2.3 Network characterisation
We created a contact network where premises (farms, traders, industrials, and markets) became the nodes and movements as their edges, excluding movements to abattoirs as they are dead-ends for disease transmission (Guinat et al., 2016). The network was represented by its adjacency matrix, where the number of neighbours of a node (degree), and its value was calculated directly from the adjacency matrix (A), where Aij = 1 if there was a connection between nodes i and j; otherwise, Aij = 0 (Newman, 2010). Swine movements were represented as a directed network (networks in which the direction of movement was taken into account).
Annual and monthly networks of premises were created and computed the following metrics: betweenness, closeness, clustering coefficient, degree, density, diameter, average shortest path, and giant weakly and strongly connected components (Table 1). Then, we analysed seasonal activity in network metrics and features such as shortest path length and small-world characteristics (highly clustered networks) (Strogatz, 2001) (James et al., 2009) that may facilitate rapid disease transmission.
Table 1. Network terminology used to characterise the Ecuadorian pig movement network.