity (e.g. within Pteropodidae) (Simsek et al. 2021). Continued monitoring in Australian bats, including targeted sampling of genera Rhinolophus, Nyctophilus, and family Pteropodidae, is needed to map this diversity. Given the established links between ecological stressors (e.g. food shortages) and viral shedding in flying foxes (Becker et al. 2023; Eby et al. 2023), future RVA surveillance should incorporate environmental variables to improve zoonotic risk modeling.

Spirit vaults as viral archives: unlocking historical viromes from natural history collections

Our recovery of viral and host RNA from historical formalin-fixed and ethanol-preserved Chiroptera specimens demonstrates the ability of global vouchered wildlife specimens for studying historical viral diversity, and the cumulative benefit for specimens under the extended specimen concept, along with advancing the growing field of museum metagenomics (Speer et al. 2022). This study marks the first characterization of RVA in Australian bat populations and promotes the further study of zoonotic RVA in Chiroptera in Australia under a One Health approach. The simultaneous recovery of host and microbial RNA demonstrates the viability of formalin-fixed collections for broad molecular analysis, including eukaryotic, prokaryotic, and viral targets.

Crucially, historical viral RNA has value beyond pathogen detection. It enables viral molecular clock calibration and refinement of evolutionary rate estimates (Worobey et al. 2008; Duggan et al. 2016; Duchêne et al. 2020; Düx et al. 2020; Patrono et al. 2022), reveals insights into historical viral taxonomy (Duggan et al. 2020), and has the potential to improve our understanding of viral evolution. Importantly, our sampling approach was blind to symptomatic disease status, as specimens were not selected based on signs of illness, nor did our species selection target members of Chiroptera with known bat-borne viral associations (Wallau et al. 2023). This approach enabled unbiased recovery of viruses in both symptomatic and asymptomatic hosts. Moreover, formalin fixation inactivates pathogens (Chinabut et al. 2023; Seeburg et al. 2023), thereby eliminating biosafety risks and facilitating higher-throughput study of high-risk pathogens (e.g. avian influenza), without the need for specialized containment facilities.

By characterizing the partial genomes of two historical RVA strains, we demonstrated how biorepositories can resolve phylogenetic gaps, clarify interspecies transmission dynamics, and apply retrospective surveillance, allowing researchers to map virus diversity across time and space. This capacity is particularly valuable for contextualizing modern outbreaks and complements real-time modern surveillance in wildlife populations.

Natural history collections are now positioned to fill critical data gaps for viral phylogenetics, host-virus coevolution, and zoonotic viral prevalence in wildlife populations. By leveraging formalin-fixed and ethanol-preserved wildlife specimens, we can reconstruct the evolutionary dynamics of viruses that shaped, and continue to shape, wildlife and human health. In addition to pandemic preparedness, historical viromes may inform conservation policy, endangered species recovery (where disease plays a role in species decline), landscape-scale microbe mapping, and zoonotic spillover risk models (Thompson Cody et al. 2021).

This work builds on a growing body of evidence that natural history collections and collections research has the ability to grow and expand with molecular technology to advance our knowledge in taxonomy, biodiversity, and evolution of species, including microorganisms (Hykin et al. 2015; Harmon et al. 2019; Card et al. 2021; Raxworthy and Smith 2021; Davis and Knapp 2025). Novel scientific contributions of biorepositories, such as understanding the microbial diversity enabled by the preservation of RNA molecules within formalin-fixed and ethanol-preserved specimens, are essential for ensuring their continued power to deliver impact for present and future generations of scientists.

Methods

Specimen selection

We selected seven specimens from the family Chiroptera, collected between 1964 and 1999, from both the frozen tissue (ethanol-preserved) and whole specimen formalin-fixed collections at the ANWC (Table 2). Specimens ANWC M24834 (R. megaphyllus) and ANWC M16448 (O. ridei), both collected in the 1990s, were preserved in formalin following dissection and ethanol storage of a subset of internal organs (including liver) at −80 °C. The remaining five specimens, ANWC M05058 (N. geoffroyi), ANWC M15981 (S. greyii), ANWC M16474 (H. diadema), ANWC M02438 (N. geoffroyi), and ANWC M19055 (M. jobensis), collected between 1964 and 1992, were fixed whole in formalin (historical records are lacking, but likely 10% unbuffered formalin was used), and subsequently preserved in 70% ethanol.

RNA extraction

We extracted RNA from approximately 5 mm² of liver tissue per specimen. We subsampled tissue from frozen ethanol-preserved samples, and dissected tissue from formalin-fixed whole specimens. We homogenized tissue in 200 µL DNA/RNA Shield (Zymo) using sterilized silicon carbide beads (1 mm, filled to halfway in an Eppendorf tube) and physical disruption on ice using the Omni Bead Ruptor “skin” protocol (Speed = 7 m/s, 30 s active, 20 s dwell). We added 400 µg Proteinase K and incubated lysates at 55 °C for 4 h. To pellet the beads and debris, we centrifuged the lysates gently, then processed the supernatant with the Zymo RNA Miniprep kit (Cat# R1057), including DNAse treatment, per manufacturer's instructions. We assessed RNA fragment length, integrity, and concentration using an Agilent 2100 Bioanalyzer with the Eukaryote Total RNA Pico kit, per manufacturer's instructions.

Library preparation and sequencing

We used two different methods of library preparation for the two preservation methods. For ethanol-preserved tissues (ANWC M24834 and M16448), we prepared libraries using the Illumina Stranded Total RNA with Ribo-Zero kit. For the formalin-fixed tissues (ANWC M05058, M16474, M16474, M02438, and M19055), we prepared total nucleic acid libraries with target enrichment using Twist Bioscience custom RNA probes (details in Supplementary Methods, Table S6). The targeted enrichment libraries were sequenced on a Novaseq X plus 10b lane 300 cycle (150 PE). The Illumina TruSeq Stranded total RNA libraries were sequenced on a NovaSeq S4-300 cycle (150 PE). We sequenced all seven libraries at the Australian Genome Research Facility.

Sequence analysis

The scripts used for analyzing the RNAseq data are available online (https://github.com/ashporter-44/Viral_discovery_pipeline_museum).

Quality control

We trimmed adaptor sequences and low-quality bases using Trimmomatic v0.39 (Bolger et al. 2014) with default settings. We assessed read quality pre- and post-trimming with FastQC v.0.12.1 (Andrews 2010).

Identifying viral fragments within libraries

We assembled contiguous sequences de novo using both Megahit v1.2.9 (97) and Trinity v2.15.1 (Grabherr et al. 2011; Li et al. 2015) with default settings. We similarity-searched the contigs against the NCMI nr database using an unbiased DIAMOND v2.0.15 blastx search via the blast+package v 2.14.0 (Buchfink et al. 2021). We considered contigs with viral hits (length >250 bp, percent identity >50%, query cover >50%, e-value < 1 × 10⁻⁵) as viral-like sequences. We mapped raw reads to candidate contigs using the kalign function of the ngskit4b suite version 200218 (https://github.com/kit4b) with options -c25 -l25 -d100-U4 to measure relative abundance, and to fill in any gaps. We calculated mapped read abundance and depth as the total number of reads mapping to the contig divided by the number of total trimmed reads.

Viral genome annotation

We annotated RVA contigs using BLAST nucleotide and protein searches. We assigned genotypes to the viral genome segments (n = 16) based on nucleotide identity thresholds described in (Matthijnssens et al. 2011b) and manually annotated open reading frames in Geneious v2024.

Phylogenetic relationships

We aligned nucleotide sequences (open reading frame regions) for 13 rotavirus segments using MAFFT v7.526 with the L-INS-I algorithm (Katoh et al. 2002; Katoh and Standley 2013). We inferred maximum likelihood phylogenetic trees from nucleotide alignments, using model testing with 1,000 non-parametric bootstrap replicates in IQtree2 v2.2.0.5 (Minh et al. 2020).

Host genome analysis

For specimen ANWC M24834, we mapped reads to the closely related R. affinus reference genome (CM093080.1) (Zhao et al. 2024) using STAR v2.7.4a (Dobin et al. 2013) and visualized the results in Geneious v2024. For other specimens lacking a closely related host genome, we used conserved host genes to estimate host read content (Tables S3 and S4). We calculated host read abundance as the proportion of reads mapped to host chromosomes/genes relative to the total trimmed read count. We also used CCmetagen v1.4.0 (Clausen et al. 2018; Marcelino et al. 2020) with the NCBI nt database to estimate overall host RNA representation.

Author contributions

Project design: C.H., M.A. Funding: C.H., M.A., I.S., E.H. Ethics: M.A., C.H. Collection of specimen tissue: M.A., C.H., T.H., C.W., I.S. Experimental work: RNA extractions, Twist panel design: M.A., Method design: M.A., E.H., Preservation media survey: E.H. Data analysis: Bioinformatics, Quality control, Genome assembly, Genome annotation, Phylogenetic trees: A.P. Writing: Original outline; A.P., Figures; A.P., Editing: A.P., E.H., C.H., C.D., T.H., C.W., I.S., M.A.

Supplementary material

Supplementary material is available at Molecular Biology and Evolution online.

Funding

We acknowledge that this work was funded by Commonweath Scientific and Industrial Research Organisation Environomics Future Science Platform (Grant number R-10011 to CH), Commonweath Scientific and Industrial Research Organisation Heath and Biosecurity Acorn Grant (Grant number R-16935 to MA), Commonweath Scientific and Industrial Research Organisation Early Career Researcher Fellow Program (Grant number R-20592 to CH and AP), Commonweath Scientific and Industrial Research Organisation National Research Collections Australia Strategic Operating (Grant number R-90579 to CH), and an Australian Department of Agriculture Fisheries and Forestry Bright Ideas Grant (Grant number R-19122 to MA).

Data Availability

The viral sequences for all segments discussed in this manuscript are available at NCBI under Genbank accession numbers PX868887–PX868902. The raw sequence data for RNA sequencing of all specimens has been uploaded to the Sequence Read Archive under BioProject ID PRJNA1445360.