Predicting crown width and length using nonlinear mixed-effects models: a test of competition measures using Chinese fir (Cunninghamia lanceolata (Lamb.) Hook.)

Table 5 Lack-of-fit statistics for CL—DBH basic models

Function no	Fitting statistics of the candidate crown length models						Cross-validation
	− 2LL (df)	AIC	BIC	Bias (m)	RMSE (m)	R²	NMSE	PRESS
[CR1]^b	5353.92 (3)	5360	5376	0.00	1.49	0.4772	0.5319	332.3461
[CR2]^c	5362.14 (4)	5370	5391	0.00	1.49	0.4786	0.5315	331.9982
[CR3]^c	5376.44 (5)	5386	5413	0.00	1.49	0.4794	0.5316	331.9573
[CR4]^b	5346.14 (3)	5352	5368	0.01	1.50	0.4903	0.5302	332.3337
[CR5]^b	5353.00 (3)	5359	5375	0.02	1.50	0.5139	0.5359	334.8563
[CR6]^b	5392.51 (3)	5399	5414	0.03	1.52	0.5153	0.5493	343.4263
[CR7]^b	5366.53 (3)	5373	5388	− 0.01	1.51	0.4412	0.5435	339.3601
[CR8]^b	5366.53 (3)	5373	5388	− 0.01	1.51	0.4412	0.5441	338.1158
[CR9]^b	5366.53 (3)	5373	5388	− 0.01	1.51	0.4412	0.5435	339.3601
[CR10]^b	5343.36 (4)	5351	5373	− 0.00	1.50	0.4707	0.5339	333.5384

CR, crown length; − 2LL(df), − 2 log-likelihood (degree of freedom); AIC, Akaike information criterion; BIC, Bayesian information criterion; Bias, absolute bias; RMSE, root mean square error; R², coefficient of determination; NMSE, normalized mean square error; PRESS, predicted the error sum of squares
^bP value for each coefficient is lower than 0.05 in a T-test
^cP values for one or more than one coefficient are higher than 0.05 in a T-test

ISSN: 1297-966X