Fig 1.
Posterior distributions for the lattice model.
Histograms of the number of simulation accepted by the ABC procedure for local-to-global ratio, L, transmission parameters for regular, β, and founder infections, latent period
and delay, δ.
Table 1.
Mode parameter values for the lattice model (using a Gaussian kernel density estimator) and the 90% Smallest Credible Interval/Highest Density Interval from the posterior distributions with an error threshold of 0.025.
Fig 2.
Posterior predictive check of the lattice model.
100 simulation runs of the lattice model (Eqs (1)–(9)) using parameters randomly drawn from the joint posterior distributions in Fig 1 shown in red. The blue dots mark the data points from the BBSC experiments. Panel (a) shows the total proportion infected, panel (b) the proportion of Infected-Infected neighbouring pairs.
Fig 3.
Stochastic realisations of the lattice-fitted model using the joint posterior distributions.
(a) Time-courses from 100 stochastic simulation runs alongside the data (blue circles). (b) The month when infection arrives on each row. The colormap shows the proportion of trees infected from the data at each time-point, with blue colours meaning low prevalence and yellow colours high prevalence. The white dots are the the average of the 100 simulations for the median (8th) tree on the row being infected and the lines for the 4th and 11th trees.
Fig 4.
Snapshots of the grid showing the spatial spread of the disease in one stochastic realisation of the lattice-fitted model.
Blue denotes a cell with a susceptible plant, purple a cell with an exposed plant and red a cell with an infected plant, with the row of founder trees in black. In this simulation run, L = 0.98, months−1,
months−1,
months and
months.
Fig 5.
Posterior distributions for parameters from the dispersal model.
Histograms of the number of simulation runs kept by the ABC procedure for dispersal (α), regular (β) and founder () transmission, latent period
and delay, δ.
Table 2.
Mode parameter values for the dispersal model (using a Gaussian kernel density estimator) and the 90% Smallest Credible Interval/Highest Density Interval from the posterior distributions with an error threshold of 0.025.
Fig 6.
Stochastic implementations of the dispersal model using the joint posterior distributions.
Left: time-courses from 100 stochastic simulation runs using parameter values selected from the joint posterior distributions alongside the data (blue circles). Right: the month when infection arrives on each row. The colormap shows the proportion of trees infected from the data at each time-point, with blue colours meaning low prevalence and yellow colours high prevalence. The white dots are the the average of the 100 simulations for the median (8th) tree on the row being infected and the lines for the 4th and 11th trees.
Fig 7.
Snapshots of the grid showing the spatial spread of the disease in one stochastic realisation of the dispersal model.
Blue denotes a cell with a susceptible plant, pink a cell with an exposed plant and red a cell with an infected plant, with the row of founder trees in black. In this simulation run, ,
,
,
and
.