Since 1985, avian influenza virus surveillance has been conducted annually from mid-May to early June in charadriiform species from the families Scolopacidae and Laridae (shorebirds and gulls) at Delaware Bay in the northeast United States. The mass migrations of shorebirds, gulls and horseshoe crabs (Limulus polyphemus) coincide at that time, and large numbers of migrating birds pause at Delaware Bay to feed on horseshoe crab eggs deposited at the high-tide line. Influenza viruses are consistently isolated from charadriiform birds at Delaware Bay, at an overall rate approximately 17 times the combined rate of isolation at all other surveillance sites worldwide (490 isolates/9474 samples, 5.2% versus 49 isolates per 15 848 samples, 0.3%, respectively; Proportion test, p < 0.0001). The likelihood of isolating influenza viruses at Delaware Bay is dependent on the presence of ruddy turnstone (Arenaria interpres) at the sampling site (G-test of independence, p < 0.001). The convergence of host factors and environmental factors results in a unique ecological ‘hot spot’ for influenza viruses in Charadriiformes.
The spread of avian influenza virus (AIV), especially highly pathogenic avian influenza (HPAI) virus, by wild aquatic birds has attracted much recent attention. At the time of the outbreak of HPAI H5N1 virus in Hong Kong SAR, China in 1997 , AIV surveillance was being conducted in wild birds only in Russia , Japan , Italy  and the United States [4,5]. The knowledge gained from these studies and from earlier investigations [6–15] established the following principles of the ecology of AIV and its natural hosts: (i) wild aquatic birds are the natural reservoirs of all influenza A viruses; (ii) influenza viruses in their natural reservoirs are in evolutionary stasis and do not cause disease; (iii) interspecies transmission to new hosts can be followed by rapid virus evolution, creating a potentially pathogenic strain; and (iv) most transfers result in unstable virus lineages, but if a ‘mixing vessel’ (e.g. pig, chicken, quail) is involved a stable lineage may be established. Examples of the latter principle are H1N1 viruses descended from the 1918 ‘Spanish flu’ pandemic, the H2N2 1957 ‘Asian flu’ pandemic strain, the 1968 ‘Hong Kong flu’ strain, the highly pathogenic H5N1 ‘bird flu’ virus and the current 2009 H1N1 pandemic strain.
Until 1997 AIV studies dealt primarily with anseriform birds (particularly dabbling ducks) and gave little attention to charadriiform species. Early studies in charadriiform birds (shorebirds and gulls) at Delaware Bay in the northeastern USA  suggested that a portion of the gene pool of influenza viruses in those birds differed from that of viruses in wild ducks. Subsequent studies showed that the full complement of AIV subtypes is not found in shorebirds, gulls and ducks [5,17,18]. Since 1998 investigators have conducted AIV surveillance in charadriiform species in many regions of the world: Alaska (USA) [19,20]; the USA (Arkansas, Delaware, Florida, Georgia, Kansas, Louisiana, Massachusetts, Missouri, New Jersey, New York, South Carolina, Texas), Argentina, Bermuda and Chile ; Barbados ; Alaska and Siberia ; Europe, Asia, Africa, North America, South America, Antarctica and the Arctic ; sub-Saharan Africa ; Italy ; and Australia . The overall prevalence of AIV in charadriiform birds at all of these sites combined was lower than that found annually at Delaware Bay during spring migration [5,21], which is characterized by large numbers of shorebird and gull species foraging on the eggs of horseshoe crabs (Limuluspolyphemus). This high prevalence at a specific location and a specific time indicates a ‘hot spot’ (defined as a site where influenza virus isolation rates are higher than elsewhere in the world) for virus–host interaction. The specific host, virus and environmental factors that contribute to its generation are unknown, but virus is isolated more frequently from Charadriiformes at that time, and primarily from the charadriiform member ruddy turnstone (Arenaria interpres) [16,21].
Here, we review the results of recent AIV surveillance studies in charadriiform species at Delaware Bay and elsewhere to examine whether the creation of the hot spot is likely to be directly related to the convergence of large numbers of migrating shorebirds and spawning horseshoe crabs (Limulus polyphemus), and to assess the role of the ruddy turnstone.
2. Ecological convergence at Delaware Bay: the shorebird migration phenomenon
Each year with the new or full moon in mid-May one of the world's largest aggregations of shorebirds converges on Delaware Bay on the northeast coast of the USA (figure 1). At the same time each year, horseshoe crabs are signalled by the tidal pull of the full or new moon to migrate en masse from the bay floor onto the shore to spawn. At high tide, the female lays her eggs in a shallow hole at the water's edge, where they are fertilized by the male. Spawning continues for six weeks, leaving billions of eggs buried or strewn about the beach.
Charadriiform birds of the family Scolopacidae (sandpipers) arrive at Delaware Bay during their spring migration from South America to their arctic breeding grounds in Canada. Predominant species are Calidris canutus (red knot), A. interpres (ruddy turnstone), Calidris alba (sanderling), Calidris pusilla (semipalmated sandpiper) and Calidris alpina (dunlin). Other prominent Charadriiformes arrivals are members of the family Laridae–Larus argentatus (herring gull) and Larus atricilla (laughing gull). As many as 1.5 million shorebirds and gulls are estimated to feed at Delaware Bay during the horseshoe crab spawning period, with peak counts of more than 400 000 birds and a density as high as 210 birds m−2 [27–29]. These long-distance migrants (e.g. the red knot migrates nearly 9000 miles from Tierra del Fuego at the southern tip of Argentina) rely on their stopover at Delaware Bay to refuel. On arrival, some birds have flown non-stop for 5000 miles , and the next leg of their journey is a nearly 2000 mile flight to the Arctic. Because the migrating shorebirds are primarily dependent on the availability of horseshoe crab eggs for survival during this leg of their journey  it is crucial that their arrival at Delaware Bay coincide precisely with horseshoe crab spawning. From mid-May to early June the swarms of charadriiform shorebirds and gulls feast on horseshoe crab eggs. Most birds double their body weight in two weeks, consuming as many as 135 000 eggs  and gaining nearly 5 g d−1 .
Disruption of the spawning cycle and other environmental factors leading to subsequent depletion of the available crab eggs could result in devastating consequences for the migrating birds. It is believed that nearly the entire population of the American race of red knot (Calidris canutus rufa) migrates through Delaware Bay, and that the entire race could be endangered if its primary food source is deleteriously altered. Over-harvesting of horseshoe crabs as bait for the eel and conch fishing industry—and to a lesser extent for biomedical research and medical testing (e.g. Limulus Amebocyte Lysate test)—has been linked to observed declines in both horseshoe crab egg densities and populations of red knot, semipalmated sandpiper, sanderling and dunlin [28,33,34]. Mean peak counts of red knots observed at Delaware Bay have fallen from greater than 50 000 in 1986–1991 to less than 20 000 in 2003–2007. Peak counts from 2003 to 2007 were 66 per cent lower than those from 1998 to 2002. The other shorebird species showed an overall decline of 50 per cent during the period 1998–2007. The declining number of shorebirds raises the question of whether the prevalence of AIV is correspondingly being attenuated.
3. AIV and the shorebirds of Delaware Bay
Charadriiform birds were first found to be AIV hosts after an H5N3 outbreak (A/tern/South Africa/61) in 1961 that killed approximately 1300 common terns (Sterna hirundo) . Further studies indicated that a variety of charadriiform species are natural hosts of AIV [9,10,36–38] and provide a reservoir for these viruses. To date, surveillance studies in the Scolopacidae (shorebirds) and Laridae (gulls) families throughout the world have yielded isolates of 15 of the 16 haemagglutinin subtypes with the exception of H14, and all nine of the neuraminidase subtypes. Similarly, in longitudinal surveillance studies at Delaware Bay from 1985 to 2008, all HA and NA subtypes except H14 and H15 were isolated from shorebirds and gulls [5,21,39]. Interestingly, the H1N1 subtype has never been detected in shorebirds and gulls at Delaware Bay, and H5N1 virus (a low-pathogenic strain) was isolated only once, in 2007. However, a large number of viruses of subtype H7N3 isolated in 2006 were found to be a low-pathogenic form of the virus that caused outbreaks of highly pathogenic H7N3 infection in the Fraser Valley of British Columbia, Canada, in 2004  and Saskatchewan, Canada, in 2007 .
AIV has been consistently isolated from birds (primarily from faecal samples) each year during the spring migration at Delaware Bay. As previously reported , most sample vials contained multiple faecal swabs (each from a different bird), and an equation was developed to normalize the prevalence rate according to the number of swabs per vial. From 1988 through to 2002, this formula was applied to all sample vials containing multiple swabs. Starting in 2003, each sample vial contained a single swab. The annual rate of virus isolation from 1988 through to 2008 (total 7343 samples) ranged from 1.0 to 14.7 per cent (figure 2), with a mean rate of 6.3 per cent over the 25 seasons.
Table 1 summarizes rates of virus isolation from various AIV surveillance studies in Charadriiformes species since 1998 [18–24,26]. Our own surveillance studies show an overall AIV isolation rate of 6.3 per cent during the spring migration at Delaware Bay from 2003 through to 2008 (table 1). This figure is likely to be accurate, as it includes only years in which one sample was collected per vial. In the global studies overall, influenza viruses were rarely isolated from Charadriiformes species except at two locations—Delaware Bay in the USA and Fullerton Cove in the Hunter estuary, New South Wales (NSW), Australia. It is noteworthy that one of the two influenza viruses detected by RT-PCR in wader species by Munster et al. was from a red knot at Delaware Bay in May 2005. The collective surveillance studies conducted during the period 1998–2008 isolated 539 influenza viruses from charadriiform species, predominantly from the Scolopacidae family (491 isolates). Of these 539 isolates, 490 were obtained at Delaware Bay, and of the 491 viruses isolated from Scolopacidae species, all but 13 were recovered at Delaware Bay. Interestingly, of the 48 influenza viruses isolated from Laridae species in six combined studies, only 12 were obtained from gulls at Delaware Bay. Nevertheless, the rate of isolation of influenza viruses from gulls was higher at Delaware Bay (3%) than at any of the other sites (proportion test, χ12 = 31.29, n = 7124, p < 0.0001; www.R-project.org), although a 1 year surveillance period in Alaska identified virus in 2.0 per cent of tested gulls.
The data in table 1 indicate that the overall rate of virus isolation from Charadriiformes species from 1999 through to 2008 was highest at Delaware Bay—5.2 per cent at Delaware Bay (; this report), while the overall rate at all of the remaining sites was only 0.3 per cent (proportion test, χ21 = 670.72, n = 25 322, p < 0.0001). Influenza virus was isolated from seven red-necked stints (Calidris ruficollis) at Fullerton Cove, NSW, Australia (isolation rate, 4% of 173 samples collected from Scolopacidae in November 2004) . However, follow-up studies in NSW from 2005 through 2007 yielded no isolates or positive RT-PCR tests among 841 specimens obtained from Charadriiformes species . Therefore, further investigation is warranted to determine whether the relatively high isolation rate at Fullerton Cove, NSW in November 2004 was a singular event.
4. Long-term AIV prevalence and optimal sample sites at Delaware Bay
As noted above, the density of horseshoe crab eggs and the number of shorebirds at Delaware Bay have declined dramatically since the early 1990s. Therefore, we reviewed whether there has been a corresponding effect on the prevalence of AIV. The overall prevalence of AIV in charadriiform birds at Delaware Bay from 1988 to 2008 was 6.3 per cent (7343 samples; this report). The annual prevalence rates are shown in figure 2. The trend line in figure 2 indicates a gradual increase in the rate of AIV isolation from charadriiform species at Delaware Bay during that period. Although prevalence estimates for the six most recent sampling years—2003–2008—are consistently at or below the trend line, the isolation rate for the period 2003–2008 is 6.1 per cent (n = 3154), nearly identical to the overall prevalence estimate of 6.3 per cent for the 21 seasons surveyed. However, the pattern of AIV isolation rates from Charadriiformes at Delaware Bay is highly irregular, and additional surveillance data will be needed to ascertain whether AIV prevalence is decreasing in proportion to the decline in horseshoe crab egg density (and the number of shorebirds).
During the years 1988–2008, we conducted surveillance at a total of 12 Delaware Bay sampling sites (sites varied from year to year). We grouped these sites into three geographical regions: Region 1 (Mispillion River/Slaughter Beach, DE, USA); Region 2 (Gandy's Beach, Fortescue, East Point, Thompson's Beach, Moore's Beach); and Region 3 (Reed's Beach, Cooke's Beach, Kimble's Beach, Pierce's Point, Norbury's Landing/Villas; figure 1). The highest rate of AIV isolation was in Region 3 (8.2%, n = 4339), followed by Region 1 (6.2%, n = 375) and Region 2 (3.2%, n = 2629), suggesting that AIV is more prevalent in the southern part of Delaware Bay. This is further supported by a lack of significant differences between Regions 1 and 3 (proportion test, χ21 = 1.61, p < 0.20), yet the prevalence at both Region 1 and Region 3 is higher than that observed at Region 2 (proportion test, χ21 = 7.72, p < 0.006; and χ21 = 68.35, p < 0.0001, respectively). It should be noted that isolation rates are based on cumulative totals for the entire study period and that each geographical region was not sampled every year. The environmental conditions affecting bird and horseshoe crab prevalence can vary annually with corresponding changes in the virus isolation rate. Therefore, the virus prevalence for each region should be viewed as a long-term average only. The highest rate of isolation within Region 3 (9.8%) occurred in an area less than 1 mile long encompassing Reed's, Kimble's and Cooke's Beaches. Sixty-five per cent of all viruses detected at Delaware Bay during 1988–2008 came from this area.
We observed that ruddy turnstones (Scolopacidae family members) were usually present on beaches with the highest virus prevalence and hypothesized that the presence of ruddy turnstones is associated with the prevalence of AIV. The isolation rate was 7.5 per cent on beaches where ruddy turnstones were present, but only 1.7 per cent on beaches where they were absent (table 2). Virus isolation from a sampling site at a specific time was dependent on the presence of ruddy turnstones (G-test of independence: G1 = 16.91, n = 93, p < 0.0001). Further, the G-test of independence showed that the likelihood of isolation of AIV was dependent on the presence of ruddy turnstones when either the number of positive samples (G1 = 86.12, n = 7343, p < 0.00001) or the percentage of positive samples (G1 = 3.91, p < 0.048) was analysed. However, this analysis was limited by the inclusion of multiple swabs in single sample vials, which may have obscured the species sampled.
5. Conclusion: what makes the Delaware Bay an AIV hot spot?
Delaware Bay becomes an AIV hot spot each year in May, during the stopover of vast numbers of migrating shorebirds and the concurrent spawning of horseshoe crabs. The high rate of influenza virus isolation in Charadriiformes species is unique to Delaware Bay. During the period 1999–2008 the overall rate of AIV isolation from charadriiform birds at Delaware Bay was more than 17 times the combined rate at all other sites worldwide: 5.2 per cent (490 of 9474 samples) versus 0.3 per cent (49 of 15 848 samples; proportion test, χ12 = 670.72, p < 0.0001; table 1). Although high prevalence of AIV at Delaware Bay remains unexplained, a convergence of interacting factors, including host (physiology, behaviour and immune status) and ecological factors are likely to underlie the phenomenon.
During May at Delaware Bay, AIV has consistently been recovered from Charadriiformes species since surveillance began in 1985. The annual spring migration of shorebirds and the ecosystem of Delaware Bay provides a unique setting in which to study the biology and ecology of AIV in its natural host. At this time of year the teeming shorebird population affords an abundant and easily accessible source of samples (faecal deposits) for virus detection and laboratory characterization of isolates. AIV isolates have been obtained from surveillance sites all around Delaware Bay (figure 1), with the highest isolation rate at or near Reed's Beach along the southern shore of the bay in New Jersey. AIV has been isolated more frequently from ruddy turnstones than from any other Scolopacidae and Laridae species at Delaware Bay [16,21]. Here, we have demonstrated that surveillance sites at which ruddy turnstones are present are significantly more probable than other sites to yield virus isolates (table 2). Because there is no evidence that ruddy turnstones, or any of the other Charadriiformes species, are subject to high rates of AIV infection except during their spring migration through Delaware Bay, the ruddy turnstone–horseshoe crab association may be a key factor in the spread of AIV. Delaware Bay is ecologically unique based on the large numbers of shorebirds and their apparent dependence on horseshoe crab eggs, and it remains to be determined how the ruddy turnstone–horseshoe crab association, as well as the interaction of different species at this site, may contribute to the generation of a hot spot.
There is limited data about the prevalence of AIV in shorebirds during or after their return migration to South America. However, a small number of faecal samples (n = 576) were collected from shorebirds at Delaware Bay during early September to mid-November 1985–1988, and five samples (approx. 0.9%) yielded AIV in 1986 . A recent study (June 2006–December 2007) of the wintering grounds along the central coast of Peru isolated three AIVs from the faecal samples of 804 shorebirds and gulls (prevalence approx. 0.3%) . Interestingly, two of the three isolates were from ruddy turnstones (subtype H10N9, October 2006), and the remaining one (subtype H13N2, November 2007) was recovered from a Dominican gull (Larus dominicanus). Although these birds may not be members of the same flocks that migrate through Delaware Bay, AIV is clearly present at the wintering grounds in South America, and ruddy turnstones (and gull species) may be a major source of AIV infection for shorebirds in that region during that time period.
It remains to be determined whether AIV transported by migrating shorebirds from their South American wintering grounds plays a significant role in the infection of shorebirds and gulls at Delaware Bay, and whether the primary source of virus resides in an undetermined ecological niche at Delaware Bay or elsewhere. The role of ducks and geese in the spread of AIV at Delaware Bay is not known, and additional surveillance in these species at Delaware Bay is required in order to answer this question. It is also not known whether Delaware Bay is the world's only hot spot for AIV in Charadriiformes species or whether other hot spots remain to be discovered.
We thank David Walker and Angela Ferguson for excellent technical support, James Knowles for assistance with manuscript preparation, and Sharon Naron for editorial assistance. This work was supported by the National Institutes of Health, Department of Health and Human Services contracts HHSN266200700005C and AI-95357, and Cancer Center Support (CORE) grant CA-21765, and by the American Lebanese Syrian Associated Charities (ALSAC).
- Received May 21, 2010.
- Accepted June 24, 2010.
- © 2010 The Royal Society