Do warning calls boost survival of signal recipients? Evidence from a field experiment in a group-living bird species

Table 1 Model selection for the factors affecting survival of 28 Siberian jay non-breeders, using logistic regression

Model	QIC	ΔQIC	Weight
Warning call	30.73	0	0.24
Warning call + group size	31.72	0.99	0.15
Warning call + habitat	32.92	2.19	0.08
Warning call + habitat + warning call x habitat	33.22	2.49	0.07
Warning call + group size + habitat	33.31	2.58	0.07
Warning call + group size + warning call x group size	33.53	2.8	0.06
Warning call + age	33.65	2.92	0.06
Warning call + group size + age	33.68	2.95	0.05
Warning call + kinship	33.75	3.02	0.05
Group size + age	33.86	3.13	0.05
Age	34.21	3.48	0.04
Kinship	34.44	3.71	0.04
Age + kinship	35.4	4.67	0.02
Group size	35.69	4.96	0.02
Predictor	Sum of weights
Warning call	0.83
Group size	0.40
Age	0.23
Habitat	0.22
Kinship	0.11

All models with a ΔQIC < 5 are shown. QIC = quasi-likelihood information criterion[34], ΔQIC = difference between focal model and model with lowest QIC value, weight = relative probability of focal model within candidate set. Warning call = warning call heard or not; group size; age: age of non-breeder; habitat = proportion unmanaged forest on territory; kinship: kinship of non-breeder in relation to breeder.

ISSN: 1742-9994