Table 1.

Details of the landscape attributes hypothesized to influence rabies virus spread in the Serengeti district, Tanzania. Village areas ranged from 9 to 220 km², and all landscape attributes were scaled to a 100 m resolution (100 × 100 m grid cells). Resistance values were assigned to each grid cell to represent the presumed effect of each attribute on rabies virus diffusion, that is, as a facilitator or barrier to spread. A barrier effect is represented by high values denoting greater resistance to movement, whereas facilitators are assigned small resistance values denoting greater ease of movement (calculated as the reciprocal of a presumed conductance value, e.g., a conductance of 100 is represented by a resistance value of 0.01)

Mechanism	Attribute	Hypothesized effect on dispersal	Rationale	Measurement	Range of resistance values	Data Source
Host demography	Dog density	Facilitator	Density‐dependent transmission often assumed for directly transmitted pathogens such as RABV (Cross et al., 2013; Ferrari, Perkins, Pomeroy, & Bjørnstad, 2011; Morters et al., 2013).	Isotropic Gaussian smoothing kernel applied to census dog counts in grid cells.	0.034–10	Human and dog population census (Sambo et al., 2017)
	Dog presence	Facilitator	Dog population distribution and possible movement routes (Beyer et al., 2011; Bourhy et al., 2016). Areas without dogs (or humans) are expected to be occupied by wildlife, which are considered to be nonmaintenance (i.e., dead end) hosts in this system (Lembo et al., 2007, 2008).	Dog presence/absence per cell.	0.1–1	Human and dog population census (Sambo et al., 2017)
	Elevation	Barrier	Typically lower human (and dog) densities at higher elevations (Cohen & Small, 1998).	90 m resolution resampled to 100 m resolution	1,164–1,741	Digital elevation model (DEM) from NASA Shuttle Radar Topology Mission data http://srtm.usgs.gov/index.php
Host movement	Human: dog ratio (HDR)	Barrier	Measure of human intervention: in areas with higher HDR, rabid dogs may be more rapidly caught/killed.	Village‐level HDRs from human and dog counts.	3.39–12	Human and dog population census (Sambo et al., 2017)
	Rivers	Barrier	Barriers to dog movement unless movement is facilitated by human activity (60,61).	Shape file rasterized	1–1,000	http://www.glcn.org/activities/africover_en.jsp
	Roads	Facilitator	Presence of humans (and dogs) close to roads/dog behaviour influenced by roads (e.g., food, movement)/human‐mediated transport.	Shape file rasterized	0.001–1	http://www.glcn.org/activities/africover_en.jsp
	Slope	Barrier	Steepness acts as a physical impediment to host movement.	90 m resolution DEM resampled to 100 m resolution	1–1.24	Estimated from resampled DEM (see above)
	Uniform landscape	Barrier	Dog movements expected to follow an isolation‐by‐distance pattern (Wright, 1943), that is, a null model for comparison.	Uniform grid	1	NA
Host susceptibility	Average vaccination coverage	Barrier	Vaccination coverage increases herd immunity, reducing transmission and disease incidence	Annual vaccination coverage from 2004 to 2013 averaged and aggregated at village level	6.43–100	This study
	Campaigns over 10‐year period	Barrier	High coverage, repeat campaigns are most effective for reducing transmission and for disease elimination (Ferguson et al., 2015; Townsend et al., 2013).	Number of vaccination campaigns with at least 10% coverage per village from 2004 to 2013	2–14	This study
	Susceptible host density	Facilitator	Resistance surface incorporating vaccination of the dog population.	Same as total density (see above)	0.037–10	This study