Table 2.

Log-likelihood Ratio tests to estimate inter- and intra-individual variability on the two principal components (PCs) of the principal component analysis (PCA).

	df	AIC	Loglik	L.Ratio	P
(a)
Random intercept model PC1
LM	5	262.67	−126.34
LME_1|colony	6	228.64	−108.32	7.08	0.008
LME_1|colony/ID	7	254.48	−120.24	5.11	0.024
Random intercept model PC2
LM	5	239.54	−114.77
LME_1|colony	6	237.84	−112.92	3.70	0.054
LME_1|colony/ID	7	225.13	−105.57	14.72	<0.001
(b)
Random slope model PC1
LME_1|colony/ID	7	242.57	−114.29
LME_0+array|colony/ID	6	235.93	−111.96	4.64	0.031
Random slope model PC2
LME_1|colony/ID	7	201.92	−98.46
LME_0+array|colony/ID	6	227.93	−107.92	19.00	<0.001

(a) To study inter-individual variability we compared a linear model (LM) built using each PC as a response variable and age, body size and experimental array as fixed variables with two mixed effect models (LMEs) using colony or individual nested in colony as random effects. (b) To study intra-individual variability we compared the random intercept model (LME_1|colony/ID) previously built using each PC with a random intercept and slope model (LME_0+array|colony/ID). Degree of freedom (df), Akaike Information Criterion (AIC), Log-likelihood values (Loglik) and Log-likelihood ratio test (L.Ratio) are presented with the corresponding p-values. Significant effects are highlighted in bold.