Fig. 3

Achieving elimination in two connected patches by varying connectedness between the two populations (x-axis) and MDA coverage in the 2nd patch (y-axis) of each subplot. Columns represent different sets of MDA coverage in patch 1 (70%, 80%, and 90%, respectively). Each row represents the relative incidence level between the two patches. a MDA coverage in patch 1 is 70%, and patch 2 has higher pre-intervention incidence. No visually distinguishable assembly effect is found. b MDA coverage in patch 1 is 80%, and patch 2 has a higher pre-intervention incidence than patch (1) Patch 1 should achieve elimination on its own but did not achieve it because of its connectedness to patch 2 (negative assembly effect from the viewpoint of patch 1). c MDA coverage in patch 1 is 90%, and patch 2 has higher pre-intervention incidence. d MDA coverage in patch 1 is 70%, and both patches have identical pre-intervention incidence. Slight negative assembly effect from the viewpoint of patch (2) e MDA coverage in patch 1 is 80%, and both patches have identical pre-intervention incidence. Slight positive assembly effect from the viewpoint of patch 2. The red asterisk represents the combination of parameter values matching the MDA trial implementation described in Parker et al. Panel F: MDA coverage in patch 1 is 90%, and both patches have identical pre-intervention incidence. Increased positive assembly effect from the viewpoint of patch 2 compared to panel E. Panel G, H, I: MDA coverage in patch 1 is 70%, 80%, and 90% respectively. Patch 2 has lower pre-intervention incidence, and its baseline MDA coverage threshold is low. From the viewpoint of patch 2, there is always a negative assembly effect but its magnitude diminishes as the MDA coverage in patch 1 is increased from 70 to 90%