Highly Pathogenic Avian Influenza Virus H5N1 in Double-crested Cormorants (Nannopterum auritum) of the Chesapeake Bay, USA Open Access

While low pathogenic (LP) avian influenza viruses (AIVs) are endemic in wild waterfowl and gulls, limited evidence of infection or IAV antibodies has been reported from other waterbird species, including cormorant species (family Phalacrocoracidae; Artois et al. 2002; Stallknecht 2003; Ishtiaq et al. 2012; Cross et al. 2013; Klimaszyk and Rzymski 2016). Although cormorants have been identified in previous mortality events associated with the A/goose/Guangdong/1/1996 (Gs/Gd) H5N1 lineage of highly pathogenic (HP) AIV, these events have been uncommon relative to their frequency in waterfowl (Liu et al. 2005; Muzyka et al. 2019). Following the emergence of clade 2.3.4.4b Gs/Gd HPAI H5N1 (hereafter HPAI H5N1) in 2021, increased wild bird mortality has been reported (Tian et al. 2023). Of the 508 wild avian species recognized as impacted by HPAI H5Nx viruses by the Food and Agriculture Organization of the United Nations, 281 have been added since 2021 (UNFAO 2024). Unprecedented disease impacts have been seen across numerous waterbird species, particularly colonial nesting birds, but scavenging and raptor species also have been affected (Harvey et al. 2023). Wild birds are now serving as a viral reservoir and suffering a high number of mortalities due to infection. This includes many newly affected species, such as Peruvian Pelicans (Pelecanus thagus; Leguia et al. 2023) and Humboldt Penguins (Spheniscus humboldti; Muñoz et al. 2024). Mass mortality events have also been observed in cormorants, such as Cape Cormorants (Phalacrocorax capensis; Molini et al. 2023; Roberts et al. 2023), Guanay Cormorants (Leucocarbo bougainvillii; Gamarra-Toledo et al. 2023; Leguia et al. 2023), and Double-crested Cormorants (hereafter DCCO; USDA 2023; Youk et al. 2023; Avery-Gomm et al. 2024).

Although these events demonstrate that numerous cormorant species are susceptible to infection with HPAI H5N1 and that mortality can occur, the capacity to recover from an HPAI H5N1infection or to develop a detectable immune response to this virus remains unclear (Wight et al. 2024). This lack of information complicates efforts to assess the risk of potential long-term population effects. Our objective was to determine if HPAI H5N1 infection or antibodies to H5 and N1 were present in DCCO on Poplar Island in the Chesapeake Bay, Maryland, US, in the summers of 2023 and 2024.

We conducted our study on the Paul S. Sarbanes Ecosystem Restoration Project at Poplar Island (hereafter, Poplar Island; 38°46′01″N, 76°22′54″W), in the Chesapeake Bay, Maryland, US. Poplar Island uses dredged material from the shipping channels leading to the Port of Baltimore, Maryland, US, to rebuild and restore remote island habitat (Erwin et al. 2007; Maryland Environmental Service 2021). This restoration effort has successfully restored breeding habitat for a variety of waterbird species, including Common Terns (Sterna hirundo) and Least Terns (Sternula antillarum), Snowy Egrets (Egretta thula) and Western Cattle Egrets (Bubulcus ibis), and Little Blue Herons (Egretta caerulea) and Tricolored Herons (Egretta tricolor), among others (Prosser 2024). Poplar Island and/or nearby Jefferson’s Island (<500 m from Poplar) have also hosted a colony of breeding DCCO annually since 2002, with records of breeding pairs as early as 1996 when monitoring began at these sites (Prosser 2024). In 2023 and 2024, U.S. Fish and Wildlife Service (USFWS) surveys estimated 2,772 and 1,852 pairs of DCCO on Jefferson Island, respectively (Prosser 2024).

From 6 June to 1 August 2023 and 2 May to 6 May 2024, USFWS personnel lethally controlled DCCO to reduce conflict between DCCO and other breeding waterbirds, such as Snowy Egrets and Glossy Ibis (Plegadis falcinellus), in targeted areas of Poplar Island. All take was performed by USFWS personnel under annually submitted federal and state depredation permits (USFWS Depredation Permit MB043544; Maryland Wildlife Damage Control Permit 58551). A subset of controlled birds, based on available time and personnel, were collected immediately upon death. Oropharyngeal (OP) and cloacal (CL) swabs were collected and placed in a single vial containing 2 mL of chilled brain heart infusion broth (Becton Dickinson and Company, Sparks, Maryland, USA) supplemented with antimicrobials: penicillin G (1,000 units/mL), streptomycin (1 mg/mL), kanamycin (0.5 mg/mL), gentamicin (0.25 mg/mL), and amphotericin B (0.025 mg/mL; Sigma-Aldrich, St. Louis, Missouri, USA). Swabs were kept on ice packs for 24–36 h until shipped to the processing laboratory. Upon receipt, swabs were placed in long-term storage at −70 C until processed. Blood samples (approximately 3 mL) were taken from the controlled birds via cardiac puncture and centrifuged within 6 h of collection. Sera were stored at −70 C until shipped to the processing laboratory where they were stored at −20 C until serologic testing. The age of all individuals sampled was recorded (all birds were after hatch year). All DCCO sampled in this study appeared healthy, except for a single individual sampled in 2023 that was easily hand collected and displayed ocular abnormalities including swelling and discharge.

We extracted RNA from all paired OP and CL swabs with the MagMAX AI/ND Viral RNA extraction kit (Ambion Inc., Austin, Texas, USA) as described previously (Das et al. 2009) and screened for the AIV matrix gene using real-time quantitative reverse transcription PCR (RT-qPCR; Spackman et al. 2002), with cycle threshold (Ct) values <40 considered positive. All RNA were also screened for 2.3.4.4b H5 via RT-qPCR; suspect positives from this assay were sent to the National Veterinary Services Laboratory (NVSL), Ames, Iowa, US, for confirmation of pathogenicity and subtype. All paired OP and CL swabs were subjected to virus isolation (VI) using 9–11-d-old embryonated chicken eggs (ECEs), as described previously (Stallknecht et al. 1990). Briefly, OP and CL swabs were vortexed and centrifuged at 1,500×G for 15 min; supernatant was inoculated (0.33 mL per egg) into three 9–11-d-old ECEs via the allantoic route. The ECEs were incubated at 37 C for 96 h; amnio allantoic fluids were then harvested and tested by hemagglutination assay using 0.5% chicken red blood cells (Killian 2008).

All sera were heat-inactivated at 57 C for 30 min and tested using a commercial blocking ELISA (bELISA; IDEXX AI MultiS-Screen Ab Test, IDEXX Laboratories, Westbrook, Maine, USA) for antibodies to the IAV nucleoprotein (NP) according to the manufacturer’s instructions. Sera were reported as positive for antibodies to AIV if the serum-sample-to-negative-control (S/N) absorbance value was <0.7 (Brown et al. 2009; Tolf et al. 2013; Shriner et al. 2016). Samples (all in 2023, all with S/N <0.7 in 2024) were also tested by hemagglutination inhibition (HI) and virus microneutralization (VN) for antibodies reactive to HPAI H5N1 and LPAI North American lineage H5 IAV as described by Stallknecht et al. (2020, 2022). The LP North American antigen was a reverse genetics construct that includes the hemagglutinin (HA) and neuraminidase (NA) from A/blue-winged teal/Texas/AI12-4150/2012 (H5N2) on a A/Puerto Rico/8/1934 (PR8) backbone (hereafter referred to as BWT). The representative clade 2.3.4.4b HPAI antigen IDCDC-RG71A (H5N8) includes the HA and NA from A/Astrakhan/3212/2020-like virus on a PR8 backbone (hereafter referred to as AST); the HA contains a modified protease cleavage site characteristic of LP AIV. Sera were also assessed for reactivity to the N1 subtype using an enzyme-linked lectin assay (ELLA) as described previously, with A/ruddy turnstone/New Jersey/AI13-2948/2013 (H10N1) used as antigen (Stallknecht et al. 2022). Positive threshold titers for HI, VN, and ELLA were 8, 20, and 80, respectively. For geometric mean titer (GMT) calculations, performed in the R statistical platform (R Core Team 2024), negative titers and endpoint titers of <8 and ≥1,024 (HI) and <20 and ≥2,560 (VN) were transformed as 4 and 1,024 for HI, and 10 and 2,560 for VN (Wong et al. 2016). The GMTs were compared between antigens (AST and BWT) for both HI and VN using linear models where GMT values were considered a function of antigen type and year. Pairwise comparisons were calculated for each model (HI and VN) via the emmeans package (Lenth 2021). Seroprevalence between years was compared via a logistical regression with a binomial distribution where antibody presence was considered a function of year. All calculations were performed within the R statistical platform (R Core Team 2024). No comparisons were made across years for the presence of active infections due to differences in time of year when samples were collected.

There were no detected mortality events believed to be related to HPAI H5N1 on Poplar Island during the summers of 2023 or 2024. However, there was a mortality event from 17 July 2023–16 October 2023 and 29 July–5 October 2024, suspected to have been caused by avian botulism (NWHC cases 203434 and 31480), which included Mallards (Anas platyrhynchos), Semipalmated Sandpipers (Calidris pusilla), and Black-necked Stilts (Himantopus mexicanus). Only a single DCCO was observed associated with the 2023 event, and none were associated with the 2024 event.

We collected paired OP and CL swabs from 32 DCCO in 2023. Two samples tested positive by IAV matrix RT-qPCR, with Ct values of 32.0 (sample 31) and 33.3 (sample 39); sample 31 also tested positive via 2.3.4.4b H5 RT-qPCR. This H5-positive sample was confirmed by RT-qPCR as HP H5N1 at NVSL but did not meet the NVSL criteria (Ct<35) for viral sequencing. No IAVs were isolated from the 32 swab samples tested in 2023. Similarly in 2024, we collected paired OP and CL swabs from 29 DCCO. However, no samples tested positive by AIV matrix or 2.3.4.4b H5 RT-qPCR.

In 2023 we collected 31 sera samples, of which 21 (68%) tested positive by bELISA at an S/N value of <0.7 (Table 1). Three of these tested positive at an S/N value between 0.5 and <0.7. Cumulative results from HI, VN, and ELLA testing indicated that 17/18 (94%) individuals with bELISA <0.5 and 19/21 (90%) with bELISA <0.7 were seropositive for H5 and N1 (Table 1). None of the bELISA-negative birds (n=10; S/N≥0.7) tested positive by HI or VN, although one tested positive by ELLA. Antibody titers for both HI and VN were biased toward the clade 2.3.4.4b HPAI H5N1 representative antigen (AST). Based on antibody titers that tested positive for H5 antibodies, the GMT for the HI BWT (North American H5 antigen) and AST (clade 2.3.4.4b H5 antigen) were comparable at 2.4 and 3.3, respectively (P=0.105; SE=0.402; n=20). However, GMT was significantly lower via VN (n=21) for BWT compared to AST (5.0 and 9.2, respectively; P<0.001; SE=0.579; n=21). A high GMT of 8.7 also was observed with results from ELLA (n=22).

In 2024 we collected 25 sera samples, of which 10 (40%) tested positive by bELISA at an S/N value of <0.7. Four of these tested positive at an S/N value between 0.5 and <0.7 (Table 1). Cumulative results from HI, VN, and ELLA testing indicated that all individuals with bELISA <0.7 were seropositive for H5 and N1 (Table 1). Antibody titers for both HI and VN were biased toward the clade 2.3.4.4b HP H5N1 representative antigen (AST). Based on sera that tested positive for H5 antibodies, the GMT for the HI BWT (North American H5 antigen) and AST (clade 2.3.4.4b H5 antigen) were comparable at 3.1 and 4.2, respectively (P=0.105; SE=0.402; n=10). However, GMT was significantly lower via VN for BWT compared to AST (3.8 and 7.9, respectively; P<0.001; SE=0.579; n=10). A high GMT of 9.0 also was observed with results from ELLA (n=10). We found no differences between GMT within BWT or AST across yr (HI: P=0.275; SE=0.426; n=30; VN: P=0.188; SE=0.620; n=31) for seropositive birds. The overall seroprevalence observed in 2024 (40%) was a significant decrease from the 68% observed in 2023 (P=0.041; SE=0.5606; n=56).

The single bird with eye lesions collected in 2023 tested negative for IAV and 2.3.4.4b H5 by RT-qPCR, and IAV was not isolated. It tested positive on bELISA (S/N=0.45) as well as AST VN and ELLA, indicating that it had antibodies to both H5 and N1; however, it is unknown if the clinical condition of this bird was related to prior IAV infection or a separate condition.

All data supporting these results are published openly in Harvey et al (2024).

In this study of wild DCCO of Chesapeake Bay, tested in 2023 and 2024, we detected HPAI H5N1 in only a single bird in the 2023 sample population. Serologic data, however, suggested that HPAI H5N1 exposure was more common, with 61% of 31 and 40% of 25 birds testing positive for antibodies to H5 and N1 in 2023 and 2024, respectively. These seropositive DCCO had presumably survived previous AIV infections. The serologic evidence supports that these previous infections were associated with HPAI H5N1 (AST), given not only the high prevalence of antibodies to H5 and N1, but also the higher prevalence observed in both HI and VN results for this antigen as compared to the North American LPAI H5 antigen (BWT). Furthermore, HI and VN GMTs were higher with the 2.3.4.4b HPAI H5 antigen, reflecting an antigenic bias toward this antigen rather than the North American LPAI H5. Finally, 94% and 100% of the H5 positive samples as detected by HI and/or VN also tested positive to N1 in 2023 and 2024, respectively.

While the exposure of DCCO to HP H5N1 in the Chesapeake Bay was expected given the observation of this pathogen within the wild avifauna of this ecosystem (Prosser et al. 2022; USDA 2024), the lack of observed mortality for DCCO associated with this exposure was unexpected. Historically, DCCO and closely related species have tested negative for antibodies to AIVs (Artois et al. 2002; Stallknecht 2003; Ishtiaq et al. 2012; Cross et al. 2013; Klimaszyk and Rzymski 2016) but have experienced large-scale mortalities associated with Gs/Gd HPAI H5Nx exposure (Liu et al. 2005; Muzyka et al. 2019; Molini et al. 2023; Roberts et al. 2023; Gamarra-Toledo et al. 2023; Leguia et al. 2023; USDA 2024). For instance, in South America, the Guanay Cormorant was one of the most heavily impacted species during a highly pathogenic avian influenza outbreak in November 2022 (Gamarra-Toledo et al. 2023; Leguia et al. 2023). Similarly, while cormorant mortalities in the US have been much lower through the course of the ongoing outbreak than for species such as Snow Geese (Anser caerulescens), there were still several reports of large-scale mortality events for wild DCCO, with as many as 900 estimated dead from a single event (USGS 2024), and mortalities reported across 13 US states and all four North American flyways (USDA 2024).

The level of protection offered by the antibodies identified in this study was not evaluated, but experimental studies have demonstrated that previous infections with related AIVs can result in partial or complete protection and reduced shedding during subsequent infections with both LP and HP viruses (Swayne and Kapczynski 2008; Fereidouni et al. 2009; Latorre-Margalef et al. 2013). The lack of previous reports of AIV infection in DCCO and historic lack of evidence of antibodies to endemic LPAIV suggest that DCCO populations were fully susceptible to HPAI H5N1 infection when these viruses were introduced into North America, but currently, population immunity to HPAI H5N1 may be higher relative to pre-outbreak levels in the DCCO population. This is supported by national mortality data, which indicated 10 mortality events involving DCCO from April to December 2022 (ranging from 2 to 900 dead or estimated-dead DCCO per mortality event, with >200 estimated-dead DCCO from each of six of these events), but only three events involving DCCO since then (the highest number of estimated dead DCCO from these three events was 16), with the most recent being in November 2023 (USGS 2024). The levels of antibodies to HPAI H5N1 observed in our study, paired with declines in mortality events in this species, suggest that, at least for the short term, large-scale mortalities with population-level impacts driven by HPAI H5N1 may be reduced for DCCO in this region. Additional work is needed to determine the status of other North American DCCO populations.

Long-term projections of DCCO susceptibility to HPAI H5N1 are difficult to make based on our limited knowledge of immunology in this and other wild avian species. For instance, it remains unclear how long AIV antibodies persist (Hoye et al. 2011; Wight et al. 2024). The reduction of overall antibody levels from 2023 to 2024 observed in this study indicates that antibody prevalence is declining and immunity within the population is waning. However, this should be interpreted with caution given our limited sample sizes. The potential for subsequent infections with both LPAI and HPAI H5N1, and an associated immunologic boost, also cannot be estimated (Latorre-Margalef et al. 2017). Finally, it is not clear how population immunity in the surviving adult breeding population will affect potential infection in hatch-year birds. Maternal transfer of AIV antibodies has been found to vary depending on maternal exposure and body condition, and across the clutch (Dirsmith et al. 2018), although this is understudied in nestlings, with most data coming from egg yolk antibody studies (Hammouda et al. 2012; Qi et al. 2021).

A notable limitation of our study was the overall sample size and study extent. While data from a single breeding site are informative, the data are not necessarily indicative of an entire regional population, and inference should be limited accordingly. Additionally, our sampling represented less than 0.01% of the population on site; therefore AIV strains with low prevalence were likely to be undetected. Methods to increase sample size all come with inherent tradeoffs. For instance, environmental sampling could improve detection of viral strains but cannot be paired with assessment of seroprevalence in individuals. Similarly, increasing the number of paired swab and serum samples requires either extensive effort from trapping or increased lethal take.

This study was also limited by the fact that we do not know the timing of initial infection for seropositive individuals. Thus, initial infection could potentially have occurred in a preceding season, with only survivors reaching the breeding grounds surveyed in this study. Nevertheless, the presence of an apparently healthy individual infected with HPAI H5N1 indicated that transmission was occurring at the time of sampling in 2023. While it was possible that mortalities did occur and were not detected, we believe this was unlikely given the numerous biologists on site during the breeding season and the high visibility of breeding sites, both of which contribute to rapid detection of carcasses. Additionally, given the lack of detected mortalities and the high level of exposure indicated by the serologic data, DCCO may have played a role in disease transmission to conspecific and to heterospecific individuals with shared ranges and habitat, suggesting the potential role of DCCO as HPAI H5N1 spreaders. Conversely, it is possible that the decline in seroprevalence observed from 2023 to 2024 is a function of differences in the time of year samples were collected (6 June–1 August 2023 and 2 May–6 May 2024). However, the lack of identified active infections in 2024 suggests that additional seroconversion of recently infected individuals was unlikely during this year. Finally, the lack of successful virus isolation leaves us with little information regarding LPAI viruses circulating in this population.

Few studies have examined population-level seroprevalence of HPAI antibodies in wild bird species since the late-2021 introduction of HPAI into North America, particularly among waterbirds or species ecologically similar to DCCO. Although much of the research has focused on virus detection or observed mortality event outbreaks, paired virologic and seroprevalence studies remain limited to waterfowl and raptors (Huang et al. 2024; Rayment et al. 2024). Expanding serosurveillance to other abundant species that share similar ecological niches with DCCO would provide valuable insights into HPAI exposure rates and transmission dynamics. Such studies are critical to understanding the full extent of HPAI H5N1 impact and the potential for species-specific differences in susceptibility and immune response.

While our data were limited to a single sampling site, these data suggested that this species can recover from infection with HPAI H5N1. Additionally, DCCO that have recovered from such infection may be partially or fully protected, and if so, the risk of large-scale mortality could be reduced. Additional work to understand the duration of antibodies and impacts of infection on juvenile DCCO is necessary to project possible impacts of continued persistence of HPAI H5N1 on the landscape.

Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government. Laboratory support was provided by Lyndon Sullivan-Brügger. Funding for project execution was provided by the U.S. Fish and Wildlife Service Zoonotic Disease Initiative (K00B4600075). Partial funding was also provided by the National Institute of Allergy and Infectious Diseases, National Institutes of Health, Department of Health and Human Services, under contract 75N93021C00016. Partial support for JDS was provided by the U.S. Geological Survey, Ecosystems Mission Areas Biological Threats and Invasive Species Program and Environmental Health Program as well as Chesapeake Bay Priority Ecosystem Science.