Table 2.

Results of the 10-fold cross-validation.

		Method	Obs. Vs. pred	Prediction errors
				Root mean squared errors	Relative root mean squared errors (%)	Mean absolute errors	Relative mean absolute errors (%)
Species richness	Training	GLM	0.56	42.73	29.07	33.14	22.54
		GAM	0.78	32.82	22.33	25.44	17.31
	Validation	OK	0.61	40.97	27.87	32.61	22.18
		GLM	0.54	43.61	29.67	33.89	23.05
		GAM	0.64	39.36	26.78	30.50	20.75
		GLM OK	0.54	43.40	29.52	33.73	22.95
		GAM OK	0.64	39.27	26.71	30.43	20.70
Shannon diversity	Training	GLM	0.56	31.81	30.01	24.64	23.25
		GAM	0.79	23.81	22.46	18.32	17.28
	Validation	OK	0.63	29.57	27.90	23.39	22.07
		GLM	0.53	32.62	30.77	25.20	23.77
		GAM	0.66	28.97	27.33	22.35	21.08
		GLM OK	0.54	32.40	30.57	25.03	23.61
		GAM OK	0.66	28.85	27.22	22.25	20.99
Simpson diversity	Training	GLM	0.55	24.82	32.66	19.20	25.26
		GAM	0.77	19.22	25.29	14.67	19.30
	Validation	OK	0.63	22.94	30.18	18.05	23.75
		GLM	0.52	25.45	33.49	19.66	25.87
		GAM	0.64	22.87	30.09	17.59	23.14
		GLM OK	0.53	25.24	33.21	19.49	25.64
		GAM OK	0.64	22.76	29.95	17.49	23.01

Notes: Obs. vs. Pred.: results of Pearson correlation between observed and predicted values.