Horizontal-vertical movement relationships: Adélie penguins forage continuously throughout provisioning trips

Table 4 Results of move persistence mixed effects models for behavioural-environmental model configurations

Model formula	AIC	LogLik	Fixed effects	Coefficients
Model formula	AIC	LogLik	Fixed effects	Est	SE	z-value	p value	CV QR	Est trend (%)
Guard
~ Wiggles + SIC + BS + Sex + (1\| id)	− 12,464	6241	Intercept	3.52	0.20	17.71	< 0.0001	3.05–4.17	100
			Wiggles	0.65	0.01	6.84	< 0.0001	0.59–0.93	100
			SIC	0.07	0.01	2.84	< 0.001	– 0.03–0.26	91
			BS	1.06	0.01	12,138.9	< 0.0001	0.89–1.79	100
			Sex	0.20	0.20	0.89	0.38	– 0.21–1.29	82
Crèche
~ Bottom duration + BATH + SST + Sex + (Bottom duration + BATH \| id)	− 14,245	7132	Intercept	3.67	0.10	108.23	< 0.0001	1.69–4.25	100
			Bottom duration	0.20	0.01	5.51	< 0.0001	– 0.22–0.81	88
			BATH	− 0.20	0.60	− 3.85	< 0.0001	– 0.54–0.62	38
			SST	0.83	0.20	52.01	< 0.0001	– 0.24–1.68	88
			Sex	1.75	0.1	150.19	< 0.001	1.32–2.12	100

Move persistence is modelled in relation to the best supported behavioural and environmental predictors separately for guard and crèche, as determined through AIC ranking (see Methods and Tables S2 and S3 for details). Estimated means ± SE are presented for wiggles, bottom duration and bathymetry on the natural log scale and logit scale for SIC. All behavioural and environmental predictors, except for SIC, were centred and scaled. Females are the reference level for the categorical Sex term. Model terms with a significant p-value at the 0.05 level are highlighted in bold text. 95% quantile range (QR) of coefficients from the “leave-one-out” cross validation (CV) are also presented alongside the percentage of CV model fits in which the estimated coefficients show a sign consistent with the final full model coefficients (refer to Methods for details)

ISSN: 2051-3933