Statistical Analyses
The statistical effect of edge and nest height was measured using generalized linear model, using a binomial distribution which measured the overall nest predation rates over the 2 week period. The response variable was the proportion of days a nest was predated, and the explanatory variables were nest height and nest location. Given the previous inconsistency between nest predation experiments, it seemed appropriate to calculate a comparable effect size of an edge effect for future reference. Cohen's d was calculated from the generalized linear model output using the sos package in R . Mantel tests were performed to examine the role spatial autocorrelation plays in nest predation at the edge and interior. All statistical analyses was performed in R . A significance level of p < 0.05 was used in all tests.