Introduction
The circulation of avian influenza viruses (AIV) across Asia is complex, involving the interplay of numerous risk factors. Reviewing the scientific literature, Gilbert & Pfeiffer (2012) identified three classes of variables associated with the incidence and spatio-temporal trends of HPAI H5N1 across studies and regions: domestic waterfowl, anthropogenic variables including factors such as human population density and distance to roads, and indicators of water presence. Other investigators reported associations with poultry population density (chicken as well as ducks), human population and rice cropping intensity (Pfeiffer et al., 2007; M. Gilbert et al., 2008; Loth et al., 2010, 2011). Once established, HPAI H5N1 virus spread is believed to be influenced primarily by local trade patterns, density and practices in wet markets or live bird markets (LBMs), the poultry production structure and value chain, and disease prevention and control systems (M. Gilbert et al., 2008). H5N1 HPAI outbreaks are highly seasonal in areas of South and South East Asia. This observed seasonality has been associated with higher temperature, poultry species, product movements and trade, and other anthropogenic factors (Pfeiffer et al., 2007; Minh et al., 2009; Tran et al., 2013).
Like many viral pathogens, HPAI H5N1 persists in multiple avian host species. Domestic ducks have been implicated as an important virus reservoir. Previous studies on Java reported them to be more than ten times as likely as chicken flocks to have seropositive birds, with scavenging being identified as a specific risk factor (Joerg Henning et al., 2010, 2016; J. Henning et al., 2013). A consequence of such a multihost system is that the identification reservoir hosts of infection, and knowledge of how these hosts enable and maintain persistence, is essential to inform targeted interventions (Haydon et al., 2002; Viana et al., 2014). The investigation of multihost systems is complex because they incorporate a number of interacting species, populations, and production systems (Viana et al., 2014). Moreover, they are influenced by a range of management practices, human behaviours and trade patterns. Very little is known about the year-round persistence of the H5N1 HPAI virus.
In Indonesia, Java has the highest density of both human and poultry (Anon., 2019). HPAI is endemic across many parts of the country. Central and West Java have been identified as hotspots for HPAI (Farnsworth et al., 2011). The poultry production system in Indonesia is highly diverse and is classified into four sectors. Sector 1 is comprised of vertically integrated commercial poultry farms that implement high-level biosecurity measures and always market the bird/products commercially. Sector 2 represents somewhat smaller-scale commercial poultry farms with moderate to high biosecurity; birds/products are sold through slaughterhouses or poultry markets. Sector 3 consists of small commercial poultry businesses which typically have lower levels of biosecurity, with birds that are usually sold through live bird markets. Sector 4 is classified as backyard poultry farms, often a subsistence or side-business enterprise, with minimal biosecurity and products mainly consumed locally (Azhar et al., 2010; Farnsworth et al., 2011; Hendra Wibawa et al., 2018).
Trade, the value chain and human behaviours have been recognised as influential drivers for the transmission of HPAI. In Indonesia, Indrawan et al. (2018) applied value chain analysis (VCA) to investigate factors related to biosecurity and HPAI control in Western Java. They identified the co-existence of four different poultry value chains, ranging from high levels of coordination, hierarchical governance, and regulated marketing and distribution mechanisms (Sectors 1 and 2) to informal, smaller-scale chains (Sectors 3 and 4) which had weaker HPAI prevention and biosecurity measures. Sector 3 and Sector 4 production often occur in close geographic proximity (Sims et al., 2005; Azhar et al., 2010), and birds are likely to move between these two systems. It is therefore suspected that HPAI is transmitted between these sectors, contributing to the difficulty in controlling the disease (Farnsworth et al., 2011; Indrawan et al., 2018). In a questionnaire-based study of poultry traders at LBMs in Bali and Lombok, Kurscheid et al. (2015) found that knowledge of viral transmission and biosecurity was generally low. Two thirds of respondents were reluctant to report sudden or suspicious bird deaths to authorities. The investigators established that there was a correlation between knowledge of HPAI transmission and prevention and the number of birds sold; education was strongly associated with better knowledge but did not influence positive reporting behaviour. Elsewhere in South East Asia, Fournié et al. (2012) recorded traders’ practices in Vietnam and Cambodia likely to influence the virus circulation in LBMs. These included factors related to trade patterns (e.g. the number of days during which traders were active, the length of time they spent at market in a day, the number of poultry sold within a day, the type of poultry etc.) as well as factors related to supply management (purchase and sale volumes, frequency and quantity of the surplus at the end of trading days, management of unsold poultry including sale to traders operating in other LBMs ). They showed that poultry traders with high surplus frequency and volume represented a risk group for perpetuating HPAI H5N1 in LBMs.
For a complex and multi-faceted disease such as HPAI H5N1, investigative approaches focusing on a discrete area of interest are unlikely to effectively generate conclusive insights. Hence, an integration of methodologies incorporating elements including population studies, surveillance techniques and data, pathogen genetics and an understanding of production systems (including high risk points for transmission) presents the best scope to further develop understanding. A consequence of such an approach is that multiple sources of data and information need to be combined. These may be extant, or such data may be collected for the specific purpose of the investigation; these data may be qualitative, quantitative or both. Integration of findings can be performed using techniques such as meta-analysis, mathematical modelling or mixed methods research. Triangulation of multiple sources is performed to assess validity, synthesize findings and generalise inferences (Farmer et al., 2006). Such methodologies are increasingly being applied in the field of animal health. Commonly, qualitative approaches include stakeholder interviews, questionnaires or focus group discussions whereas quantitative techniques include data generated by observational studies, animal health surveillance or disease control programmes (e.g. Limon et al., 2017, 2014; Mosimann et al., 2017; Walker et al., 2015; Wu, Kagoli, Kaasbøll, & Bjune, 2018).
The overall objectives of this study were to determine the drivers of H5N1 HPAI persistence in endemic areas in Indonesia. Specific objectives were to determine where H5N1 HPAI is maintained in poultry populations in Indonesia, and how contact patterns between and within poultry production systems and enterprise types contribute to the maintenance of virus, including interactions between these systems and the populations of birds. We also aimed to investigate the co-circulation of HPAI virus with H9 LPAI viruses. Finally, the study aimed to profile and describe the value chain and marketing systems of live bird marketing, including collector yards and live bird markets. The ultimate objective of this work was to provide the relevant authorities with information that could be applied to develop recommendations for effective disease control.