Table 2.

Results of the spatio‐temporal GAMMs fitted to eggshell reflectance (PC1) and spot patterning (PC2)

Dependent variable	Fixed effects	df	AICc	∆AICc	w i	$R_{\mathrm{m}}^2$	$R_{\mathrm{c}}^2$
PC1	Year + s (latitude)	6	450.59	2.24	0.206	–	–
	Year + latitude	5	448.35	0.00	0.632	0.24	0.39
	Latitude	4	453.16	4.81	0.057	0.18	0.39
	Year	4	452.00	3.61	0.104	0.17	0.37
	Null model	3	464.12	16.12	0.000	0.00	0.37
PC2	Year + s (latitude)	6	361.12	4.24	0.061	–	–
	Year + latitude	5	358.88	2.00	0.167	0.03	0.27
	Latitude	4	358.68	1.80	0.168	0.01	0.27
	Year	4	358.11	1.22	0.223	0.01	0.27
	Null model	3	356.89	0.00	0.381	0.00	0.27

Parametric coefficients for latitude were estimated when the degrees of freedom for the spline estimates (s) equals 1. Collection site was entered as an intercept‐only random effect in both models. The difference between the lowest AIC and the AICc score of each model (ΔAICc), the Akaike weight (w _i), the variance explained by the fixed factors ($R_{\mathrm{m}}^2$ ), and the variance explained by the entire model ($R_{\mathrm{c}}^2$ ) are presented.