Table 2.
Results of the influence of traits on species association strength calculated by the social network analysis
| Habitats | Variables | Within-MSF level | Whole-MSF level | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Number of significant transects | Number of Bonferroni-corrected significant transects | Mean Estimate | P(β ≥ r) | P(β ≤ r) | P(|β| ≤ |r|) | Number of significant transects | Number of Bonferroni-corrected significant transects | Mean estimate | P(β ≥ r) | P(β ≤ r) | P(|β| ≤ |r|) | ||
| Primary forest (n = 8) | Body length dissimilarity | 3 | 1 | −0.75 | ≥0.003 | ≥0.98 | ≤0.03 | 8 | 5 | −0.29 | ≥ 0 | ≥ 0.99 | ≤ 0.03 |
| Diet similarity | 0 | 0 | / | / | / | / | 2 | 0 | 0.02 | ≥ 0.98 | ≥ 0.02 | ≤ 0.02 | |
| Plumage color similarity | 0 | 0 | / | / | / | / | 1 | 1 | 0.02 | 0.99 | 0.001 | 0.001 | |
| Plumage pattern similarity | 1 | 0 | 0.04 | 0.97 | 0.04 | 0.05 | 1 | 0 | 0.02 | 0.98 | 0.02 | 0.02 | |
| Phylogenetic distance | 4 | 3 | −1.78 | ≥0 | ≥0.99 | ≤0.03 | 1 | 0 | −0.43 | 0.005 | 1 | 0.005 | |
| Secondary forest (n = 6) | Body length dissimilarity | 1 | 0 | −0.14 | 0.02 | 0.99 | 0.04 | 4 | 4 | −0.28 | ≥ 0 | ≥ 0.99 | ≤ 0.002 |
| Diet similarity | 0 | 0 | / | / | / | / | 0 | 0 | / | / | / | / | |
| Plumage color similarity | 2 | 0 | 0.03 | ≥0.98 | ≥0.005 | ≤0.02 | 1 | 0 | 0.02 | 0.97 | 0.03 | 0.04 | |
| Plumage pattern similarity | 3 | 2 | 0.01 | ≥0.98 | ≥0 | ≤0.03 | 3 | 1 | 0.01 | ≥ 0.99 | ≥ 0 | ≤ 0.02 | |
| Phylogenetic distance | 4 | 3 | −0.47 | ≥0.001 | ≥0.99 | ≤0.02 | 2 | 0 | −0.53 | 0.004 | 1 | 0.007 | |
Each transect was analyzed at the whole-MSF and the within-MSF levels. “P(β ≥ r)” represents the probability of the regression coefficient in the model being equal or larger than the permuted value; “P(β ≤ r)” represents the probability of the regression coefficient being equal or less than the permuted value; “P(|β| ≤ |r|)” represents the probability that the absolute value of the regression coefficient (β) being equal or less than the absolute permuted value. P(|β| ≤ |r|) was considered significant if it was ≤0.05. We also show how many results would be considered significant if the dataset was Bonferroni-corrected, given that we made 14 tests, one for each transect.