TABLE 3.

AUC values of the variables selected after the univariate scale selection, conducted on different predictors at each of the nine scales considered

Category	Variables	Class	QLSNP		QMLNR
			Best scale	AUC	Best scale	AUC
Linear and Point features	DENS_riv		4,800	0,627	19,200	0,794
	DENS_rd		19,200	0,613	28,800	0,748
	DENS_set		28,800	0,71	9,600	0,769
Topographic	CTI		9,600	0,741	14,400	0,726
	SLP		2,400	0,801	28,800	0,839
	ROUGH		600	0,717	19,200	0,78
	DISS		600	0,756	14,400	0,826
	ELEV		19,200	0,722	1,200	0,819
Climatic	TEMP		300	0,732	300	0,843
Landcover (Landscape level)	CWED		4,800	0,647	28,800	0,856
	AI		28,800	0,675	19,200	0,776
	PD		28,800	0,673	19,200	0,819
Landcover (Class level)	PLAND	Br	14,400	0,631	28,800	0,873
	AREA_AM		9,600	0,669	28,800	0,878
	GYR_AM		9,600	0,639	28,800	0,908
	PLAND	Sn	9,600	0,626	9,600	0,715
	AREA_AM		9,600	0,627	28,800	0,796
	GYR_AM		9,600	0,629	28,800	0,788
	PLAND	Gr	14,400	0,686	14,400	0,819
	AREA_AM		2,400	0,628	19,200	0,807
	GYR_AM		4,800	0,638	19,200	0,823
	PLAND	Shr	19,200	0,595	14,400	0,759
	AREA_AM		19,200	0,613	28,800	0,827
	GYR_AM		19,200	0,608	28,800	0,826
	PLAND	NLF	19,200	0,568	14,400	0,754
	AREA_AM		28,800	0,573	19,200	0,767
	GYR_AM		28,800	0,574	14,400	0,808

In bold the predictors selected after collinearity test (r ≥ 0.7).

Abbreviations: QLSNP, Qilianshan National Park; QMLNR, Qomolangma National Nature Reserve.