Recent improvements in molecular diagnostic tests have facilitated high-throughput screening of wild birds for influenza A virus. Despite the introduction of the molecular tests and the wide range of bird species tested, only a few “new” influenza A virus hosts were identified: barnacle, bean, brent, and pink-footed goose, bewick's swan, common gull, and guillemot. It is thus reassuring that the use of classical methods for virus detection in numerous surveillance studies has not resulted in an apparent biased detection toward viruses that can be isolated easily in embryonated hens' eggs; therefore, it remains a viable approach for virus detection. However, in our hands, virus detection by RT-PCR was more sensitive than using classical tests, since viruses were isolated from only one-third of the RT-PCR positive samples. Even under ideal conditions of transport, storage, and processing, not all RT-PCR positive samples yielded virus isolates.
We confirmed the high virus prevalence of dabbling ducks in fall as observed in previous studies in the Northern Hemisphere [1
]. Our data indicate that timing relative to migration, instead of the absolute time point, is a determinant of virus prevalence. High virus prevalence early in fall migration likely declines gradually as the migration proceeds, thus forming a north–south gradient of virus prevalence even within single species. This explains prevalence differences in earlier surveillance studies [4
]. Influenza A virus prevalence was generally higher in juvenile ducks as compared with adults, as reported for North America [5
]. The estimated yearly turnover of mallards in Northern Europe is roughly one-third; 56% of the juvenile mallards die during their first year and the mortality in adult birds is ~40% [24
]. Thus, one-third of the mallard population consists of juvenile birds, which are immunologically naïve and therefore probably more susceptible to influenza A virus[1
The influenza A virus prevalence in mallards was comparable to that of other dabbling ducks. Viruses were also detected in ducks belonging to other guilds, but the prevalence was lower. Influenza A virus was detected in 811 of 13,297 dabbling ducks, but in only six of 440 other ducks (Pearson χ2
= < 0.001). Analysis of 7,130 samples from 11 goose and swan species revealed that virus prevalence was also low, ranging from 0.7% to 2.4%. Several factors could contribute to the high virus prevalence in dabbling ducks as compared with other species. The dabbling behavior itself is likely an important factor; virus excreted in surface waters via feces may efficiently transmit viruses to other ducks that feed on the same waters. Influenza A virus can remain infectious for prolonged periods in surface water depending on temperature, salinity, and pH [25
]. The prolonged presence of influenza A viruses in surface water may enable the spread of viruses in different host sub-populations that otherwise would be separated in time and space. In contrast, diving ducks forage deeper under the surface and more often in marine habitats, and most goose and swan species graze in pastures and agricultural fields. Such differences in feeding behavior could lead to less-efficient virus transmission and thus account for the differences in prevalence. Population size and age structure could be additional important factors enabling the annual co-circulation of multiple virus subtypes within the same (meta-) populations [26
]. The dabbling duck populations are estimated at between 5,000,000 and 10,000,000 in Northern Europe alone [27
]. Mallards are the most abundant species, followed by Eurasian wigeon and common teal [27
]. We observed the highest virus prevalence in two of these species: mallards and common teals. The population estimates for goose species in Northern Europe are significantly lower as compared with the dabbling ducks [27
]. The smaller population sizes could limit in general the potential of perpetuation of influenza A virus in these species and in particular the continuous co-circulation of multiple virus subtypes. When we plotted the influenza A virus prevalence in duck, goose, and gull species as a function of population size (), population size did not appear to be the main correlate of virus prevalence (R2
= 0.0001). The relative clustering of the data points from the duck, goose, and gull species () suggests that other factors (taxonomy, behavior, etc.) could determine virus prevalence.
Relation between Influenza A Virus Prevalence in Avian Hosts and Their Population Sizes
The population size of 2,000,000 black-headed gulls in Northern Europe seems to be sufficiently large for the continuous circulation of influenza A viruses. Behavioral factors influencing influenza A virus ecology in gulls could include colony breeding, gregariousness during migration and wintering, feeding patterns, and the mixing of different populations of birds. From our data, it appears that virus prevalence in gulls peaks shortly after they have left their breeding grounds.
Surveillance studies performed along the East Coast of North America suggested a distinct role for wader species in the perpetuation and maintenance of certain influenza A virus subtypes [4
]. It was suggested that different families of wetland birds are involved in perpetuating influenza viruses and that waders may carry the virus north to the duck breeding grounds in spring. We detected influenza A virus in one shorebird sample obtained from Delaware Bay (United States) and one from South Korea. The differences in prevalence between our American wader data and those described by others [4
] could be due to sampled species, sampling procedures, and timing. Within our European surveillance study, not a single influenza A virus was detected in waders. Although the majority of our wader samples were collected during fall migration, a reasonable sample size was collected during spring migration. Thus, there is no evidence that waders play a role in the perpetuation of influenza A virus in Europe. The recently intensified surveillance in waders, including serological data collection, may allow a definitive conclusion about their role in the influenza virus ecology in Europe.
Although historically influenza A viruses have been obtained from more than 105 species of 26 different families [2
], we did not detect significant influenza virus prevalence in species other than those belonging to the orders Anseriformes and Charadriiformes. We therefore suggest that although multiple bird species can be infected, their contribution to the overall virus ecology could be limited; infections of these hosts, although potentially with high peak prevalence, may be only transient.
Influenza A virus subtypes H1–H12 were isolated frequently from mallards, and several of these subtypes were also detected in Eurasian wigeons, common teals, gadwalls, and northern shovelers. The absence of subtypes H14 and H15 in our collection was probably due to the geographical separation of virus hosts [28
]. Because all HA subtypes isolated from Anseriformes were also isolated from mallards, it is likely that mallards play a pivotal role in the perpetuation of influenza A virus subtypes H1–H12 in Europe. Subtypes H5 and H7 were rarely detected in longitudinal studies of ducks in Canada, whereas in this European study, they represent 7.5% and 11.1% of viruses obtained from ducks. The most common virus subtypes in Europe—H3, H4, H6—were also common subtypes in Canada.
The predominant isolation of H13 and H16 viruses from gull species confirms the common notion that these viruses belong to the influenza A virus “gull lineage.” H13 and H16 viruses are genetically distinct from viruses from other hosts and seem to have adapted to replication in gull hosts in particular [3
Interestingly, viruses of the H6 subtype seem to have a broader host range compared with that of other virus subtypes in our study. Of the influenza viruses obtained from birds other than dabbling ducks and gulls, 79% were H6 viruses. H6 viruses were isolated from gulls, auks, swans, and geese. The relative abundance of this subtype in ducks does not explain the large variety of species from which these viruses were isolated; H4 and H7 viruses were also detected frequently in ducks, but rarely in other birds. H6 viruses have been transmitted from wild birds to poultry on several occasions [30
], providing further evidence for the ability of these viruses to be transmitted between different bird species.
HPAI H5N1 viruses have caused large-scale outbreaks in poultry in Southeast Asia since 1997 and have also been transmitted to a variety of mammalian species, including humans [17
] . Until 2005, wild migratory birds probably did not play a significant role in the epidemiology and spread of HPAI H5N1, although the virus was detected sporadically in wild birds. A large-scale outbreak in wild migratory birds occurred in April–June 2005 at Lake Qinghai in China [33
], after which the HPAI H5N1 virus rapidly spread westward across Asia, Europe, the Middle East, and Africa. Since then, affected wild birds have been reported in several countries [36
], but even in areas with significant outbreaks in poultry, the virus prevalence in wild birds is low and their role in spreading the disease is unclear. It is likely that the influenza A surveillance studies in wild birds such as those presented here could provide “early warning” signals for the introduction of HPAI H5N1 into new regions [37
]. The current increased interest in influenza virus surveillance in wild and domestic birds provides a unique opportunity to increase our understanding not only of HPAI epidemiology but also of the ecology of low pathogenic avian influenza viruses in their natural hosts.