Table 7.
Weight of predictors for each response variable.
| N | Stim | Day | Loc | -Behav Prox | Stim *Day | Stim *Loc | Stim *-Behav Prox | Loc *-Behav Prox | Int Index | Stim *Int index | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Total Number of grunt-Poisson | 2 | 1.00 | 1.00 | 1.00 | 1.00 | – | 1.00 | 1.00 | 0.47 | NA | NA |
| Grunt rate (log) | 2 | 1.00 | – | – | 0.30 | – | – | – | – | – | |
| Grunt duration (log) | 6 | 1.00 | 1.00 | 1.00 | 1.00 | – | 1.00 | 0.24 | 0.25 | 1.00 | 0.51 |
| Acoustic spectral score (PC1ac.) | 9 | 1.00 | 1.00 | 1.00 | 1.00 | 0.36 | 1.00 | 0.44 | 0.13 | 1.00 | 0.24 |
The number of equivalent best models (N), and each term of the full initial model are indicated in columns: “Stim” for stimulus type, Day, “Loc” for location in the room (Close to or Away from the stimulus), “Int. Index” for interval index, “Behav Prox” for behavioral proximity score, as well as relevant interactions between explanatory variables. Only weights different from zero are indicated. For the total number of grunts, since the variable is a sum of all intervals per day, location, individuals and type of stimulus, the interval index was not included in the full model and referred as “NA.” The best predictors are the one consistently appearing in all equivalent selected models so the ones having a weight of “1”.