Skip to main content
NCBI home page
PLOS One logoLink to PLOS One
. 2025 Sep 5;20(9):e0319880. doi: 10.1371/journal.pone.0319880

Genetic data and meteorological conditions suggesting windborne transmission of H5N1 high-pathogenicity avian influenza between commercial poultry outbreaks

Alexander Nagy 1,*, Lenka Černíková 1, Kamil Sedlák 1
Editor: Muhammad Ahmad,2
PMCID: PMC12412978  PMID: 40911554

Abstract

Understanding the transmission routes of high-pathogenicity avian influenza (HPAI) is crucial for developing effective control measures to prevent its spread. In this context, windborne transmission, the idea that the virus could travel through the air over considerable distances, is a contentious concept, and documented cases have been rare. Here, though, we provide genetic evidence supporting the feasibility of windborne transmission. During the 2023−24 HPAI season, molecular surveillance identified identical H5N1 strains among a cluster of unrelated commercial farms about 8 km apart in the Czech Republic. The episode started with the abrupt mortality of fattening ducks on one farm. This was followed by disease outbreaks at two nearby high-biosecurity chicken farms. Using genetic, epizootiological, meteorological and geographical data, we reconstructed a mosaic of events strongly suggesting wind was the most probable mechanism of infection transmission between poultry in at least two independent cases. By aligning the genetic and meteorological data with critical outbreak events, we determined the most likely time window during which the transmission occurred and inferred the sequence of infected houses at the recipient sites. Our results suggest that the contaminated plume emitted from the infected fattening duck farm was the critical medium of HPAI transmission, rather than the dust generated during depopulation. Furthermore, our results also strongly implicate the role of confined mechanically-ventilated buildings with high population densities in facilitating windborne transmission and propagating virus concentrations below the minimum infectious dose at the recipient sites. These findings underscore the importance of considering windborne spread in future outbreak mitigation strategies.

Introduction

Poultry farms are important emitters of air pollutants into the environment [1,2]. The airborne particles they emit consist of a complex mix of gases, liquid droplets, and organic and inorganic matter [3]. Larger particles, visible to the naked eye, settle on surfaces as dust. The finest particles and droplets, referred to as particulate matter (PM), levitate in the air for extended periods and include size fractions that are inhalable and respirable.

During an outbreak of HPAI, viral particles were regularly detected in the air and dust inside poultry houses or and markets [4–10]. Although airborne transmission was not considered the primary mode of spread in poultry [11,12], contaminated air may contribute to the spread of the disease [13–15]. The question is to what extent does this occur.

Wind has long been considered a plausible factor in the long-distance transmission of the avian influenza A virus (IAV) [16]. However, field reports on this mode of spread are rare and air-sampling studies conducted around poultry or pig houses did not support long-distance airborne transmission. Such studies reliably detected IAV only up to ~150 meters away from infected farms, but with low viral loads and positivity rates [4,5,7,8,10,17], and even fewer isolated viruses [4,5,17]. This suggests that virus load in the air was inversely proportional to the distance covered [18,19]. Furthermore, the aerosol is a dynamic system where the interactions of environmental stressors, such as sunlight and ultraviolet radiation, temperature, relative humidity, and evaporation, with dispersed chemical and organic components, may rapidly inactivate IAV particles [20]. Consequently, with increasing distance, the contaminated plume emitted from a farm with an ongoing IAV outbreak is progressively diluted, and the proportion of infectious virions decreases.

Despite this, some reports proposed that windborne spread played a significant role in IAV transmission over longer distances under suitable weather conditions. In studies of the severe H7N7 HPAI outbreak in the Netherlands in 2003, one study estimated that wind spread accounted for 18% [19] and another 24% [21] of transmission events up to 25 km. Similarly, during the 2007 equine H3N8 influenza outbreak in Australia, 81% of infections within a cluster of 437 horse farms were attributed to windborne spread over a distance of 1−2 km [22]. Around the same time, the serological screening of turkeys in Minnesota in 2007−08 revealed that turkey premises within a 1.9 km radius of swine farms were most likely to test seropositive for H3N2 and H1N1 IAVs, thereby suggesting windborne transmission [23]. And during the 2014−15 multistate H5N2 HPAI outbreak in the USA, it was estimated that up to 39% of farms in Iowa alone could have experienced windborne infection within a radius of 8.5 km [24]. However, the disparity between air-sampling studies and transmission data raises critical questions about the potential windborne spread of avian IAV—specifically, the distances it can travel and whether even trace concentrations in the air can trigger infection in poultry flocks. Genetic data may prove crucial to answering such questions, but studies of windborne IAV transmission incorporating such data remain sparse. Batalie et al. [25] utilized partial genomic data to trace the transmission pathways of H7N7 HPAI during the 2003 Netherlands outbreak. For the same outbreak, Ypma et al. [19] used statistical correlations between the inferred routes of spread and weather conditions as genetic evidence of windborne transmission. Overall, though, there is a significant lack of genetic data mapped to the prevailing weather conditions for such outbreaks.

In this study, we present empirical genetic evidence supporting the windborne transmission of the H5N1 HPAI virus over a distance of 8 km. Molecular surveillance of HPAI outbreaks in the Czech Republic during the 2023−24 season identifies identical H5N1 strains in a cluster of unrelated commercial farms. This evidence, combined with a detailed epizootiological investigation and strong correlation with weather conditions, suggests that wind can transport infectious virus particles over substantial distances and, thereby, facilitate the spread of HPAI between poultry farms. These findings underscore the importance of considering windborne transmission as a significant factor in the management and control of HPAI outbreaks.

Materials and methods

Whole-genome sequencing

The presented study does not involve any animal experiments. All specimens were collected by regional veterinary inspectors of the State Veterinary Administration of the Czech Republic as part of the HPAI outbreak management and not specifically for this study. The specimens were investigated in the State Veterinary Institute Prague, Czech Republic. Animal protection and welfare were supervised by the State Veterinary Administration of the Czech Republic. This study adheres to the STROBE reporting guidelines [26].

Pooled organ suspensions, taken individually from each duck carcass collected on farm B and each chicken carcass collected from farms C1 and C2, were homogenized in RNA Later solution (Invitrogen) and used for the detection and typing of H5N1 HPAI, as well as for whole-genome sequencing. Total nucleic acids were extracted from 200 µl supernatants by MagNA Pure 24 (Total NA Isolation Kit), MagNA Pure 96 (DNA and Viral NA Small Volume Kit) (both from Roche) or MaxWell RSC (Maxwell® RSC Whole Blood DNA Kit, Promega) and eluted into 50 µl (MagNA Pure 24/96) and 60 µl (MaxWell RSC), respectively. The H5N1 genome was amplified in a two-step RT-PCR protocol. Reverse transcription was conducted using SuperScript III or SuperScript IV RT kits (Thermo Fisher Scientific) with a universal-tagged forward primer TTTCTGTTGGTGCTGATATTGCAGCRAAAGCAGG and utilizing a quarter of the recommended enzyme amount. Amplification was carried out with Q5 HiFi DNA polymerase (New England Biolabs; primers available on request) in a final volume of 25 µl (20 µl of master mix and 5 µl of RT reaction). The sequencing libraries were purified by SPRIselect magnetic beads (Beckman-Coulter) and quantified by QIAxpert (Qiagen). End preparation, native barcoding and sequencing adapter ligation were performed by Native Barcoding Kit (SQK-LSK-114.96, Oxford Nanopore Technologies (ONT)), according to the manufacturer’s instructions. Sequencing runs were carried out on the ONT MinION Mk1B using R10.4.1 flow cells and operated via MinKNOW software (v2.3.04.3 or higher, ONT). Basecalling was conducted with Dorado software (v0.5.0 or higher, ONT) using the super accurate basecalling model (dna_r10.4.1_e8.2_400bps_sup@v4.3.0 or higher model). The run was monitored in real time by the RAMPART (Read Assignment, Mapping and Phylogenetic Analysis in Real Time) module of the ARTIC bioinformatic pipeline [27] set to the concatenated H5N1 genome as a reference. Demultiplexing and adapter and barcode trimming were performed by Porechop and Dorado. Consensus sequences were generated using various tools: Bcftools mpileup [28], NGSpeciesID [29], Decona [30], Amplicon sorter [31], and ViralConsensus [32]. The resulting consensus sequences were aligned using MAFFT (Multiple Alignment using Fast Fourier Transform) [33] and manually curated against the corresponding SAM files to obtain a final consensus genome. The whole-genome sequencing was carried out in five independent runs using different primer sets. For further details, please refer to the Supporting Information. Finally, the obtained genomes were annotated using the Influenza Virus Sequence Annotation Tool [34]. All H5N1 genomes were submitted to the GISAID EpiFlu [35] database with the accession codes listed in Table 1.

Table 1. Chronological list of H5N1 strains detected in the affected farms.

Strain ID Collection Date Farm/House Name Accession no.
2268/24 2024-02-04 B A/domestic_duck/Czech_Republic/2268-1orig/2024 EPI_ISL_18928997
A/domestic_duck/Czech_Republic/2268-1/2024 EPI_ISL_19338428
A/domestic_duck/Czech_Republic/2268-2orig/2024 EPI_ISL_18928998
A/domestic_duck/Czech_Republic/2268-2/2024 EPI_ISL_19338429
A/domestic_duck/Czech_Republic/2268-3/2024 EPI_ISL_19338430
A/domestic_duck/Czech_Republic/2268-4/2024 EPI_ISL_19338431
A/domestic_duck/Czech_Republic/2268-5/2024 EPI_ISL_18928999
A/domestic_duck/Czech_Republic/2268-7/2024 EPI_ISL_18929000
A/domestic_duck/Czech_Republic/2268-9/2024 EPI_ISL_19338432
2894/24 2024-02-12 C1A A/chicken/Czech_Republic/2894-1/2024 EPI_ISL_19338427
A/chicken/Czech_Republic/2894-4/2024 EPI_ISL_18946721
A/chicken/Czech_Republic/2894-5/2024 EPI_ISL_18946722
2895/24 2024-02-12 C2B A/chicken/Czech_Republic/2895-1/2024 EPI_ISL_18946713
A/chicken/Czech_Republic/2895-2/2024 EPI_ISL_18946712
A/chicken/Czech_Republic/2895-3/2024 EPI_ISL_18946716
A/chicken/Czech_Republic/2895-4/2024 EPI_ISL_18946715
A/chicken/Czech_Republic/2895-5/2024 EPI_ISL_18946714
2896/24 2024-02-13 Backyard A/chicken/Czech_Republic/2896-1orig/2024 EPI_ISL_18946728
A/chicken/Czech_Republic/2896-2orig/2024 EPI_ISL_18946730
A/chicken/Czech_Republic/2896-2/2024 EPI_ISL_18946732
A/chicken/Czech_Republic/2896-3/2024 EPI_ISL_18946729
A/chicken/Czech_Republic/2896-4/2024 EPI_ISL_18946731
A/chicken/Czech_Republic/2896-5/2024 EPI_ISL_18946727
3299/24 2024-02-19 C1B A/chicken/Czech_Republic/3299-1/2024 EPI_ISL_19000406
3529/24 2024-02-22 C1B A/chicken/Czech_Republic/3529/2024 EPI_ISL_19044783
3627/24 2024-02-23 C1B A/chicken/Czech_Republic/3627-3/2024 EPI_ISL_19000408
3549/24 2024-02-26 C1B A/chicken/Czech_Republic/3549-1/2024 EPI_ISL_19000411
A/chicken/Czech_Republic/3549-3/2024 EPI_ISL_19338424
A/chicken/Czech_Republic/3549-4/2024 EPI_ISL_19044780
A/chicken/Czech_Republic/3549-7/2024 EPI_ISL_19338422
A/chicken/Czech_Republic/3549-8/2024 EPI_ISL_19338426
A/chicken/Czech_Republic/3549-9/2024 EPI_ISL_19044781
A/chicken/Czech_Republic/3549-10/2024 EPI_ISL_19044782
A/chicken/Czech_Republic/3549-11/2024 EPI_ISL_19000412
A/chicken/Czech_Republic/3549-12/2024 EPI_ISL_19338425
A/chicken/Czech_Republic/3549-13/2024 EPI_ISL_19338423
A/chicken/Czech_Republic/3549-14/2024 EPI_ISL_19000413
A/chicken/Czech_Republic/3549-15/2024 EPI_ISL_19000414
Open in a new tab

Sequence and phylogenetic analysis

The concatenated H5N1 genomic sequences were aligned using MAFFT, with alignment trimming and format conversion (Phylip full names and padded) by AliView [36]. A maximum likelihood (ML) tree (IQ-TREE multicore version 2.2.0-beta for Linux 64-bit [37]; best fit moder GTR + F + G4 according to the Bayesian information criterion) was calculated from the concatenated genomic segments. Segment concatenation was performed using the Union program from EMBOSS [38] in the following order: PB2, PB1, PA, H5, NP, N1, MP and NS. A median-joining network was calculated using PopART (Population Analysis with Reticulate Trees) [39,40] software with the ε set to zero.

Meteorological data

Meteorological data (wind direction and speed, average temperature, and relative humidity), were obtained from three Czech Hydrometeorological Institute monitoring stations with the following IDs: B7TREB01, B2VMEZ01, and B2SEDC01.

Results

HPAI outbreak cluster of interest

On 4 February 2024, a commercial farm with ~50,000 fattening ducks (hereafter referred to as B) experienced a sudden increase in mortality affecting ~800 ducks in two of its eight houses. Within two days, the mortality rate increased to ~5,000 deaths. The ducks were kept in identical houses on a littered floor with natural ventilation. The farm was situated on the bank of a ~ 15 ha lake [41] – a significant landscape element relatively frequently visited by wild birds. The farm lacked enhanced fencing or a protective net, so it was easily accessible to wild birds. To control the HPAI outbreak, the entire flock was depopulated between 7 and 9 February, and a 3 km protection zone and 10 km surveillance zone was established.

One week later, on 12 February 2024, H5N1 HPAI was detected in two chicken farms (hereafter referred to as C1 and C2) owned by another company, C. These farms were situated within the surveillance zone and, together with B, formed three vertices of an imaginary triangle (Fig 1). Company C is internationally recognized for its unique accredited breeding program, which supplies a variety of chicken hybrids with different plumage and egg colors to customers worldwide. This activity is conducted in accordance with high biosecurity standards.

Fig 1. Map of outbreak locations.

Fig 1

Open in a new tab

Commercial poultry farms are marked by red dots and a backyard poultry farm by an orange dot. Czech Hydrometeorological Institute monitoring stations are indicated by blue numbers. Reprinted with permission from the Czech Office for Surveying, Mapping, and Cadaster under a CC BY license, original copyright 2025. An interactive version of the map is available online at: https://www.google.com/maps/d/edit?mid=1rlhK5gts6bwoNhAsamBt54wsvr8iBvc&ll=49.24410201219534%2C16.024409179698264&z=13.

Both C1 and C2 consisted of two houses, with birds kept in indoor cages without litter. C1 housed ~24,500 birds: C1A ~ 11,500 laying hens and C1B ~ 13,000 birds of parent and grandparent breeds. C2 housed ~45,000 birds: C2A ~ 5,500 birds and C2B ~ 39,500 birds intended for local sale. Both farms used treated water from their own wells and feed purchased from various companies. All houses were equipped with a negative pressure tunnel ventilation system (Fig 2). Air was drawn in through the inlet grille to capture larger mechanical impurities, but it was not subjected to any further treatment. Since the farms were located within the surveillance zone, the Viusid Vet nutritional supplement (Catalysis, Spain) was administered starting from 7 February 2024 to enhance chicken immunity .

Fig 2. Infected premises.

Fig 2

Open in a new tab

Layout of affected houses on farms B (A), C1 (B), and C2 (C). In figures B and C, tunnel ventilation system airflow is indicated with orange arrows for inflow and blue arrows for outflow. Reprinted with permission from the Czech Office for Surveying, Mapping, and Cadaster under a CC BY license, original copyright 2025.

H5N1 HPAI first manifested itself in C1A and C2B with an observed slight increase in mortality. For at least a week before HPAI was confirmed, both C1 and C2 experienced a gradual, though insignificant, decrease in water and feed consumption, with this decline initially attributed to the administration of Viusid. It is noteworthy that in the affected houses, the infection and subsequent mortality started in the areas closest to the air inlets. The situation in C1 was particularly serious as C1B contained ~2,000 birds representing a crucial genetic reserve for the breeding program. To protect this indigenous gene pool, C1A was immediately depopulated. However, on 19 February, the virus was also detected in C1B. In an effort to limit the culling of this valuable population the progression of the infection was monitored (Fig 3). Ultimately, though, this house also had to be depopulated. C2A remained unaffected throughout.

Fig 3. Timeline of events.

Fig 3

Open in a new tab

Timeline of key dates related to the outbreak, including outbreak identification, depopulation of affected premises, and specific sampling events carried out at the locations.

Molecular epizootiology

To determine the relationships between the affected farms, whole-genome sequencing and phylogenetic analysis were performed. A total of 38 H5N1 genomes were obtained (Table 1), including nine genomes from B, three from C1A, and five from C2B, each collected during a single sampling occasion. Another 15 were obtained from C1B during the course of four consecutive sampling events. An additional six genomes were obtained from the samples collected from the outbreak among backyard poultry near B.

Phylogenetic analysis showed that all H5N1 strains belonged to genotype EA-2024-DI [42] and formed a common sub-clade (Fig 4A). Interestingly, the H5N1 viruses from B and C1A and the index strain from C1B did not show farm-specific clustering but were placed in statistically unsupported branches. In particular, three H5N1 strains from B were 100% identical at the nucleotide level to all strains from C1A and the index strain of C1B. In contrast, H5N1 strains from C2B and the later strains from C1B showed clear farm-specific clustering. Finally, H5N1 viruses from backyard poultry, identified nine days apart near B (Fig 1), showed a close relationship to the B, C1A and index C1B strains.

Fig 4. Phylogenetic analysis.

Fig 4

Open in a new tab

A maximum likelihood tree (A) was constructed using concatenated genomic data. Each branch is annotated with bootstrap values (derived from 1000 replicates) expressed as percentages. Genomes with 100% sequence identity are indicated by a square bracket. Median-joining network (B) of the same dataset. The sizes of the vertices are proportional to the number of genomes included. Nucleotide differences are shown as hatch marks on the edges.

To further elucidate the relationship between B, C1 and C2, a median-joining network was calculated. This revealed that all but one of the H5N1 strains from B, all strains from C1A and the index virus from C1B were grouped into a single central node (Fig 4B). From this node, three main branches emerged: the first two represented C1B- and C2B-specific clusters, and the third grouped H5N1 strains from backyard poultry. These results are consistent with those obtained from the genomic tree. The genetic overlap between B and C suggests a common origin, with B likely being the primary source from which the disease spread to C1 and C2.

Tracing the potential infection routes

Companies B and C were only about 8 km apart and active in different market segments. Interviews with the veterinary inspector and general manager of C revealed the complete absence of any interaction between the companies. The companies even used different third parties involved in feed replenishment, waste disposal, and the transport of carcasses to rendering plants. All farms used their own supplies of well water. In addition, the employees of C were not allowed to keep their own poultry. Therefore, in view of the biosecurity measures in place, the possibility of human-associated secondary spread from B to C1 and C2, or even between C1 and C2 was excluded.

Because C1 and C2 were located in an area without significant water bodies, H5N1 transmission from wild birds also appeared highly unlikely. The C1 and C2 were enclosed in a well-maintained clean environment with reinforced fencing (Fig 3) that effectively prevented the entry of rodents and small carnivores known to be susceptible to H5N1 infection [43–46]. Furthermore, the combination of the outbreak season and tunnel ventilation system ruled out the involvement of flies or other insects in infection transmission.

Thus, the genetic data and field investigations suggest an abiotic route of H5N1 HPAI transmission from B to C1 and C2.

Local landscape and meteorological conditions

All farms were situated at similar elevations: B at 438 m, C1 at 448 m and C2 at 408 m (Fig 5A). The landscape between the farms consisted mainly of open countryside, fields, meadows and sparse tree cover with the highest local hill being 510 m in elevation. Moreover, there were direct routes between these locations that do not require significant traversal of higher land. Therefore, the local landscape variation was slight and unlikely to hinder windborne virus transmission.

Fig 5. Landscape and meteorological conditions.

Fig 5

Open in a new tab

Landscape (A) characterizes the ruggedness of the terrain between the donor and recipient farms according to elevation in meters above sea level. Average temperature (B), wind speed (C) and wind direction (D) were all recorded at hourly intervals. The entire graph shows a summary of data recorded at three Czech Hydrometeorological Institute stations. Temperature was recorded at two meters above ground, and wind speed/ direction at ten meters above ground. Time data from 0:00 to 23:00 h are shaded in a gradient of grey. The secondary y-axis (C) indicates the time to reach recipient farm C1 at a given wind speed. The numerical values used to construct the graphs, as well as detailed data recorded individually for each measuring station at ten-minute intervals, are provided in the Supporting Information file.

The prevailing weather conditions during the critical period included extensive cloud cover, minimal precipitation and intermittent sunshine. Humidity levels averaged from 77 to 81%. Apart from the night of 7–8 February, temperatures did not fall below 6 °C, peaking at 8–11 °C between 4 and 11 February. Wind conditions were remarkable between 4 and 7 February (Fig 5B–5D), with continuous wind from the west or southwest (250–300 degrees). Generally, wind speeds were around 4 m/s (~14 km/h), but there were periods during which they increased to over 6 m/s (~22 km/h) on 4 February and reached 8–10 m/s (~29–34 km/h) on 5 February. From 7 February the wind velocity started to decrease, and between 8 and 9 February there was minimal wind with directions shifting frequently, including wind currents moving away from the recipient farms. Overall, weather conditions showed a strong correlation with the transmission route inferred from the genetic data, with the optimal period for infection spread identified as noon on 4 February to midnight on 5 February.

Synthesis of key findings suggests windborne transmission

The H5N1 HPAI outbreak started at B on 4 February 2024. Although the exact source of the infection remains unidentified, it most likely originated from mallards at a nearby pond. A nearby backyard flock was infected with a genetically closely related H5N1 virus strain approximately nine days later.

Alignment of the events timeline with wind direction and speed suggest a six-day window (from outbreak identification on 4 February to depopulation completion on 9 February) during which the infection was spread from B to C1 and C2. For this window, the theory of windborne spread, which began with an order of magnitude increase in the number of infected birds at B, is supported by four key observations: i) the relative proximity of B to C, ii) ideal weather conditions, iii) slow disease progression at C with the location of the affected birds in sections closest to the air inlets [24], and iv) genetic identity between the H5N1 strains from the donor and recipient farms. Moreover, all possible alternative routes of infection during this period were excluded by our field investigation.

Thus, the C1A and index C1B H5N1 strains, which were genetically identical to B, were most likely seed strains carried by the wind. These strains then evolved locally, forming a cluster of farm-specific variants [25] that were detected in C2B and later in C1B. While the critical early phase of infection was not captured for C2B, the timeline of events and H5N1 sequence diversity suggest infection in the following order: C2B, C1A and C1B. However, it remains unclear whether the infection of C1B was an independent event or a secondary infection (Fig 2).

Discussion

Understanding the routes of IAV transmission is crucial for developing strategies to control its spread and to prevent both inter-species and cross-species infections. While the windborne spread of IAV is considered a possible route for infecting poultry, pigs and even horses, the unpredictable nature of infection outbreaks, the specific weather conditions required, and the lack of samples from crucial transmission time points make this mode of spread difficult to investigate. Consequently, our ability to infer windborne spread in the field has largely been limited to correlation or simulation studies performed retrospectively [21–24,47,48]. Moreover, there is a lack of robust genetic data to support these inferences.

In the current study, we have supported previous findings [19,25] by providing direct genetic evidence for the windborne transmission of H5N1 HPAI based on whole-genome sequence data. Our analysis indicated that the H5N1 genomic identity between the unrelated poultry farms could only be attributed to an abiotic route. The combination of landscape, meteorological, clinical and epizootiological factors pointed to wind-mediated spread as the most likely explanation.

However, as a retrospective study, our investigation faced several limitations inherent to its design. Notably, the absence of air or dust specimens prevented direct confirmation of windborne transmission during the outbreak. While the genetic and epidemiological evidence strongly supports this route, the inability to collect real-time environmental samples represents a key constraint. This limitation highlights the need for future studies to employ real-time sampling methodologies to corroborate transmission mechanisms and enhance the robustness of evidence in similar outbreaks.

The windborne spread of H5N1 HPAI in the cluster of interest raises significant questions about the infection dynamics of housed poultry, particularly with respect to the role of tunnel ventilation systems in facilitating such infection. Tunnel ventilation systems in poultry houses act like high-volume air samplers; powered by exhaust fans, they create a negative pressure that pulls in ambient air [49]. The maximum ventilation capacity, estimated in cubic meters per hour per kilogram of body weight, usually ranges between 4.5 and 7 m3 [50]. For a poultry house with 10,000 chickens, each weighing 1 kg, this results in the influx of a colossal volume of 45,000–70,000 m3 of ambient air per hour, which is equivalent to a cube with an edge ranging from 34 to 41 meters. This enables a significant amount of PM to enter the chicken house. During stormy periods, the influx can be even greater [51]. Consequently, active air ventilation not only provides housed chickens with a continuous influx of fresh ambient air but also exposes them to a considerable amount of PM from diverse sources [51], including that originating from wild birds [52]. Indeed, the prolonged exposure of immunologically naive chickens to contaminated air, even at IAV concentrations below the minimal infectious dose (MID), has been concluded to potentially increase the risk of airborne infection [24,48].

During the infection period, the ventilation rates in C1A, C1B and C2B were set to a staggering 154111, 134 228 and 257244 m3/h, respectively. The exposure of a dense population of chickens to such high air volumes might well have led to minute virus amounts being captured and subsequently enriched to levels sufficient to trigger illness. This was consistent with the observed slow disease progression at the recipient farms C. Thus, the specific settings of the poultry houses at recipient site C likely facilitated the wind-mediated spread of HPAI from the nearby source. Conversely, this also explained why other backyard poultry in villages located between the donor and recipient farms, as well as the poultry in C2A, were unaffected; they lacked sufficient population density and/or the booster effect of tunnel ventilation. Furthermore, the population-dependent and booster effects also aligned with the deduced order of infected farms, indicating that C2B, with the largest bird population and highest ventilation rate, was likely infected first. Therefore, while further studies are needed, it does appear that wind can transmit infectious HPAI particles over considerable distances and that densely populated tunnel-ventilated poultry houses can promote the propagation of sub-MID HPAI virus concentrations.

However, detecting viral particles in such minute concentrations is challenging. Simply put, at greater distances, virus amounts were below the detection limit of existing air samplers, making reliable estimation extremely difficult, if not impossible, as discussed by Zhao et al. [24]. In addition, given the invisibility of the plume and the unpredictability of air currents, measurements have to be performed at the right time and place under weather conditions conducive to windborne spread. Accordingly, and consistent with Zhao et al. [24], negative results from air samples collected at greater distances may not accurately reflect an IAV-negative air environment. Thus, the failure to detect IAV particles away from infected farms [4,5,7,8,10] must not be taken as evidence of the infeasibility of windborne spread. Indeed, when sampling was correlated with careful estimation of wind direction, IAV particles were possible to detect in air collected up to 1.5 and 2.1 km from affected farms, indicating true wind-mediated dispersal [17].

By aligning the timeline of events with wind velocity and a detailed knowledge of the clinical progression, we were able to infer with high confidence the time window during which the infection was transmitted by the wind from B. The period from outbreak identification to the end of depopulation (4‒9 February) left about six days during which the virus could have been carried to C1 and C2. Taking into account the wind velocity, with an almost constant easterly flow between 4 and 7 February, this window could be further reduced to four days. However, if the observed decrease in water and feed intake was considered to be the first sign of infection [53], then infection occurred between 4 and 5 February. This period corresponded to the rapid increase in the number of infected birds on the donor farm, which must have acted as a massive source of aerosolized virus particles. Additionally, between 4 and 5 February, the highest wind speeds were recorded, lasting approximately 36 hours—from noon on 4 February to midnight on 5 February—enabling the virus to reach C1 and C2 within 13–22 minutes Furthermore, correlating the observed genomic diversity with the sampling dates showed that C1 and C2 were apparently infected independently, highlighting the underlying infectious dynamics. On the other hand, between 8 and 9 February when the depopulation took place, there was minimal wind with directions shifting frequently, including wind currents moving away from the recipient farms. The only period with a slightly stronger wind blowing from the appropriate direction was on the morning of 7 February. However, this day was largely dedicated to preparatory and logistical work for depopulation which, according to the depopulation scheme, started in the afternoon in the first of the affected halls.

Overall, it seemed that both recipient farms were infected relatively very early after the outbreak on the donor farm and before it was depopulated. These findings stand in stark contrast to the conventional theory [54,55], which considers dust generated during depopulation as the primary risk factor for the secondary windborne spread of HPAI. Rather, in the presented case, PM that emerged before depopulation served as the virus-carrying medium, suggesting that a dense duck population housed under natural ventilation generated a sufficient amount of PM to facilitate airborne transmission.

Windborne transmission is a controversial concept in relation to the spread of avian influenza in commercial poultry. While some studies suggested that it could play a significant role, others considered the potential for airborne transmission low. The specific, often ambiguous, conditions required make the spread over kilometric distances seem counterintuitive. As a result, it is questionable whether the windborne route has been adequately considered during outbreak investigations, meaning that, compared with other transmission routes, it might well have been underestimated. However, when factors – such as suitable weather conditions, relative proximity to the source population, and densely populated tunnel-ventilated houses – align, windborne transmission becomes a feasible mode of spread over kilometric distances. This was demonstrated by at least two independent events observed in our study. In the European Union, air filtration in commercial poultry farms is not a standard practice, and current guidelines do not generally recommend it as a necessary measure [56]. This represents a potential gap in existing biosecurity systems. It is therefore worth considering the integration of airborne control technologies [57] with existing on-farm biosecurity procedures to effectively limit the introduction and spread of pathogens via the windborne route. Continued research and enhanced surveillance are essential to fully understand the risks associated with such transmission. The knowledge gained could then be used to develop more effective preparedness strategies and control measures, better equipping poultry farms to combat future outbreaks.

Supporting information

S1 Data. Supporting information.

(XLSX)

pone.0319880.s001.xlsx (366.3KB, xlsx)

Acknowledgments

We wish to thank Kristína Oreničová, B.Sc., Michaela Štěrbová, B.Sc., Lada Hofmannová, DVM, PhD, Ondřej Horák, DVM, PhD, Roman Maleček and all our colleagues at the State Veterinary Institute Prague and the State Veterinary Institute Jihlava for their excellent technical support in conducting the HPAI survey. We would also like to thank Marie Nováková, DVM, from the Regional Veterinary Administration of the Vysočina Region, and Pavel Dobrovolný, Managing Director of company C, for providing valuable information from the field. In addition, we are grateful for the bioinformatics assistance of Petr Buřič, M.Eng.

Data Availability

All relevant data are within the manuscript and its Supporting Information files.

Funding Statement

This work was funded by the National Agricultural Research Agency (NAZV) of the Czech Republic, project QL24010383. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

References

  • 1.Cambra-López M, Aarnink AJA, Zhao Y, Calvet S, Torres AG. Airborne particulate matter from livestock production systems: a review of an air pollution problem. Environ Pollut. 2010;158(1):1–17. doi: 10.1016/j.envpol.2009.07.011 [DOI] [PubMed] [Google Scholar]
  • 2.Le Bouquin S, Huneau-Salaün A, Huonnic D, Balaine L, Martin S, Michel V. Aerial dust concentration in cage-housed, floor-housed, and aviary facilities for laying hens. Poult Sci. 2013;92(11):2827–33. doi: 10.3382/ps.2013-03032 [DOI] [PubMed] [Google Scholar]
  • 3.Bist RB, Chai L. Advanced Strategies for Mitigating Particulate Matter Generations in Poultry Houses. Appl Sci. 2022;12(22):11323. doi: 10.3390/app122211323 [DOI] [Google Scholar]
  • 4.Jonges M, van Leuken J, Wouters I, Koch G, Meijer A, Koopmans M. Wind-Mediated Spread of Low-Pathogenic Avian Influenza Virus into the Environment during Outbreaks at Commercial Poultry Farms. PLoS One. 2015;10(5):e0125401. doi: 10.1371/journal.pone.0125401 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Torremorell M, Alonso C, Davies PR, Raynor PC, Patnayak D, Torchetti M, et al. Investigation into the Airborne Dissemination of H5N2 Highly Pathogenic Avian Influenza Virus During the 2015 Spring Outbreaks in the Midwestern United States. Avian Dis. 2016;60(3):637–43. doi: 10.1637/11395-021816-Reg.1 [DOI] [PubMed] [Google Scholar]
  • 6.Zhou J, Wu J, Zeng X, Huang G, Zou L, Song Y, et al. Isolation of H5N6, H7N9 and H9N2 avian influenza A viruses from air sampled at live poultry markets in China, 2014 and 2015. Euro Surveill. 2016;21(35):30331. doi: 10.2807/1560-7917.ES.2016.21.35.30331 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Scoizec A, Niqueux E, Thomas R, Daniel P, Schmitz A, Le Bouquin S. Airborne Detection of H5N8 Highly Pathogenic Avian Influenza Virus Genome in Poultry Farms, France. Front Vet Sci. 2018;5:15. doi: 10.3389/fvets.2018.00015 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Wei J, Zhou J, Cheng K, Wu J, Zhong Z, Song Y, et al. Assessing the risk of downwind spread of avian influenza virus via airborne particles from an urban wholesale poultry market. Build Environ. 2018;127:120–6. doi: 10.1016/j.buildenv.2017.10.037 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Filaire F, Lebre L, Foret-Lucas C, Vergne T, Daniel P, Lelièvre A, et al. Highly Pathogenic Avian Influenza A(H5N8) Clade 2.3.4.4b Virus in Dust Samples from Poultry Farms, France, 2021. Emerg Infect Dis. 2022;28(7):1446–50. doi: 10.3201/eid2807.212247 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.James J, Warren CJ, De Silva D, Lewis T, Grace K, Reid SM, et al. The Role of Airborne Particles in the Epidemiology of Clade 2.3.4.4b H5N1 High Pathogenicity Avian Influenza Virus in Commercial Poultry Production Units. Viruses. 2023;15(4):1002. doi: 10.3390/v15041002 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Spekreijse D, Bouma A, Koch G, Stegeman JA. Airborne transmission of a highly pathogenic avian influenza virus strain H5N1 between groups of chickens quantified in an experimental setting. Vet Microbiol. 2011;152(1–2):88–95. doi: 10.1016/j.vetmic.2011.04.024 [DOI] [PubMed] [Google Scholar]
  • 12.Tsukamoto K, Imada T, Tanimura N, Okamatsu M, Mase M, Mizuhara T, et al. Impact of Different Husbandry Conditions on Contact and Airborne Transmission of H5N1 Highly Pathogenic Avian Influenza Virus to Chickens. Avian Diseases. 2007;51(1):129–32. doi: 10.1637/0005-2086(2007)051[0129:iodhco]2.0.co;2 [DOI] [PubMed] [Google Scholar]
  • 13.Yee KS, Carpenter TE, Farver TB, Cardona CJ. An evaluation of transmission routes for low pathogenicity avian influenza virus among chickens sold in live bird markets. Virology. 2009;394(1):19–27. doi: 10.1016/j.virol.2009.08.017 [DOI] [PubMed] [Google Scholar]
  • 14.Spekreijse D, Bouma A, Koch G, Stegeman A. Quantification of dust-borne transmission of highly pathogenic avian influenza virus between chickens. Influenza Other Respir Viruses. 2013;7(2):132–8. doi: 10.1111/j.1750-2659.2012.00362.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Bertran K, Balzli C, Kwon Y-K, Tumpey TM, Clark A, Swayne DE. Airborne Transmission of Highly Pathogenic Influenza Virus during Processing of Infected Poultry. Emerg Infect Dis. 2017;23(11):1806–14. doi: 10.3201/eid2311.170672 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Davison S, Eckroade RJ, Ziegler AF. A review of the 1996-98 nonpathogenic H7N2 avian influenza outbreak in Pennsylvania. Avian Dis. 2003;47(3 Suppl):823–7. doi: 10.1637/0005-2086-47.s3.823 [DOI] [PubMed] [Google Scholar]
  • 17.Corzo CA, Culhane M, Dee S, Morrison RB, Torremorell M. Airborne detection and quantification of swine influenza a virus in air samples collected inside, outside and downwind from swine barns. PLoS One. 2013;8(8):e71444. doi: 10.1371/journal.pone.0071444 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Boender GJ, Hagenaars TJ, Bouma A, Nodelijk G, Elbers ARW, de Jong MCM, et al. Risk maps for the spread of highly pathogenic avian influenza in poultry. PLoS Comput Biol. 2007;3(4):e71. doi: 10.1371/journal.pcbi.0030071 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Ypma RJF, Jonges M, Bataille A, Stegeman A, Koch G, van Boven M, et al. Genetic data provide evidence for wind-mediated transmission of highly pathogenic avian influenza. J Infect Dis. 2013;207(5):730–5. doi: 10.1093/infdis/jis757 [DOI] [PubMed] [Google Scholar]
  • 20.Longest AK, Rockey NC, Lakdawala SS, Marr LC. Review of factors affecting virus inactivation in aerosols and droplets. J R Soc Interface. 2024;21(215):18. doi: 10.1098/rsif.2024.0018 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Ssematimba A, Hagenaars TJ, de Jong MCM. Modelling the wind-borne spread of highly pathogenic avian influenza virus between farms. PLoS One. 2012;7(2):e31114. doi: 10.1371/journal.pone.0031114 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Davis J, Garner MG, East IJ. Analysis of local spread of equine influenza in the Park Ridge region of Queensland. Transbound Emerg Dis. 2009;56(1–2):31–8. doi: 10.1111/j.1865-1682.2008.01060.x [DOI] [PubMed] [Google Scholar]
  • 23.Corzo CA, Gramer M, Lauer D, Davies PR. Prevalence and risk factors for H1N1 and H3N2 influenza A virus infections in Minnesota turkey premises. Avian Dis. 2012;56(3):488–93. doi: 10.1637/10037-121211-Reg.1 [DOI] [PubMed] [Google Scholar]
  • 24.Zhao Y, Richardson B, Takle E, Chai L, Schmitt D, Xin H. Airborne transmission may have played a role in the spread of 2015 highly pathogenic avian influenza outbreaks in the United States. Sci Rep. 2019;9(1):11755. doi: 10.1038/s41598-019-47788-z [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Bataille A, van der Meer F, Stegeman A, Koch G. Evolutionary analysis of inter-farm transmission dynamics in a highly pathogenic avian influenza epidemic. PLoS Pathog. 2011;7(6):e1002094. doi: 10.1371/journal.ppat.1002094 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP, et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. J Clin Epidemiol. 2008;61(4):344–9. doi: 10.1016/j.jclinepi.2007.11.008 [DOI] [PubMed] [Google Scholar]
  • 27.Arctic Network. Available from: https://web.archive.org/web/20221021094307/ https://artic.network
  • 28.Danecek P, Bonfield JK, Liddle J, Marshall J, Ohan V, Pollard MO, et al. Twelve years of SAMtools and BCFtools. Gigascience. 2021;10(2):giab008. doi: 10.1093/gigascience/giab008 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Sahlin K, Lim MCW, Prost S. NGSpeciesID: DNA barcode and amplicon consensus generation from long‐read sequencing data. Ecol Evol. 2021;1–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Oosterbroek S, Doorenspleet K, Nijland R, Jansen L. Decona: From demultiplexing to consensus for Nanopore amplicon data. ARPHA Conference Abstracts. 2021;4:e65029. doi: 10.3897/aca.4.e65029 [DOI] [Google Scholar]
  • 31.Vierstraete AR, Braeckman BP. Amplicon_sorter: A tool for reference-free amplicon sorting based on sequence similarity and for building consensus sequences. Ecol Evol. 2022;12(3):e8603. doi: 10.1002/ece3.8603 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Moshiri N. ViralConsensus: a fast and memory-efficient tool for calling viral consensus genome sequences directly from read alignment data. Bioinformatics. 2023;39(5):btad317. doi: 10.1093/bioinformatics/btad317 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Katoh K, Rozewicki J, Yamada KD. MAFFT online service: multiple sequence alignment, interactive sequence choice and visualization. Brief Bioinform. 2019;20(4):1160–6. doi: 10.1093/bib/bbx108 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Hatcher EL, Zhdanov SA, Bao Y, Blinkova O, Nawrocki EP, Ostapchuck Y, et al. Virus Variation Resource - improved response to emergent viral outbreaks. Nucleic Acids Res. 2017;45(D1):D482–90. doi: 10.1093/nar/gkw1065 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Shu Y, McCauley J. GISAID: Global initiative on sharing all influenza data - from vision to reality. Euro Surveill. 2017;22(13):30494. doi: 10.2807/1560-7917.ES.2017.22.13.30494 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Larsson A. AliView: a fast and lightweight alignment viewer and editor for large datasets. Bioinformatics. 2014;30(22):3276–8. doi: 10.1093/bioinformatics/btu531 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Trifinopoulos J, Nguyen L-T, von Haeseler A, Minh BQ. W-IQ-TREE: a fast online phylogenetic tool for maximum likelihood analysis. Nucleic Acids Res. 2016;44(W1):W232-5. doi: 10.1093/nar/gkw256 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Rice P, Longden I, Bleasby A. EMBOSS: The European Molecular Biology Open Software Suite. Trends Genet. 2000;16:276–7. [DOI] [PubMed] [Google Scholar]
  • 39.PopART. [cited 29 November 2024]. Available from: https://popart.maths.otago.ac.nz/download/
  • 40.Bandelt HJ, Forster P, Röhl A. Median-joining networks for inferring intraspecific phylogenies. Mol Biol Evol. 1999;16(1):37–48. doi: 10.1093/oxfordjournals.molbev.a026036 [DOI] [PubMed] [Google Scholar]
  • 41.Opatsky Rybnik. Available from: https://www.virtualtravel.cz/valdikov/opatsky-rybnik
  • 42.Fusaro A, Zecchin B, Giussani E, Palumbo E, Agüero-García M, Bachofen C, et al. High pathogenic avian influenza A(H5) viruses of clade 2.3.4.4b in Europe-Why trends of virus evolution are more difficult to predict. Virus Evol. 2024;10(1):veae027. doi: 10.1093/ve/veae027 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Romero Tejeda A, Aiello R, Salomoni A, Berton V, Vascellari M, Cattoli G. Susceptibility to and transmission of H5N1 and H7N1 highly pathogenic avian influenza viruses in bank voles (Myodes glareolus). Vet Res. 2015;46(1):51. doi: 10.1186/s13567-015-0184-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Root J, Shriner S. Avian Influenza A Virus Associations in Wild, Terrestrial Mammals: A Review of Potential Synanthropic Vectors to Poultry Facilities. Viruses. 2020;12(12):1352. doi: 10.3390/v12121352 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Chestakova IV, van der Linden A, Bellido Martin B, Caliendo V, Vuong O, Thewessen S, et al. High number of HPAI H5 virus infections and antibodies in wild carnivores in the Netherlands, 2020-2022. Emerg Microbes Infect. 2023;12(2):2270068. doi: 10.1080/22221751.2023.2270068 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Kutkat O, Gomaa M, Moatasim Y, El Taweel A, Kamel MN, El Sayes M, et al. Highly pathogenic avian influenza virus H5N1 clade 2.3.4.4b in wild rats in Egypt during 2023. Emerg Microbes Infect. 2024;13(1):2396874. doi: 10.1080/22221751.2024.2396874 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Guinat C, Rouchy N, Camy F, Guérin JL, Paul MC. Exploring the Wind-Borne Spread of Highly Pathogenic Avian Influenza H5N8 During the 2016-2017 Epizootic in France. Avian Dis. 2019;63(sp1):246–8. doi: 10.1637/11881-042718-ResNote.1 [DOI] [PubMed] [Google Scholar]
  • 48.Nguyen XD, Zhao Y, Lin J, Purswell JL, Tabler T, Voy B, et al. Modeling long-distance airborne transmission of highly pathogenic avian influenza carried by dust particles. Sci Rep. 2023;13(1):16255. doi: 10.1038/s41598-023-42897-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.The Poultry Site. Key Factors for Poultry House Ventilation. Available from: https://www.thepoultrysite.com/articles/key-factors-for-poultry-house-ventilation
  • 50.Poultry World. Tips and tricks for proper ventilation. Available from: https://www.poultryworld.net/specials/tips-and-tricks-for-proper-ventilation
  • 51.Elbers ARW, Gonzales JL, Koene MGJ, Germeraad EA, Hakze-van der Honing RW, van der Most M, et al. Monitoring Wind-Borne Particle Matter Entering Poultry Farms via the Air-Inlet: Highly Pathogenic Avian Influenza Virus and Other Pathogens Risk. Pathogens. 2022;11(12):1534. doi: 10.3390/pathogens11121534 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 52.Bossers A, de Rooij MM, van Schothorst I, Velkers FC, Smit LA. Detection of airborne wild waterbird-derived DNA demonstrates potential for transmission of avian influenza virus via air inlets into poultry houses, the Netherlands, 2021 to 2022. Euro Surveill. 2024;29(40):2400350. doi: 10.2807/1560-7917.ES.2024.29.40.2400350 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53.Adlhoch C, Fusaro A, Gonzales JL, EFSA (European Food Safety Authority), ECDC (European Centre for Disease Prevention and Control), EURL (European Reference Laboratory for Avian Influenza). Scientific report: Avian influenza overview September–December 2022. EFSA J. 2023;21(1):7786. doi: 10.2903/j.efsa.2022.7786 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 54.Bowes VA, Ritchie SJ, Byrne S, Sojonky K, Bidulka JJ, Robinson JH. Virus characterization, clinical presentation, and pathology associated with H7N3 avian influenza in British Columbia broiler breeder chickens in 2004. Avian Dis. 2004;48(4):928–34. doi: 10.1637/7218-060304R [DOI] [PubMed] [Google Scholar]
  • 55.Swayne D. Epidemiology of Avian Influenza in Agricultural and Other Man-Made Systems. In: Swayne DE, editor. Avian Influenza. 1st edition. Ames: Blackwell Publishing; 2008. p. 59–85. [Google Scholar]
  • 56.World Organisation for Animal Health. Biosecurity procedures in poultry production. World Organisation for Animal Health. 2022. [Google Scholar]
  • 57.Ouyang H, Wang L, Sapkota D, Yang M, Morán J, Li L, et al. Control technologies to prevent aerosol-based disease transmission in animal agriculture production settings: a review of established and emerging approaches. Front Vet Sci. 2023;10:1291312. doi: 10.3389/fvets.2023.1291312 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

S1 Data. Supporting information.

(XLSX)

pone.0319880.s001.xlsx (366.3KB, xlsx)

Data Availability Statement

All relevant data are within the manuscript and its Supporting Information files.

Articles from PLOS One are provided here courtesy of PLOS

RESOURCES