We obtained stranding data from the archives of the Quebec Marine Mammal Emergency Response Network, which monitors mortality and morbidity of marine mammals in the St. Lawrence River, Estuary and Gulf (48°23′N, 69°07′W). We compared the number of stranded (dead, ill, or injured) harbor seals, gray seals, and seals of unspecified species during the second and third quarters of 2022 (April 1–September 30) with the average of the 10 previous years for the same period using a 1-sample t-test. We evaluated goodness-of-fit of the stranding distribution of 2012 to 2021 with a Shapiro-Wilk test for both groups of seals. We considered distribution normal if p>0.05. We combined seals of unidentified species with harbor seals for this comparison to control for the differences in species identification rates in 2022 compared with previous years (data not shown).

We based our selection of carcasses to be examined on the state of decomposition (23) and field access. All carcasses were submitted frozen and, after thawing, were examined by veterinary pathologists with experience in marine mammal pathology. Animals were classified into 2 age groups, <1 year old or adult, on the basis of total length (24). Animals with no evidence of muscular or fat depletion were considered to be in good nutritional condition. Tissue samples of major organs (lung, heart, kidney, brain, intestines, lymph nodes, liver, spleen, pancreas, tongue, adrenal gland, esophagus, bladder, stomach, thyroid gland, mammary gland, and thymus) were processed for histopathologic evaluation by light microscopy using standard laboratory procedures. For each necropsy case, separate nasal and rectal swab samples were collected using a sterile polyester-tipped plastic applicator (UltiDent Scientific, https://www.ultident.com) placed in UT medium (Micronostyx, https://micronostyx.com). In addition, rectal and nasal swab samples were collected from seals found stranded for which the carcass could not be examined either because of poor preservation state or logistical limitations. All samples were first tested for IAV by PCR at the provincial Animal Health Laboratory (Laboratoire de santé animale, Ministère de l’Agriculture, des Pêcheries et de l’Alimentation du Québec; MAPAQ). All IAV H5–positive samples were subsequently sent to the National Centre for Foreign Animal Disease (NCFAD) laboratory (Winnipeg, MB, Canada) for confirmatory testing. Samples of brain or lung were also submitted for PCR for cases that had lesions suggestive of IAV but tested negative on swab samples.

Both laboratories (MAPAQ and NCFAD) used the same methods for RNA extraction and PCR. Total RNA was extracted from clinical specimens (swabs and tissues) and virus isolates using the MagMax AM1836 96 Viral RNA Isolation Kit (ThermoFisher Scientific, https://www.thermofisher.com) according to manufacturer recommendations, using the KingFisher Duo Prime, KingFisher Flex, or Apex platforms (ThermoFisher Scientific). Spiked enteroviral armored RNA (Asuragen, https://asuragen.com) was used as an exogenous extraction and reaction control. The extracted RNA samples were tested for IAV genomic material by using matrix gene–specific real-time reverse transcription PCR. Samples positive with the matrix primer set underwent repeat PCR with H5- and H7-specific primer sets, as described previously (25,26). Cycle threshold values <36.00 were considered positive and values 36.00–40.00 suspicious. For virus isolation, PCR-positive samples were inoculated into 9-day-old embryonated specific pathogen-free chicken eggs via the allantoic route.

To determine the clade, lineage, and clusters of each positive sample, the full genome segments of IAVs were amplified directly from clinical specimens or isolates using reverse transcription PCR, as described previously (27). Nanopore sequencing was performed on a GridION sequencer (Oxford Nanopore, https://nanoporetech.com) with an R9.4.1 flowcell after library construction using the rapid barcoding kit (SQKRBK004 or SQK-RBK110.96). The raw nanopore signal data was basecalled and demultiplexed with Guppy version 5.1.12 (Oxford Nanopore) using the high accuracy or super-accurate basecalling model on each run. Basecalled nanopore reads were analyzed and assembled with a BLAST search (https://blast.ncbi.nlm.nih.gov) of Iterative Refinement Meta-Assembler assembled genome segment sequences against all sequences from the National Center for Biotechnology Information Influenza Virus Sequence Database (n = 959,847) (https://www.ncbi.nlm.nih.gov/genomes/FLU/Database/nph-select.cgi) and influenza virus sequences from the 2021–2022 HPAI H5N1 outbreaks. We deposited whole-genome sequences of the HPAI H5N1 viruses from seals into the GISAID database (https://www.gisaid.org; accession nos. EPI_ISL_18916928–37).

We combined the HPAI H5N1 genomes sequenced from seals (10 samples) with 40 closely related sequences (as determined by BLAST similarity search) collected from wild birds during April–September 2022. We trimmed individual viral segments (polymerase basic 1 and 2, polymerase acidic, hemagglutinin, nucleoprotein, neuraminidase, matrix, nonstructural) of regions flanking the open reading frames and concatenated. We removed duplicate sequences from the dataset, leaving 43 complete viral genomes, totaling 13,112 nt in length, which we aligned using MAFFT version 7.49 (https://mafft.cbrc.jp). We used that alignment to estimate a time-scaled phylogenetic tree using BEAST version 1.10.4 (28). We performed tree estimations under the best fitting model of nucleotide substitution as determined by ModelFinder (generalized time-reversible model, empirical base frequencies, invariant sites, 4-category gamma distribution of rate heterogeneity) (29), a relaxed molecular clock with log-normal distribution, and a Gaussian Markov Random Field Bayesian skyride tree prior (30). We also ran 2 independent Markov chain Monte Carlo chains (200 million steps, sampled every 20,000 steps), discarding the first 10% of samples from each chain as burn-in, and assessed the chains for convergence (based on effective sample size >200) using Tracer version 1.7.2 (31). We combined post–burn-in samples using LogCombiner version 1.10.4 and produced maximum clade credibility (MCC) trees by using TreeAnnotator version 1.10.4 (28).

For immunohistochemistry (IHC), we quenched paraffin tissue sections for 10 minutes in aqueous 3% hydrogen peroxide. We then retrieved epitopes using proteinase K for 15 minutes and rinsed. The primary antibody applied to the sections was a mouse monoclonal antibody specific for IAV nucleoprotein (F26NP9, produced in-house) used at a 1:5,000 dilution for 30 minutes. We visualized the primary antibody binding by using a horseradish peroxidase labeled polymer, the Dako EnVision+ system (anti-mouse) (Agilent, https://www.agilent.com), reacted with chromogen diaminobenzidine. We then counterstained the section with Gill hematoxylin.

During April 1–September 30, 2022, a total of 209 dead or sick seals were reported in the waters bordering Quebec: 127 harbor seals, 47 gray seals, 6 harp seals (Pagophilus groenlandicus), 1 hooded seal (Cystophora cristata), and 28 seals of undetermined species. The number of dead or sick stranded harbor seals and seals of unknown species combined during that period of time (n = 55) was 3.7 times higher than the average annual number in the previous 10 years for the same period (n = 41.6), representing a statistically significant increase in number of strandings (t = 5.55, d.f. = 9; p<0.001) (Figure 1, panel A). A statistically significant increase of similar magnitude was noted for the number of gray seals found dead or sick during the second and third quarters of 2022 (47 in 2022 compared with an average of 12 in 2012–2021; t = 4.35, d.f. = 9; p = 0.002) (Figure 1, panel B). We observed no increase in mortality or morbidity for the other species of pinnipeds compared with the previous 10 years.

Carcasses of 22 harbor seals, 3 gray seals, 1 harp seal, and 1 hooded seal found during April 22–October 6, 2022, were submitted for postmortem examination. On the basis of molecular detection of H5 and presence of lesions suggestive of IAV infection, HPAI H5N1 was identified as cause of death of 14 harbor seals and 1 gray seal (56%) (Table 1). Combined nasal and rectal swab samples were obtained on the shore from an additional 11 harbor seal and 1 gray seal carcasses, and HPAI H5N1 was identified in 6 of the harbor seals (50% of those sampled) (Table 1).

All 21 infected seals were found during May 30–July 8, 2022, in the estuarine segment of the St. Lawrence waterway, mainly on the south shore, between the towns of Baie-Comeau (49.2213°N, 68.1504°W) and Notre-Dame-du-Portage (47.7630°N, 69.6096°W) (Figure 2). Infections were detected in both <1 year old and adult seals with no age predilection. All the infected adult harbor seals (n = 9) were female, and 6 had evidence of recent parturition (active lactation, asymmetric uterine horns without the presence of a fetus, or both). The infected adult gray seal was male. There were 3 male and 7 female (and 1 nondetermined) infected <1 year old seals (Table 2). One of the infected seals was found alive with profound lethargy and neurologic signs. In addition, anecdotal observations of weak and dyspneic harbor seals were reported during the outbreak.

Carcasses were attributed decomposition scores (21), varying from code 2–2.5 (fresh to mild decomposition) in 12 cases to code 3 (moderate decomposition) for 3 cases. Twelve seals were in excellent nutritional condition, 2 were thin, and 1 was emaciated. Postmortem findings (Table 3) showed notable gross lesions limited to lymphadenomegaly (submandibular and mesenteric lymph nodes), red-tinged foam in the tracheal lumen, and pulmonary congestion. Relevant histologic lesions (Table 3) included acute multifocal to diffuse mixed meningoencephalitis (Figure 3, panel A), characterized by a predominantly neutrophilic infiltrate with lymphocytes in Virchow-Robin spaces, the meninges, the submeningeal neuropil. Neuronal necrosis, satellitosis, and gliosis were also regularly observed. Acute fibrinosuppurative alveolitis, consisting of neutrophils with fibrinous aggregates in the alveolar lumen (Figure 3, panel B), was observed in 9/15 seals. Interstitial pneumonia, characterized by a mild to moderate mixed inflammatory infiltrate within the alveolar septa, was seen either concurrently or distinctively from the alveolitis. In addition to the inflammatory changes, alveolar emphysema, mild acute alveolar damage with hyaline membranes, and necrotic type 1 pneumocytes were sometimes present. Six animals presented multifocal necrotic foci of the adrenal cortex with infiltrates of degenerate neutrophils (Figure 3, panel C). Acute necrotizing thymitis, lymphadenitis, and splenitis were also observed and consisted of multifocal to coalescing zones of necrosis, mainly centered on cortical lymphoid follicles along with extensive lymphoid depletion (Figure 3, panel D). Lymph nodes were often reactive with an accumulation of histiocytic cells in the lymphatic trabeculae and medullary sinus, macrophages in the germinal centers, and an infiltration of the subcapsular sinus with granulocytes and macrophages. Other microscopic changes were seen in 6 cases with mild membranous glomerulonephritis, 4 cases with necrotizing fibrinous hepatitis of variable severity, 1 case of mild multifocal myocarditis with degenerate myofibers, and 1 animal with mild neutrophilic perivascular infiltration in the perimysium.

Immunoreactivity for IAV antigen was observed in tissues from 12 of the 13 cases that were tested (Table 2). Variable abundance of antigens, from mild to extensive, was detected in different tissues, including the neuropils and neurons (Figure 4, panel A), pulmonary alveolar septa and glandular bronchial cells (Figure 4, panel B), renal glomeruli, spleen, pancreas, liver, vascular walls of skeletal muscle, lymph nodes, tracheal vessels, and adrenal glands. Those antigens were often associated with, but not limited to, necrotic foci.

The presence of IAV H5 RNA was confirmed by PCR in 15 necropsied seals and in 6 seals that were swabbed in the field (n = 21) (Table 2). All samples were negative for H7. All 15 necropsied seals had lesions suggestive of IAV on histology. In 4 cases, NCFAD confirmatory PCRs for H5 performed on combined rectal/nasal swab samples were negative but H5 RNA was detected by subsequent PCR on frozen lung or brain tissues. Virus isolation (using embryonated specific pathogen-free chicken eggs) was successful in 16 of those 21 cases. The 16 isolates were sequenced to determine the subtype and lineage of the H5 virus. All isolates belonged to Gs/GD lineage H5N1 clade 2.3.4.4b. Five isolates had fully Eurasian gene segments similar to the Newfoundland-like clade 2.3.4.4b H5N1 viruses that emerged in Canada in late 2021. The other 11 isolates were reassortant H5N1 viruses containing gene segments polymerase basic 1 and 2, polymerase acidic, and nucleoprotein, belonging to the North American lineage IAVs and hemagglutinin, neuraminidase, matrix, and nonstructural gene segments belonging to the Newfoundland-like clade 2.3.4.4b H5N1 viruses. Accordingly, the phylogenetic tree estimated from those sequences is bifurcated at the root between fully Eurasian and reassortant viruses (Figure 5). In both the fully Eurasian and reassortant lineages, sequences derived from wild birds are ancestral to the (well-supported and monophyletic) clades that include seals. That finding provides strong support for independent bird-to-seal spillover events for each viral lineage, rather than a single spillover and subsequent reassortment event in seals.