Table 1. Detailed results for generalized additive mixed models testing for the effect of the spatial location of heterospecifics on the distance between the carnivore and the reindeer carcass, based on 1,643 simultaneous locations for simultaneous (real-time) locations (A) and a time delay between predator locations of 30 min (B).

	WOLVERINE				LYNX
	estimate	se	edf	p	estimate	se	edf	p
A) Simultaneous (real-time) predator locations
Parametric coefficients:
Intercept	3.07	0.05		<0.001	3.1	0.06		<0.001
Smooth terms:
ƒ (Distance between predators)			5.67	<0.001			3.86	<0.001
ƒ (Distance competitor-carcass)			3.79	<0.001			5.15	<0.001
R2 (adjusted)	0.21				0.26
BIC	1721.9				1442.2
B) 30-min time delay between predator locations
Parametric coefficients:
Intercept	3.07	0.05		<0.001	3.02	0.07		<0.001
Smooth terms:
ƒ (Distance between predators)			4.33	<0.001			2.12	<0.001
ƒ (Distance competitor-carcass)			3.89	<0.001			2.73	<0.001
R2 (adjusted)	0.14				0.16
BIC	1865.2				1451.3
edf = estimated degrees of freedom.

For each species, models (A and B) were compared by using Bayesian Information Criterion (BIC). BIC tend to be more conservative and less prone to overparameterization than Akaike’s Information Criterion⁵⁵. BIC of the best models are denoted in bold.