Investigating high pathogenicity avian influenza virus incursions to remote islands: Detection of H5N1 on Gough Island in the South Atlantic Ocean

24 Understanding the mechanisms underlying the emergence and spread of high pathogenicity avian 25 influenza virus (HPAIV) is critical for tracking its global dissemination, particularly via migratory 26 seabirds, given their role in transmission over long distances. Scavenging seabirds, such as skuas, may 27 act as both reservoirs and vectors, and have been linked to multiple outbreaks since 2021. Here, we 28 report the detection of HPAIV clade 2.3.4.4b H5N1 in three Tristan skua (Stercorarius antarcticus 29 hamiltoni) carcasses on Gough Island in the central South Atlantic Ocean. To investigate potential 30 incursion routes, we combined genomic analyses with year-round tracking data from global location 31 sensors. Although migratory movement patterns suggested southern Africa as the most obvious 32 pathway, the strain detected on Gough Island was more closely related to that found in South Georgia, 33 suggesting infection may have occurred during the pre-laying exodus when skuas disperse into frontal 34 waters south of the island. No further cases have been confirmed for Gough, but further systematic 35 monitoring is needed to understand the dynamics of virus infection. The detection of HPAIV H5N1 in 36 skuas on Gough Island highlights the importance of continued vigilance, coordinated surveillance, 37 and proactive biosecurity across the South Atlantic and Southern Ocean, alongside efforts to reduce 38 other pressures on globally important seabird populations to help strengthen their resilience. 39 40

(5-8) High pathogenicity avian influenza, Gough Island, Brown Skua, migration, 41 transmission, emerging 42

43 The ongoing expansion of high pathogenicity avian influenza virus (HPAIV) poses a major threat to 44 wildlife, livestock, and, through its zoonotic potential, human health. Since first emerging in poultry 45 in China in 1996, the A/goose/Guangdong/1/96 (GsGd)-lineage of HPAIV H5Nx has evolved to 46 spread efficiently among a broad range of bird species globally, causing mass mortality events in 47 unprecedented numbers [1-6]. Following its dissemination across Europe by migratory species, H5N1 48 .CC-BY-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 11, 2025. ; https://doi.org/10.1101/2025.09.06.674618doi: bioRxiv preprint reached North America via the trans-Atlantic flyway and, by October 2022, had expanding rapidly 49 throughout South America, with devastating impacts on both seabird and pinniped populations [7,8]. 50 In the austral spring of 2023, HPAIV reached the sub-Antarctic islands and Antarctica, first detected 51 in brown skuas (Stercorarius antarcticus) on Bird Island, South Georgia [9]. Soon after, cases were 52 reported in southern fulmars (Fulmarus glacialoides) and black-browed albatrosses (Thalassarche 53 melanophris) in the Falkland Islands [9], and in multiple species across the Antarctic Peninsula [10]. 54 Genetic analyses indicate that the detections in these regions were linked to strains originating in 55 South America, likely introduced via long-distance movements of scavenging seabirds such as skuas, 56 gulls, and giant petrels [9, 11]. In the following breeding season 2023/2024, the virus had spread 57 further east across the sub-Antarctic and into the southern Indian Ocean, causing significant die-offs 58 in wandering albatross (Diomedea exulans) chicks on Marion Island [12], before it was reported for 59 the Crozet and Kerguelen Archipelagos, where mass mortality events predominantly affected southern 60 elephant seals (Mirounga leonina) [11]. Genetic evidence suggests that the Crozet and Kerguelen 61 outbreaks incurred from independent introductions linked to South Georgia rather than from the 62 nearer southern African coast [11]. 63 64 Gough Island in the central South Atlantic is situated approximately 3,600 km northeast of South 65 Georgia and 2,450 km northwest of Marion Island and considered one of the most important seabird 66 breeding sites globally [13]. Despite its small size (~65 km²), the island supports an estimated eight 67 million birds of at least 24 species, many of them endemic or near-endemic to the island or island 68 group of Tristan da Cunha, and several of global conservation concern, including the Critically 69 Endangered Tristan albatross (Diomedea dabbenena), MacGillivray’s prion (Pachyptila 70 macgillivrayi), and Gough finch (Rowettia goughensis) [13-16]. Consequently, the potential for 71 arrival of HPAIV into this sensitive environment had remained a significant conservation concern. A 72 risk assessment in 2022 considered the likelihood of the emergence of HPAIV on Gough Island as 73 ‘low’ [17]. However, the spread of the HPAIV epizootic into the southern hemisphere and its 74 continued expansion across the sub-Antarctic region immediately increased the risk of incursion and, 75 given the island’s position along major migratory flyways and marine corridors, linking the Atlantic 76 .CC-BY-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 11, 2025. ; https://doi.org/10.1101/2025.09.06.674618doi: bioRxiv preprint and Southern Oceans [18], means Gough was vulnerable to viral introduction from multiple directions 77 [18]. 78 79 Evidence of antibodies against AIVs was first detected in 2009, when a serosurvey on Gough Island 80 revealed exposure in brown skuas (S. a. hamiltoni; hereafter referred to as Tristan skua) and northern 81 rockhopper penguins (Eudyptes moseleyi), demonstrating that the virus had previously reached the 82 island and circulated among these species [19, 20]. Given their scavenging behaviour, skuas represent 83 both key reservoirs and vectors for disease transmission, making the study of their movements critical 84 for identifying introduction pathways [20, 21]. 85 86 Here, we describe the detection of HPAIV clade 2.3.4.4b H5N1 in three Tristan skuas from Gough 87 Island from September 2024 and investigate potential incursion routes by using a combination of 88 genetic analyses of the virus and year-round tracking data from global location sensors. 89 90

& Methods 91 Study site 92 Gough Island (40°19′12″S 09°56′24″W) is part of the British Overseas Territory of St Helena, 93 Ascension, and Tristan da Cunha in the central South Atlantic Ocean (Fig.1). Gough Island lies 94 approximately 380 km south-southeast from the Tristan da Cunha archipelago, which consists of three 95 main islands: Tristan da Cunha (96 km²), Inaccessible (14 km²), and Nightingale (4 km²), along with 96 its satellite islets Middle (or Alex) and Stoltenhoff (both ~0.1 km²), all within 40 km of each other 97 (Fig. 1). Gough is uninhabited except for a small year-round team of 7-9 people operating the research 98 station. The station is re-supplied once a year in September/October by the South African research 99 vessel SA Agulhas II. A field team of 2-3 biologists is on the island year-round as part of the island’s 100 long-term term monitoring and science programme, trained to provide surveillance for unusual avian 101 mortality events. 102 103 104 .CC-BY-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 11, 2025. ; https://doi.org/10.1101/2025.09.06.674618doi: bioRxiv preprint Tracking data 105 Migrate Technology Intigeo‑C330 geolocation tags (17 x 19 x 8 mm, 3.3 g) were deployed on 10 106 incubating adult Tristan skuas near the Gough Island research station in October-November 2017. The 107 tags were attached to standard stainless steel SAFRING rings. Eight tags were retrieved during the 108 following breeding season; the two other devices were not recovered. 109 110 Light-level data were processed in R version 4.4.1 using the SGAT package (v0.1.3) [22], 111 implementing a Bayesian state-space model to estimate probable locations based on the timing of 112 twilight transitions, while accounting for uncertainties in light detection and movement behaviour. 113 Raw light data were pre-processed using the BAStag package (v0.1.3) [23] to identify twilight events 114 (sunrise and sunset) based on light intensity thresholds. These twilight times were then manually 115 reviewed and edited to correct for anomalies caused by shading, cloud cover, or behavioural factors 116 e.g., incubation. The resultant data were used to infer two positions per day, corresponding to daily 117 twilight transitions. 118 119 To quantify space use, we calculated kernel utilisation distributions (KUDs) using the adehabitatHR 120 package (v0.4.22) [24]. SGAT-derived location estimates were projected onto an equal-area coordinate 121

system to allow for unbiased spatial analysis. For each bird, 50% and 95% KUDs were 122 calculated to represent core use areas and home ranges, respectively, across the pre-laying, breeding, 123 and non-breeding periods. Location data during incubation in 2017 and 2018 were pooled for KUD 124 analysis. To evaluate population-level space use, individual KUDs were averaged to assess spatial 125 habitat use. Migration pathways between breeding and non-breeding areas were excluded from the 126 KUD analysis and instead visualised using raw location data. For each individual, the onset of north-127 east migration away from the breeding colony and that of south-west return migration, was defined as 128 the point at which the individual's distance from the colony increased or decreased continuously, 129 without subsequent decreases/increases. The end point of each migration phase was defined as the 130 point at which the distance from the colony stabilised, indicating arrival at a new residency area. All 131 spatial operations were conducted using the sp, sf, and raster packages in R. 132 .CC-BY-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 11, 2025. ; https://doi.org/10.1101/2025.09.06.674618doi: bioRxiv preprint Sample collection 133 Swab samples from the brain, trachea, and cloaca of dead birds were collected, frozen at −20°C 134 within one hour of collection, and shipped to the World Organisation for Animal Health/Food and 135 Agriculture Organisation (WOAH/FAO) International Reference Laboratory for Avian Influenza at 136 the Animal and Plant Health Agency (APHA) in Weybridge, UK, for diagnostic evaluation. 137 Surveillance for suspicious wildlife mortality continued with special attention given to any spatial or 138 temporal clustering of cases or unusual behaviours in wildlife. 139 140 Molecular methods 141 Total nucleic acid was extracted from all swab samples as described previously [25] and viral RNA 142 was screened using four real-time reverse transcription polymerase chain reaction (rRT-PCR) assays 143 including: i) a Matrix (M)-gene assay for generic influenza A virus detection [26], ii) a HPAIV H5 144 clade 2.3.4.4b specific assay for HA subtype and pathotyping [25], iii) an N1-subtype specific rRT-145 PCR to confirm the neuraminidase type [27], and iv) an avian paramyxovirus type-1 (APMV-1) large 146 polymerase (L)-gene assay [28]. All rRT-PCRs were undertaken on the AriaMx qPCR System 147 (Agilent, United Kingdom). Material from positive brain tissue swabs was used for virus isolation in 148 10-day-old specific pathogen-free (SPF) embryonated fowls’ eggs according to internationally 149 recognised methods (Codes and Manuals - WOAH - World Organisation for Animal Health). 150 151 Viral Sequencing 152 Whole-genome sequencing (WGS) was undertaken on brain tissue swabs. Extracted vRNA was 153 converted to double-stranded cDNA and amplified using a one-step RT-PCR using SuperScript III 154 One-Step RT-PCR kit (Thermo Fisher Scientific) as previously described [29]. PCR products were 155 purified with Agencourt AMPure XP beads (Beckman Coultrer) prior to sequencing library 156 preparation using the Native Barcoding Kit (Oxford Nanopore Technologies) and sequenced using a 157 GridION Mk1 (Oxford Nanopore Technologies) according to manufacturer’s instructions. Assembly 158 of the influenza A viral genomes was performed using a custom in-house pipeline as previously 159 described [9]. 160 .CC-BY-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 11, 2025. ; https://doi.org/10.1101/2025.09.06.674618doi: bioRxiv preprint Phylogenetic Analysis 161 To identify the genetic relatedness of Gough Island viruses to other reported strains, H5N1 HPAIV 162 clade 2.3.4.4b sequences from Europe, Africa, Antarctica and South America available in the EpiFlu 163 database between 1st January 2020 and 28th July 2025 were collated. To remove over-represented 164 groups, the sequences were subset to cover 0.5% sequence divergence using PARNAS [30]. The 165 remaining dataset was separated by segment and aligned using Mafft v7.525 [31] and manually 166 trimmed to the open reading frame using Aliview v.2021 [32]. The trimmed alignments were used to 167 infer maximum-likelihood (ML) phylogenetic trees using IQ-Tree v2.4.0 [33] with model finder, 1000 168 ultrafast bootstraps and SH-like approximate likelihood ratio test. Clade classification of Gough 169 Island sequences was confirmed with GenoFLU-multi https://github.com/moncla-lab/GenoFLU-170 multi. 171 172 To produce a time scale phylogeny, all H5 HA sequences from South America and Antarctica were 173 used alongside a subset of North American sequences to improve temporal signal. The dataset was 174 aligned, trimmed, and an ML phylogenetic tree was inferred using the same approaches as described 175 above. The HA ML phylogeny was then used to infer a time scaled phylogeny using TreeTime v0.11.4 176 [34]. Discrete trait analysis of transitions between locations was carried out using the mugration 177 inference model in TreeTime with default settings. 178 179

180 The first suspected HPAIV-related death of an adult Tristan skua was observed on Gough Island on 12 181 September 2024 at a club of 50-200 non-breeding skuas at the helipad near the research station. The 182 following day, another skua at the same location exhibited clinical signs consistent with HPAIV , 183 including drooping head and wings, lethargy, and an inability to walk or fly, and was found dead later 184 the same day [35]. Two additional carcasses were discovered on 15 September. Swab samples were 185 collected from three of the four carcasses on 20 September 2025 and the carcasses were buried 186 afterwards. The fourth carcass had disappeared by the time of sampling and was presumed to have 187 been scavenged by other skuas. 188 .CC-BY-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 11, 2025. ; https://doi.org/10.1101/2025.09.06.674618doi: bioRxiv preprint Diagnostic evaluation of samples 189 All three birds sampled on Gough Island tested positive for HPAIV H5N1 in oropharyngeal and brain 190 swabs but tested negative in cloacal swabs (Table 1). Brain samples indicated a high viral load with 191 HPH5 rRT-PCR values ranging from 19.5 - 23.8 Cts. Virus isolation was attempted on two of the 192 three brain samples and both yielded hemagglutinating virus. 193 Whole genome sequences (WGS) obtained from two samples confirmed the viruses belonged to HA 194 clade 2.3.4.4b, genotype B3.2, as previously detected in the Antarctic region (Fig. 2a) [9]. The two 195 WGS shared >99% identity across all genes. Phylogenetic analysis shows that these viruses are part of 196 monophyletic clade within the Antarctic Peninsula and sub-Antarctic region, including ancestral 197 viruses from South Georgia and their descendants in Crozet and Kerguelen (Figure 2). This suggests 198 an expansion of the introduction described by Banyard et al. [9], rather than a new incursion from 199 Africa. Analysis of the asymmetric transition rate matrix reveals notable patterns of viral movement 200 among key sub-Antarctic and South Atlantic islands (Figure 2b). South Georgia appears to be a 201 significant ancestral source, exhibiting outward transitions to Crozet, Kerguelen, Gough Island and 202 the Falkland Islands. Gough Island demonstrates moderate connectivity across the region, particularly 203 with Crozet and Kerguelen. However, transition rates from Gough Island to South Georgia and back 204 suggest less frequent bidirectional exchange. Overall, the data supports a model in which South 205 Georgia serves as a primary source, seeding transmission across the sub-Antarctic and South Atlantic 206 islands (Figure 2b). 207 208 Ringing and tracking data 209 Tracking durations ranged from 343 to 353 days (mean ± SD: 345 ± 3.3 days), yielding 687 ± 7.8 210 locations per individual (Fig. 3 and Fig. 4). 211 212 Following the breeding season, individuals departed the foraging grounds around Gough Island 213 between 12 January and 12 February 2018, en route to non-breeding foraging grounds off the coasts 214 of South Africa and Namibia. Arrival at these non-breeding areas occurred between 20 January and 25 215 February 2018, with the north-eastward migration taking 11.9 ± 4.9 days (range 4–17 days). 216 .CC-BY-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 11, 2025. ; https://doi.org/10.1101/2025.09.06.674618doi: bioRxiv preprint Individuals remained within the non-breeding wintering grounds for 141 ± 19.5 days (range 109–165 217 days). Departures from these areas occurred between 4 June and 15 July 2018, with arrival back at the 218 pre-laying foraging area near Gough Island occurring between 13 June and 1 August 2018. The south-219 westward return migration took 13.8 ± 6.1 days (range 7–23 days). 220 Two of the three dead skuas sampled for HPAIV had been ringed on Gough Island in 2016 and 2017. 221 The individual ringed in 2017 (SAFRING 888666) was a known breeder on the island and had been 222 equipped with a GLS device (BH142) as part of the tracking study (Table 2, Figure 4d). In contrast, 223 the skua ringed in 2016 (SAFRING 888054) was caught as a non-breeder at the skua club near the 224 helipad. 225

226 The arrival of HPAIV H5N1 on Gough Island represents one of the most geographically isolated 227 detections to date, highlighting the interconnectivity of pelagic seabird populations and their role in 228 assisting global transmission in the current HPAIV H5N1 epizootic into some of the world’s most 229 remote (and sensitive) ecosystems. Confirmation of HPAIV in Tristan Skuas on Gough prompted 230 suspicions that individuals had become infected by viruses originating from Africa due to the species’ 231 extensive connectivity with southern African coastal habitats outside their breeding season [36]. The 232 GLS data analysed in this study corroborated this migratory pattern, demonstrating high temporal and 233 spatial consistency in the species’ use of their wintering foraging grounds and suggesting that this 234 behaviour is likely representative of the wider Gough Island skua population. 235 236 Multiple outbreaks of clade 2.3.4.4b H5 HPAIV among seabird and Cape fur seal (Arctocephalus 237 pusillus) colonies in southern Africa [37-40] could have provided opportunities for transmission to 238 scavenging species such as skua. However, a phylogenetic analysis of global 2.3.4.4b H5 HA 239 sequences places the Gough Island viruses within a cluster of South American strains, distinct to the 240 ones reported for Africa. Further analysis identified the Gough strains as belonging to genotype B3.2, 241 previously detected in the Antarctic region [9]. While terrestrial aggregations such as breeding 242 colonies represent clear and obvious hotspots for virus transmission, at-sea connectivity may play an 243 .CC-BY-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 11, 2025. ; https://doi.org/10.1101/2025.09.06.674618doi: bioRxiv preprint equally critical role with overlapping foraging grounds and shared marine habitats increasing the 244 potential for interspecies contact, facilitating viral spread between geographically distant populations. 245 The Benguela upwelling region, stretching from Cape Agulhas in South Africa north to southwest 246 Angola, is one of the most productive marine ecosystems globally [41] and plays a vital role as a 247 migratory corridor and seasonal feeding area for numerous migratory species, breeding in the 248 Antarctic or sub-Antarctic (e.g. White-capped/Shy albatross (T. cauta) [36]; Indian yellow-nosed 249 albatrosses (T. carteri) and northern giant petrels (Macronectes halli) [42]; black-browed albatrosses 250 [43]. The frequency and likelihood of interspecies interactions at shared foraging grounds, whether 251 through direct or indirect contact, make these zones high-risk areas for the persistence and 252 transmission of infectious diseases. However, any spatial pattern needs to be considered in a temporal 253 context i.e., the timing of seasonal/migratory movements within and across species and how it 254 coincides with outbreaks. Based on tracking records for 2018, individuals completed their south-255 westward migration in one to three weeks, arriving near Gough Island between 13 June and 1 August. 256 While the incubation period for HPAIV can vary considerably depending on species susceptibility and 257 viral genetic characteristics, it is unlikely symptoms would only present 2–4 months after infection 258 (assuming broadly similar migratory timing in 2024) and thus improbable that the skuas picked up the 259 virus during their winter migration. 260 261 It is therefore more plausible that the virus reached Gough via broader Southern or Atlantic Ocean 262 flyways, potentially during the species’ pre-laying exodus period in August and September, when 263 Tristan skuas disperse into the frontal region south of the island. These nutrient-rich boundary waters 264 between the sub-tropical and Antarctic polar fronts are known to support large congregations of 265 pelagic seabirds and serve as foraging grounds for marine mammals [18, 44, 45]. Seabird species 266 breeding on Gough Island use both the Southern and Atlantic Ocean Flyways [45-47] which overlap 267 considerably between 30-60°S, linking Antarctic seabird populations to those on islands in the South 268 Atlantic [18]. For example, populations of southern elephant seals, brown skua, and black-browed 269 albatross on South Georgia have all been found to utilise waters near Gough Island [43, 48-50]. 270 271 .CC-BY-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 11, 2025. ; https://doi.org/10.1101/2025.09.06.674618doi: bioRxiv preprint Ultimately the timing and location of transmission of HPAIV to the Tristan skuas in this study remains 272 unknown along with the uncertainty about whether in fact skuas introduced the virus to the island or 273 other avian species or seals, vagrants or breeding on Gough (45, 51-53]. However, scavenging species 274 are more likely than other seabird species to interact with carcasses, spend extended periods of time 275 sitting on the water, and come into close contact with other foraging individuals [35, 49, 54]. Such 276 behaviours increase the opportunities for viral transmission under conditions where environmental 277 dilution of virions would otherwise hinder infection, supporting the role of skuas and giant petrels as 278 the most likely vectors of AIV across the Southern Ocean [20]. Despite these unknowns, the detection 279 of HPAIV H5N1 originating from the Antarctic peninsula illustrates a novel pathway by which 280 influenza viruses from the Americas could also be introduced back into wild bird populations in Afro-281 Eurasia. Therefore, understanding the spatial and temporal dynamics of at-sea dispersal and 282 interactions, especially in areas of high biodiversity and migratory fly- and swim-ways, is essential for 283 anticipating and mitigating future outbreaks. 284 285 Initial concerns that HPAIV detection could constitute the onset of a mass outbreak on Gough Island, 286 have not materialized; no further symptomatic birds or mortalities related to HPAIV have been 287 confirmed (aside from the suspicious death of a breeding female Tristan albatross in April 2025). 288 Although mass mortality has occurred in other skua populations as a result of HPAIV , notably in the 289 UK where great skua (Stercorarius skua) declined by 73% [4, 35], the apparent failure of the virus to 290 cause a large-scale outbreak on Gough may be due to several factors. Cloacal swabs taken during the 291 initial investigation returned negative results for AIV , suggesting minimal environmental shedding. 292 Additionally, while one carcass was presumed scavenged, the remaining three were buried following 293 sampling, preventing further scavenging. Furthermore, while Gough hosts large populations of 294 seabird species, many of these species nest at low densities (e.g., Tristan albatross) or in burrows, such 295 as petrels and prions [13], reducing opportunities for direct contact and environmental transmission. A 296 serological study in 2009 demonstrated some seropositivity against influenza in northern rockhopper 297 penguins and brown skua on Gough [19, 20], however, how far pre-existing immunity against 298 influenzas could have influenced transmission dynamics in this outbreak remains unknown. 299 .CC-BY-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 11, 2025. ; https://doi.org/10.1101/2025.09.06.674618doi: bioRxiv preprint While there have been no further mortalities attributed to HPAIV on Gough Island or in the 300 neighbouring Tristan da Cunha Archipelago, the impact of HPAIV in the region could be substantially 301 underestimated. Large parts of these remote islands are inaccessible, limiting timely monitoring to 302 detect carcasses or individuals exhibiting clinical signs. While there has been consistent presence of a 303 field team of biologists on Gough year-round since 2008, there is less systematic surveillance at the 304 Tristan da Cunha Archipelago. In February 2024, the island’s fishery observer reported a dead skua 305 floating at sea off Nightingale Island, but no further unusual mortality of skuas (or other seabirds) was 306 observed on Nightingale and Inaccessible islands when visited in March and September 2024, 307 respectively (PGR pers. obs.). Therefore, the absence of confirmed cases should not be taken as 308 evidence of the absence of the disease. The foraging behaviour of Tristan skuas during the breeding 309 season could increase the risk for both intra- and interspecific transmission among susceptible seabird 310 taxa, contributing to the spread of the virus, locally and regionally. Other scavenging species such as 311 southern (M. giganteus, resident) and northern giant petrels (vagrants) are also frequent visitors to the 312 Tristan da Cunha Archipelago, where they depredate northern rockhopper penguins and generally 313 scavenge on carcasses [55]. If HPAIV were to reach the Tristan archipelago, it would heighten the 314 threats faced by already vulnerable populations endemic to the islands. 315 316 Only few mitigating activities are available to prevent the movements of migratory birds, especially 317 over large distances, and as such, avoidance of incursions of diseases like that described here is 318 seemingly not possible. Understanding the spatiotemporal dynamics of seabird movements -in 319 particular scavenging seabirds- disease ecology, and host-pathogen interactions is essential for 320 mitigating the impact of HPAIV and other emerging disease threats, particularly in high-risk and 321 ecologically sensitive areas [56]. This emphasises the need for continued vigilance, proactive 322 biosecurity measures, as well as an integrated research framework linking surveillance efforts across 323 the South Atlantic and Southern Oceans to design and implement adaptive disease management 324 strategies. Finally, it is a stark reminder of the vulnerability of remote ecosystems to infectious 325 diseases and for the need to alleviate already persisting pressures on globally important seabird 326 .CC-BY-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 11, 2025. ; https://doi.org/10.1101/2025.09.06.674618doi: bioRxiv preprint populations including habitat degradation and disturbance, invasive species, and climate-driven 327 stressors to strengthen their resilience. 328 329

330 All samples for HPAI testing were collected with the permission of the Tristan da Cunha government. 331 We are extremely grateful to Dr Laura Roberts, Western Cape Government and Dr Gretna DeWit, 332 Directorate of Animal Health, South Africa for their invaluable help issuing the South African import 333 permit. A special thank you to Ashley Bennison and Jennifer Forster Davidson, British Antarctic 334 Survey, for their generous and insightful advice and guidance on the Gough HPAIV response plan. We 335 would also like to thank the crews of the SA Agulhas II for their kind support and collaboration to 336 receive the samples onboard the vessel. The engagement, shipment to the UK, testing and generation 337 of the viral sequences was funded by the Department for Environment, Food and Rural Affairs (Defra, 338 UK) and the Devolved Administrations of Scotland and Wales, through the following programmes: 339 SV3006, SV3032 and SE2227. This work was also supported by the Biotechnology and Biological 340 Sciences Research Council (BBSRC) and Department for Environment, Food and Rural Affairs 341 (Defra, UK) research initiative ‘FluTrailMap’ [grant number BB/Y007271/1]. Funded by the 342 European Union (EU) under grant agreement (101084171) - (Kappa-Flu). Views and opinions 343 expressed are however those of the author(s) only and do not necessarily reflect those of the EU or 344 REA. Neither the EU nor the granting authority can be held responsible for them. The Science Animal 345 Ethics Committee (SFAEC) of the University of Cape Town approved the research protocols for the 346 deployments of GLS devices on seabirds including skuas (UCT SFAEC 2014\V10\PR), with 347 permission from the Tristan government. 348 Declaration of interest statement 349 The authors declare that they do not have any competing interests to disclose.350 .CC-BY-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 11, 2025. ; https://doi.org/10.1101/2025.09.06.674618doi: bioRxiv preprint

351 [1] Caliendo V , Lewis NS, Pohlmann A, et al. Transatlantic spread of highly pathogenic avian 352 influenza H5N1 by wild birds from Europe to North America in 2021. Sci Rep. 2022;12(1):11729. 353 doi:10.1038/s41598-022-13447-z 354 [2] Falchieri M, Reid SM, Ross CS et al. Shift in HPAI infection dynamics causes significant losses in 355 seabird populations across Great Britain. Veterinary Record. 2022;191(7):294-6. 356 [3] Lane JV , Jeglinski JW, Avery‐Gomm S, et al. High pathogenicity avian influenza (H5N1) in 357 Northern Gannets (Morus bassanus): Global spread, clinical signs and demographic consequences. 358 Ibis. 2024;166(2):633-50. doi:10.1111/ibi.13275 359 [4] Tremlett CJ, Cleasby IR, Bolton et al. Declines in UK breeding populations of seabird species of 360 conservation concern following the outbreak of high pathogenicity avian influenza (HPAI) in 2021–361 2022. Bird Study. 2024;71(4):293-310. doi:10.1080/00063657.2024.2438641 362 [5] Kuiken T, Vanstreels RE, Banyard A, et al. Emergence, spread, and impact of high‐pathogenicity 363 avian influenza H5 in wild birds and mammals of South America and Antarctica. Conserv Biol. 364 2025;e70052. doi:10.1111/cobi.70052 365 [6] Pardo-Roa C, Nelson MI, Ariyama N, et al. Cross-species and mammal-to-mammal transmission 366 of clade 2.3. 4.4 b highly pathogenic avian influenza A/H5N1 with PB2 adaptations. Nat Commun. 367 2025;16(1):2232. doi:10.1038/s41467-025-57338-z 368 [7] Leguia M, Garcia-Glaessner A, Muñoz-Saavedra B, et al. Highly pathogenic avian influenza A 369 (H5N1) in marine mammals and seabirds in Peru. Nat Commun. 2023;14(1):5489. 370 doi:10.1038/s41467-023-41182-0 371 [8] Gamarra-Toledo V , Plaza PI, Angulo F, et al. Highly pathogenic avian influenza (HPAI) strongly 372 impacts wild birds in Peru. Biol Conserv. 2023;286:110272. doi:10.1016/j.biocon.2023.110272 373 .CC-BY-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 11, 2025. ; https://doi.org/10.1101/2025.09.06.674618doi: bioRxiv preprint [9] Banyard AC, Bennison A, Byrne AM, et al. Detection and spread of high pathogenicity avian 374 influenza virus H5N1 in the Antarctic Region. Nat Commun. 2024;15(1):7433. doi:10.1038/s41467-375 024-51490-8 376 [10] León F, Ulloa-Contreras C, Pizarro EJ, et al. Skuas mortalities linked to positives HPAIV A/H5 377 beyond Polar Antarctic Circle. bioRxiv. 2025;1. doi:10.1101/2025.03.02.640960 378 [11] Clessin A, Briand FX, Tornos J, et al. Mass mortality events in the sub-Antarctic Indian Ocean 379 caused by long-distance circumpolar spread of highly pathogenic avian influenza H5N1 clade 2.3. 4.4 380 b. 2025. BioRxiv;2025-02. doi:10.1101/2025.02.25.640068 381 [12] Ryan PG, Gill R, Roberts L, Stephen V . Avian flu reaches Marion Island. African Birdlife. 382 2025;13(5):14-15. 383 [13] Caravaggi A, Cuthbert RJ, Ryan PG, et al. The impacts of introduced House Mice on the 384 breeding success of nesting seabirds on Gough Island. Ibis. 2019;161(3):648-61. 385 doi:10.1111/ibi.12664 386 [14] Oppel S, Clark BL, Risi MM, et al. Cryptic population decrease due to invasive species predation 387 in a long‐lived seabird supports need for eradication. J Appl Ecol. 2022;59(8):2059-70. 388 doi:10.1111/1365-2664.14218 389 [15] Jones CW, Risi MM, Osborne AM, et al. Mouse eradication is required to prevent local 390 extinction of an endangered seabird on an oceanic island. Anim Conserv. 2021;24(4):637-45. 391 doi:10.1111/acv.12670 392 [16] BirdLife International. Rowettia goughensis. The IUCN Red List of Threatened Species 2024: 393 e.T22723149A252044473. 2024 [cited on 07 August 2025]. doi:10.2305/IUCN.UK.2024-394 2.RLTS.T22723149A252044473.en 395 .CC-BY-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 11, 2025. ; https://doi.org/10.1101/2025.09.06.674618doi: bioRxiv preprint [17] Dewar ML, Vanstreels RET, Boulinier T, et al. Biological Risk Assessment of Highly Pathogenic 396 Avian Influenza in the Southern Ocean. Scientific Committee on Antarctic Research. Cambridge UK. 397 SCAR Antarctic Wildlife Health Network. 2023. 398 [18] Morten JM, Carneiro AP, Beal M, et al. Global Marine Flyways Identified for Long‐Distance 399 Migrating Seabirds From Tracking Data. Glob Ecol Biogeogr. 2025;34(2):e70004. doi: 400 10.1111/geb.70004 401 [19] Abad, F.X., N.Busquets, A.Sanchez, et al. Serological and virological surveys of the influenza A 402 viruses in Antarctic and sub-Antarctic penguins. Antarct. Sci. 2013;25:339-344. 403 doi:10.1017/S0954102012001228 404 [20] Gittins O, Grau-Roma L, Valle R, et al. Serological and molecular surveys of influenza A viruses 405 in Antarctic and sub-Antarctic wild birds. Antarct Sci. 2020;32(1):15-20. 406 doi:10.1017/s0954102019000464 407 [21] Gamble A, Bazire R, Delord K, et al. Predator and scavenger movements among and within 408 endangered seabird colonies: Opportunities for pathogen spread. J Appl Ecol. 2020;57(2):367-78. 409 doi:10.1111/1365-2664.13531 410 [22] Sumner MD, Wotherspoon SJ, Hindell MA. Bayesian estimation of animal movement from 411 archival and satellite tags. PLoS One. 2009.13;4(10):e7324. doi:10.1371/journal.pone.0007324 412 [23] Colchero F, Jones OR, Rebke M. BaSTA: an R package for Bayesian estimation of age‐specific 413 survival from incomplete mark–recapture/recovery data with covariates. Methods Ecol Evol. 414 2012;3(3):466-470. doi: 10.1111/j.2041-210x.2012.00186.x 415 [24] Calenge C. Home range estimation in R: the adehabitatHR package. Office national de la classe 416 et de la faune sauvage: Saint Benoist, Auffargis, France. 2011. 417 .CC-BY-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 11, 2025. ; https://doi.org/10.1101/2025.09.06.674618doi: bioRxiv preprint [25] James J, Seekings AH, Skinner P, et al. Rapid and sensitive detection of high pathogenicity 418 Eurasian clade 2.3. 4.4 b avian influenza viruses in wild birds and poultry. J Virol Methods. 419 2022;301:114454. doi: 10.1016/j.jviromet.2022.114454 420 [26] Nagy A, Černíková L, Kunteová K, et al. A universal RT-qPCR assay for “One Health” detection 421 of influenza A viruses. PloS one. 2021;16(1):e0244669. doi:10.1371/journal.pone.0244669 422 [27] Payungporn S, Chutinimitkul S, Chaisingh A, et al. Single step multiplex real-time RT-PCR for 423 H5N1 influenza A virus detection. J Virol Methods. 2006;131(2):143-7. 424 doi:10.1016/j.jviromet.2005.08.004 425 [28] Sutton DA, Allen DP, Fuller CM, et al. Development of an avian avulavirus 1 (AAvV-1) L-gene 426 real-time RT-PCR assay using minor groove binding probes for application as a routine diagnostic 427 tool. J Virol Methods. 2019;265:9-14. doi:10.1016/j.jviromet.2018.12.001 428 [29] Slomka MJ, Reid SM, Byrne AM, et al. Efficient and informative laboratory testing for rapid 429 confirmation of H5N1 (Clade 2.3. 4.4) high-pathogenicity avian influenza outbreaks in the United 430 Kingdom. Viruses. 2023;15(6):1344. doi:10.3390/v15061344 431 [30] Markin A, Wagle S, Grover S, Vet al. PARNAS: objectively selecting the most representative 432 taxa on a phylogeny. Syst Biol. 2023;72(5):1052-63. doi:10.1093/sysbio/syad028 433 [31] Katoh K, Standley DM. MAFFT multiple sequence alignment software version 7: improvements 434 in performance and usability. Mol Biol Evol. 2013;30(4): 772-780. doi:10.1093/molbev/mst010 435 [32] Larsson A. AliView: a fast and lightweight alignment viewer and editor for large 436 datasets. Bioinformatics 30.22. 2014:3276-3278. doi:10.1093/bioinformatics/btu531 437 [33] Minh BQ, Schmidt HA, Chernomor O, et al. IQ-TREE 2: new models and efficient methods for 438 phylogenetic inference in the genomic era. Mol Biol Evol 2020;37(5):1530-1534. 439 doi:10.1093/molbev/msaa015 440 .CC-BY-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 11, 2025. ; https://doi.org/10.1101/2025.09.06.674618doi: bioRxiv preprint [34] Sagulenko P, Puller V , Neher RA. TreeTime: Maximum-likelihood phylodynamic analysis. Virus 441 Evol. 2018;4(1):vex042. doi:10.1093/ve/vex042 442 [35] Banyard AC, Lean FZ, Robinson C, et al. Detection of highly pathogenic avian influenza virus 443 H5N1 clade 2.3. 4.4 b in great skuas: a species of conservation concern in Great Britain. Viruses. 444 2022;14(2):212. doi:10.3390/v14020212 445 [36] Carneiro APB, Dias MP, Clark BL, et al. The BirdLife Seabird Tracking Database: 20 years of 446 collaboration for marine conservation. Biol Conserv. 2024:299;110813. 447 doi:10.1016/j.biocon.2024.110813 448 [37] Abolnik C, Phiri T, Peyrot B, et al. The molecular epidemiology of clade 2.3.4.4B H5N1 high 449 pathogenicity avian influenza in Southern Africa, 2021–2022. Viruses. 2023;15(6):1383. 450 doi:10.3390/v15061383 451 [38] Abolnik C, Roberts LC, Strydom C, et al. Outbreaks of H5N1 High Pathogenicity Avian 452 Influenza in South Africa in 2023 Were Caused by Two Distinct Sub-Genotypes of Clade 2.3. 4.4 b 453 Viruses. Viruses. 2024;16(6):896. doi:10.3390/v16060896 454 [39] Roberts LC, Abolnik C, Waller LJ, et al. Descriptive epidemiology of and response to the high 455 pathogenicity avian influenza (H5N8) epidemic in South African coastal seabirds, 2018. Transbound 456 Emerg Dis. 2023;2023(1):2708458. doi:10.1155/2023/2708458 457 [40] World Organisation for Animal Health (WOAH), 2025. WAHIS Event 5034 Influenza A viruses 458 of high pathogenicity (Inf. with) (non-poultry including wild birds) (2017-). 459 https://wahis.woah.org/#/in-event/5034/dashboard 460 [41] Hutchings L, Van der Lingen CD, Shannon LJ, et al. The Benguela Current: An ecosystem of 461 four components. Prog in Oceanogr. 2009;83(1-4):15-32. doi:10.1016/j.pocean.2009.07.046 462 .CC-BY-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 11, 2025. ; https://doi.org/10.1101/2025.09.06.674618doi: bioRxiv preprint [42] Reisinger RR, Raymond B, Hindell MA, et al. Habitat modelling of tracking data from multiple 463 marine predators identifies important areas in the Southern Indian Ocean. Divers Distrib. 464 2018;24(4):535-50. doi:10.1111/ddi.12702 465 [43] Phillips RA, Silk JR, Croxall JP, et al. Summer distribution and migration of nonbreeding 466 albatrosses: individual consistencies and implications for conservation. Ecology. 2005;86(9):2386-96. 467 doi:10.1890/04-1885 468 [44] Bost CA, Cotté C, Bailleul F, et al. The importance of oceanographic fronts to marine birds and 469 mammals of the southern oceans. J Mar Syst. 2009;78(3):363-376. doi:10.1016/j.jmarsys.2008.11.022 470 [45] Dias MP, Oppel S, Bond AL, et al. Using globally threatened pelagic birds to identify priority 471 sites for marine conservation in the South Atlantic Ocean. Biol Conserv. 2017;211:76-84. 472 doi:10.1016/j.biocon.2017.05.009 473 [46] Schoombie S, Dilley BJ, Davies D, et al. The foraging range of Great Shearwaters (Ardenna 474 gravis) breeding on Gough Island. Polar Biol. 2018;41(12):2451-8. doi:10.1007/s00300-018-2381-7 475 [47] Green CP, Green DB, Ratcliffe N, et al. Potential for redistribution of post‐moult habitat for 476 Eudyptes penguins in the Southern Ocean under future climate conditions. Glob Chang Biol. 477 2023;29(3):648-67. doi:10.1111/gcb.16500 478 [48] Reisinger RR, Bester MN. Long distance breeding dispersal of a southern elephant seal. Polar 479 Biol. 2010;33(9):1289-91. doi:10.1007/s00300-010-0830-z 480 [49] Carneiro AP, Manica A, Clay TA, et al. Consistency in migration strategies and habitat 481 preferences of brown skuas over two winters, a decade apart. Mar Ecol Prog Ser. 2016;553:267-81. 482 doi:10.3354/meps11781 483 .CC-BY-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 11, 2025. ; https://doi.org/10.1101/2025.09.06.674618doi: bioRxiv preprint [50] Hindell MA, McMahon CR, Bester MN, et al. Circumpolar habitat use in the southern elephant 484 seal: implications for foraging success and population trajectories. Ecosphere. 2016;7(5):e01213. 485 doi:10.1002/ecs2.1213 486 [51] Bester MN, Wege M, De Bryn, et al. Ranging and fiving behaviour of Subantarctic fur seals from 487 the Tristan da Cunha Islands, South Atlantic. Tristan da Cunha Islands Seal Research Report (2009 – 488 2019). 2019 489 [52] Ryan PG, Dilley BJ, Risi MM, et al. Three new seabird species recorded at Tristan da Cunha 490 archipelago. Seabird. 2019;32:122-5. doi:10.61350/sbj.32.122 491 [53] Bester MN, Somers MJ. Monitoring of vagrant seals on mid-oceanic islands of the South Atlantic 492 Ocean. Polar Biology. 2025;48(3):79. doi:10.1007/s00300-025-03397-3 493 [54] Sinclair JC. Subantarctic Skua Catharacta antarctica predation techniques on land and at 494 sea. Mar Ornithol. 1980;8:3-6. doi:10.5038/2074-1235.8.1.27 495 [55] Ryan PG, Sommer E, Breytenbach E. Giant petrels Macronectes hunting northern rockhopper 496 penguins Eudyptes moseleyi at sea. Ardea. 2008;96(1):129-34. doi:10.5253/078.096.0116 497 [56] Sacristán C, Ewbank AC, Ibáñez Porras P, et al. Novel epidemiologic features of high 498 pathogenicity avian influenza virus A H5N1 2.3. 3.4 b panzootic: a review. Transbound Emerg Dis. 499 2024;2024(1):5322378. doi:10.1155/2024/5322378 500 501 502 .CC-BY-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 11, 2025. ; https://doi.org/10.1101/2025.09.06.674618doi: bioRxiv preprint Tables 503 Table 1: RT-PCR results, virus isolation and viral whole genome sequencing on samples collected from three Tristan skuas (Stercorarius antarcticus 504 hamiltoni) on Gough Island on the 20th September 2024. 505 Bird number Sample Type (Swab) M Gene RT-PCR H5HP RT-PCR N1 RT-PCR Interpretation Strain GISAID Accession number Isolate 1 Tracheal Positive Positive Positive H5N1 HPAIV A/Brown_Skua/Gough_Island/047354/2024_|H5N1|_2024- 09-20 EPI_ISL_201 51596 Yes Cloacal Negative Negative Negative Brain Positive Positive Positive 2 Tracheal Positive Positive Positive H5N1 HPAIV A/Brown_Skua/Gough_Island/047355/2024_|H5N1|_2024- 09-20 EPI_ISL_201 51597 NA Cloacal Negative Negative Negative Brain Positive Positive Positive 3 Tracheal Positive Positive Positive H5N1 HPAIV A/Brown_Skua/Gough_Island/ 050656/2024_|H5N1|_2024-09-20_EP1 NA Yes Cloacal Negative Negative Negative Brain Positive Positive Positive 506 .CC-BY-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 11, 2025. ; https://doi.org/10.1101/2025.09.06.674618doi: bioRxiv preprint Figures 507 508 509 510 511 512 .CC-BY-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 11, 2025. ; https://doi.org/10.1101/2025.09.06.674618doi: bioRxiv preprint 513 514 515 516 517 518 519 .CC-BY-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 11, 2025. ; https://doi.org/10.1101/2025.09.06.674618doi: bioRxiv preprint Figure Captions 520 Figure 1. a) Map indicating the location for Gough Island in the context of locations with ongoing 521 HPAIV H5N1 outbreaks reported to WAHIS (red dots), b) Gough Island with the research station 522 located in the southeast of the island and the island helipad’s (yellow diamond), c) view of research 523 station and helipad, d) Tristan skua (Stercorarius antarcticus hamiltoni). 524 525 Figure 2. a) Maximum-likelihood tree of HA-gene including the two Gough Island sequences 526 identified from this study and 944 HA-gene sequences from Antarctica, South and North America 527 previously identified as 2.3.4.4b. Sequences from South America coloured according to country, all 528 sequences from North America in black, b) a source-sink heat map of transmission rate estimates 529 between geographic regions. 530 531 Figure 3. Time series showing the distance from the colony of GLS-tracked Tristan skuas (8) during 532 the 2017/18 and 2018/19 breeding seasons, across key stages of the breeding cycle. 533 534 Figure 4. Kernel utilisation distributions (KUDs) representing home-range (95% UD) and core (50% 535 UD) foraging areas of all tracked individuals during the a) pre-laying exodus, b) breeding, and c) non-536 breeding periods, overlaid on major frontal features from the 2017/18 and 2018/19 breeding seasons. 537 Dashed lines indicate individual migration routes from the Gough Island region to non-breeding 538 foraging grounds on the South African and Namibian continental shelves, and their return paths. The 539 solid line represents the movement of one individual (GLS tag: BH142) that tested positive for 540 HPAIV in 2024. 541 542 .CC-BY-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 11, 2025. ; https://doi.org/10.1101/2025.09.06.674618doi: bioRxiv preprint