Table 8. Covariates used for Generalized Additive Modelling analysis.
| Covariate | Sp.1 | Method of Calculation | Values2 | Method of data capture (and source)3 |
| Vegetation type | E,G,C | Majority vegetation type within 1km circle radius | Gilbertiodendron forest; mixed forest; Savannah; Agriculture; Swamp; Raphia swamp; Earth/Roads; Water; Marantaceae forest | 9-class land-cover reclassification of a partial coverage 18-class vegetation grid derived from Landsat 7 ETM+ imagery (WCS Congo/Woods Hole Research Center, USA) |
| Distance to bais | E,G | Euclidean distance (km) to all bais | 13.2(±0.9) | Digitized from 1∶200,000 Topographic map and Landsat 7 ETM+ satellite imagery (WCS-Congo) |
| Density of yangas | E,G | Density of yangas within a 5km moving window | 0.03(±0.008) | See ‘DISTBAIS’ |
| Distance to roads | E,G,C | Euclidean distance to all roads accessible by vehicles at time of and in year preceding surveys | 18.2(±1.3) | GPS data for logging roads in Kabo, Pokola, Loundougou & Toukoulaka (CIB-Pokola); digitized from 1∶1million map (WCS-Congo) |
| Distance to rivers | E,G,C | Euclidean distance to all rivers navigable by canoes | 14.7(±0.8) | Digitized from 1∶200,000 map (WCS Congo) |
| Distance to settlements | E,G,C | Euclidean distance to permanent towns, villages and camps | 18.5(±0.9) | GPS data for camps/logging towns; digitized from 1∶200,000 and 1∶1 mil. maps (WCS Congo) |
| Logging history | E,G,C | Number of years since start of commercial logging operations | Unlogged; <5 years; 5–10 yrs; 10–15 yrs; 15–20 yrs; 25–30 yrs; >30 yrs | Spatial limits defined by VMA (Maximum Annual Wood Volume) (CIB-Pokola/WCS Congo archives) |
| Distance to NNNP boundary | E,G,C | Euclidean distance from NNNP boundary4 | 64.3(54.0) | Spatial limits defined in legal decree (CNIAF, National Monitoring and Inventory Agency, Congo) |
| Management status | E,G,C | Status of formal management planning per unit | Underway; Adopted; None | WCS Congo Project archives |
1
E = Elephant, G = Gorilla, C = Chimpanzee.
2
Mean values (with standard errors) shown for continuous variables (grid cell values corresponding to 2km transect mid-point); categories shown for factor variables.
3
We provide the original method of data capture, where known. For some covariates, data were collated from different sources to ensure landscape-wide coverage.
4
Negative distance values inside the NNNP boundary.