Fagus sylvatica seedlings show provenance differentiation rather than adaptation to soil in a transplant experiment

Table 2 Model selection and levels of significance in the linear mixed model analysing seedling aboveground biomass

Aboveground biomass
Explanatory variable	numDF	F-value	p-value		ΔAIC
*Intercept*		1668.74	< 0.0001	***
*Height.0*	1	165.20	< 0.0001	***
*Drought*	1	3.44	0.08	n.s.
*Soil.origin*	2	2.84	0.06	n.s.
*Plant.origin*	2	254.28	< 0.0001	***
*Drought : plant.origin*	2	3.78	0.02	*
*Soil.origin : plant.origin*	4	4.81	0.0009	**
Inoculum	1		0.24	n.s.	− 0.61
Soil.origin : inoculum	2		0.10	n.s.	+ 0.60
Drought : soil.origin	2		0.36	n.s.	− 1.96
Drought : inoculum	1		0.65	n.s.	− 1.79
Drought : soil.origin : inoculum	2		0.45	n.s.	− 2.42
Drought : soil.origin : plant.origin	4		0.80	n.s.	− 6.35
Plant.origin : inoculum	2		0.91	n.s.	− 3.81
Drought : plant.origin : inoculum	2		0.80	n.s.	− 3.54
Soil.origin : plant.origin : inoculum	4		0.80	n.s.	− 6.38
Drought : soil.origin : plant.origin : inoculum	4		0.86	n.s.	− 6.69

Variables are shown in reverse order of the backward variable selection (i.e. variables at the bottom of the table were deleted first from the full model). Variables retained in the minimal model are shown in bolditalics on top. numDF specifies the number of degrees of freedom for each variable. The difference in Akaike’s Information Criterion (ΔAIC) is shown for the variables that were deleted from the model. Both the p value from the likelihood ratio test (p-value) and the ΔAIC values were considered and variables were only retained in the model if their deletion resulted in a significant p-value and a ΔAIC > 2. In the minimal model, fixed factors alone explained 66% of the deviance (R²m = 0.66), while both fixed and random increased the deviance explained by 4% (R²c = 0.70)

ISSN: 1472-6785