Fig 1.
Map of the study area, showing size and location.
Households from the census included in this modelling study are indicated as white circles, with Mfuwe airport highlighted in the north (produced using Landsat 7 imagery from USGS).
Fig 2.
(A) A section of the 11 m spatial resolution land classification image, illustrating bare land areas around the airport (brown), cropland (yellow), bush/forest (green), and finer scale digitised features including roads (grey) and river (blue). (B) Example path produced using the A* algorithm and land classification between arbitrary points. Arbitrary points were used to emphasise how the algorithm diverts the path around a prominent obstacle; in this case, the river itself (after [37]). Produced using Landsat 7 imagery from USGS and Bing Aerial Imagery.
Fig 3.
Spatial distribution of human and cattle populations in the study area, identified in the census (produced using Landsat 7 imagery from USGS).
Fig 4.
Estimation of tsetse distribution in the study area, using kernel density estimation (KDE).
Numbers in brackets represent estimated values based on the number of caught flies, prior to scaling-up to account for an estimation of total population size. Reproduced from [38].
Table 1.
Additional human population required per scenario.
Fig 5.
A section of the edited version of the land classification, highlighting the buffer (dark grey) used to simulate developed land encroaching on bush areas, and the location of new villages (gold) in the region.
Fig 6.
Spatial distribution of the increasing human population for scenarios 1–4 (produced using Landsat 7 imagery from USGS).
Fig 7.
Spatial distribution of the increasing cattle population for scenarios 1–4 (produced using Landsat 7 imagery from USGS).
Fig 8.
Spatial distribution of the increasing population of historically indigenous and migrant tribes in the study area, between the current population levels, and those in the most extreme scenario–scenario 4 (produced using Landsat 7 imagery from USGS).
Fig 9.
Heat surface of the region representing aggregate number of infections (Left: Human, Right: Cattle) across the 100 repeats of the control simulations, with pixel values taking the units of infections per square kilometre (produced using Landsat 7 imagery from USGS).
Fig 10.
Mean approximate incidence rates (per 1000 agent-years) for human and cattle agents, plotted against population sizes.
Fig 11.
Mean daily population size of adult tsetse flies by sex (+/- standard error) for all scenarios, overlain on mean monthly temperature and precipitation for the three-year simulation period.
Fig 12.
Heat surfaces representing the aggregate number of human infections across the 100 repeats, with pixel values taking the units of infections per square kilometre (produced using Landsat 7 imagery from USGS).
Fig 13.
Heat surfaces representing the aggregate number of cattle infections across the 100 repeats, with pixel values taking the units of infections per square kilometre (produced using Landsat 7 imagery from USGS).
Table 2.
Average proportion of human infections attributable to each activity, for the control and each scenario, represented as a percentage.
Table 3.
Approximate incidence rates of different age groups, for each scenario.
Table 4.
Approximate incidence rates of different sexes and cattle ownership, for each scenario.
Table 5.
Approximate incidence rates by immigrant/non-immigrant tribes, for each scenario.