Table 2.
Summary of the effects of mesocosms with and without migration possibilities (percent change compared to the undisturbed samples) on the abundance of springtails and mites estimated from log-linear regression models
| Effect size (%) | Variance explained | |||
|---|---|---|---|---|
| Functional group | Mesocosms with open lids | Mesocosms with closed lids | cR2 | Power |
| All springtails | −4.5 (−53.6, 96.5) | −11.5 (−56.9, 82.2) | 0.007 | 0.06 |
| Surface-living springtails | −3.7 (−54.7, 105) | −12.5 (−58.9, 86.5) | 0.009 | 0.06 |
| Soil-living springtails | −1.7 (−62.7, 177) | −18.8 (−70.2, 121) | 0.02 | 0.07 |
| All mites | −0.4 (−52.7, 110) | −18.0 (−61.1, 72.6) | 0.02 | 0.07 |
| Saprotrophic mites | 19.1 (−44.2, 155) | −0.5 (−53.4, 113) | 0.02 | 0.07 |
| Predatory mites | −59.8 (−80.8, −15.9) | −64.0 (−82.8, −24.7) | 0.38 | 0.78 |
Effect sizes (median and 95% confidence interval) of the different mesocosm types (with open lids allowing migration, and with closed lids preventing migration) is expressed as percentage change in abundance relative to the undisturbed controls. When the 95% confidence interval is not including zero, it is significantly different from controls, and is highlighted in bold. The values in each row have been computed from separate log-linear regression models fitted to the different functional groups of soil fauna with sample type as fixed effect. The total variance explained (cR2 = conditional R2) and statistical power of each model are shown