Our study was conducted following approval by the Chinese Center for Disease Control and Prevention Ethical Review Committee (201118). All participants provided signed informed consent forms.

Wild bird sample collection sites were designated by the Forestry Administration and Hunan East Dongting Lake Nature Reserve. Fresh fecal samples from wild birds and water samples were collected every month from November 2011 to April 2012. We collected one to four fresh fecal samples at each sampling point in the flood plain. Water samples were collected from locations where a clear footprint attributable to a waterfowl was seen. There was at least 10 m between two sampling points.

We collected fresh fecal and water samples from duck farms surrounding the Eastern Dongting lake area. These farms were within about a 5 km straight-line distance to the lake area. Inclusion criteria for duck farms included: ducks were being raised during our field survey; and there was a possible shared water source between domestic ducks and wild birds. We referred to fecal and water samples collectively as environmental samples.

Two cross-sectional serological surveys were conducted in November 2011 and April 2012. Staff of the Eastern Dongting Lake Wetland Nature Reserve, local staff on duck farms, and individual duck breeders residing along the Eastern Dongting Lake were enrolled. Venous blood was collected by registered nurses. Inclusion criteria of participants were: they were local residents or individuals that had been in the area longer than 2 weeks; they signed the informed consent form; they were aged 18 years or older; and they had been exposed to wild birds and/or domestic ducks.

Fecal samples were placed in tubes containing M199 solution [0.5% (w/v) bovine serum albumin (BSA), 2×106 U/L penicillin, 200 mg/L streptomycin, 2×106 U/L polymyxin B, 250 mg/L gentamycin, 60 mg/L levofloxacin hydrochloride, and 5×105 U/L nystatin] [32]. Water samples were placed in empty tubes. Serum and environmental samples were transported to the influenza surveillance laboratory of the local center for disease control (CDC) at 4°C within 24 h of collection. Serum and environmental samples were stored at −20 and −70°C, respectively. Samples were transported in accordance with national biosafety regulations; all samples were tested by the Chinese National Influenza Center.

Total RNA was extracted and purified from sample aliquots (200 µL) using a QIAamp One-For-All Nucleic Acid Kit (Qiagen, Germany) and the BioRobot Universal System (Qiagen), following the manufacturer’s protocol. The RNA of AIVs was detected using quantitative polymerase chain reaction (qPCR) assays (AgPath, Foster City, CA, USA) targeting the Matrix gene. Samples that were positive were inoculated in the allantoic cavity of 9-day-old embryonated chicken eggs (ECEs). The ECEs were incubated at 35°C for 48 h, then chilled at 4°C overnight. Allantoic fluid was harvested and hemagglutination assays were performed with 1% turkey red blood cells to confirm the presence of viruses. Viral RNA was extracted using an RNeasy Mini Kit (Qiagen) following the manufacturer’s instructions. RNA was reverse transcribed into cDNA with a SuperRT cDNA Kit (CWBIO, Beijing, China) using the Uni12 primer (5′-AGC AAA AGC AGG-3′). Identification of isolate subtypes was performed by PCR using 16 sets of primers specific for hemagglutinin (HA; H1–H16) and nine sets specific for neuraminidase (NA; N1–N9) that were designed by the Chinese National Influenza Center.

Extracted viral RNA was reverse transcribed into cDNA using SuperRT cDNA Kit (CWBIO) and the aforementioned Uni12 primer. Reactions were incubated at 25°C for 10 min and then 42°C for 80 min. A 2×Es Taq MasterMix Kit (CWBIO) was used to amplify virus genes with the aid of eight gene segment-specific primers. Amplified products were purified using a QIAquick PCR purification kit (Qiagen). Sequencing reactions were carried out with a BigDye Terminator v3.1 Cycle sequencing kit (AB, Foster City, CA, USA) on an ABI PRISM 3700xl DNA sequencer following the manufacturer’s instructions.

Lasergene 8.1 software (DNASTAR) was used to assemble gene sequences and Mafft 6 was used for the comparison of sequences. The neighbor joining (NJ) algorithm with a bootstrap value of 1,000 was used within MEGA 4.0 software. Phylogenetic trees were created for eight genes of AIV H5N1. All gene sequences for comparison were acquired from the GenBank influenza database.

We conducted hemagglutination-inhibition (HI) assays using horse red blood cells for preliminary screening of HPAI H5N1-specific antibodies. Microneutralization (MN) assays were used to confirm results when an HI titer of 20 or greater was obtained [31]. We used two H5N1 viruses [A/Anhui/1/2005 (H5N1) virus (clade 2.3.4) and A/domestic duck/Yueyang/C0816/2012 (H5N1) virus (clade 2.3.2.1)] as antigens in this study for HI and MN assays.

We initially entered our data into a customized database (Epidata3.0) and then transferred it to SPSS (SPSS Inc., Chicago, IL, USA) for statistical analysis. Differences between proportions were analyzed using the χ2 test. A P-value less than 0.05 was considered significant.

We collected 6,621 environmental samples from November 2011 to April 2012, of which 5,419 were from wild birds and 1,202 from domestic ducks (Table 1 and Figure 1). We detected AIVs in 5.19 and 5.32% of samples from wild birds and domestic ducks, respectively (χ2 = 0.0384, P = 0.84). We observed two peaks for virus prevalence among wild birds, in December 2011 and March 2012. A peak in prevalence was seen for domestic ducks in March 2012 (Figure 2). There was a greater proportion of AIV-positive fecal samples than water samples for wild birds (χ2 = 30.4556, P<0.001), however this was not significantly different for ducks (χ2 = 0.3841, P = 0.54). There was a higher proportion of AIV-positives among the water samples associated with ducks than those with wild birds (χ2 = 15.9069, P<0.001); the same trend was not seen with fecal samples (χ2 = 1.1357, P = 0.29) (Table 2). The prevalence of AIV positives varied among the different wild bird species, with the greatest proportion seen in Anatidae birds (5.15%), followed by Charadriiformes (0.44%) and then members of the Ardeidae family (0.18%) (Table 3).

We isolated 39 AIV strains from fecal specimens of wild birds (Table 3), including 22 H9N2, 5 H1N5, 3 H5N2, 2 H1N8, 2 H6N1, 2 H6N2, 1 H1N2, 1 H7N7, and 1 H12N8; of these, 32 strains were isolated from Anatidae birds, 2 from Ardeidae birds, 2 from Charadriiformes, and 3 from unidentified species. We isolated five avian influenza virus strains from fecal samples of domestic ducks (Table 3), including 1 H5N1, 2 H4N2, and 2 H3N6 strains. No viruses were isolated from water samples. There was no statistical differences in the proportion of virus strains isolated from wild birds and domestic ducks (0.72% and 0.42%, respectively). Among different species of wild birds, virus isolation rates differed among the Anatidae (0.73%), Charadriiformes (0.05%), and the Ardeidae (0.05%). The number of AIVs was more in December and March. The trends in numbers of virus strains isolated were similar to those for the prevalence of AIV in samples. (Figure 3).

The A/Duck/Yueyang/C0816/2012 (H5N1) strain was isolated from the feces of a domestic duck. Phylogenetic analysis showed that all eight genes were most closely related to clade 2.3.2.1 H5N1 viruses [A/Hubei/1/2010 (H5N1)-like] currently circulating in China (Figure S1, Figure S2). The cleavage site of the HA protein for A/Duck/Yueyang/C0816/2012 contained multiple basic amino acids (PQIERRRRKR↓GL). The HA receptor binding sites contained 226Q and 228G (H3 numbering), indicating preferential binding for α-2,3 sialic acid. A 20-amino acid deletion (amino acids 49–68) was observed in the NA protein, and the virus was sensitive to NA inhibitors based on the sequence of the NA protein (275H).

We collected 1,050 human serum samples from 515 men and 535 women; approximately 82% of subjects were older than 40 years. Of the subjects, 981 had been exposed to healthy poultry, 55 to sick and/or dead poultry, 76 to healthy wild birds, and 22 to sick and/or dead wild birds. When we used A/Anhui/1/2005 (H5N1) as the detecting antigen, the H5N1 antibody titer in 106 samples was greater than or equal to 20. When we used A/Duck/Yueyang/C0816/2012 (H5N1) as the detecting antigen, the H5N1 antibody titer was greater than 10 in 76 samples with the highest titer being 80. The female participant (1∶80) was a 63-year-old involved in the fishing industry who had lived on a boat in the lake area for 17 years (Table 4). However, none of these positive results were confirmed by MN assays.