Appendix E. Logistic regression results for the effect of canopy fungi (presence/absence of fungi on nearest mature conspecific), juvenile density (no. juveniles in 3-m radius around adult), and percentage canopy openness on the proportion of juvenile Cleistanthus myrianthus colonized by epifoliar fungi at Cape Tribulation, Australia.
Akaike Information Criterion (AIC), Likelihood Ratio (LR), P values of Wald test chi-square estimates for each factor, and percentage concordance were used to select the best model. Intercept was always significant (P ≤ 0.042 for all).
P values |
|||||||
Model |
AIC |
LR Chi-square |
LR |
Can Fungi |
Juv Dens |
% open |
Conc.% |
1 |
55.7 |
11.6 |
0.0089 |
0.211 |
0..1366 |
0.0166 |
77 |
2 |
61.3 |
5.18 |
0.0749 |
0.0969 |
0.0446 |
63.9 |
|
3 |
56 |
9.29 |
0.0096 |
0.4875 |
0.0083 |
75.2 |
|
4 |
55.2 |
10.06 |
0.0065 |
0.2681 |
0.0101 |
72.6 |
|
5 |
62.5 |
0.85 |
0.3570 |
0.3498 |
23.8 |
||
6 |
60.9 |
2.42 |
0.1199 |
0.1211 |
47.4 |
||
7 |
54.5 |
8.82 |
0.0030 |
0.007 |
72.6 |
||
Model 7 was the best fit (lowest AIC with high concordance),
taking the form p(juvenile colonized) = expy / (1 + expy) where:
y = -2.8624 + 0.2729(% Canopy Openness).
Fit of this model against the observed data are shown in Fig. E1.
|
| FIG. E1. Increased canopy openness increases the likelihood of epifoliar fungi on understory juveniles of Cleistanthus myrianthus. Each point is the percentage of juveniles within 3 m of an adult that were colonized by epifoliar fungi. The solid line indicates the probability that a juvenile would be colonized based on best-fit logistic regression (logit(probability juvenile colonized) = -2.8624 + 0.2729 (canopy openness). |