-Electronic Supplementary Material- Optimizing surveillance for livestock disease spreading through animal movements

The data describing bovines’ displacements are recorded on a daily basis, making it possible to construct different network representations according to different assumptions or aggregating on different time window lengths [1] [2]. In particular, we simulated the spread of an infectious disease on networks aggregated on a daily, weekly, monthly and yearly scale. The choice of the aggregating time window length ∆t affects the underlying mobility structure, leading to denser displacement networks for longer time windows, while the time step used in the numerical simulations of the spreading dynamics is kept fixed to 1 day. In this perspective, when the spreading process takes place on the daily dynamical networks, at every time step of the spreading the snapshot of the static network is different, while for a longer aggregating window length the network topology remains unchanged for exactly ∆t time steps. In Figure 2 of the main paper, the unfolding of the spreading for different aggregating time window lengths is followed by plotting the temporal evolution of the number of infected premises for every spreading time step (=1 day). In the following, if not otherwise specified, we present results corresponding to spreading phenomena simulated on networks aggregated on time windows of length ∆t = 1 day, starting at t0 =Jan 1, and with an infectious period μ−1 = 7 days.