Disturbance | Estimated effect | P value |
---|---|---|

Intercept (AA) | 0.0163 | NA^{b} |

T1: D | −0.0450 | 0.3262 |

T1: S | 0.0565 | 0.2159 |

T2: D | 0.0659 | 0.1510 |

T2: S | 0.0276 | 0.5472 |

T1: D × T2: D | −0.143 | 0.0282* |

T1: S × T2: D | −0.228 | <0.001*** |

T1: D × T2: S | 0.022 | 0.7350 |

T1: S × T2: S | −0.176 | 0.0065** |

Random effect | ||

OTU | 0.0459 | NA |

↵a Results of the linear mixed model using disturbances at T1 and T2 as predictors of the OTU-level variability (as given by the residuals of OTU CVs) of bacterial communities from the nine experimental treatments. As in Table 1, disturbance effect estimates are given in comparison to the undisturbed treatment, AA. There were no significant effects of single disturbances on the variability of OTUs at T1 or T2. However, there were significant negative interactions between three doubly disturbed treatments, such that treatments DD, SD, and SS were less variable than would have been expected. No

*P*value was calculated for the random effect, because we were not interested in testing how much variability was explained by differences between taxa. *,*P*< 0.05; **,*P*< 0.01; ***,*P*< 0.001.↵b NA, not applicable.