Fig. 3.

Metapopulation: comparing predictions to simulations for variance: (a) for $M=2,$ over 50 realisations; (b) for $M=2,$ over a moving window after detrending using the mean of the subpopulations; (c) for $M=3,$ over a moving window after detrending using the mean of the subpopulations, (d) for CV, $M=2,$ over a moving window after detrending using the mean of the subpopulations; (e) ROC curve calculated over 50 realisations at various timepoints by thresholding in variance, detrending using the mean the subpopulations; and (f) ROC curve calculated over 50 realisations at various timepoints by thresholding in variance, detrending using the mean of 50 realisations. Timeseries figures show: simulation of NExt (red line); simulation of Ext (blue line); dynamic predictions (purple line); steady state predictions (green line). For repeated simulations each line is the mean value obtained over the simulations and the shaded area represents one standard deviation about the mean. Note that for ROC curves no simulation had reached eradication at t=450, although all Ext simulations had achieved eradication by the end of the simulation. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)