Table 3.

Model selection statistics (Akaike Information Criterion ΔAIC and Akaike weights w_i) for various candidate models describing the RGR of tree height as a function of initial tree height (H), treatment (T), and species composition (C) effects of European beech (Fagus sylvatica), Sessile oak (Quercus petraea), and Douglas fir (Pseudotsuga menziesii).

Model	Fixed effects							Fagus sylvatica		Quercus petraea		Pseudotsuga menziesii
	H	T	C	H × T	H × C	T × C	H × T × C	ΔAIC	wi	ΔAIC	wi	ΔAIC	wi
1	×	×						23.99	0.00	4.47	0.08	11.57	0.00
2	×		×					15.72	0.00	7.94	0.01	2.38	0.09
3		×	×					73.76	0.00	9.46	0.01	7.21	0.01
4	×	×	×					19.45	0.00	7.25	0.02	0.01	0.30
5	×	×	×	×				13.67	0.00	10.08	0.00	0.00	0.30
6	×	×	×		×			10.82	0.00	0.13	0.50	1.20	0.16
7	×	×	×			×		25.29	0.00	8.32	0.01	7.38	0.01
8	×	×	×	×	×			0.00	0.91	3.51	0.07	1.75	0.12
9	×	×	×	×		×		19.64	0.00	8.55	0.00	7.81	0.01
10	×	×	×		×	×		14.00	0.00	0.00	0.26	8.42	0.00
11	×	×	×	×	×	x		4.74	0.08	2.55	0.05	9.46	0.00
12	×	×	×	×	×	×	x	15.61	0.00	5.20	0.00	15.21	0.00

The best-supported models with the highest Akaike weights are highlighted in bold. For Douglas fir the more parsimonious model that included a marginal significant height-treatment interaction (P = 0.09) was rejected, since the main effects-only model fit the data equally well (ΔAIC = 0.01, w_i for both models = 30%).