Table 2.

Regression models for the effect of density (number of females per enclosure) on variables of reproduction, space use and physiology in enclosure experiments on bank voles.

study	variable	model	R2	n	F	p	coefficient a	constant b
moderate densities (4-20 bank vole females/ha)
	breeding females	inverse	0.230	38	12.1	0.001	-3.37	3.78
	breeding females	linear	0.265	38	14.3	0.001	0.43	1.1
moderate to high densities (12-42 bank vole females/ha)
	breeding females	linear	0.817	12	50.2	< 0.001	0.83	-2.33
	breeding females	inverse	0.674	12	23.7	0.001	-70.6	14.2
	mean litter size	linear	< 0.01	12	< 0.01	0.931
	mean home range	inverse	0.789	12	37.5	< 0.001	2420	364
	interaction (nr. of females)	linear	0.653	12	18.9	0.001	0.45	0.58
	captures/animal	linear	0.3	12	4.2	0.065	-0.21	8.7
	Log(FGM 1stweek	inverse	0.035	6	0.1	0.723
	Log(FGM 2ndweek)	inverse	0.8	6	16	0.016	-4.6	2.9
	Log(FGM 3rd week)	inverse	0.562	6	5.1	0.086

We tested linear model (y = ax+b) and/or an inverse models (y = a/x+b), the latter indicating a threshold value (constant b) for high densities. Analyses were based on populations (rates or enclosure means). 38 Populations at moderate densities were investigated over 4 years in different earlier experiments (see text), 12 populations in moderate to high densities were investigated during the 2-year experiment reported here.