Morphological variation of fine root systems and leaves in primary and secondary tropical forests of Hainan Island, China

Table 3 Regression results from least-squares linear models in the form of trait ~ variable. Table to accompany Fig. 5. Models were fit separately by forest type

Trait (units)	Forest type	Variable (units)	β	se	p	F(1,148)	R²	RMSE
Root diameter (mm)	Secondary	Soil BS (%)	− 0.004	0.002	*	4.39	0.03	0.29
	Primary	Soil BS (%)	− 0.010	0.004	*	4.96	0.03	0.65
	Secondary	Soil P (g kg⁻¹)	− 0.703	0.792	n.s.	0.79	< 0.01	0.29
	Primary	Soil P (g kg⁻¹)	− 2.350	1.002	*	5.50	0.04	0.26
Specific root length (m kg⁻¹)	Secondary	Soil BS (%)	0.262	0.188	n.s.	1.94	0.01	31.14
	Primary	Soil BS (%)	1.171	0.551	*	4.53	0.03	33.79
	Secondary	Soil P (g kg⁻¹)	23.42	84.80	n.s.	0.08	< 0.01	31.33
	Primary	Soil P (g kg⁻¹)	324.04	128.32	*	6.38	0.04	33.58
Root tissue density (g cm⁻³)	Secondary	Soil BS (%)	0.001	0.001	n.s	0.07	< 0.01	0.20
	Primary	Soil BS (%)	− 0.002	0.003	n.s	0.73	< 0.01	0.16
	Secondary	Soil P (g kg⁻¹)	0.837	0.540	n.s.	2.40	0.02	0.20
	Primary	Soil P (g kg⁻¹)	− 0.523	0.613	n.s.	0.73	< 0.01	0.16
Root branching intensity (tips cm⁻¹)	Secondary	Soil BS (%)	− 0.001	0.004	n.s	0.02	< 0.01	0.72
	Primary	Soil BS (%)	0.002	0.011	n.s	0.02	< 0.01	0.65
	Secondary	Soil P (g kg⁻¹)	3.188	1.922	n.s	2.75	0.02	0.65
	Primary	Soil P (g kg⁻¹)	− 0.48	1.922	n.s	2.75	0.02	0.72

Model coefficient estimates (β), standard errors (se), and associated probabilities (p) are given for each variable by forest type (intercept terms are not shown). Regression F-statistics (F) and coefficients of determination (R²) and root-mean-squared error (RMSE) are given for each model. The F(1,148) critical value at α = 0.05 is 3.905. Italicized model coefficients show significant ANCOVA interaction terms between forest type and soil variable (p < 0.05)
n.s. non-significant
Probabilities are denoted as follows:
*p < .05
**p < .01
***p < .001

ISSN: 1297-966X