Table 4. Results of generalised linear mixed-effects models predicting the duration of observed working behaviour.

The format of the models was: Duration of working behaviour ∼  Colony|Individual + Fixed Effect Where Duration of working behaviour is the duration of working behaviour observed during the ten-minute observation period, Colony and Individual are random effects with Individual nested within Colony, and Fixed Effect was omitted in the null model and one of sex, age (months), weight (grams) and rank (0–1, scaled) was included in the corresponding models. The scripts used to run the models are available in File S3. Relative likelihoods are calculated with exp(-($\frac{1}{2}$)ΔAICc). Nakagawa’s R2 estimates the variance explained by the fixed effects (marginal variance) and both fixed and random effects (conditional variance) (Nakagawa & Schielzeth, 2013).

Variable	Model
	Null		Sex (Base = F)		Age		Weight		Rank
	Co-ef	S.E.	Co-ef	S.E.	Co-ef	S.E.	Co-ef	S.E.	Co-ef	S.E.
Intercept	259.86	13.68	261.86	17.48	263.18	18.80	306.22	51.44	259.07	14.37
Sex			−4.74	26.28
Age					−0.09	0.31
Weight							−1.66	1.73
Rank Rank2									282.05 −649.10	200.03 200.68
Model AICc	3622.44		3624.48		3624.44		3623.71		3615.23
AICc versus Null model	–		+2.04		+2.00		+1.28		−7.21
Relative likelihood versus null model			0.36		0.37		0.53		36.78
Nakagawa’s R2:
Conditional Marginal	0.169 0.000		0.169 0.000		0.171 0.000		0.175 0.005		0.182 0.056