Table 2.

Statistical parameters of the PLSR model validation data set

The lowest RMSEP-CV was used to choose the number of components in the model. NC, number of components; REP, relative error of prediction; RMSEP CV, root mean square error of prediction from cross validation with PLSR; Tr, training set; Val, validation or test data.

Traits	N Tr	N Val	RMSEP CV	NC	R2 Tr	R2 Val	REP Val (%)	Bias Val (%)
Narea	282	243	0.22	21	0.92	0.93	7.6	0.73
LMA	282	243	4.50	21	0.86	0.89	7.0	-0.23
SPAD	342	272	3.16	16	0.87	0.81	6.8	-0.34
V cmax	262	226	31.53	23	0.79	0.74	18.7	0.20
J	262	226	25.44	18	0.82	0.70	13.0	-0.73
Nmass	342	273	3.30	24	0.86	0.70	10.5	1.3
Pmass	219	212	0.93	19	0.54	0.65	25.8	3.3
V cmax25	262	226	20.68	18	0.76	0.62	15.9	0.17
A	307	253	3.93	15	0.64	0.49	17.7	0.49
V cmax25/Narea	262	226	10.62	14	0.40	0.48	17.5	1.9
Parea	219	212	0.04	19	0.40	0.42	23.5	4.2
g s	307	253	0.15	11	0.50	0.34	33.5	3.3