Table 3.

Results of the environmental GAMMs fitted to eggshell reflectance (PC1) and spot patterning (PC2)

Model	df	AICc	∆AICc	w i	$R_{\mathrm{m}}^2$	$R_{\mathrm{c}}^2$
PC1
Year + sv + s (insolation) + s (temperature)	9	451.83	22.85	0.000	–	–
Year + s (insolation) + s (temperature)	8	452.36	23. 38	0.000	–	–
Sv + s (insolation) + s (temperature)	8	449.81	20.83	0.000	–	–
Year + sv + s (insolation)	7	447.14	18.16	0.000	–	–
Year + sv + s (temperature)	7	447.81	18.83	0.000	–	–
Year + sv + insolation3	8	442.02	13.04	0.001	0.34	0.47
Year + sv + insolation4	9	428.98	0.00	0.998	0.45	0.48
Null model	3	464.12	35.14	0.000	0.00	0.37
PC2
s (insolation) + s (temperature)	7	363.76	6.88	0.014	–	–
s (insolation)	5	360.06	3.18	0.089	–	–
s (temperature)	5	361.23	4.35	0.049	–	–
Insolation	4	357.87	0. 98	0.266	0.02	0.26
Temperature	4	359.04	2. 15	0.148	0.00	0.27
Null model	3	356.89	0.00	0.434	0.00	0.27

Parametric linear effects were applied for surface/volume (sv) of eggs and collection year. Smoothing spline functions (s) were used both for temperature and solar radiation, then linear and polynomial (cubic and quartic) models were also fitted. Collection site was entered as a random effect in both models.

ΔAICc, the difference between the lowest AICc and the AICc score of each model; w _i, Akaike weight; $R_{\mathrm{m}}^2$ , marginal R ²; $R_{\mathrm{c}}^2$ , conditional R ².