Table 3.

Results of the ordination models (direct gradient analysis—RDA) exploring and testing the relations between large-scale predictors and plant species richness/cover in the Kruger National Park.

Forward selection	p-value (without spatial effects )	Explained variance (%)	p-value (with spatial effects)	Explained variance (%)
Mean temperature (tempMean)	0.006	20.1	-	-
Variation in Enhanced Vegetation Index (eviSD)	0.032	5.8	-	-
Distance from gravel roads (distDirt)	0.002	7.3	-	-
Covariables: bedrock, habitat			-	-
tempMean, eviSD, distDirt	0.002	30.5	0.008	12.7
All predictors	0.118	26.7	0.182	10.9
Rain1	0.158	8.4	0.588	-0.3
Distances2	0.472	3.8	0.086	7.4
Temperature3	0.008	27	0.006	10.2
Surface water4	0.846	− 2.6	0.304	0.2
Enhanced Vegetation Index (EVI)5	0.078	12.1	0.162	2.8
Fire6	0.998	− 3.5	0.68	− 0.8

In all ordination models, total plant richness, richness of herbs, richness of shrubs, cover of grasses and cover of shrubs were used as the response variables. Spatial effects are represented by the three main principle coordinates (PCO1, PCO2, PCO3). Significant effects are displayed in bold, marginally significant effects in italics. The significance values refer to individual predictors in case of forward selection, and to the significance of all canonical axes in the remaining ordination models. See Supplementary Table 1 for detailed description of variables.

¹Summary effect of rainfall (rainSum, rainMean, rainSD).

²Summary effect of distance from potential sources of propagules—roads, rivers, rest camps and KNP boundary (DistBnd, distCamp, distTar, distDir, distRiv, distStrm; see Supplementary Table 1).

³Summary effect of temperature and its variation over time (tempMean, tempSD, tempMin, tempMax).

⁴Summary effect of surface water occurrence (waterSum, waterMean, waterSD).

⁵Summary effect of Enhanced Vegetation Index (eviSum, eviMean, eviSD).

⁶Summary effect of fire (FireSum, FireMean, FireSD).