Table 4.
Randomization analysis of variances showing each effect and each response variable based on position
| Effect | Position | df | MS | F | p |
|---|---|---|---|---|---|
| Chemical cues | Top | 1 | 0.001 | 0.29 | .60 |
| Bottom | 1 | 0.000 | 0.033 | .85 | |
| In/on refuge | 1 | 0.0002 | 0.023 | .88 | |
| Infection status | Top | 1 | 0.0344 | 16.91 | .0001 |
| Bottom | 1 | 0.1506 | 33.31 | <.0001 | |
| In/on refuge | 1 | 0.3296 | 40.81 | <.0001 | |
| Cues:parasitism | Top | 1 | 0.007 | 0.84 | .3 |
| Bottom | 1 | 0.011 | 1.18 | .30 | |
| In/on refuge | 1 | 0.000 | 0.15 | .71 | |
| Residuals | Top | 34 | 0.0020 | ||
| Bottom | 35 | 0.0045 | |||
| In/on refuge | 35 | 0.0080 |
We have grouped results from each position according to the different effects for easier comparison. The response variable for each ANOVA is the proportion of larvae in one of the three positions. Chemical cues indicate whether predatory cues were present or absent, while parasitism status refers to larvae infected with Ascogregarina barretti or uninfected. Behavior indicates which response value was used for the particular randomization ANOVA. We show degrees of freedom (df), mean squares (MS), F statistics (F), and p‐values (p) for each effect test in the model, and we have bolded model terms that are significant at the .016 level.