Table 4.
Model selection results of the analysis of breeding territories (n = 56) vs. abandoned territories (n = 20)
| Hypothesis | Variables in model | LL | K | AICc | ΔAICc | Weight |
|---|---|---|---|---|---|---|
| Forest structure | ||||||
| (a) Ground variables | Number of tussocks, year, number of tussocks x year | −22.315 | 7 | 60.277 | 0 | 0.202 |
| Number of tussocks, number of bushes2 | −22.774 | 7 | 61.194 | 0.917 | 0.128 | |
| Number of tussocks, number of bushes2, cover of herb layer | −21.562 | 8 | 61.273 | 0.996 | 0.123 | |
| Number of tussocks, year, number of tussocks x year, number of bushes | −21.798 | 8 | 61.745 | 1.468 | 0.097 | |
| Number of tussocks, year, number of tussocks x year, cover of herb layer | −21.986 | 8 | 62.122 | 1.845 | 0.080 | |
| … | ||||||
| Null | −31.868 | 4 | 72.299 | 12.022 | 0.000 | |
| (b) Tree variables | Number of trees2, tree dbh2, tree species diversity2 | −17.092 | 10 | 57.569 | 0 | 0.091 |
| Number of trees, tree species diversity2 | −21.187 | 7 | 58.020 | 0.452 | 0.072 | |
| Number of trees, tree dbh2, tree species diversity2 | −18.652 | 9 | 58.031 | 0.462 | 0.072 | |
| Number of trees2 | −22.514 | 6 | 58.246 | 0.677 | 0.065 | |
| Number of trees, tree species diversity2, tree dbh | −20.060 | 8 | 58.269 | 0.701 | 0.064 | |
| Number of trees, tree dbh2, tree species diversity2, sky visibility | −17.695 | 10 | 58.776 | 1.207 | 0.050 | |
| Number of trees2, tree dbh2, tree species diversity2, sky visibility | −16.336 | 11 | 58.797 | 1.229 | 0.049 | |
| Number of trees2, tree species diversity2 | −20.364 | 8 | 58.877 | 1.308 | 0.047 | |
| Number of trees2, tree dbh2, tree species diversity2, sky visibility2 | −15.234 | 12 | 59.420 | 1.852 | 0.036 | |
| … | ||||||
| Null | −31.868 | 4 | 72.299 | 14.730 | 0.000 | |
| (c) Tree species composition | Proportion conifers | −29.723 | 5 | 70.303 | 0 | 0.265 |
| Proportion conifers, proportion beech | −29.533 | 6 | 72.283 | 1.981 | 0.098 | |
| Proportion conifers, proportion other deciduous trees | −29.539 | 6 | 72.295 | 1.992 | 0.098 | |
| … | ||||||
| Null | −31.868 | 4 | 72.299 | 1.996 | 0.098 | |
| Rodent-avoidance | Null | −31.868 | 4 | 72.299 | 0 | 0.575 |
| Rodent numbers | −31.023 | 5 | 72.903 | 0.605 | 0.425 | |
| Disturbance | Distance to forest edge, distance to path2 | −20.719 | 7 | 57.084 | 0 | 0.694 |
| Distance to forest edge2, distance to path2 | −20.453 | 8 | 59.055 | 1.97 | 0.259 | |
| … | ||||||
| Null | −31.868 | 4 | 72.299 | 15.214 | 0.000 | |
| Topography | Slope steepness2, elevation2, southness, eastness | −13.541 | 10 | 50.467 | 0 | 0.692 |
| … | ||||||
| Null | −31.868 | 4 | 72.299 | 21.832 | 0.000 | |
| Across hypothesesa | Slope steepness, distance to forest edge, number of trees | −14.985 | 7 | 45.457 | 0 | 0.734 |
| … | ||||||
| Null | −31.868 | 4 | 72.299 | 26.842 | 0.000 | |
For each hypothesis, the top-ranked model (ΔAICc = 0), the models with ΔAICc < 2 to the top-ranked model and the null model (referred to as “null”) are shown. “…” refers to additional models examined, but not listed in detail to avoid overlong table
LL log-likelihood; K number of parameters in the model (including intercept), weight Akaike weight (chance of the model to be the best one, given the candidate models)
The quadratic effect of a variable x, composed of a linear and a quadratic component (x ± x2), is denoted as x2
Each model included x- and y-coordinates (and their interaction) of territories to account for spatial autocorrelation
a Only linear terms of variables from best models per hypothesis and at most three habitat variables jointly used due to convergence problems with quadratic terms and more than three habitat variable per model