Are current avian influenza vaccines a solution for smallholder poultry farmers?

Vaccination against highly pathogenic avian influenza (HPAI) viruses, along with other measures, was successful in eradicating AI in very few countries where the competence of national veterinary services or the geography and bird density have contributed favorably to the outcome. The main constraints to an effective AI vaccination are vaccine composition matching field strains, reliable cold chain and logistics to target all poultry smallholders, constraints related to the availability of sufficient financial and human resources. When not conducted properly, vaccination can also contribute to the emergence of new field viral strains, through genetic drifts of HPAI viruses. While new technologies have improved the possibility to produce high quality vaccines matching field strains, recurrent issues like post-vaccination field surveillance and vaccination coverage continue to limit the relevance of AI vaccination in smallholder settings. A “game-changer” vaccine targeting smallholders should be universal to protect against all field viral strains and reduce significantly, if not totally eliminate, the need for costly post-vaccination surveillance. The ease of administration of this vaccine (eye drop or one single injection) would further contribute to its relevance in the field. These characteristics are considered essential for the product profile of an AI vaccine that can contribute in a meaningful way to the livelihoods of poultry smallholders.


Introduction
Although more than 30 epizootics of high pathogenicity Avian Influenza (HPAI) have been reported in poultry (Gallus domesticus) and other birds since 1959, vaccination of Poultry has only been added as a control tool since 1995 (Swayne et al., 2014). In recent outbreaks, vaccination has been used only in about 19% of the countries experiencing HPAI 15 countries out of 80 countries), showing that vaccination is not the most common and immediate response to an outbreak (Swayne et al., 2011). The nature of the AI virus, with the rapid emergence of new field viral strains, through genetic drifts of HPAI viruses, has affected the effectiveness of vaccines. Vaccination along with other measures has been successful in eradicating AI in very few countries where the competence of national veterinary services (e.g. France, the Netherlands) or the geography and bird density (Hong Kong) have favorably contributed to the outcome. In the four countries (China, Egypt, Indonesia and Vietnam) where massive vaccination campaigns have been initiated since 2004 (accounting for more than 99% of the use of AI vaccines), HPAI virus is now endemic, outlining the difficulty to eradicate AI in countries with a high percentage of backyard poultry (Swayne et al., 2011). A number of factors are limiting the effectiveness of AI vaccination in smallholder settings and are discussed in the following sections along with the characteristics of the ideal vaccine to target the poultry smallholder segment.

AI -etiology and epidemiology
Influenza viruses belong to the Orthomyxoviridae family, causing respiratory disease of the upper respiratory tract in humans, avian species, and a variety of mammal species. Orthomyxoviruses are classified as Types A, B or C with Avian influenza caused by a highly mutable Type A influenza virus.
The replication of this single stranded RNA virus is highly variable, resulting in a constantly evolving and highly mutable virus (Swayne et al., 2013). The lipid envelope of the virus makes it unstable and relatively susceptible to environmental destruction from ultraviolet light, chemicals, or desiccation. The hemagglutinin (HA or H) antigen is a protein providing the mechanism for the entry into the host cell while the neuraminidase (NA or N) protein allows the exit of newly replicated virions from the host cells. The HA protein is the major antigen stimulating the host immune response with protective antibodies against clinical signs and mortality. There are at least 16 H strains and 9 N strains, which can result in 144 possible HN combinations. This forms the basis of the serological classification using the hemagglutinin inhibition and neuraminidase inhibition tests. The nomenclature of the virus is based on the HN subtype, influenza type, host species, sample location, strain number and year of isolation, e.g. H5N1 A/goose/Guangdong/1/1996. The antigenic variation of the HA and NA surface glycoproteins occurs at a high frequency through minor "drift" changes and may be associated with the immune pressure exerted by the vaccination of birds. Major antigenic "shift" in the HA and NA coding proteins, the result of genetic re-assortment between gene segments of two different influenza virus strains (subtypes) in host cells, commonly occurs especially when domestic waterfowl and poultry are in close proximity as often seen in developing countries under smallholder settings. Ducks are often silent carriers of the AI virus, constituting an additional challenge for the control of AI (Swayne et al., 2013). AI viruses in poultry are classified as either low pathogenic (LPAI) or highly pathogenic (HPAI), based on the clinical signs and mortality, using the definition established by the World Organization for Animal Health (OIE, 2018). During the period January 2013 to February 2019, HPAI in domestic poultry was reported in 68 countries and territories with a total of 7,270 outbreaks (OIE, 2019). In most of the countries in Asia and Africa, the poultry sector is dominated by smallholder farmers.
During the same period, a total of 12 subtypes of AI viruses were reported (OIE, 2019), with subtypes H5N1, H5N2 and H5N8 widespread and more frequently reported (Table 1).
LPAI outbreaks with viruses of H5 and H7 subtypes are also reportable to the OIE as there is a risk of the viruses becoming highly pathogenic by mutation. During the period January 2013-February 2019, the vast majority of these outbreaks were reported in Asia, Europe, the Americas and a few in Africa (mainly in South Africa). H5N1 HPAI viruses have been sub-classified according to Clades, or groups of AI viruses that share a common ancestor (WHO, 2011). There are at least 10 Clade groups currently identified by the OIE and FAO network of expertise on animal influenza (OFFLU). Both antigenic shift and drift are important mechanisms for the evolution of the virus. The presence of co-circulating subtypes among dense populations of birds adds extra pressure for antigenic shift. Intrinsic subtype specific antigenic drift is associated with the frequency and distribution of infection in a poultry population as naive populations become exposed to new variants. Vaccination is also believed to exert a selection pressure on the virus by increasing the mutation rate by several orders of magnitude (Swayne et al., 2014). In addition, the evolution of new clade types can change the morbidity and mortality. Prior to 2012, clade 2.2 (Indonesia) and clade 1.1 (Cambodia) were predominant, with mortality in domestic ducks of less than 10%. Following the introduction of clade 3.2.1, the mortality reported in ducks is greater than 40% and up to 90% depending on the age of The type of poultry production is also critical to assess the risk of exposure. Birds kept for longer periods (laying hens, breeder birds and slow-growth meat birds) have a longer duration of potential exposure to AI virus than short-lived poultry (broiler-type meat birds). The concurrent presence of immunosuppressive agents and conditions for example, reduce the infectious dose and has been associated with morbidity and mortality due to both LPAI and HPAI infections. Therefore, general health screening for concurrent diseases in poultry, often difficult to achieve under smallholder settings, is important during immunization and for the overall control of avian influenza.
AI outbreaks during the period January 2013-February 2019 resulted in the loss of approximately 128 million birds, with more than half (57.6%) of the reported losses in Asia, followed by the Americas (22.1%) and Europe (13.4%) (OIE, 2019). Losses in Africa accounted for only 6.5% of the total losses ( Table 2).  , 2015) and presents the advantage of being safe to use in BSL-2 production facilities, more readily available in developing countries.

AI vaccines and vaccination
The second alternative applies the reverse technology to produce live ND vector vaccine viruses, often using the more aggressive LaSota vaccinal strain as the backbone, which express HA genes (H5, H6, H7 and H9 inserts). One of the primary benefits of this alternative is that the live virus replicates on mucosal surfaces, can be administered by mass application such as water or aerosol application, thus reducing the overall cost of vaccination and may be administered to other species of poultry, especially ducks. Conversely, one of the main concerns is related to the level of maternal ND antibodies in chicks which may interfere with the replication of the recombinant ND viruses and overall efficacy of the vaccine. In addition to being considered as genetically modified organisms, the potential of live ND vectored vaccines to spread to non-target species and to unvaccinated flocks often raises concerns during the regulatory review and vaccine licensing process (Suarez et al., 2017).
There are several new experimental AI vaccine approaches not currently licensed for commercial use including wild type or attenuated LPAI; use of various vectors e.g. adenovirus, salmonella, avian leukosis and vaccinia; eukaryotic systems e.g. plants; and DNA vaccines, with a main objective being a universal vaccine covering against all subtypes of HPAI. Of course, this would be an ideal scenario. However current opinion is that there is little expectation that such vaccine will be developed to licensure in the foreseeable future.

Multivalent vaccines commercially available
Review of the vaccine database from The Center for Food Security and Public Health, Iowa State University showed that of 43 manufacturers listing vaccines for the control of Avian Influenza, 31 (72%) are currently manufacturing vaccines combining AI with other antigens (CFSPH, 2018). These manufacturers are based in 11 different countries on five continents but the majority of them (71%) are located in Asia (22 of the 31) where AI vaccination has been practiced since 1995. Manufacturers located in Europe and North America are producing vaccines mostly for export and for the setting-up of emergency stocks. There is currently just one manufacturer of AI vaccines on the African continent, although it is believed that others are at a planning stage.
The combination vaccines produced are either AI + ND only (26 manufacturers, 84%), AI +ND + other antigens (15 manufacturers, 48%) and AI + another antigen but not ND (3 manufacturers, 10%), these numbers accounting for the fact that some manufacturers are producing more than one type of AI combination vaccine.

Use of AI combination vaccines by the poultry sector
Based on the list of antigens combined with AI presented previously, it is evident that these vaccines are mostly used in the egg laying and breeder poultry sectors and not by poultry smallholders. As might be expected from this wide range of vaccines and the disease challenges experienced in the field, there are a multitude of vaccination schedules used by poultry producers.
As an example, a recent study in Indonesia by Tarigan  were not protected throughout production. Without vaccination after 19 weeks, the birds failed to be protected after 38 weeks of age; with vaccination after 19 weeks, the birds failed to be protected around 58 weeks of age. The data clearly demonstrated showed that with laying or breeder birds the duration of immunity is one of the main limitation issues for life cycle protection in the field, especially in the smallholder setting.
Among the recombinant viral-vectored vaccines, the rHVT-AIV has the advantage that it can be used in day-old chicks at the hatchery. This vaccine has shown some good results in field studies in Egypt and is currently licensed in five countries (Bangladesh, Egypt, Mexico, Vietnam and the USA (von Dubschuetz, 2013).

Need for regular AI vaccine reformulation
Due to the antigenic drift of the AI virus in the field, there is a need to constantly update the composition of AI vaccines. This can be achieved only through constant monitoring of the virus strains in the field, rapid regulatory review and approval process and, good manufacturing processes. As the largest user of AI vaccines, the evolution over time of the vaccinal strains used in China is quite indicative of the need for regular reformulation, both for killed and live vaccines ( The key learnings from these national AI vaccination campaigns are summarized below: • Antigenic drift occurs with all AI viruses and can reduce the effectiveness of vaccination over time. It is essential to use vaccinal strains with sufficient quantities of antigens reasonably well matched with circulating strains of AI virus. • Vaccination with high-quality, registered vaccines, according to established protocols reduces resistance to infection, decreases viral shedding and decreases the probability of infection to poultry, other animals and humans.
• Vaccination only partially reduces viral shedding and can promote, especially if not conducted correctly, depending on the choice of vaccine, mode and frequency of administration, the selection of mutation in the circulating virus.
• Vaccination alone cannot eliminate the virus and is only meant to be part of an integrated, holistic control program adapted to local conditions • It is difficult to maintain high level flock immunity in some poultry populations, especially at small production and backyard levels.
• Vaccination is logistically demanding, and additional costs will be initially incurred in countries lacking efficient cold chain distribution networks required for most vaccines.
• Vaccination is expensive due to the need to conduct high quality post-vaccination surveillance to monitor the genetic and antigenic characteristics of circulating field viruses.
• The return on investment of vaccination must be carefully considered before adding vaccination as part of an AI control programme.
In Indonesia, a survey of commercial egg producers (Brum, 2013) showed that AI vaccinations were ineffective for the following reasons: poor selection of vaccinal strains vs. field challenge, first AI vaccine given too late and, not enough booster vaccinations, especially during the period of egg production (when vaccination will cause a reduction in egg production). Farmers also lacked independent high-quality technical support and 55% of the respondents said that they needed better information on vaccine selection and the vaccination schedule.
Regarding the situation in countries with large backyard poultry production, Alders et al. • Backyard poultry are extremely hard to reach, and the exercise remains extremely costly. Thus, vaccination should not be started if adequate funding for the preparation, implementation and monitoring cannot be guaranteed.
• As a rule, it is not recommended to vaccinate backyard poultry since maintaining adequate levels of immunity is extremely difficult.
• Community vaccinators are often reluctant to vaccinate backyard birds under 2 months of age for fear of killing them, thereby missing about 50% of the birds at risk.
• Vaccination is not a substitute for other important measures like biosecurity and can give farmers a false sense of security.
• Poor vaccination practices, insufficient vaccination coverage and lack of post-vaccination monitoring (as often seen in developing countries with large backyard poultry populations) can contribute to an endemic situation.

Overview on government AI control
Information on national regulations and policies related to AI vaccination is available via the World Animal Health Information System (WAHIS) interface hosted by the OIE (OIE, 2019). This database is built from information submitted to the OIE by its 184 country members. Among the parameters recorded, the following information is relevant to this paper: vaccination prohibited and official vaccination. It must be noted that official vaccination is not necessarily the converse of vaccination prohibited. Vaccination against poultry diseases may be the result of official vaccination and / or voluntary vaccination.
According to the OIE website (OIE, 2019), the countries with official AI vaccination programs are Mexico, Egypt, Pakistan, Kazakhstan and Russia. China also has official AI vaccination. Vaccination is also allowed in Bangladesh where import permits are delivered to the private sector and killed AI vaccines imported (no combination vaccines used). Vietnam, Egypt and Indonesia have transitioned from mass vaccination to targeted vaccination (OFFLU, 2013). The cost of AI vaccination is often shared between governments and the private sector. For instance, in Vietnam, farmers with >2,000 birds pay the full cost of vaccination while farmers with <2,000 birds receive some provincial subsidies.
Regional organizations like the African Union and the Association of South-East Asian Nations do not have specific policies regarding AI except for the desire to strengthen the linkages within countries and across borders, for the sharing of information and knowledge and for the development of partnership between all stakeholders in public and private sectors and civil society.
At the intergovernmental level, the Network of Expertise on Animal Influenza (OFFLU), a joint OIE-FAO technical committee with a worldwide network of contributors, issued in 2013 a series of recommendations on vaccination and vaccines (OFFLU, 2013).

The ideal AI vaccine candidate for poultry smallholders
The desired attributes (product profile) for AI vaccines targeting the smallholder segment are presented in Table 5.
Based on the views of a number of experts from academia, governments, intergovernmental organizations and the private sector, the top desired attributes for the smallholder segment are listed in Table 6. Maternal antibody to AIV hemagglutinin or virus vector inhibits primary immune response. Initial vaccination must be timed for declining maternal antibody titers to allow optimal primary immune response Given at 1 day of age in hatchery or in ovo Inactivated vaccine provides poor protection when given at 1 day of age. Vectored vaccines can be given at 1 day of age, but generally require a boost with inactivated vaccine 10 days or more later

Universal vaccine
The majority of inactivated whole AIV vaccines use reverse genetic generated vaccine seed strains to antigenically match field viruses. The vaccinal strain of virus should also be a strong immunogen Thermostable Killed AI vaccines, rNDV-AI and rFPV-AI vaccines require refrigeration and rHVT-AI vaccine must be stored in liquid nitrogen Table 6. Desired attributes of AI vaccines addressing the smallholder segment.

Universal vaccine
The antigenic drift of the AI viruses in the field requires a constant surveillance usually poorly performed by national veterinary services in developing countries due to the lack of proper facilities, human and financial resources. In addition, the availability of universal vaccines would eliminate the need to regularly update vaccines, often affected by inefficient national regulatory review and approval processes for vaccines

Use in multiple species
Ducks often act as silent carriers, shedding AI virus without expressing clinical signs. An effective vaccine in ducks as well as chickens would be of greater value, especially in South East Asia

Single dose protection
The biggest logistical hurdle and highest cost of AI vaccination is related to the handling and injection of each bird. A vaccine conferring at least a 6-month protection after one single administration would maximize the value of mass vaccination campaigns Due to the genetic drift of the AI viruses, the current commercial vaccines have a limited geographical coverage and use. The development of a universal vaccine would allow for the same commercial vaccine to be used in all major countries currently vaccinating against AI (China, Egypt, Indonesia, Vietnam and Mexico). Such a vaccine would be also available in countries considering adding vaccination to their AI control measures even though the current trend shows that fewer countries are willing to initiate AI vaccination.

Conclusion
Vaccination against HPAI has been used as one of the tools in national AI control programmes with varying degrees of successes. Few countries have been able to eradicate HPAI and these countries have typically relied on extremely competent national veterinary services and / or had limited poultry population at risk. Current vaccines, when well-matched to field strains (in quantity and quality), administered properly (timing, frequency and method of administration) and combined with a wide range of other measures have prevented financial losses in poultry farms and reduced the risks to human populations.
The cost of AI vaccines is only a small part of the overall cost of vaccinating poultry against AI, especially when targeting poultry smallholders in lower-and middle-income countries. The current vaccine production technologies and vaccines have not been able to address the cost inherent to their use. In order to represent a significant advance, the profile for vaccine candidates must include protection against all viral strains (universal vaccine), efficacy in ducks with reduction in viral shedding and, protection after a single administration. Without the inclusion of these properties in the vaccine profile, it is unlikely that any vaccine candidate would add significant value to the current AI vaccine market and thus warrant the investment of funds for its development.

Data availability
No data are associated with this article.

Disclaimer
The views expressed in this article are those of the authors. Publication in Gates Open Research does not imply endorsement by the Gates Foundation.

Leslie D. Sims
Asia Pacific Veterinary Information Services Pty. Ltd, Montmorency, Vic, Australia The paper provides an interesting commentary and suggestions on vaccination against highly pathogenic avian influenza. In general, the recommendations are reasonable that better vaccines with the properties described would be valuable. However, there are areas where revision should be considered given the population of farmers targeted has not been fully defined, all reasons for vaccination have not been covered, the authors have not always provided sufficient balance between the problems associated with vaccination and the solutions available, and could have provided more information on new developments in vaccination. In addition, availability of suitable vaccines is not necessarily the main reason for not applying vaccination.

○
The paper focuses on mass public sector vaccination programmes, in particular vaccination against Gs/GD/96-lineage highly pathogenic H5Nx avian influenza. However, in several countries where these viruses are endemic and vaccination is used against this disease, no formal government programme is in place (the role of government is to approve vaccines for use). It is up to individual farmers to purchase and administer the vaccine (or get someone to do it for them).

○
As the authors point out, vaccination of small flocks against avian influenza can be difficult due to logistical difficulties, and rapid turnover in multi-age flocks. It has been demonstrated in Egypt and Indonesia that it is not sustainable as a government-sponsored programme. However, there is no technical reason why a well-matched, relatively inexpensive, killed antigen vaccine cannot be administered to birds owned by small scale producers and interested in protecting their flock provided there is a suitable veterinary professional or paraprofessional to perform the work and the owner is willing to pay for vaccination. This is being done already or is potentially available in both Vietnam and China -two countries that first introduced mass vaccination against Gs/GD/96-lineage H5Nx viruses. Issues that arise are those of application, including availability of a well-matched vaccine, providing sufficient doses at an appropriate time, a lack of interest/demand for vaccination from some very small scale, low input producers. Field observations and reports ○ on mass vaccination in places with large numbers of smaller scale producers do suggest it is not always done well, based on levels of antibodies reported in places where these have been measured post-vaccination. Nevertheless, a community-based veterinary paraprofessional could provide this service to any household in much of SE Asia if the demand was there and vaccination was allowed (which it isn't in a number of countries where Gs/GD/96-lineage H5Nx viruses are endemic or where cases recur).
The term "small scale producers" is not defined. If this refers to household flocks of, say, 10 to 30 birds reared extensively then there are many logistical problems in delivering vaccine to these flocks. Many are multi-age, owners are not interested in vaccination and delivering vaccine to all birds at the appropriate age is difficult. One problem is that vaccine is produced for larger flocks in multidose bottles (see recommendations from OFFU meeting in Beijing). Another part of the problem is that the amount paid for administering vaccination in some places is too low to justify the expense of travel by veterinary paraprofessionals. However, in places where highly pathogenic avian influenza occurs seasonally and regularly, it would be possible to provide vaccination coverage to flocks in an area if required/demanded.

○
In places where outbreaks occur sporadically the signals and need for vaccination are weak.

○
Much of the infection with Gs/GD/96-lineage virus occurs in domestic ducks and for many strains the level of mortality produced is relatively low and age-dependent. There is little incentive for duck farmers to vaccinate unless they have experienced losses from this disease previously or have experienced losses due to government action when an outbreak occurs on nearby farms (such as whole village culls).

○
Even if a universal HPAI vaccine for poultry could be developed there is no guarantee it would be used by all producers, especially those rearing poultry in low input systems or for short-lived birds, especially if there is an outlet for birds in the event of an outbreak. Many farmers will sell birds when they see early signs of mortality and those with very small flocks are less likely to vaccinate than those with larger flocks (Delabouglise et al., 2020 1 ). In low input systems the loss of some birds is expected.

○
Vaccination is a cost to producers and in many countries rearing of broilers is a low margin business. This will result in some farmers choosing not to vaccinate even if effective vaccines are available.

○
The paper highlights the potential downsides of vaccination without necessarily balancing this with information on methods to overcome the concerns. For example, much is made of the antigenic variation restricting the effective life of vaccines. This has been the case in several countries in which the virus remains poorly controlled but not in others. Action has been taken to ensure appropriate vaccines are available in most countries albeit more slowly than required. Several vaccines already offer broad cross protection against different clades within the Gs/GD/96-lineage.

○
The paper does not cover one of the biggest constraints to vaccination which is the right for farmers to access vaccines. In a number of countries, H5 avian influenza vaccine cannot be used legally despite the virus being endemic.

○
There is also no mention of ring vaccination, as used in some countries (but discouraged in others) around outbreaks. There is also no discussion on the benefits of having immune birds delivered to live bird markets as has been seen with both H5Nx and H7N9 viruses reducing the likelihood of infection in birds and spill over to humans.

○
The following points on the authors text (in italics) should be considered if/when revising the paper. Note that some of the references cited were published after this paper was submitted but should still be considered. Papers cited on advances in vaccines are only examples.

Title:
The article is discussing vaccination against highly pathogenic avian influenza so this should be included in the tile.

Vaccination against highly pathogenic avian influenza (HPAI) viruses, along with other measures, was successful in eradicating AI in very few countries where the competence of national veterinary services or the geography and bird density have contributed favorably to the outcome.
The objective of vaccination against highly pathogenic avian influenza in most places was not elimination but, rather, to reduce levels of infection and the threat to humans. This decision was made because it was not possible to eliminate the virus using the methods generally applied, based on stamping out and movement management. Vaccination was used in Hong Kong and Mexico to help eliminate particular strains of highly pathogenic H5 avian influenza virus and in Hong Kong to supplement measures in place to prevent infection in markets.

○
Eradicate vs eliminate? WHO uses the former for global "elimination". I prefer use of elimination.

○
The main constraints to an effective AI vaccination are vaccine composition matching field strains, reliable cold chain, and logistics to target all poultry smallholders, constraints related to the availability of sufficient financial and human resources.
Matching vaccines to field strains is not as important as proposed here in all places where vaccine is being used (it is an issue that can be overcome, and methods are available for rapid production of matched vaccines). In southern Vietnam, the same vaccine remained suitable from 2005 for at least 5 years against clade 1 viruses until the dominant strain changed due to introduction of a new clade. Cold chains are important but not a constraint for delivering other poultry vaccines that require similar storage in SE Asia. Vaccination does not need to target all small-scale producers if the goal is disease containment. The cost of vaccination is only one of the issues. Other important constraints to use of vaccine are the life span of birds (e.g. very short life span for broilers in places such as Indonesia), interference from maternal antibody and as mentioned above not having the right to use vaccines due to national bans (e.g. not currently permitted in a number of Asian countries even in countries where the virus is endemic).

When not conducted properly, vaccination can also contribute to the emergence of new field viral strains, through genetic drifts of HPAI viruses.
Note that antigenic variation also occurs when vaccination is not used -e.g. H6 in China (Wang et al., 2014 2 ). Antigenic change will occur slowly in places where vaccinated or immune birds are rarely exposed to virus. In Vietnam and Indonesia, the bigger driver of vaccination failure was the introduction of a novel strain of virus from outside the country. Antigenic drift was not a factor for Clade 1 viruses and vaccines in southern Vietnam over a number of years. Circulation of multiple strains of virus is also a constraintmatching becomes more difficult when more than one antigenic variant is circulating.
○ Would use of vaccine be regarded as improper in places where infection is endemic and high-level population immunity across areas is hard to achieve but good immunity at the individual farm level is achievable? Much depends on the purpose of the vaccination programme. It would be good to have such a vaccine, but we can still vaccinate smallholder flocks even with existing products, provided care is taken to use the most appropriate vaccine. The main post-vaccination surveillance required is detection of antigenic variant strains. This is not expensive. Surveillance is only costly if the goal is virus elimination and detection of all infected flocks. This is not the case in places where viruses remain endemic. Surveillance programmes recommended for use in Europe are expensive and can be a barrier for use of vaccines.

Although more than 30 epizootics of high pathogenicity Avian Influenza (HPAI) have been reported in poultry (Gallus domesticus) and other birds since 1959, vaccination of poultry has only been added as a control tool since 1995 (Swayne et al., 2014). In recent outbreaks, vaccination has been used only in about 19% of the countries experiencing HPAI 15 countries out of 80 countries), showing that vaccination is not the most common and immediate response to an outbreak (Swayne et al., 2011).
Vaccination is not (in most situations) the most common response. However, the reason for this low uptake is that in high income countries, if virus is detected early, stamping out is used successfully to eliminate the virus. There are some exceptions to this rule. Trade restrictions can also inhibit use of vaccines. There are also situations where vaccination could be used as an early response to an outbreak, but trade issues inhibit its use.  ., 2011).
The virus was endemic before vaccination commenced in these countries and was used because virus elimination was not possible owing to the nature of the poultry production and marketing sector (Sims, 2016). Vaccination did not result in endemic infection -it was a response to it. Note also that certain strains of virus have disappeared (e.g. clade 7 viruses in China). Note also that vaccination has been used in Bangladesh for a number of years now. Vaccination has also been used illegally in a number of countries where it is banned. The presence of a high proportion of smallholder flocks is not necessarily an impediment to elimination (e.g. Thailand) even when vaccination is not used. Note that there is considerably more usage of H9N2 vaccination across Eurasia than against H5Nx so when referring to AI vaccines ensure this states it is for HP viruses.

Orthomyxoviruses are classified as Types A, B or C with Avian influenza caused by a highly mutable Type A influenza virus.
And Type D. All type A viruses are highly mutable, not just avian influenza viruses

Vaccination is also believed to exert a selection pressure on the virus by increasing the mutation rate by several orders of magnitude (Swayne et al., 2014).
This is probably true in places where there is a high likelihood of infection in vaccinated flocks (e.g. Egypt) but not in situations where there is a good match between circulating virus and the vaccine antigen or where incursions of virus are infrequent. Experimentally it is possible to select for antigenic variants using antiserum suggesting a role for antibodies, but we still do not know the precise contribution of vaccination vs natural infection, especially in ducks.

The type of poultry production is also critical to assess the risk of exposure. Birds kept for longer periods (laying hens, breeder birds and slow-growth meat birds) have a longer duration of potential exposure to AI virus than short-lived poultry (broiler-type meat birds).
Also important is the level of biosecurity in the premises in which birds are reared. Longlived breeder birds tend to be a lower risk because of the higher biosecurity measures in place. It is easier to obtain high level immunity in long lived meat birds than in white feathered broilers given the difficulty in giving 2 doses of vaccine during the short life span of the latter.

○
In addition to being considered as genetically modified organisms, the potential of live ND vectored vaccines to spread to non-target species and to unvaccinated flocks often raises concerns during the regulatory review and vaccine licensing process Often?    Table 3 does not contain multiple antigens introduced recently -now up to Re-11 and Re-12 ○ for H5 vaccines. Three H7N9 antigens have been introduced -the last one Re-3 only a few weeks ago The most common bivalent or trivalent vaccine against HPAI is H5/H7 vaccine used in China. ○ Table 5: There has been consensus for a number of years that an improved vaccine specifically for ducks that also addresses another disease (DVE) would be preferable. One of the vaccines that does offer broad cross clade protection in chickens (HVT-AI) does not protect domestic ducks.

○
The OFFLU Beijing meeting concluded "Any country considering vaccination of household poultry should take into account the difficulties that have been encountered in implementing and sustaining these programs in some countries. Intensive programs have been implemented successfully for limited time period in some countries where public health was a priority." This should be included in the paper.

On universal vaccines:
The

AI -etiology and epidemiology:
"…16 H strains and 9 N strains…" -The H and N refer to subtypes, not strains.
○ "AI viruses in poultry are classified as either low pathogenic (LPAI) or highly pathogenic (HPAI)…" -Importantly, this applies to gallinaceous poultry (the testing is done in chickens and ducks are considered poultry but as stated by the authors, they are often silent carriers even for strains that are deadly for chickens).
○ "H5N1 HPAI viruses have been sub-classified according to Clades, or…" -The clade system for the goose/Guangdong/1996 lineage refers to the H5 HA only, the NA subtypes and lineages vary.

AI vaccines and vaccination:
"…3) HA DNA vaccine adjuvanted;…" -Although licensed in the US this vaccine platform has not been utilized in the field (as of C. 2020).

○
Need context as to why the NDV vector is the focus. It's not widely used in the field because of interference from immunity to the vector as discussed. Of the vectored vaccines, HVT and Fowlpox are most common.
○ "…rHVTAIV has the advantage that it can be used in day-old chicks at the hatchery." -And can be also be administered in ovo.
○ "Vaccination with high-quality, registered vaccines, according to established protocols reduces resistance to infection…" -Should this be "increases" resistance? There is not much work on vaccination and infectious dose for AIV, what data are available (for inactivated vaccines) suggest that vaccination doesn't affect the infectious dose, which is consistent with how the vaccine induces immunity.
○ "Vaccination is expensive due to the need to conduct high quality post-vaccination surveillance to monitor the genetic and antigenic characteristics of circulating field viruses." -This is true of long-term vaccination, if the outbreak can be halted vaccine use may be stopped. However, surveillance may be conducted regardless of vaccination.
○ "As a rule, it is not recommended to vaccinate backyard poultry since maintaining adequate levels of immunity is extremely difficult." -Alternatively, if the birds have some immunity it might contribute to better food security and cost-benefit by improving survival and maintaining production if there is a high level of virus challenge in the field. ○ Table 5: "Mass application" -The only AIV vaccine which might be given by spray application are avian paramyxovirus type 1 (NDV) vaccine vectored products which is a very small proportion of vaccines that have been utilized (i.e. well below 4.5%). ○ Reviewer Expertise: Virus evolution, AI molecular epidemiology, vaccine strain selection.
I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.
Reviewer Report 12 October 2020 https://doi.org/10.21956/gatesopenres.14370.r29700 © 2020 Busani L. This is an open access peer review report distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Luca Busani
Dipartimento di Malattie Infettive, Istituto Superiore di Sanità, Rome, Italy In the manuscript "Are current avian influenza vaccines a solution for smallholder poultry farmers?" the authors present the situation of the Highly pathogenic avian influenza, a list of limiting factors for the vaccination in smallholder settings and their opinion on what the characteristics of the ideal vaccine to target the poultry smallholder segment should have.
The paper is presented as an overview, without any kind of analysis or original data presented.
The paper is well written and the items are clearly presented, but I would suggest to the authors to add some details about the focus of the manuscript, that is the smallholder poultry farms: in particular a clear definition (FAO presents some descriptions in the Chapter 1 "The socio-economic Importance of Family Poultry" http://www.fao.org/3/y5169e/y5169e02.htm#TopOfPage) and descriptions (backyard flocks are often multispecies and multiage, with poor or no biosecurity measures, a complex setting for vaccination). This is in my opinion important, to give to the reader a clear picture of the challenges of vaccinating smallholder flock.
Moreover, the background presents information on avian influenza epidemics and vaccine schedules, but mainly for commercial poultry. Among the positive key factors for the successful control of AI in poultry, the veterinary services capacity is mentioned, but I would also suggest to add the need of lab capacity for diagnosis of poultry diseases and the regular use of the lab by vet public services, whenever possible.
In table 6, at the universal vaccine rationale a possible interpretation is if such universal vaccine would become available, also the surveillance could be avoided. I would suggest to clarify this sentence pointing out the need of basic surveillance to ensure detection of the circulating viruses