Fig 1.
Structure of agent connections in the model.
Shows 1-phase (low-specialization), 2-phase (mid-specialization), and 3-phase (high-specialization) connectivity heuristics. Also indicates livestock transfer age conditions.
Table 1.
Parameter values used in the experiment.
Fig 2.
Sample network map as displayed on the model dashboard.
Shows agents as nodes and inter-agent contacts (both potential and active) as edges. Key provides an overview of connectivity heuristics for each agent type.
Table 2.
Python / NetworkX code used in experiment two.
Fig 3.
Histograms showing distribution of dependent variables.
Infection duration data appear in the left column and proportion of infected agents in the right column, with color indicating producer specialization level. Low density runs were those with 0 < Np ≤ 500, mid-density 500 < Np ≤ 1000, and high-density 1000 < Np ≤ 1500. Data were split into 40 bins.
Fig 4.
Right-censored histograms showing distribution of dependent variables.
These plots are parallel to those in Fig 3, yet include only datapoints in which the infection duration was ≥ 3000 model days. Data were split into 11 bins.
Fig 5.
Scatter plots showing full model-output dataset for both dependent variables.
Proportion of agents infected (cumulative) appears in the top row, and network-level infection duration in the bottom. Each point represents one of the 45,000 model runs (15,000 for each level of specialization).
Table 3.
Kruskal-Wallis equality-of-populations rank test statistics.
Fig 6.
Percolation threshold visualizations.
Lines plot average values for the 100 runs at each of 150 Np levels, with corresponding color fields indicating 95% CI. Top left plot shows infection duration. Top right shows mean proportion infected (cumulative). Bottom left shows the fraction of runs resulting in a systemic network-level infection lasting the full duration of the model run (4135 model days). Bottom right shows the fraction of runs in which 95% or more of the agents became infected.
Fig 7.
Finding percolation points numerically.
Upper plots show raw model output data with LOESS-smoothed curves (span length = 0.45 × N). Lower plots show the slope of each LOESS curve, with maximum-slope points annotated. Green represents 1-phase, blue 2-phase, and red 3-phase treatments.
Table 4.
Key contact network metrics for each producer specialization level, stratified across three producer density categories.
Fig 8.
Correlating Np with contact network metrics.
Key contact network metrics, calculated for each treatment. Lines plot averages for each Np value; color fields show 95% CI. Green represents 1-phase, blue 2-phase, and red 3-phase treatments.
Fig 9.
Visualizations of sample networks generated by the model under each level of producer specialization.
Np = 500 for each network. Nodes were positioned using a spring layout, and sized according to total number of contact events. Blue nodes are producers; yellow are feed mills, and red are slaughter plants.
Table 5.
Key infected component network metrics for each producer specialization level, stratified across three producer density categories.
Fig 10.
Correlating Np with infected component network metrics.
Key infected component network metrics, calculated for each treatment. Bars plot averages for five Np ranges; whiskers show 95% CI. Green represents 1-phase, blue 2-phase, and red 3-phase treatments.