Introduction

In late March 2013, when cases of human disease caused by an influenza A virus of the H7N9 subtype were reported from eastern China (Gao et al. 2013), it was soon evident that the outbreak bore many similarities to events in Hong Kong in 1997. At that time, a highly pathogenic avian influenza (HPAI) virus (of the H5N1 subtype) crossed the species barrier from poultry to cause a severe, often fatal, zoonotic disease in humans (Shortridge et al. 1998).

Genetic analysis demonstrated that the influenza A(H7N9) virus was essentially an ‘avian’ virus (Gao et al. 2013), subsequent testing in markets revealed high levels of viral contamination (see, for example, Chen et al. 2013), as was the case in Hong Kong in 1997 (Shortridge 1999), and most of the human cases had visited or worked in live bird markets some of which were found to be housing infected live poultry. The Hong Kong outbreak in 1997 led to concerns about the potential for a severe human pandemic to arise directly from an avian virus (Shortridge et al. 1998) and similar concerns were expressed in 2013 (Uyeki and Cox 2013).

The number of human influenza A (H7N9) cases and the geographic extent increased dramatically during April 2013 prompting calls for actions to be taken to prevent the disease. Given the similarities between the two outbreaks, it was not surprising that measures taken in markets to prevent exposure of people to infected poultry were followed by temporary cessation of new human cases in both the H5N1 and H7N9 outbreaks. H5N1 HPAI virus was not eliminated from the broader region in 1997 and nor was the H7N9 virus in early 2013. In both outbreaks the initial activities did not prevent re-emergence of human cases months (H7N9) or years (H5N1) later (CDC 2003; Chen et al. 2013).

Although there were some differences between the two viruses [for example, the H5N1 virus was highly pathogenic, as defined by the World Organization for Animal Health (OIE 2013), whereas the H7N9 virus was not; humans appeared to be more readily infected with the H7N9 virus than the H5N1 virus] this outbreak demonstrated the importance of understanding and learning from the history of AI viruses (AIVs).

In January 2014, outbreaks of HPAI caused by a virus of the H5N8 subtype were reported in the Republic of Korea. Once again, it appeared that history was repeating itself with the outbreak having many similarities to the four previous HPAI outbreaks from 2003 onwards involving H5N1 HPAI viruses in this country (see Kim et al. 2012 for details of the earlier outbreaks).

To understand these and other AI outbreaks, not only is it necessary to understand the biology of influenza viruses but also the natural history of the hosts in which these viruses multiply and the different environments in which the hosts and viruses interact. This includes the anthropogenic factors that have influenced where, whether and how avian influenza (AI) viruses can replicate and transmit between wild birds and poultry, and between poultry and mammals, including factors influencing uptake and application of appropriate control and preventive measures for AI.

The history of AI, and in particular HPAI, has been well documented (see, for example, Alexander and Brown (2009); Halvorson (2009); Kaleta and Rülke (2008); Morens et al. (2013a); Sims and Brown (2008); Swayne (2007)). These reviews provide valuable information on specific events that have occurred over the past 140 years. However, the history cannot be fully appreciated unless it is viewed through an ecohealth prism that integrates the many elements that influence and contribute to the disease and infection complex caused by AIVs.

The three main components are influenza A viruses, their hosts (wild birds, particularly, but not only, Anseriformes and Charadriiformes, and acquired hosts, domestic poultry), the environments they inhabit and the anthropogenic forces that affect these three elements. In particular, it is important to understand the way the environment has changed over time and the effects this has on the biology of the viruses, and the diseases they cause. This environment is influenced by decisions by public and private sector actors affecting farming and marketing practices. As all of these elements are intimately linked, it is not possible to view any of these factors in isolation when considering this disease.

It is noteworthy that HPAI was eliminated in 1925 from the US even before much was known about the causative agent (except that it was a filterable agent) using classical disease control measures based on first principles of quarantine, hygiene, stamping out and movement control (Halvorson 2009). Yet, despite remarkable improvements in knowledge about AI since 1925, it has not been possible to eradicate HPAI viruses of the H5N1 subtype that were first detected in Guangdong province in 1996 (FAO 2011). This difference demonstrates the importance of understanding the factors that permit these viruses to survive and flourish and the many factors that determine whether appropriate control and preventive measures are or can be introduced. It also demonstrates that some elements have changed between 1925 and the early 2000s that made control of this disease more difficult.

AI represents one of the best examples of the need for a ‘One Health’ approach to understand and tackle disease. A study conducted in 1969–1970 bleeding Australian shearwaters (Puffinus pacificus) for evidence of infection with influenza viruses demonstrates that ‘One Health’ is not a new idea. The study was funded by the World Health Organisation (WHO) because of interest in determining the origins of human pandemic influenza viruses (Dasen and Laver 1970).

This issue of Ecohealth contains a series of papers covering aspects of AI that demonstrate the importance of adopting an Ecohealth approach to the study and control of this disease.

Understanding Avian Influenza Viruses

The term AI has been used to describe any infection or disease in birds caused by type A influenza viruses. AIV describes influenza A viruses found customarily in birds. Most AIV do not cause significant disease in their avian hosts but some do, either alone or in combination with other infectious agents.

The term ‘AI’ has been complicated by the decision by the OIE in 2013 to narrow the definition under the Terrestrial Code to ‘an infection of poultry caused by any influenza A virus of the H5 or H7 subtypes or by any influenza A virus with an intravenous pathogenicity index (IVPI) greater than 1.2.’ (OIE 2013). Previously the Code used this definition for the term ‘notifiable AI’. Here we refer to AI in its broader sense described in the previous paragraph.

Disease caused by AIV has been recognised as a distinct entity since the late nineteenth century and was shown to be caused by a filterable agent in the early twentieth century (Kaleta and Rülke 2008). It was not until the middle of the twentieth century that AIV were isolated and several decades later before the widespread carriage of AIV by clinically normal wild birds, and the link between AIV in wild birds and those in domestic poultry, and other species, became apparent. Disease caused by AIV only represents the ‘tip of a very small iceberg of infection’. Much of the infection in birds is sub clinical or at least extremely mild (Kuiken 2013).

The division of AIV into pathotypes—low pathogenicity avian influenza (LPAI) and HPAI—provides a simple system for characterising these viruses that remains relevant for terrestrial poultry, although it provides little indication of likely virulence of viruses in humans. The system was originally based on the capacity of these viruses to cause disease in experimentally infected birds inoculated intravenously, and later, on the amino acid sequence of the virus at the haemagglutinin (HA) cleavage site (OIE 2013). It has been demonstrated that viruses of the H5 and H7 subtype can convert from LPAI to HPAI strains, although this process is not inevitable or predictable. For example, low pathogenicity influenza A (H7N9) viruses have been circulating in poultry in China since early 2013 without conversion to a HPAI strain (to the time of writing in early 2014).

Once influenza viruses were cultured in laboratories it was not long before their remarkable genetic plasticity was demonstrated. Because of their capacity to change the nature of individual genes, these viruses (like many other RNA viruses) evolve quickly. This is achieved mainly through point mutations in individual genes. The poor fidelity of the polymerase complex results in errors in replication whenever influenza A viruses multiply—a process that facilitates production and selection of fitter variants. Evolution is also brought about occasionally by recombination in which partial genes are incorporated into existing genes, as has been seen with conversion of some LP viruses of the H7 subtype to HPAI viruses (Suarez et al. 2004). Even more dramatic is genetic reassortment made possible because of the segmented genome of influenza A viruses that allows incorporation of entire genes from other influenza A viruses when cells are co-infected with two or more different influenza A viruses. This process has been crucial in the genesis of both the H5N1 and H7N9 subtype viruses that have caused severe zoonotic disease in humans and has been a major factor in human influenza pandemics that arise as a result of so-called antigenic shift.

Over time, these processes have produced amazing diversity in AIVs. So far, at least 16 different H and 9N subtypes in various combinations have arisen, with marked variations within the H and N genes in each subtype, plus a broad range of different combinations of genes coding for internal proteins (Morens et al. 2013a). Additional subtypes have been identified in bats (Tong et al. 2013).

Understanding the genetics of AIVs has mirrored improvements in capacity to sequence influenza viruses and owes as much to studies of influenza viruses in other species, especially in humans, as it does to studies in birds. Some of the early work unravelling the antigenic and genetic code of influenza viruses appears remarkable today given the limited tools available to virologists and molecular biologists in the 1970s, in the era before polymerase chain reactions and sophisticated equipment for gene sequencing (see, for example, Laver et al. 1979; Hay et al. 1979).

Using rates of evolution as a guide, researchers have attempted to determine when influenza viruses and the various subtypes first emerged (see, for example, Suzuki and Nei 2002). Although it is not possible to pin point exactly when this occurred, it is likely to be tens of thousands of years ago, suggesting that influenza A viruses would have been present in ducks before domestication.

Genotypic and phenotypic changes are influenced by the hosts in which the virus multiplies and environmental factors. Since the advent of rapid sequencing, it has been possible to assess rates of evolution of genes and to determine whether certain genes are under selection pressure.

For example, genetic analysis has been used to assess potential effects of vaccination on rates of evolution (Cattoli et al. 2011). Changes in the viruses that tend to reflect the hosts in which they multiply include deletions in the stalk of the N protein and conversion of LP to HP AIV, both of which appear to be associated with multiplication in gallinaceous birds (Abdelwhab et al. 2013. Li et al. 2011) Other changes occur when AIV infect mammals including changes to the amino acid at position 627 of the PB2 protein that enhances pathogenicity in mammals. This change has been reproduced in a few passages of an avian virus in a mammalian host (de Jong et al. 2013). Gene sequencing also facilitates epidemiological studies and provides information on the evolution of viruses during outbreaks (Ypma et al. 2012).

Receptor specificity of the viruses that determine the cell types they bind to can also change, although the drivers for these changes are not always clear. Viruses of the H9N2 and more recently emerged H7N9 viruses have changes that facilitate binding to receptors that predominate in the upper respiratory tract of mammals (Gao et al. 2013; Matrosovich et al. 2001). AIV in ducks and other waterfowl are usually shed via the cloaca but some avian species infected experimentally shed more virus via the oropharyngeal route (Costa et al. 2011), and some influenza virus strains are shed in larger quantities via the respiratory tract than via the cloaca across a range of species including those of the H7N9 subtype (Pantin-Jackwood et al. 2013). The predominant route of shedding is important because it influences modes of transmission. HPAIV has also been detected in feathers from infected birds, and this has potential implications for transmission of virus over longer distances (Yamamoto et al. 2008).

Systemic infection, which is usually seen with HPAI viruses, results in wide dissemination of virus in infected poultry, although in some species the virus localises to the brain. Systemic infection is associated with the multiple amino acids at the HA cleavage site found in HPAI viruses that allows a broader range of proteases to cleave the virus HA, an important precursor to cell invasion, although other genetic factors unrelated to the HA also play a role (Abdelwhab et al. 2013).

The presence of systemic infection also resulted in transmission of H5N1 HPAI virus to carrion-eating birds as occurred in crows in a number of locations (Mase et al. 2005) and to mammals fed on uncooked poultry infected with H5N1 HPAIV, including tigers and other large felids in zoological/conservation parks (Keawcharoen et al. 2004).

Studies on survival of AIV in the environment have shown that under certain conditions these viruses could overwinter (Lebarbenchon et al. 2012) providing opportunities for infection of migratory birds returning to summer breeding grounds. Nevertheless some strains of virus die out in wild bird populations such as clades of H5N1 HPAI viruses that have infected migratory birds. These have not persisted for more than a few years before being replaced by other distantly related clades. Influenza viruses also vary in their tolerance to temperature (Brown et al. 2007).

Under hot conditions influenza A virus does not survive in the environment for more than few days which means that there must be sufficient hosts to maintain the virus in areas with a warm climate, as appears to be the case with Clade 1.1 H5N1 HPAI virus in the Mekong Delta, covering southern Vietnam and Cambodia. Soil type can also influence survival time of virus outside the host (Gutiérrez and Buchy 2012).

AIV have developed resistance to amantadine, and this was identified in places where this anti-viral drug has been used illegally for disease control in poultry (He et al. 2008). Concerns have been expressed about low levels of neuraminidase inhibitors such as oseltamivir in sewage following treatment with these drugs. Drug residues then find their way into waterways frequented by waterfowl. Experimentally, resistant influenza viruses can develop in mallards exposed to low levels of these drugs equivalent to those found in contaminated water (Järhult 2012). Illegal use of ribavirin has also occurred in poultry in China and this could eventually lead to development of resistance to these valuable drugs used in some cases for treatment of severe human influenza infections.

In summary, an understanding of the virus and the way the hosts and ‘environment’ in which the agent multiples and survives shapes the organism is essential. It demonstrates why adopting an ecohealth approach to AI is so important.

Understanding the Hosts

The seminal work of Easterday, Laver, Webster, Slemons and others identifying wild birds as the primary host of influenza A viruses was critical in understanding these viruses (Easterday et al. 1968; Laver and Webster 1973; Slemons et al. 1974). It is likely that most if not all influenza A viruses have an avian host somewhere in their past (the only exception at present are the H17 and H18 viruses detected only in bats). This applies even to those viruses that have evolved and are now well adapted as human, equine, canine or porcine pathogens.

The avian orders recognised as the main natural hosts of influenza A viruses are Anseriformes and Charadriiformes with some AIV serotypes more likely to be found in one or other order (Tønnessen et al. 2013). Infection has also been recorded in avian species from other orders (Fuller et al. 2010), involving both high and low pathogenicity viruses, and experimental studies have demonstrated a potentially broad host range for these viruses (Stallknecht and Shane 1988). In many cases, such as infection of humans with Influenza A viruses of the H5N1 or H7N9 subtypes found in poultry, the infection results from spillover of virus to aberrant hosts, rather than extension of the natural host species. A major public health concern is that one of these virus subtypes will adapt to mammals leading to efficient inter-mammalian transmission. Fortunately, this has not occurred yet.

Studies have shown that most wild birds do not remain infected with influenza virus for more than a few weeks. This means that for AIV to survive they either have to do so in a favourable environment and/or have sufficient susceptible birds in which to multiply.

The major host orders are migratory and this results in AIVs moving along migratory pathways. Some locations are now recognised as hot spots for influenza virus detection such as Delaware Bay (Bahl et al. 2013) where Charadriiformes congregate, in particular ruddy turnstones that are frequently infected. In some places limited introduction of new serotypes (or variants of existing serotypes) has occurred despite being on migratory pathways, as has been demonstrated in Australia with H7 viruses. The H7 viruses in Australia form a monophyletic (Bulach et al. 2010) and have done so for at least the past 40 years. This contrasts with viruses of the H10 subtype for which newly introduced viruses have been identified both wild birds and poultry (Vijaykrishna et al. 2013).

As a rule wild birds are not regarded as the natural hosts of HPAI viruses. Until 2002, the only situations in which wild birds were known to have been infected with HPAI viruses were (i) an outbreak in terns in South Africa, for which the origin of the virus remains unknown (Becker 1966), and (ii) occasional infected wild birds detected in or near to poultry farms experiencing outbreaks of HPAI. The latter were generally regarded as being the result of spillover from infected poultry (Alexander 1987).

From 2002 onwards, it became apparent that wild birds were capable of being infected with and transferring H5N1 HPAI viruses over relatively long distances as the virus moved initially from China to South Korea and Japan (2003), and in 2005 out of western China to Mongolia, Russia, Kazakhstan and onwards to the Middle East, Egypt, Europe and West Africa (Sims and Brown 2008). The virus also caused mortality in infected wild birds in a number of locations. The manner in which wild birds transfer these viruses over long distances remains unclear although it is apparent that birds can travel very long distances in a very short time during migration. Therefore, if infected just prior to these events, then the birds could carry a virus from one site to another. There are some reports of potential effects of influenza virus on migratory capacity but the circumstantial, biological evidence is clear that viruses have moved over long distances to places with no poultry, demonstrating the role of wild birds in virus distribution (Gilbert et al. 2012). Relay transmission may have played a role in the movement of virus across continents (Gaidet et al. 2010).

In attempts to understand the role played by wild birds much has been learned about the migratory pathways and interactions between different species at points along migratory paths (Cui et al. 2011). Weather conditions have been shown to correlate with movement of birds and viruses (Ottaviani et al. 2010). Farming of wild birds may also play a role in propagation of virus especially in places where cross infection of domestic ducks could occur.

Spillover of AIVs from wild birds occurs occasionally to domestic poultry, and in some cases viruses become established either temporarily or permanently in different species. For example, AIV viruses of the H9N2 subtype have become well adapted to poultry and are endemic across much of Asia. Other strains are also established in poultry (Pepin et al. 2013).

Quail and pheasants have been shown to play a potential role in modifying receptor specificity because they carry receptors for both avian and mammalian adapted viruses in the upper respiratory tract (Yu et al. 2011). Pheasants have also been shown to be able to shed some influenza virus subtypes for an extended period, up to 45 days for an H10 subtype virus (Humberd et al. 2007).

A range of mammals has been infected with AIV. One of the striking points regarding AIV is that they do not cross easily to humans. It has been recognised for some time that there have been millions of human exposures to H5N1 HPAI viruses but few productive human infections (Morens et al. 2013b). Genetic factors may play a role in whether infection in humans causes disease but additional research is needed in this area (Horby et al. 2013). There are also variations in the pathogenicity of viruses of the same subtype (e.g. H5N1 viruses) in mammalian models.

Given concerns about human exposures to AIVs, especially those that can cause severe disease, it is essential to understand how and where humans are being exposed and to implement appropriate measures to reduce this risk.

Understanding the Role of Environment and Environmental Factors

As discussed in the previous section wild birds play an important role in the maintenance and propagation of AIVs as well as the genesis of novel AIV variants through reassortment of genes from the large pool they carry. Wild bird migration depends on availability of wetlands and breeding sites that provide stopover points for the birds. Human development in many areas has resulted in shrinkage and fragmentation of wetlands and in some cases important sites have been lost. In addition a number of areas with high concentrations of poultry have developed in areas that play host to migratory birds, providing opportunities for transmission of influenza A viruses in both directions. Some areas where this is known to have occurred are Poyang lake in Jiangxi province of China, Fraser Valley in British Columbia, Northern Italy and the Republic of Korea. Much of this development was largely unplanned and occurred because these areas provided advantages for poultry production such as access to feed and water or ease of access to markets.

The available evidence suggests that migratory wild birds are not capable of sustaining H5N1 HPAI virus for more than a few years (Gilbert et al. 2012) but new strains of H5N1 HPAI have been introduced to wild bird populations since 2002, presumably from poultry at places where the two populations interact either directly or indirectly (via waste water from farms). As long as H5N1 HPAI viruses exist in poultry the opportunity for wild birds to become infected and transfer virus over long distances remains (APEIR 2013). This has major implications for surrounding countries that have experienced incursions via wild birds in the past. The role of farmed wild birds and song birds in this process remains unknown but is worthy of further exploration.

One factor that appears to have played a crucial role in influencing transmission of AIV has been the rapid expansion of poultry production, especially, but not only, in Asia. From the mid-1980s to mid-1990s the reported population of domestic chickens, ducks and geese all increased about three fold in China and some other Asian countries. Some of this expansion occurred in farms practicing appropriate biosecurity measures to prevent or minimise the risk of infection with AIV; much of it didn’t.

The expansion depended at least in part on the availability of cheap animal feed, successful selection of poultry for rapid growth poultry and demand for animal protein in diets driven by increased incomes. These events have been referred to as the ‘livestock revolution’ which describes the rapid evolution that occurred in animal production from about the 1990s (Delgado et al. 1999). Although this ‘revolution’ has been described as demand driven, many of the supply side factors to facilitate production of cheap animal protein via poultry were already in place, without which the marked expansion of the sector would not have been possible (Sumberg and Thompson 2012). Regardless of the underlying causes, the end result was massive expansion of poultry production providing an environment that was conducive for propagation and transmission of AIV.

This process first played out in western countries from the middle of the twentieth century (although intensive duck production was occurring in the US on Long Island in the early part of the twentieth century, well before this time) and is now being replicated on a massive scale in many other parts of the developing world. At present, a patchwork of production systems is in place in the emerging economies such as Vietnam, Indonesia, China and Egypt in which traditional production and marketing systems co-exist with intensively reared poultry raised indoors and sent direct to slaughterhouses rather than through wet markets where live poultry are kept, sold and slaughtered. Traditional marketing systems facilitate dissemination and persistence of AIV. There is a trend towards increased intensification of production that is seeing many small farms disappear, which has been recorded in Eastern China (Sims 2012).

In some cases, hybrid systems have developed in which production is intensive but marketing still occurs through live poultry markets, a system for which the rapidly growing broiler chicken is probably ill suited given the bird has been bred for western-style production and marketing/processing systems.

Many large farms have excellent biosecurity systems in place that limit the likelihood of infection with AIV. However, as examples from Mexico and the Republic of Korea show, these do not always prevent infection from occurring, even in modern breeder flocks which usually have highly sophisticated biosecurity measures in place. In many countries in Asia, biosecurity systems on many farms, especially but not only on smaller farms, are weak and provide opportunities for entry of AIV (APEIR 2013).

Duck production systems in many parts of Asia provide ideal environments for transmission of AIV (both LPAI and HPAI). Ducks share the same habitat as wild birds and in some places they are moved over long distances to graze on harvested rice fields. In places such as Cambodia in the dry season, ducks are concentrated on the few remaining watercourses providing ideal conditions for transmission of virus via contaminated water if infected birds are introduced to an area.

The early outbreaks of HPAI in Europe and the US in the late nineteenth and early twentieth century demonstrated that places where poultry congregate including markets and poultry exhibitions were important for persistence and/or transmission of AIV. As pointed out in the introduction measures taken to control movement of poultry to these markets were used effectively as part of the overall control and eradication programmes for HPAI.

Many human cases of influenza A caused by viruses of the H5N1 and H7N9 subtype occurred as a result of direct or indirect contact with poultry in live poultry markets. As a result of these cases, AI has shaped the nature of production and marketing systems. For example, the systems of rearing and marketing in place in Hong Kong today bear little resemblance to those in 1997 because of measures introduced successfully to reduce the risk of infection of poultry and humans with H5N1 HPAI virus (Sims and Peiris 2013). Changes have also been applied in other countries, including China and Vietnam, but in many developing countries the conditions in markets still provide ideal opportunities for persistence and transmission of influenza viruses. Places where poultry from different sources congregate and are allowed to remain for more than 24 h provide opportunities for viral infection and reassortment and persistence of virus if measures are not taken to break the infection cycle in which birds in the market infect newly arrived poultry.

As with many diseases, the tools are available to prevent terrestrial poultry from getting infected with AIV, as demonstrated by the many well-managed farms and markets that remain free from infection. However, these measures are not applied in all cases. Understanding the factors that motivate or interfere with implementation of these measures is crucial. Unless producers and traders see benefits in implementing measures and have the resources to do so they will not be introduced. A good example is the recommendation to restrict scavenging poultry in villages. This changes the production system and requires provision of feed to the housed poultry therefore raising the cost of rearing and has been resisted by most producers. Difficulties faced with implementing changes in a market in Bali, Indonesia are discussed by Naysmith (2014) in this issue. He concludes that further qualitative research is needed to understand why at-risk individuals fail to adopt biosecurity measures, even after recently experiencing an outbreak of AI.

The H7N9 virus that emerged in China in 2013 is also having major effects on marketing systems. A number of large cities in China are in the process of changing the way poultry are sold to prevent zoonotic transmission of this virus from poultry in markets to humans.

One major change has occurred in the past 15 years regarding HPAI. Prior to the emergence of HPAI viruses of the H5N1 subtype in Asia in the 1990s, all other HPAI viruses have been eradicated as a result of the measures implemented, and in some cases, this was facilitated because the virus had insufficient hosts in which to sustain infection. H5N1 HPAI has now been endemic in Asia for over 16 years and the prospects for eradication of this virus remain remote in the foreseeable future.

Until this event, HPAI was regarded as an eradicable disease and all contingency plans for HPAI were based on the premise that it would be eradicated. For a variety of reasons, classical methods of control based around stamping out were unable to eradicate the disease in a handful of countries including China, Vietnam, Bangladesh, Egypt and Indonesia whereas other countries were apparently successful including Thailand and Nigeria as well as more than fifty other countries (Food and Agriculture Organization 2011).

Understanding the factors that allowed the virus to persist in these places depends on adopting an ecohealth approach. Three broad factors have been identified as being common to and important in these countries (Food and Agriculture Organization 2011; Sims 2012). The first is the complex structure of the poultry sector with very large populations of poultry reared under different production systems, the second is the quality of veterinary and animal production services, and the third is the degree of commitment to virus elimination at all levels from central government to producers. These issues will not be overcome for a number of years.

By contrast, Thailand has been successful in controlling the virus but much of this can be attributed to the marked differences in the poultry production systems. The large export-oriented intensive industry had very strong incentives to control the virus, and the Thai poultry sector has greater integration of production, processing and marketing than that in other countries in the region. Differences in the structure of the poultry sector must be considered when building appropriate control and preventive programmes, and it is now evident that it will not be possible to eliminate H5N1 HPAI virus from some of the countries with enzootic infection in the foreseeable future.

As described earlier, the poultry sector is made up of many different, often overlapping production and marketing systems. The contribution of each of these to the overall burden of AI varies. Alders et al. (2014) argue in this issue that the role of village poultry production has been misunderstood. With improved understanding of the epidemiology of the disease, it is now recognised that village poultry raised under extensive conditions pose less of a threat than intensively raised poultry of homogeneous genetic stock with poor biosecurity. Nevertheless, many of the human cases of influenza A (H5N1) in Cambodia occur at the village level so if measures to protect public health are to be implemented the village sector must be included through implementation of practical preventive programmes.

Slaughter of sick poultry still occurs at village level in a number of counties and in places where this continues, alternative practices may need to be explored to reduce the risk as proposed by Rimi et al. (2014) in this issue, based on observations and studies in Bangladesh. Some progress is being made in assisting villages to adopt appropriate culturally sensitive measures so as to reduce the risk posed by AI to poultry and human health. Programmes such as the Healthy Livestock, Healthy Village, Better Life program in Cambodia aim to improve overall disease control in villages and reduce the incentive for slaughter of sick poultry or preparation of diseased poultry for food because more poultry survive.

Urbanisation is occurring rapidly in many of the countries where H5N1 HPAI is enzootic. Finucane et al. (2014) in this issue conclude urbanisation improves some aspects of life but simultaneously poses significant challenges for poultry farming and disease management.

Vaccination is one of the measures introduced for control of H5N1 HPAI and is discussed in this issue by Swayne et al. (2014). Vaccination was used rarely in the past for control and prevention of HPAI because alternative methods based around stamping out were effective in eliminating the disease. Vaccination was introduced as a response to enzootic infection when it was evident that virus eradication was a distant goal and that vaccination could help to reduce the quantities of circulating virus, therefore assisting in protection of public health.

Vaccination has been much maligned as a preventive measure for HPAI because it does not produce sterilising immunity but much of the criticism of vaccination stems from a failure to understand the purposes of vaccination programmes. It was never expected that vaccination alone would eliminate the virus. As with any control measure if vaccination is applied improperly, then its effectiveness will be limited, and there are examples where this has been the case (Peyre et al. 2009). Swayne et al. argue that improper vaccination has been more important that antigenic changes in vaccination failures. Understanding why vaccination is not applied correctly is also an ideal subject for investigations based on ecohealth principles.

One of the reasons why stamping out was not effective in eliminating H5N1 HPAI was that surveillance systems were not sufficiently well developed to detect all cases of disease. Mariner et al. (2014) in this issue discusses the role of participatory surveillance as a useful tool for gathering better information on disease occurrence.

A favourable outcome of the investments made into control and prevention of H5N1 HPAI by governments and international donors has been the marked improvements in laboratory capacity and quality standards especially, but not only, in places where this virus has occurred. Daniels et al. (2014) in this issue discusses the importance of laboratory networks and the role of OFFLU in the diagnosis of emerging infectious diseases, in particular influenza.

Much remains to be done to prevent AI especially those strains that pose a direct zoonotic and potential severe human pandemic threat. Risk assessment is one tool that can assist authorities to determine appropriate control and preventive strategies as discussed in this issue by Costard et al. (2014). Forster (2014) also points out in this issue that the solutions used in the past for AI management based around traditional veterinary services may no longer be the most appropriate mechanisms for dealing with complex diseases for which novel control, and preventive measures involving societal changes are probably required.

The H7N9 outbreaks in China since early 2013, which also threaten neighbouring countries, already demonstrate that changes are needed in poultry marketing systems if further human cases are to be avoided. Making these changes depends on understanding and taking into account all of the factors that influence decisions on how and where poultry are sold (FAO 2013), built on an understanding of the changing biology of AIVs. An ecohealth approach is required if appropriate solutions to these difficult and complex problems are to be found and implemented.