Table 2 Comparisons of the relationships between proportion of dead seeds and seedlings caused by animal attack/pathogen infestation towards increasing DTNCT among different fragment types at the seedling-emergence stage, using GLMMs assuming binomial distribution of residuals. P-values of all tests were shown and Bonferroni correction was used to evaluate the significance for all pair-wise tests

From: Distance-dependent seed‒seedling transition in the tree Castanopsis sclerophylla is altered by fragment size

Response variable

Fixed effects

Model parameters

LR tests

Z-tests

  

Residual deviance

df of residuals

df

LR (p-value)

Pair-wise comparisons (slope (mean ± S.E.))

z-value (p-value)

Proportion of dead seeds and seedlings caused by animal attack

Fragment type × DTNCT (LN-transformed)

1479.5

1215

3

6.90 (0.075)NS

Mainland (−0.40 ± 0.20*) vs. large island (−0.43 ± 0.18*)

0.11(0.915)NS

      

Mainland vs. medium islands (−1.65 ± 0.49***)

2.39 (0.017)NS

      

Mainland vs. small islands (−0.51 ± 0.20*)

0.39 (0.698)NS

      

Large island vs. medium islands

2.38 (0.017)NS

      

Large island vs. small islands

0.30 (0.765)NS

      

Medium islands vs. small islands

−2.19 (0.028)NS

Proportion of dead seeds and seedlings caused by pathogen infestation

Fragment type × DTNCT (LN-transformed)

1298.1

1215

3

28.69 (<0.001)***

Mainland (−0.85 ± 0.25***) vs. large island (−0.66 ± 0.20**)

−0.61 (0.542)NS

      

Mainland vs. medium islands (−1.72 ± 0.94NS)

0.90 (0.366)NS

      

Mainland vs. small islands (0.46 ± 0.18*)

−4.25 (<0.001)***

      

Large island vs. medium islands

1.12 (0.263)NS

      

Large island vs. small islands

−4.11 (<0.001)***

      

Medium islands vs. small islands

−2.28 (0.022)NS

  1. NSNot significant, *p < 0.05, **p < 0.01, ***p < 0.001