Surveillance for Influenza Viruses in Poultry and Swine, West Africa, 2006–2008

West Africa might be an animal influenza–free zone.

R elatively little is known about the emergence, prevalence, and circulation of animal infl uenza viruses in Africa. There is no recent evidence of infl uenza infection in pigs in West Africa. In 2007, Gaidet et al. (1) found a 3.5% prevalence of avian infl uenza virus in wild birds in Africa; the highest prevalence in Mauritania and Senegal, and the most frequently infected species were Eurasian and African ducks. In addition, low-pathogenic avian infl uenza viruses of subtypes H1N8, H3N8, H4N2, H4N6, H4N8, H5N1, H5N2, H5N8, H6N2, H7N7, H9N1, and H11N9 have now been detected in wild birds in Nigeria, Egypt, Zambia, and South Africa (2)(3)(4)(5)(6)(7).
We performed a systematic active surveillance study of animal infl uenza in Côte d'Ivoire, Benin, and Togo. These 3 West African countries reported cases of HPAI (H5N1) only in 2006, 2007, or 2008 (13,14). We aimed to confi rm the current absence of HPAI (H5N1) from the region and determine whether any other infl uenza virus strains might circulate in domestic birds and pigs.

Sampling Sites
Samples were collected exclusively in live-bird markets and backyard farms. The latter were preferred to commercial farms because the outbreaks of HPAI (H5N1) reported during 2006-2008 occurred most often in backyard fl ocks (for 11/12, 4/5, and 2/3 outbreaks in Côte d'Ivoire, Benin, and Togo, respectively) or on small farms (308-7,771 birds per farm) (13,14). Sampling sites were selected in the 3 countries for 1) their density of poultry farms (backyard and commercial, even though we focused  Figure 1). Specimens were collected in the district of Abidjan (i.e., Bingerville, Marcory, Treichville, Port-Bouet, Koumassi, Yopougon), the Middle-Comoé region (i.e., Agnibilékro, Takikro, Abengourou, Niablé), and the South-Comoé region (i.e., Aboisso, Adiaké). In Benin, samples were collected during November 2008-September 2010 in live-poultry markets in Malanville, Gogounou, and Dérasi in the provinces of Borgou and Alibori in the north of the country (Figure 1). A total of 200 swab samples were collected from birds in Lokossa (Mono Province) in 2009. The specimens from pigs were collected from animals in slaughterhouses in Parakou (Borgou Province).
In each region in Côte d'Ivoire, a minimum of 5 villages were randomly selected among those willing to participate. Birds from live-bird markets were randomly selected before sampling (5 randomly selected birds per vendor, number of vendors randomly selected depending on the total number of specimens to collect in a given market).

Sample Collection
At each sampling site, >25 birds were clinically examined, and tracheal and cloacal swab samples were collected at least monthly. In Côte d'Ivoire, nasal swab samples from pigs were collected monthly in 2009 and every 3 months in 2010. In backyard fl ocks in Côte d'Ivoire, serum was collected every 3 months. Each selected market was visited 1×/month in Togo and 2×/month in Benin.
The samples were collected in viral transport media as described (15) and then stored in liquid nitrogen or on ice during sampling and transportation to the laboratory, which never exceeded 1 day. Swabs were then immediately stored either at -80°C in Côte d'Ivoire and Benin or in liquid nitrogen in Togo before further processing. Serum was stored at -20°C before further processing.

Serologic Testing
Serum was screened for infl uenza antibodies by performing ELISAs and/or hemagglutination inhibition (HI) assays. ELISAs were performed by using the FlockChek AI MultiS-Screen Ab Test Kit (Idexx, Westbrook, ME, USA) according to the manufacturer's instructions. HI assays to detect infl uenza virus were performed as described (15,16)

Molecular Testing
Tracheal and cloacal swabs were processed as described (17,18). The samples were screened either individually or in pools of 2 or 5 swabs. RNA was isolated by using the RNeasy mini kit (QIAGEN, Valencia, CA, USA), the QIAmp viral RNA minikit (QIAGEN), or the MagMAXTM-96 AI/ND viral RNA isolation kit (Applied Biosystems/Ambion, Austin, TX, USA) with a Kingfi sher Flex magnetic particle processor (Thermo Scientifi c, Rockford, IL, USA). RNA was eluted in 50 μL of nucleasefree water.
The swab samples from Côte d'Ivoire were tested by using 2-step reverse transcription PCRs (RT-PCRs). The RT step was performed by using random hexamers (Invitrogen, Carlsbad, CA, USA) with 10 μL of extracted RNA and the First-Strand cDNA Synthesis kit (GE Healthcare Europe GmBH, Orsay, France,) according to the manufacturer's protocol. Next, 5 μL of the cDNA obtained was used as the template for the PCR step. The PCR was performed by using the Gene Amp PCR System 2400 (Perkin-Elmer, Applied Biosystem, Paris, France) as described (14).
The swabs from Benin and Togo were tested by using 1-step RT-PCRs performed with the Qiagen 1-step RT-PCR kit (QIAGEN) with either an ABI 9700, ABI 2720 (Applied Biosystems, Vienna, Austria) or ABI 7500 (Stratagene; Applied Biosystems, Carlsbad, CA, USA) thermocycler. For conventional RT-PCR screenings, we used primers that target the infl uenza A matrix gene (19)

Sample Collection
We collected 25,136 swab and 1,819 serum samples from birds and 1,610 swab and 457 serum samples from pigs during the 2-year survey in the 3 countries. Of the bird samples, 70% were from live-poultry markets and 30% from backyard fl ocks (Tables 1, 2). Specimens were collected year-round, and monthly samples ranged from 20 to 160 and from 218 to 1,778 per month, from swine and poultry respectively.

RT-PCR Screening
The 26,746 total swab samples collected from birds and pigs in Côte d'Ivoire, Benin, and Togo all tested negative for infl uenza A genome by RT-PCR, irrespective of collection month or host, and the annual prevalence per country was null (95% CI 0.04-4.79%) (Tables 1, 3). To verify that cold-chain or storage problems had not simply degraded our samples' nucleotides, we screened a subset of 2,427 swab samples collected from birds during early 2009 and 2010 from Benin and Togo for other RNA avian viruses (NDV, IBV, IBDV, or aMPV) and DNA viruses (CAV, ILTV, or MDV). Of the 2,427 samples collected in Benin and Togo, the prevalence of the other viral pathogens ranged from 0 for MDV and IBDV to 4.9% for NDV (119 positive samples), 2.8% for ILTV (68), 2.1% for CAV (51), 1.4% for IBV (34), and 0.3% for aMPV (7) (Figure 2). In addition, 3,330 swab samples collected from birds in Côte d'Ivoire in 2010 were screened for NDV; NDV prevalence ranged from 0.3% to 1.4% depending on time (data not shown). Taken together, these results show the fair quality of our specimens. Cold-chain and sample quality were unlikely to account for the absence of detected infl uenza virus RNA.

Serologic Testing
Because infl uenza virus infection might last only a few days in birds and pigs, we could have missed the virus in the animals sampled. Therefore, we conducted serologic screening, which provides insight into the infection history of an animal's entire life. None of the serum samples collected in birds in Côte d'Ivoire, Benin, and Togo were positive for infl uenza antibodies by ELISA or HI assay. Although 16 of 457 pig serum samples from Côte d'Ivoire were weakly positive by ELISA, none were confi rmed positive by HI; they most likely were all negative for infl uenza antibodies. NDV antibodies were detected in 20% of serum samples from birds in Côte d'Ivoire and in 32% of bird serum samples from Togo (data not shown).

Infl uenza Seasonality and Environmental Factors
Determining the factors contributing to the seasonality of infl uenza has been diffi cult because some countries report having infl uenza activity year-round and others report having 2 peaks of activity or a combination of these patterns. To follow up on recent results (20) seemingly confi rming the year-round activity hypothesis that states that infl uenza spread in the tropics is due to contact rather than aerosol transmission, we compared the average livestock production, temperature, and relative humidity (RH) levels in Côte d'Ivoire, Benin, and Togo with those of Nigeria, Egypt, Vietnam, and Indonesia. Côte d'Ivoire, Benin, and Togo produce signifi cantly less bird and pig meat and fewer bird eggs than do Nigeria, Egypt, Vietnam, or Indonesia (Table 4). West Africa is hot and humid all year, with temperatures ranging from 22°C to 32°C and RH ranging from 63% to 82% (Table 4).  (13,14), and all the strains characterized belonged to clade 2.2 (22,23). Whether wild birds or trade brought the virus to Africa remains unclear (24)(25)(26). However, evidence suggests that the pathogen was fi rst detected in People's Republic of China and might have transited through western Asia, Russia, or Europe (27), with wild birds probably playing a role in introducing HPAI (H5N1) to Africa (27)(28)(29)(30)(31). Incidentally, Côte d'Ivoire, Benin, and Togo are on the Black Sea-Mediterranean and East Atlantic fl yways that these birds use for migration (32).
Within a year after the initial outbreaks, the clade 2.2 strains were thought to have become endemic in Nigeria, and intra-clade 2.2 reassortant viruses were characterized, highlighting the effects of the virus virus and its evolution within the country (33,34). Nigeria has not reported an HPAI (H5N1) outbreak since July 2008 and now seems free of the pathogen (13). The neighboring countries reported only isolated outbreaks in 2006, 2007, or 2008, but no sustained transmission has been reported (13).The subtype H5N1 outbreaks in backyard poultry in West Africa were associated with less severe symptoms and lower death rates than those usually described in such outbreaks.
That HPAI (H5N1) is endemic in several Southeast Asian countries and in Egypt but did not persist in West Africa, except for a couple of years in Nigeria, is intriguing. Despite the effect of viruses such as NDV, infl uenza virus was not detected in any of the swabs or serum samples collected during our active surveillance for animal infl uenza in Côte d'Ivoire, Benin, and Togo. Several factors, such as type of hosts available, animal density, and climate with its effect on virus persistence in the environment and on virus transmission, might have prevented continued circulation of the virus in the region.
Ducks in particular are a natural reservoir for infl uenza and play a major role in infl uenza transmission (35). Fewer ducks are raised in Africa than in Southeast Asia (Table 4), which might limit the virus pool and sustainability on the continent. Moreover, the few ducks in West African are backyard birds in low density fl ocks, not free-range birds in large fl ocks on lakes and rice paddies as they often are in Southeast Asia. Chickens have so far been the fi rst host infected by subtype H5N1 in Egypt (36), but duck meat production is much higher in Egypt than in Côte d'Ivoire, Benin, Nigeria, or Togo ( Table 4). The structure of livebird markets in West Africa also differs substantially from that in Southeast Asia, with fewer birds, fewer species (large majority of chickens and guinea fowls), and often lower confi nement (authors' observations).
We believe that temperature and humidity might be critical parameters for the survival of infl uenza virus in West Africa. Temperature and humidity affect the duration of virus persistence in the environment. In the guinea pig model, infl uenza transmission by the aerosol route depends on humidity and temperature, although contact transmission does not (37,38). Of the experimental conditions tested, only 5°C, 20°C, and 35% RH allowed 100% aerosol transmission; thus, warmer and more humid environments might have less aerosol transmission of infl uenza virus (37). Côte d'Ivoire, Benin, Nigeria, and Togo are hot and humid countries year-round. Therefore, one would expect shorter virus persistence in the environment, and, according to the data obtained by using the guinea pig model, one would expect contact transmission rather than aerosol transmission of infl uenza in West Africa. Still, aerosol transmission may occur in Egypt because the RH is We considered an additional factor-animal densitywhile trying to determine why infl uenza might not be sustained in West Africa. The amount of eggs and bird meat and pork produced is considerably lower in Côte d'Ivoire, Benin, Nigeria, and Togo than in Egypt, Nigeria, Vietnam, and Indonesia. Thus, we hypothesized that a high animal density might be required for sustained transmission of the virus. This high density would explain why subtype H5N1 seems to have persisted in Nigeria, with its large avian commercial sector, for a couple of years while causing only sporadic outbreaks in neighboring countries. Our hypothesis is in agreement with the recent fi nding that infl uenza prevalence in Egypt is higher in commercial fl ocks than it is in backyard fl ocks (36).
If infl uenza virus transmission is limited by climate and animal density, then these limitations should apply to other similar pathogens. In that case, the high prevalence of NDV in Benin and Togo is surprising. However, we had a sampling period bias because all of the NDV (and noninfl uenza virus) from screened specimens from Benin and Togo was from specimens collected during January-March when NDV is known to cause disease in the western African backyard sector. Moreover, Songer et al. showed that NDV aerosol transmission at 23°C is better at 10% RH than at 90% RH or 35% RH (39). Although we lack data on whether NDV transmission differs from infl uenza transmission, different viruses, even those with the same nucleic acid core, might have different sensitivities to aerosol generation, depending on the RH level (39). Further experiments are warranted to determine whether NDV and infl uenza A virus transmission patterns actually differ and whether temperature and humidity have any role in that process.
We cannot exclude the possibility that infl uenza might exist in areas that we did not check. Such infl uenza hot spots have been discovered for infl uenza, including 1 at Delaware Bay in the United States (40). The requirements for continued circulation of infl uenza virus in animals (and the role of domestic animals in maintenance and interspecies spread) are not well understood. Future surveillance in West Africa should include more collection sites and include the commercial sectors and wild bird population to survey putative faster transmission and new introductions.  Our systematic year-round active infl uenza surveillance program in the backyard sector in Côte d'Ivoire, Benin, and Togo showed a prevalence of 0 (95% CI 0-0.04% to 0-1.48% in birds and of 0-0.28% to 0-5% in pigs). We hypothesize that the combination of climate and animal density factors might be responsible for what appears to be the absence of infl uenza virus in the backyard sector of the 3 countries we studied.