Table 2 Comparison of generalized linear models used to describe corticosterone levels in all small mammal species in forest fragment samples.

Corticosterone logistic models	Rank	Df.res	AUC	BIC	AICc	ΔAICc	p value
f(Area + Genus + Trap)^a	9	97	0.87	137.9	113.6	0	< 0.0001
f(Area + Species + Trap)	12	94	0.88	146.2	115.5	1.9	< 0.0001
f(Area + Genus + Ecomorphological + Trap + Capture)	11	95	0.88	146.8	118.2	4.6	< 0.0001
f(Area + Genus + Capture)	7	99	0.83	140.7	120.9	7.3	< 0.0001
f(Area + Species + Capture)	10	96	0.85	149.1	122.7	9.1	< 0.0001
f(Species + Genus + Ecomorphological + Trap)	11	95	0.88	157.3	122.7	9.1	< 0.0001
f(Area + Genus + Ecomorphological + Capture)	8	98	0.83	145	122.9	9.3	< 0.0001
f(Species)	7	99	0.77	150.5	130.6	17	< 0.0001
f(Area + Ecomorphological + Capture)	5	101	0.75	150.1	135	21.4	< 0.0001
f(Area + Ecomorphological + Capture)	5	101	0.75	150.1	135	21.4	< 0.0001
f(Species + Genus)	7	99	0.77	164.4	138	24.4	< 0.0001
f(Capture)	2	104	0.67	146.4	138.6	25	0.0002
f(Species + Genus + Ecomorphological)	8	98	0.77	168.1	139.5	25.9	< 0.0001
f(Area + Capture)	4	102	0.70	153	140.3	26.7	0.0007
f(Genus:Area)^b	8	98	0.77	178.4	145.7	32.1	< 0.0001
f(Area + Ecomorphological)	4	102	0.66	158.7	146	32.4	0.0096
f(Area)	3	103	0.56	161.3	151.1	37.5	0.1495
f(Species + Genus + Genus:Area)	11	95	0.79	197.3	155.6	42	< 0.0001
f(Species:Area)	13	93	0.79	215.4	167.6	54	< 0.0001
f(1)	1	105	0.5	154.4	149.2	35.6	Null

Models are ranked based on ΔAICc, and models with values < 2 are considered equally valid (bold). All variables are explained in the text.
^aA nomenclature specification of the form f(first + second) indicates all the terms in first together with all the terms in second with any duplicates removed (i.e. additive effects of factors).
^bA specification of the form f(first:second) indicates the set of terms obtained by taking the interactions of all terms in first with all terms in second.

Search