Table 4.

Accuracy matrix of the best model (pseudo-absence with 3 presence and absence clusters each) among 100 bootstrap models. Accuracy matrix of the best model (pseudo-absence with 3 presence and absence clusters each) among 100 bootstrap models. The observed categories are in rows, and the predicted categories in the column. For a perfect model fit, all the numbers should be on the diagonal, with no off-diagonal entries. P1–P3 are the presence categories, and A1–A3 are the absence categories.

		Predicted Category
Observed category		P1	P2	P3	A1	A2	A3	Tot.
	P1	52	5	3	0	2	0	62
	P2	9	87	7	0	0	0	103
	P3	1	4	29	0	1	0	35
	A1	1	1	0	139	2	1	144
	A2	0	1	4	0	27	0	32
	A3	0	0	0	0	1	23	24
	Tot.	63	98	43	139	33	24	400