Potential Pathways of Spread of Highly Pathogenic Avian Influenza A/H5N1 Clade 2.3.4.4b Across Dairy Farms in the United States =============================================================================================================================== * Haley Stone * Mehak Jindal * Samsung Lim * Rebecca Dawson * Ashley Quigley * Matthew Scotch * C. Raina MacIntyre ## Abstract The emergence of Highly Pathogenic Avian Influenza (HPAI) A/H5N1 clade 2.3.4.4b has led to an unprecedented intercontinental spread, affecting a broad spectrum of avian and mammalian species. Recent outbreaks in US dairy farms, underscore the urgent need to understand the transmission pathways. The study aimed to evaluate the modes of introduction and transmission to dairy farms, through geospatial and exposure analyses. Our findings favour a singular introduction over multiple independent introductions, with non-waterfowl species exhibiting the highest dairy farm exposure, which is a major shift from historical waterfowl spread. Moreover, bidirectional spread between cattle and poultry highlights the intricate nature of disease transmission within the agricultural ecosystem. Additional factors such as livestock trade, poultry litter feed and contaminated milking machinery likely contributed to the amplification of the outbreaks throughout the United States. As large-scale outbreaks persist in the United States, the likelihood of a human pandemic increases, making it imperative to enact and sustain heightened surveillance measures across all potentially impacted species. ## Introduction Since its emergence, Highly Pathogenic Avian Influenza virus (HPAI) A/H5N1 clade 2.3.4.4b has undergone unprecedented intercontinental spread, affecting not only a wider range of wild birds, but a wide range of mammalian species that have never been infected before. Traditionally, avian influenza virus (AIV) has spread along specific migratory routes of wild waterfowl (ducks, geese and swans), which infect farmed poultry. However, with the emergence of clade 2.3.4.4b, a notable shift has occurred, with 328 new avian species, many of which are not waterfowl, and 41 mammalian species (including aquatic and terrestrial mammals) being affected, thus opening new avenues for AIV transmission [1]. Recently, outbreaks in dairy cattle have emerged in the United States. On 25 March, 2024, an outbreak of A/H5N1 was reported in dairy cattle across Texas, Kansas, and New Mexico, with symptoms including reduced milk production and mild respiratory issues [2]. As of 1 May 2024, a total of 33 dairy farms across eight states have confirmed outbreaks, with additional outbreaks suspected. Since the initial emergence, multiple theories have surfaced regarding the origins and pathways of transmission among dairy farms. One theory suggests that the virus has spread through contaminated feed or milking equipment, as the virus has been found in high concentration in the mammary glands of cows [3, 4]. Notably, the transmission has occurred rapidly from farm to farm, adding complexity to the situation. Two theories have emerged from this: firstly, that the initial exposure occurred in a single farm, with subsequent spread to other farms facilitated by livestock trade; the other suggests that multiple independent events led to introductions of the virus into different dairy farms. These events may have occurred through farming practices, trade in poultry or livestock, or infection introduced by wild birds or terrestrial animals. Farming practices such as the use of ‘poultry litter’ and inadequate sanitation of machinery used in milking of cows are potential contributors. Poultry litter often contains excrement/feathers along with bedding from the poultry farms. Spread may also be through migratory birds, both waterfowl and non-waterfowl, which, typically, spend time in the winter within the southern parts of the United States and travel north to their breeding grounds. Terrestrial animals are also a potential source of infection, particularly mammals in close proximity to human habitation, which may have contact with farmed animals or domestic pets. In this study, we aim to explain the unprecedented spread of A/H5N1 in dairy farms in the United States with consideration of the possible routes of introduction and transmission. Understanding the modes of introduction and transmission to dairy farms is essential for outbreak preparedness and response. ## Methods ### Data A comprehensive line list of all reported outbreaks in wild birds, poultry, and mammals from 2021 to 26 April 2024 in the United States was generated, as clade 2.3.4.4b was first introduced into the United States in 2021. Data for wild bird outbreaks was obtained from the United States Department of Agriculture, while data on mammalian, wild birds and poultry outbreaks were obtained from the World Organisation for Animal Health’s reference database (WAHIS) [5, 6]. Total affected wild bird species were identified after collating both WAHIS database and USDA and filtering by all unique species. Mammalian outbreaks were categorized by species, with a separate line list created for dairy farm outbreaks [Supplementary Material]. Of the 23 mammalian species affected, 8 terrestrial mammal species have close interactions with human habitation and farms i.e. Abert’s Squirrels, American Martens, Coyotes, Fishers, Red Fox, Raccoons, Striped Skunks and Virginia Opossums. Regarding wild birds, a list of affected species was created and categorised into waterfowl or non-waterfowl. Lastly, to assess the interaction between poultry and dairy farm outbreaks, all poultry and domestic bird outbreaks were filtered for date of outbreaks within 2024. ### Geospatial Analysis To explain the spread in dairy farms in the US, we performed geospatial analysis to estimate the prevalence of known infected species with consideration of each known cattle outbreak. Python v.3.11.5 and QGIS software (v.3.30.3) were used in the geospatial analysis of dairy farm outbreaks, waterfowl dispersal, non-waterfowl dispersal, terrestrial mammals’ habitats and poultry outbreaks to create static and dynamic maps. The cartographic boundaries of the United States were acquired from the United States Census Bureau and the US migration flyways were sourced from the U.S. Fish & Wildlife Service [7]. All figures are mapped using the EPSG 4326 projection, which means they are presented in coordinates of latitude and longitude. Cattle and poultry inventory per county was acquired from the National Agricultural Statistics Service [15]. Cattle density was calculated by taking the cattle including calves’ inventory estimates per county and divided by the county area. Similar to cattle density, poultry density was calculated by taking the aggregate inventory for broiler chickens, layer chickens, pullets, roosters, ducks, geese, turkeys and quail and dividing it by county area. The densities were transformed to a logarithmic scale for enhanced visualization. For wild birds, estimated weekly abundance data was individually retrieved for the affected species, encompassing abundance estimates per 14km2 area [16]. Avian species lacking suitable abundance data were excluded. These species include the Black-billed Magpie, Black-legged Kittiwake, Brandt’s Cormorant, Gannet, Glaucous Gull, Northern Fulmar, and Northern Gannet. Subsequently, a total weekly abundance was compiled for each 14km2 area in the United States, separately for waterfowl and non-waterfowl, as waterfowl are the traditional spreaders of A/H5N1, but an expanding range of infected non-waterfowl birds may now be spreading the infection. The aggregate of waterfowl and non-waterfowl affected species were calculated for each 14km2 area. To calculate terrestrial mammal density, habitat data for eight infected mammal species who have potential cattle interaction were obtained. These mammals include Abert’s Squirrel (Sciurus aberti), American Marten (Martes americana), Coyote (Canis latrans), Fisher (Martes pennanti), Red Fox (Vulpes vulpes), Raccoon (Procyon lotor), Striped Skunk (Mephitis mephitis), and Virginia Opossum (Didelphis virginiana), sourced from the U.S. Geological Survey (USGS) [8-15]. The aggregate of the habitat areas was calculated for each 14km2 area. ### Data Analysis In order to determine all potential pathways of spread to cattle, we calculated the aggregate density measurement of each potential source at the site of the cattle outbreaks for the month of March 2024, as a majority of the outbreaks have begun within the month of March. In order to reduce potential biases, we calculated z-scores to determine if the density deviates from the expected levels for each group (waterfowl, non-waterfowl, terrestrial mammal habitats, and poultry). A positive z-score signifies an exposure above the expected level, whilst a negative z-score indicates an exposure below the expected level. ## Results Of the 33 documented outbreaks in dairy farms as of 30 April 2024, 30 were reported to the WAHIS system, complete with county-specific data [Figure 1]. Within this dataset, Texas accounted for 11 outbreaks, followed by New Mexico (6), Michigan (5), Kansas (4), and single outbreaks in Ohio, Idaho, South Dakota, and North Carolina. The initial outbreak occurred on March 10, 2024, in Castro County, Texas, preceding the confirmation of A/H5N1 in dairy cattle by 15 days. Among the 30 dairy farm outbreaks, 10 reported start dates before the confirmed cases of A/H5N1 in dairy cattle. In addition, cattle densities varied, however, the first outbreak in Castro County, Texas had the third highest reported density in the United States (ρ=0.14). ![Figure 1:](http://medrxiv.org/http://medrxiv.stage.highwire.org/content/medrxiv/early/2024/05/04/2024.05.02.24306785/F1.medium.gif) [Figure 1:](http://medrxiv.org/content/early/2024/05/04/2024.05.02.24306785/F1) Figure 1: Time Series(animated) plot of the distribution of dairy farm outbreaks (n=30) layered over cattle density per county area from 10 March to 11 April 2024 [5]. We illustrate the migration pathways of wild waterfowl [Figure 2] and non-waterfowl [non-waterfowl] species, along with the density of habitats for the eight terrestrial mammals [Figure 4] that have tested positive for the virus since 2021. ![Figure 2:](http://medrxiv.org/http://medrxiv.stage.highwire.org/content/medrxiv/early/2024/05/04/2024.05.02.24306785/F2.medium.gif) [Figure 2:](http://medrxiv.org/content/early/2024/05/04/2024.05.02.24306785/F2) Figure 2: Time series(animated) map of the estimated weekly relative abundance of 36 waterfowl species affected by highly pathogenic avian influenza virus from 4 January (Week1) to 12 April (Week 14) layered with dairy farm outbreaks. ![Figure 3:](http://medrxiv.org/http://medrxiv.stage.highwire.org/content/medrxiv/early/2024/05/04/2024.05.02.24306785/F3.medium.gif) [Figure 3:](http://medrxiv.org/content/early/2024/05/04/2024.05.02.24306785/F3) Figure 3: Time series(animated) map of the estimated weekly relative abundance of 103 non-waterfowl affected by highly pathogenic avian influenza virus from 4 January (Week1) to 12 April (Week 14) layered with dairy farm outbreaks. ![Figure 4:](http://medrxiv.org/http://medrxiv.stage.highwire.org/content/medrxiv/early/2024/05/04/2024.05.02.24306785/F4.medium.gif) [Figure 4:](http://medrxiv.org/content/early/2024/05/04/2024.05.02.24306785/F4) Figure 4: Terrestrial wild mammal density per 14km2 area in the mainland USA, based on potential interactions with cattle layered with dairy farm outbreaks. A total of 139 infected wild bird species have reported cases in the United States [6]. Of the 139 bird species, 36 were categorised as waterfowl and 103 non-waterfowl. Many of these species migrate from southern wintering regions to northern breeding grounds. The estimated dairy cattle exposure to non-waterfowl species within the outbreak regions (ρ=0.723, z-score=1.162) exceeded that of waterfowl (ρ=0.499, z-score=0.410). In addition, the estimated dairy cattle exposure within the outbreak regions to the selected terrestrial mammals was 0.300 (z-score= -0.662). In Castro County, Texas, the estimated relative abundance was 0.715 for waterfowl, 0.860 for non-waterfowl, and a terrestrial wild mammal density of 0.487 per 14km2. Within the subsequent three days, three more farm outbreaks emerged: one in Hale County, Texas; another in Castro County, Texas; and the third in Cassia County, Idaho. Given the geographic spread of these outbreaks over a short timeframe, we conducted an analysis of the outbreak in Cassia County, revealing estimated relative abundances of 0.615 for waterfowl, 0.713 for non-waterfowl, and a terrestrial wild mammal density of 0.366. In addition, a total of 30 poultry and 28 domestic outbreaks have been reported within 2024 with many within a 50-mile radius of dairy farm outbreaks [Figure 5]. In Texas’s Castro County, for instance, cattle outbreaks commenced on 10 March 2024 and 13 March 2024, while poultry outbreaks in neighbouring Parmer County began on 1 April 2024. Additionally, Deaf Smith County, adjacent to Castro County, reported a backyard poultry outbreak on 13 March 2024. Similarly, in New Mexico’s Curry County, four outbreaks were declared on 16 March 2024, 28 March 2024, 29 March 2024 1 April 2024 and 2 April 2024, with most directly followed by poultry outbreaks reported in Texas’s Parmer County on 1 April 2024, with subsequent outbreaks occurring in poultry in neighbouring Roosevelt County on 9 April 2024 and 10 April 2024. In Michigan’s Montcalm County, 2 outbreaks in cattle occurred on 20 March 2024 and 30 March 2024, followed by subsequent poultry outbreaks in neighbouring Ionia County on 29 March 2024, 5 April 2024, and 12 April 2024. Moreover, neighbouring Isabella County and Ionia County itself subsequently had outbreaks in dairy farms on 10 April 2024 and 11 April 2024, respectively. In Kansas’s Kearny County, two dairy farm outbreaks were recorded on 20 March 2024 and 21 March 2024, followed by subsequent poultry outbreaks in neighbouring Grant County on 7 April 2024. Additionally, three dairy farm outbreaks occurred on 15 March 2024, 17 March 2024, and 25 March 2024, with another in Hartley County on 30 March 2024 and 03 April 2024, following an outbreak in domestic/backyard poultry in Moore County, Texas, on 6 March 2024. Moreover, a dairy farm outbreak reported in Cassia County on 13 March 2024 was in close proximity to a domestic poultry outbreak that occurred in Twin Falls County on 1 February 2024. ![Figure 5:](http://medrxiv.org/http://medrxiv.stage.highwire.org/content/medrxiv/early/2024/05/04/2024.05.02.24306785/F5.medium.gif) [Figure 5:](http://medrxiv.org/content/early/2024/05/04/2024.05.02.24306785/F5) Figure 5: Dairy, poultry and backyard farm outbreaks in the mainland USA from January to April 2024 overlayed poultry density per county area. ## Discussion This is the first study to show the importance of non-waterfowl wild birds in spreading A/H5N1 clade 2.3.4.4b. This is a major shift in our understanding of long-range spread of AIV, as historical long-range spread has been through waterfowl and their specific flyways. The unprecedented spread of A/H5N1 Clade 2.3.4.4b into new species of wild birds increases the risk of other farm outbreaks, as there are 10,727 more non-waterfowl species than waterfowl species in the world. Our analysis supports a single introduction rather than multiple independent introductions. After the introduction, the migration of wild birds continued to the Midwest during the week of 17 March, which does not support multiple independent introductions to dairy farms, given that a significant number of dairy farms still experienced outbreaks in Texas and New Mexico post 17 March. The data indicates the infection may have been first introduced to dairy farms by non-waterfowl wild birds, but that other factors amplified the spread across the US. There is a likely interplay of farming practices and shared environmental factors also contributing to disease transmission. A past epidemic of A/H5N1 and A/H5N2 in turkey farms in the US also had complex spread and was difficult to explain by wild bird pathways alone [16]. While waterfowl and terrestrial mammals may have contributed, the evidence suggests a greater involvement of non-waterfowl species. In addition, outbreaks of poultry within close proximity to the dairy farms occurred both before and after dairy farm outbreaks suggesting a bidirectional spread between cows and poultry. The combination of high cattle densities and the timing of non-waterfowl migration in early March likely predisposed the initial dairy farms to the virus. Furthermore, the observed pattern of temporal proximity between outbreaks of avian influenza in poultry and domestic animals, particularly cattle, underscores the potential interplay or shared environmental factors contributing to disease transmission simultaneously or between these livestock species. Of the 30 outbreaks, ten were within Texas and New Mexico following the week of 17 March 2024. This is inconsistent with the expected geographical dispersal of migratory birds as the highest density for both waterfowl and non-waterfowl was in the Midwest rather than the Southwest, suggesting that after initially being introduced by wild birds, local farming and trade practices contributed to further spread. However, with outbreaks clustered in densely populated dairy and beef regions and detection delays, the possibility of exposure via cattle trade still remains. Several other factors contributing to the rapid detection in farms nationwide include the continual cattle trade and relaxed early trade bans/restrictions. Cattle trade within the United States continuously occurs through country, state, and interstate borders, potentially facilitating the spread of the virus. Subsequent outbreaks following the initial exposure may have been started by cattle trade, with the initial dairy farm case in Michigan directly linked to this trade activity [14]. Initially, certain states adopted a more lenient approach regarding bans or limitations on cattle trade, only restricting cattle that were ill or from farms with infections. Despite subsequent increased restrictions, testing remains voluntary. This outbreak emphasises the need for proactive rather than reactive surveillance. The United States has previously and continues to report large poultry outbreaks, suggesting broader wild bird outbreaks. This trend is likely to persist with the current migration and future spring migration. While increased surveillance is being implemented with testing requirements for dairy cows on 24 April 2024, there have been massive poultry outbreaks, spillover events into other mammals, and neuroinvasive mutations since 2021 globally, with the Americas being a hotspot [15]. Increased surveillance measures should be implemented and continued across all potentially affected species. Limitations encompass sampling and testing biases observed in the dairy farm outbreaks and among the affected wild bird and mammalian species. The economic repercussions of outbreaks may create a substantial disincentive for AIV testing and reporting on farms. Furthermore, the prevalence of affected non-waterfowl species significantly skewed the sampling, which is why z-scores were used as it standardises the data distributions, detect outliers, and ensure fair comparisons between variables. Lastly, it’s worth noting that the counts of cattle and poultry per county, as well as the chosen terrestrial mammal habitats, were from 2022 estimates, which might affect the estimates if these have changed substantially since then. It remains possible that the initial introduction to dairy farms was from waterfowl infected terrestrial animals, but the strongest exposure measure was for non-waterfowl. ## Conclusion Among the potential wild bird or animal routes of transmission to cattle, we identified the highest exposure to non-waterfowl avian species. In addition, a significant interaction with poultry and dairy farm outbreaks was observed, indicating a possible bidirectional spread between cows and poultry. This highlights the threat of A/H5N1 spread through newly infected non-waterfowl wild bird species, which is a departure from the historical spread by waterfowl. This shift from waterfowl to non-waterfowl wild bird species in the spread of A/H5N1 to mammals has not been formally documented before. Our data supports such a shift as a factor in seeding the dairy farm outbreaks. However, additional factors such as livestock trade, poultry litter feed and contaminated milking machinery likely contributed to amplifying the epidemic. Over the past four years, the epicenter of AIV has shifted from Asia to Europe and the Americas, possibly due to new bird species being infected, therefore the risk of a human pandemic emerging is greatest in these new regions [17]. This makes it imperative to understand the rapid spread in farmed ruminants in the US. ## Supporting information Supplementary [[supplements/306785_file08.docx]](pending:yes) ## Data Availability All data produced are available online: [https://doi.org/10.6084/m9.figshare.25728336](https://doi.org/10.6084/m9.figshare.25728336) * Received May 2, 2024. * Revision received May 2, 2024. * Accepted May 4, 2024. * © 2024, Posted by Cold Spring Harbor Laboratory The copyright holder for this pre-print is the author. All rights reserved. The material may not be redistributed, re-used or adapted without the author's permission. ## References 1. 1.Food and Agriculture Organization of the United Nations. Global Avian Influenza Viruses with Zoonotic Potential situation update. Bird species affected by H5Nx HPAI 2024 2024-02-19 [cited 2024; Available from: [https://www.fao.org/animal-health/situation-updates/global-aiv-with-zoonotic-potential/bird-species-affected-by-h5nx-hpai/en](https://www.fao.org/animal-health/situation-updates/global-aiv-with-zoonotic-potential/bird-species-affected-by-h5nx-hpai/en). 2. 2.Animal and Plant Health Inspection Service, USDA Confirms Highly Pathogenic Avian Influenza in Dairy Herd in Idaho. 2024. 3. 3.Burrough, E., et al., Highly Pathogenic Avian Influenza A(H5N1) Clade 2.3.4.4b Virus Infection in Domestic Dairy Cattle and Cats, United States, 2024. Emerging Infectious Disease journal, 2024. 30(7). 4. 4.Cohen, J., Bird flu may be spreading in cows via milking and herd transport. Science, 2024. 5. 5.World Organisation for Animal Health, WAHIS. 2024. 6. 6.U.S. DEPARTMENT OF AGRICULTURE Animal and Plant Health Inspection Service, Detections of Highly Pathogenic Avian Influenza in Wild Birds. 2024. 7. 7.1. U.S.C. Bureau United States Census Bureau, Cartographic Boundary Files, U.S.C. Bureau, Editor. 2023. 8. 8.U.S. Geological Survey (USGS) - Gap Analysis Project (GAP), Abert’s Squirrel (Sciurus aberti) mABSQx\_CONUS\_2001v1 Habitat Map: U.S. Geological Survey data release. 2018. 9. 9.U.S. Geological Survey (USGS) - Gap Analysis Project (GAP), American Marten (Martes americana) mAMMAx\_CONUS\_2001v1 Habitat Map: U.S. Geological Survey data release. 2018. 10. 10.U.S. Geological Survey (USGS) - Gap Analysis Project (GAP), Coyote (Canis latrans) mCOYOx\_CONUS\_2001v1 Habitat Map: U.S. Geological Survey data release. 2018. 11. 11.U.S. Geological Survey (USGS) - Gap Analysis Project (GAP), Fisher (Martes pennanti) mFISHx\_CONUS\_2001v1 Habitat Map: U.S. Geological Survey data release. 2018. 12. 12.U.S. Geological Survey (USGS) - Gap Analysis Project (GAP), Red Fox (Vulpes vulpes) mREFOx\_CONUS\_2001v1 Habitat Map: U.S. Geological Survey data release. 2018. 13. 13.U.S. Geological Survey (USGS) - Gap Analysis Project (GAP), Raccoon (Procyon lotor) mRACCx\_CONUS\_2001v1 Habitat Map: U.S. Geological Survey data release. 2018. 14. 14.U.S. Geological Survey (USGS) - Gap Analysis Project (GAP), Striped Skunk (Mephitis mephitis) mSTSKx\_CONUS\_2001v1 Habitat Map: U.S. Geological Survey data release. 2018. 15. 15.U.S. Geological Survey (USGS) - Gap Analysis Project (GAP), Virginia Opossum (Didelphis virginiana) mVIOPx\_CONUS\_2001v1 Habitat Map: U.S. Geological Survey data release. 2018. 16. 16.Bui, C.M., L. Gardner, and C.R. MacIntyre, Highly pathogenic avian influenza virus, Midwestern United States. Emerging infectious diseases, 2016. 22(1): p. 138. 17. 17.Xie, R., et al., The episodic resurgence of highly pathogenic avian influenza H5 virus. Nature, 2023. 622(7984): p. 810–817.