Fig 1.
Map of the Udzungwa Mountains of Tanzania, showing the National Park, the Kilombero Nature Reserve and the location of six camera trap arrays placed in five different areas (MT = Matundu, MB = Mbatwa, LU = Lumemo, ND = Ndundulu, MW = Mwanihana) characterised by different habitat types (Table 1). The background layer is a Digital Elevation Model, with darker colour indicating higher elevation.
Table 1.
Survey effort for the five areas in the Udzungwa Mountains, Tanzania, sampled with camera traps. These areas hold broadly different habitat types, as indicated in table; the elevation range (in m a.s.l.) is related to sampling sites.
Table 2.
Habitat and anthropogenic covariates and species traits.
Covariates and traits used to model community occupancy (ψ) and detectability (p) across the landscape, with predictions for their effect on these parameters (see text for further details).
Table 3.
Area-specific data and model results for species richness and body mass.
Summaries of species richness, estimated richness, body mass of species detected and average estimated body mass for communities of mammals sampled by camera traps in five areas in the Udzungwa Mountains of Tanzania.
Fig 2.
Covariate effects on occupancy and detection at the landscape scale.
Mammal meta-community responses of occupancy probability to body mass (a), distance to the nearest river (b) and distance to human settlements (c), and of species detection probability to mass (d) and distance to human settlements (e) in the Udzungwa Mountains of Tanzania.
Table 4.
Posterior distributions from model results.
Summaries of posterior distributions from the multi-region hierarchical occupancy model for mammal meta-community in the Udzungwa Mountains of Tanzania. Parameters α and β are given on the logit scale and are the coefficients for detection probability and occupancy, respectively.