Table 2:

Model comparisons for the four activity metrics modelled as function of FL_j and X_p

			Yk = Aday		Yk = Umax		Yk = Uinst		Yk = ARday
Model	FLj	Xp	P-value	ΔAIC	P-value	ΔAIC	P-value	ΔAIC	P-value	ΔAIC
M0	−	−	n.a.	5.0	n.a.	13.1	n.a.	1.9	n.a.	1.9
M1	+	−	0.0083*	0*	0.026	10.1	0.94	3.9	0.0497*	0*
M1a	+	FRj	0.24	0.6	0.0005*	0*	0.015*	0*	0.37	1.2
M1b	+	SMRj	0.40	1.3	0.50	11.6	0.80	5.8	0.39	1.2
M1c	+	MMRj	0.47	1.5	0.76	12.0	0.45	5.3	0.32	1.0
M1d	+	AMSj	0.35	1.1	0.65	11.9	0.38	5.1	0.21	0.4
M1e	+	MET1j	0.59	1.7	0.88	12.1	0.65	5.7	0.20	0.4
M1f	+	MET2j	0.86	2.0	0.95	12.1	0.72	5.8	0.72	1.9
M1g	+	MET3j	0.30	0.9	0.74	12.0	0.33	4.9	0.21	0.4

Asterisks indicate optimal models. P-values represent the significance of the respective term tested with each model (i.e. FL_j in M1 and X_p in M1a–M1g) obtained using likelihood ratio tests. The ΔAICs were obtained by comparing each model with the optimal model for each respective activity measure.