Spatial and temporal network analysis of 2021 swine trading movements in the Philippines using government-issued permits.

Abstract

Animal movement plays a critical role in the spread of infectious diseases. Social network analysis can be used to understand animal movement data (AMD) by analyzing trading dynamics and identifying key nodes that are vulnerable to disease spread. This study used swine local shipping permits (LSP) in 2021 from the Philippines' AMD platform. Towns were considered nodes and arcs were defined as shipments, weighted by the number of shipments or pigs per shipment. The year was divided into four quarters to evaluate seasonal trends. R (igraph package) and ArcGISPro were used to analyze and map the network. Results showed a network structure of 685 towns from 37,138 LSPs representing 2,178,536 pigs shipped from 311 towns and delivered to 516 towns with a median travel distance of 102.35 km. Network-level density was 0.06%, transitivity 7.9%, assortativity -0.277, reciprocity 0.014, and average path length 3.73. The mean number of shipments and pigs shipped indicated seasonal variation; October-December was highest (55.5, 3167) while January-March was lowest: (29.5, 1861.5). The Walktrap algorithm identified six large trade communities, involving 21 to 282 towns. Towns with breeder and fattener farms showed high out-degree while high in-degree were associated with metropolitan cities and slaughterhouses. Seasonal trends are consistent with cultural practices with fewer shipments during January-March (Lent) and more shipments during October-December (Christmas). Network-level metrics suggest weak clustering, and one-sided trade relationships. Knowledge of the swine trading patterns is critical in controlling ASF to support risk-based surveillance and proper budgetary resource allocation.