Selection of scenarios for landscape-level risk assessment of chemicals: case studies for mammals

Environmental Sciences Europe

Table 2 Best GLM (lowest AICc among all parameter combinations marked in bold) for modelling the population survival of four species in different landscapes with one to four predictor variables

Species	Parameter no.	Model	AICc	R^b (McFadden)
Common vole (N = 51)	1	SizeLarOC	574.37	0.8114
	2	SI + SizeLarOC PropOC + SizeLarOC^a	339.54 340.20	0.8930 0.8928
	3	CompLarOC + PropOC + TotalOC	233.35	0.9304
	4	CompLarOC + DI + PropOC + TotalOC DI − OvResilInd + SI + SizeLarOC^a	150.85 151.64	0.9595 0.9593
European rabbit (N = 51)	1	TotalOC	875.93	0.6865
	2	− PropOC + TotalOC	654.09	0.7739
	3	Area − SizeLarOC + TotalOC	561.79	0.8107
	4	Area + SI − SizeLarOC + TotalOC	512.72	0.8307
Brown hare (N = 41)	1	Area	37.02	0.9833
	2	Area + CompLarOC^b	38.06	0.9865
	3	Area + CompLarOC − PropOC^b	39.50	0.9887
	4	Area + CompLarOC − PropOC + TotalOC^b	41.52	0.9894
Wood mouse (N = 51)	1	SizeLarOC	832.78	0.5414
	2	− DI + TotalOC	603.82	0.6746
	3	EC − IIC + SI	290.10	0.8568
	4	− DImultOCN + EC − IIC + SI Area − DI + PropOC + SI^a	197.99 199.63	0.9112 0.9102

The signs of the parameters represent positive or negative influence on the endpoint
^aThis model is within an AICc range of 2 and is therefore considered to be comparable to the model with the lowest AICc
^bThere were 10, 55 and 131 additional 2-, 3- and 4-parameter models within an AICc range of 2, respectively, all of which had ‘Area’ as the dominant parameter