Spatial modeling for low pathogenicity avian influenza virus at the interface of wild birds and backyard poultry

La Sala, L F and Burgos, J M and Blanco, D E and Stevens, K B and Fernández, A R and Capobianco, G and Tohme, F and Pérez, A M (2019) Spatial modeling for low pathogenicity avian influenza virus at the interface of wild birds and backyard poultry. Transboundary and Emerging Diseases, 66 (4). pp. 1493-1505.

[img]
Preview
Text
11942.pdf - Accepted Version
Available under License Creative Commons Attribution Non-commercial.

Download (1MB) | Preview

Abstract

Low pathogenicity avian influenza virus (LPAIV) is endemic in wild birds and poultry in Argentina, and active surveillance has been in place to prevent any eventual virus mutation into a highly pathogenic avian influenza virus (HPAIV), which is exotic in this country. Risk mapping can contribute effectively to disease surveillance and control systems, but it has proven a very challenging task in the absence of disease data. We used a combination of expert opinion elicitation, multicriteria decision analysis (MCDA), and ecological niche modeling (ENM) to identify the most suitable areas for the occurrence of LPAIV at the interface between backyard domestic poultry and wild birds in Argentina. This was achieved by calculating a spatially‐explicit risk index. As evidenced by the validation and sensitivity analyses, our model was successful in identifying high‐risk areas for LPAIV occurrence. Also, we show that the risk for virus occurrence is significantly higher in areas closer to commercial poultry farms. Although the active surveillance systems have been successful in detecting LPAIV‐positive backyard farms and wild birds in Argentina, our predictions suggest that surveillance efforts in those compartments could be improved by including high‐risk areas identified by our model. Our research provides a tool to guide surveillance activities in the future, and presents a mixed methodological approach which could be implemented in areas where the disease is exotic or rare and a knowledge‐driven modeling method is necessary.