Viral emergence in marine mammals in the North Pacific may be linked to Arctic sea ice reduction

Climate change-driven alterations in Arctic environments can influence habitat availability, species distributions and interactions, and the breeding, foraging, and health of marine mammals. Phocine distemper virus (PDV), which has caused extensive mortality in Atlantic seals, was confirmed in sea otters in the North Pacific Ocean in 2004, raising the question of whether reductions in sea ice could increase contact between Arctic and sub-Arctic marine mammals and lead to viral transmission across the Arctic Ocean. Using data on PDV exposure and infection and animal movement in sympatric seal, sea lion, and sea otter species sampled in the North Pacific Ocean from 2001–2016, we investigated the timing of PDV introduction, risk factors associated with PDV emergence, and patterns of transmission following introduction. We identified widespread exposure to and infection with PDV across the North Pacific Ocean beginning in 2003 with a second peak of PDV exposure and infection in 2009; viral transmission across sympatric marine mammal species; and association of PDV exposure and infection with reductions in Arctic sea ice extent. Peaks of PDV exposure and infection following 2003 may reflect additional viral introductions among the diverse marine mammals in the North Pacific Ocean linked to change in Arctic sea ice extent.

responsible for extensive mortality in European harbour seals (Phoca vitulina vitulina) in the North Atlantic, was identified in northern sea otters (Enhydra lutris kenyoni) in Alaska 8 . This finding raised the possibility that increased contact between Arctic and sub-Arctic marine mammals could result from climate change-associated reductions in Arctic sea ice extent which could alter animal movement allowing for disease transmission across the Arctic Ocean.
Phocine distemper virus was recognized as an important pathogen of phocid seals in 1988 when an outbreak caused mass mortality among European harbour seals in the North Atlantic Ocean. An early hypothesis suggested that harp seals (Pagophilus groenlandicus) from the Arctic may have been the source and reservoir of infection for other seal species in the Atlantic, and sympatric grey seals (Halichoerus grypus) may provide the link for viral transmission from Arctic species to harbour seals 10,11 . A second PDV epidemic occurred among European harbour seals from May to November in 2002, leading researchers to question whether the virus had persisted in the region in marine or terrestrial hosts, or if it was reintroduced 12 . Differences between the 1988 and 2002 PDV isolates identified through phylogenetic analysis supported introduction of a new virus. However, a PDV isolate from a harbour seal infected during a 2006 outbreak on the US Atlantic coast showed more similarity (99.3% across the genome, Genbank accession numbers NC_028249, KY229928) to the 1988 isolate, suggesting the possibility of multiple viral lineages circulating in Arctic and Atlantic seal species 13 . While Atlantic harbour seals were highly susceptible to these three strains of PDV and suffered large population losses associated with infection 14 , sympatric species, including grey seals and Arctic species such as harp and hooded (Cystophora cristata) seals, appear to have varying susceptibility. Although sporadic deaths do occur in these species, they are not on the order of magnitude seen with harbour seals (as reviewed in 9 ).
In contrast to the North Atlantic, outbreaks of PDV have not been observed in the North Pacific Ocean. Although a small number of seropositive Kuril harbour seals (Phoca vitulina stejnegeri), Steller sea lions (Eumetopias jubatus), and spotted seals (Phoca largha) were reported in 1994-1999 near Japan 15 , PDV was not molecularly confirmed in these animals and seropositive results may reflect exposure to the closely related canine distemper virus (CDV). Ringed seals (Phoca hispida) that were seropositive for PDV were detected along Canada's Northwest Territories near northern Alaska in 1993 and 1994 10 , but serologic surveys of Pacific harbour seals (Phoca vitulina richardsii), Steller sea lions, and northern sea otters off Alaska prior to 2000 showed little evidence of exposure to distemper viruses, and PDV had not been identified as a cause of illness or death [16][17][18] . PDV was not confirmed in the North Pacific Ocean until virus was detected in northern sea otters sampled in 2004, which raised a number of questions regarding the timeline of introduction into the North Pacific, how the virus reached sea otters, and what role the virus may play in North Pacific marine mammal illness and mortality.
Nomadic Arctic seals with circumpolar distributions (e.g. ringed and bearded, Erignathus barbatus, seals) and geographic ranges that intersect with those of harp seals, may be carriers of PDV to the North Pacific. Overlapping distributions of marine mammal species may then allow for transmission of the virus to other ice seal species (spotted seals and ribbon seals, Histriophoca fasciata) and sub-Arctic species such as Steller sea lions, northern fur seals (Callorhinus ursinus), and northern sea otters. We hypothesized that reduction in Arctic Ocean sea ice increased contact among Arctic seal species, leading to PDV introduction into the North Pacific Ocean and infection in sub-Arctic species. We evaluated exposure to and infection with PDV in sympatric ice-associated seals, northern fur seals, Steller sea lions, and northern sea otters sampled in the North Pacific Ocean from 2001-2016 to explore the timing of PDV introduction, patterns of transmission following introduction, and environmental and other risk factors associated with the emergence of PDV across the North Pacific Ocean. We incorporated satellite telemetry data from ongoing ecological studies of seals and Steller sea lions, which provided a unique opportunity to combine animal movement and epidemiologic data to understand the potential spread of PDV.

Methods
Marine mammal capture and sampling. Ice-associated seals (bearded seals, ribbon seals, spotted seals, and ringed seals), northern fur seals, Steller sea lions, and northern sea otters were live-captured and manually restrained, sedated, or anesthetized using established methods during ongoing field studies [19][20][21][22][23][24] . From 2001-2016, paired blood and nasal swab samples were collected when possible 25,26 . Blood and tissues were also collected from dead animals harvested for subsistence 27  Serologic and molecular analysis. Serology to detect antibodies to PDV (n = 1,227 animals tested 2001-2013) was performed by micro neutralisation 28 . Briefly, 1 in 10 dilutions of serum were made followed by 2-fold dilutions in high glucose DMEM medium (Gibco, USA). Cell monolayers were examined for cytopathic effect and end point neutralisation titres determined by the Reed and Muench method. Sera were tested against PDV/ USA2006 and PDV/DK/2002 strains. As the earliest molecularly confirmed PDV infection in the North Pacific was detected in 2004 8 , a subset of sera from Steller sea lions (n = 80) sampled 2001-2004 in Russia, the Aleutian Islands, the Gulf of Alaska, and Southeast Alaska was tested by micro neutralization against the related morbillivirus, canine distemper virus (CDV/Snyder Hill/wild type strain).
Serologic titres were log transformed and cumulative percent was plotted against the log(10) titre to determine the appropriate cut-off to classify positive animals for further analyses (i.e. the geometric mean). The cut-off for PDV seropositive animals used was ≥1:32 (log10 = 1.5, Supplementary Fig. 1). This threshold level was used to minimize the likelihood of false positives, providing a robust estimate of seropositivity. Steller sea lions tested for both PDV and CDV were classified as seropositive for one virus based on the highest titre measured in each sample.
For molecular analysis, total RNA was extracted using Tri Reagent ™ (Sigma) and complementary DNA transcribed (Superscript III, Invitrogen) with random nonamers from nasal swabs, blood, and tissues from animals sampled 2001-2002 and 2004-2016 (n = 1,994). Samples were screened for a fragment of the morbillivirus phosphoprotein (P) gene using a Real-Time Quantitative PCR designed for the study (Forward primer (RT-P2 s): 5′-CAT GCT AAT GGA GGA AGA GTT GAC T-3′; Reverse Primer (RT-P2 as): 5′-GTT CTC CCA TCC CTC TTT TGG-3′; Probe (PDV P2): 5′d FAM-CTC TGC TTG GCA CAG GCC ACA ATG-BHQ-1 3′) to quantify viral load or a heminested PCR with universal morbillivirus primers followed by PDV and CDV-specific primers to obtain products for sequencing 29 . Heminested PCR products were visualized by agarose gel electrophoresis and products of the expected size were cloned before sequencing. Sequences were edited manually in Geneious Pro v9.1.3 (Biomatters Ltd., Auckland, NZ) and compared with known sequences in the Genbank database. Viral nucleotide sequences were aligned with ClustalW and MUSCLE, and phylogenetic trees constructed comparing the corresponding P-gene fragments (389 bp) of known morbilliviruses using Bayesian (Mr. Bayes) algorithms v3.2.6 30 . A subset of duplicate samples (n = 60 northern fur seal nasal swabs) was submitted to The Pirbright Institute, Non-Vesicular Disease Reference Laboratory in the UK for confirmatory testing. Nasal swab, blood, or tissue samples with a C t < 37 and/or confirmed by sequencing were considered positive.
Assessing Arctic Ocean sea ice extent and open water routes from the north Atlantic to north Pacific Oceans. Sea ice in the Arctic Ocean reaches its minimum annual extent, which varies in size (area covered) and shape (geographic extent), during August to September. In some years, open water routes are created along its edges connecting the northern Atlantic and Pacific Oceans. Additionally, a trend in reduced minimum extent of the Arctic sea ice has been linked to long-term climate change 2 . Two potential routes of contact near the sea ice edge exist across the Arctic Ocean between the Atlantic and Pacific -(1) along the northern Russian coast from the Barents Sea and eastern North Atlantic and (2) along the coast and islands of northern Canada from the western North Atlantic (Fig. 1). Contact with potential PDV reservoir species is possible on either route, as harp seals breed in the Barents Sea, the coast of Greenland, and the eastern coast of Canada 31  Phocine distemper virus prevalence and risk factor analyses. PDV seroprevalence and prevalence of viral infection (PCR detection of viral nucleic acid) with 95% exact confidence intervals were calculated for each species and all species combined by year of sample collection. Observed prevalence was calculated as the number of positive (seropositive or PCR positive) animals divided by the total number of animals tested. To evaluate infection annually, we limited calculations for prevalence of viral infection to animals actively shedding PDV (PCR detection of PDV in nasal swabs from live captured and subsistence harvested animals). Prevalence calculations and statistical analyses described below were performed using R v3.3.2 33 with a significance level of α = 0.05.
Associations between PDV exposure (seropositive titre) or viral infection (PCR positive) and demographic variables (animal group (ice-associated seals, Steller sea lions, northern fur seals, and northern sea otters), age class, and sex), disposition (live captured or subsistence harvested vs. found dead), and presence of an open water route through Arctic Ocean sea ice were assessed in mixed effects logistic regression models using the lme4 package 34 . We evaluated three open water route variables: (1) presence of an open water route along the Russian coast, (2) presence of an open water route along the Russian coast in a year following closed sea ice, and (3) presence of an open water route along the Canadian coast. Sampling region (Bering Sea, Pribilof Islands, Southeast Alaska, Eastern Gulf of Alaska, Central Gulf of Alaska, Western Gulf of Alaska, Eastern Aleutian Islands, Central Aleutian Islands, Western Aleutian Islands, or Russia) was determined based on individual animal sampling locations and included as the random effect in mixed effects models. Risk factors associated with PDV exposure or viral infection in univariable regression models (P ≤ 0.20) were incorporated into multivariable models. Following inclusion of statistically (P ≤ 0.05) and biologically significant variables, multivariable models were compared using Akaike's Information Criterion (AIC) to select the most parsimonious model. Adjusted odds ratios with respective 95% confidence intervals were calculated for levels of each risk factor.

Satellite tag data and spatial analyses. Telemetry databases maintained by the Alaska Fisheries Science
Center's Marine Mammal Laboratory and ADF&G were queried retrospectively for bearded seals, northern fur seals, and Steller sea lions tagged at times and locations that coincided with the geographic area and, when possible, years of PDV sample collection. Location data received from Service Argos were filtered 35 using a swim speed of 2 m/s for bearded seals (n = 7), northern fur seals (n = 136), and Steller sea lions (n = 77). Filtered data were processed with a continuous-time correlated random walk (CTCRW) model to predict animal locations for time intervals (e.g. 20 min or 1 hour) corresponding to the timing of conductivity readings (i.e. wet or dry) collected for each species 36,37 . This model provided swim speed estimates (m/s) between sequential predicted locations, which were used to obtain the average median swim speed (m/s) for each species at sea.
The duration of viral shedding for PDV in pinniped species is unknown, but previous estimates of the combined latent (post-exposure, but not shedding virus) and infectious (shedding virus) periods in European outbreaks range from 11-18 days in harbour seal outbreak models 38 . Mean latent periods of 6-10 days and mean infectious periods of 9-13 days were used to model CDV transmission among wild canids in terrestrial systems 39 . A modeling study of an outbreak of another marine morbillivirus, Dolphin morbillivirus, estimated a mean infectious period of 8 days, with an upper bound of 24 days 40 . To capture variability in marine mammal latent and infectious periods and to assess the geographic distances that an animal infected with PDV could move and potentially transmit the virus (i.e. movement during latent and infectious periods), the distance moved by a PDV-infected individual of a given species was estimated for 1, 2, and 4 week periods (calculated as the median speed (m/s) of movement multiplied by the number of hours in the time period). The resulting estimated 1, 2, and 4 week movement distances in kilometers (km) were used to spatially buffer known animal sampling locations in ArcGIS v10.3.1 to map the distance the virus could be transported via animal movement. A subset of animals with PDV serologic results (seropositive and seronegative animals) and/or PCR testing results (only PCR negative animals) had telemetry data available. For seropositive animals, 1, 2, and 4 week post-sample collection movement tracks were overlaid on the spatial distance buffers to compare observed and expected movement of the virus with host movement.  Supplementary Fig. 3c), whereas the proportion of PCR positives decreased from 11.6% (7 positive/60 tested, confirmed by testing a subset of samples at The Pirbright Institute, Infected pinnipeds shedding PDV have the potential to reach nearby rookeries and haul-outs as well as more distant areas inhabited by conspecifics or other species (Fig. 4a). Median speeds of travel calculated from satellite-tagged bearded seals, northern fur seals, and Steller sea lions were 0.36 m/s, 0.89 m/s, and 0.32 m/s, respectively. Bearded seal speeds were used as a conservative movement estimate for wide-ranging spotted seals. Estimated movement distances for the 1, 2, and 4-week periods were 210 km, 420 km, and 840 km for bearded and spotted seals; 535 km, 1,070 km, and 2,140 km for northern fur seals; and 194 km, 388 km, and 776 km for Steller sea lions. Recorded movement tracks from one PDV seropositive bearded seal sampled in 2009 and one www.nature.com/scientificreports www.nature.com/scientificreports/ seropositive northern fur seal sampled in 2010 occurred within the estimated 1, 2, and 4-week movement distances for each species (Fig. 4b). Movement tracks for three additional PDV seropositive animals (two bearded seals and one northern fur seal) sampled in 2009 also fell within their species-specific 1, 2, and 4-week estimated movement zones. PCR positive spotted seals, ribbon seals, and northern fur seals sampled 2009-2010 were detected in close proximity to bearded seal or northern fur seal movement tracks (Fig. 4b).

Discussion
Serologic, PCR, and sequencing results provide evidence for (1) 32 . Open water along most of northern Russia linked the North Pacific Ocean to the eastern North Atlantic Ocean (Fig. 1), where European harbour seals were dying in an epidemic of PDV that began in May 2002. Reduction in the sea ice that previously created a barrier to animal movement across the Arctic Ocean may have allowed for increased contact between harp seals, the proposed Arctic reservoir of PDV 10,11,41 , and circumpolar seal species (including bearded seals and ringed seals). As the strain of PDV detected in North Pacific marine mammals is most similar to the strain isolated during the 2002 outbreak in European harbour seals, seals may have moved it from the North Atlantic outbreak or they may share a common source. If the 2002 PDV strain was circulating in harp seals that breed in the Barents Sea or near the coast of Greenland, it could have been carried south to grey seals and/ or sympatric European harbour seals in early 2002 and across the Arctic Ocean to the North Pacific Ocean in August-September. Evidence of widespread PDV exposure in the North Pacific was first detected in Steller sea lions in our study in 2003, with 20 two month-old pups, 10 four to five month-old pups, and 8 animals 16 months   www.nature.com/scientificreports www.nature.com/scientificreports/ outbreaks of PDV in European harbour seals appear to sweep through the population within a single year and then fade out until the virus is reintroduced likely through cross-species contact with grey seals or harp seals 41 . In combination with the proportion of susceptible marine mammals in the population at a given point in time, viral reintroduction via contact between Arctic reservoir species may also shape cycles of exposure and infection in the North Pacific Ocean. As seropositive ringed seals in the Northwest Territories of Canada were reported in the 1990's 10 , Canadian circumpolar seals could potentially serve as a source of PDV to North Pacific marine mammal species. However, sea ice extent and genetic data from our study support introduction along the Russian route. Openings in the sea ice along the Russian coast were linked to the increased likelihood of PDV exposure or infection (Table 1), and peaks of viral exposure in 2003 and 2009 occurred after a year with low sea ice extent and open water along the Russian coast, suggesting that multiple introductions may have occurred.
Alternatively, cross-species transmission resulting in PDV circulation among ice-associated seals, Steller sea lions, northern fur seals, and northern sea otters might also result in peaks of infection following introduction. The estimated number of Steller sea lions, northern fur seals, and ice-associated seals in Alaskan stocks 44 exceeds the critical community size estimated to be necessary to sustain endemic PDV transmission in European harbour seals (~300,000 animals) 38 . The probability of long-term pathogen persistence is likely to increase in systems with multiple hosts capable of cross-species disease transmission 45 . Differences in the density and spatial distribution of a single species may also enhance the potential for disease persistence. Although species like Steller sea lions and northern sea otters are distributed across the North Pacific, the patchy distribution of haul outs and rookeries could prolong the duration of viral transmission among species following PDV introduction 38 . Low density, patchy groups of terrestrial carnivores spread across a large landscape can maintain transmission of CDV 39 . Population age and sex structure, seasonal patterns of movement and intra-and inter-specific contact, and variation in host susceptibility to viral infection and disease severity could also contribute to the potential for endemic viral transmission. Spatially explicit epidemiologic modeling integrating movement data from satellite-tagged marine mammals with seroprevalence and viral infection prevalence trends may help to elucidate whether circulation among species or reintroduction associated with ice change is driving viral transmission in the North Pacific.
Linking movement data from satellite-tagged marine mammals with biological information on viral shedding illustrates that exposed animals have the potential to carry PDV long distances. Movement of PDV seropositive bearded seals and northern fur seals occurred within their species-specific predicted viral transmission distances and in close proximity to known locations of PCR positive individuals from all species tested. These data demonstrate the potential for animals exposed to PDV to carry the virus to areas with conspecifics and sympatric species. Data from satellite-tagged Steller sea lions suggest infectious animals could move over 100 km in one week, reaching nearby and distant rookeries (Fig. 4a). Bearded seals and northern fur seals have the potential to move over 200 km and 500 km in one week, respectively. Spotted seals and ribbon seals may bridge the gap between northern ice-associated seals and Steller sea lions and northern fur seals living in the southern Bering Sea (Fig. 4a). The ability to move long distances and timing of movements associated with life history cycles likely influence transmission patterns in the North Pacific Ocean and the potential to transmit PDV to species living in southern habitats.
Finally, the animal health impacts of PDV in North Pacific species are unknown, but may be quite different from the outbreaks in the Atlantic Ocean where the virus has caused extensive mortality in harbour seals 46 . Although no mortality events have been documented with PDV infection in Pacific species, the virus may have contributed to sporadic northern sea otter deaths, including those that occurred during an unusual mortality event from 2004-2006 8 , as well as to sporadic Steller sea lion deaths (PCR positive tissues in dead animals in this study). We found evidence of PDV viral infection in apparently healthy Steller sea lions, northern fur seals, and ice-associated seals sampled live or through subsistence harvest throughout the study. As with grey seals and Arctic seal species in the North Atlantic, PDV may be able to persist in Arctic and some sub-Arctic species in the North Pacific without causing widespread disease. The current impacts of PDV associated disease and deaths in marine mammal species in the North Pacific Ocean are unknown, but the potential for outbreaks in sensitive species highlights the importance of understanding spatio-temporal PDV transmission in this environment. Climate change-driven reductions in sea ice extent in the Arctic Ocean are projected to increase 47 and open water routes along the northern Russian coast have occurred every August and/or September since 2008. The health impacts of this new normal in the Arctic are unknown, but association of open water routes through Arctic sea ice with increased PDV exposure or infection suggest that opportunities for PDV and other pathogens to cross between North Atlantic and North Pacific marine mammal populations may become more common.