Table A1.
The day–maze model yielded the smallest AIC.
| Models | AIC | Estimated coefficients of variables | ||||
|---|---|---|---|---|---|---|
| β0 (Intercept) | β1 (Day) | β2 (Maze) | β3 (Cage) | |||
| 1 | [β0, β1, β2] | 14794 | 9.475460 | 0.033304 | −0.347460 | – |
| 2 | [β0, β1, β2, β3] | 14794 | 9.442871 | 0.033308 | −0.347421 | (0.065203) |
| 3 | [β0, β2] | 14812 | 9.47573 | – | −0.34738 | – |
| 4 | [β0, β2, β3] | 14813 | 9.44303 | – | −0.34736 | (0.06539) |
| 5 | [β0, β1] | 14822 | 9.301773 | 0.033290 | – | – |
| 6 | [β0, β1, β3] | 14823 | 9.269186 | 0.033299 | – | (0.065212) |
| 7 | [β0] | 14841 | 9.30201 | – | – | – |
| 8 | [β0, β3] | 14842 | 9.26935 | – | – | (0.06536) |
AICs for running distance (Experiment 1). AICs and estimated coefficients of variables were calculated for 8 models designed for running distance in Experiment 1. The models are sorted in ascending order of AIC. Coefficients in parentheses represents that the 95% confidence interval (CI) of the estimate included 0. The model [β0, β1, β2] indicates that both day and maze had significant effects, whereas the same-AIC. Model [β0, β1, β2, β3] indicates that cage was not reliable for its coefficient. Most-likely fitting formulas are indicated below:
Single foraging in the maze: X = 9.128 + (0.033304 + ris) × day + rii.
Paired foraging in the maze: X = 9.475460 + (0.033304 + ris) × day + rii.