2.5 Statistical analysis
All data used in this study met the Shapiro–Wilk test of normality and the Levene’s test of homogeneity of variance.
Two-way ANOVA of general linear models were used to assess the effects of fencing duration, shrub cover and their interaction on species richness, phylogenetic diversity and phylogenetic structure. A one-way ANOVA analysis with Tukey’s HSD test was conducted to evaluate the difference among levels of fencing duration or shrub cover when interaction were not significant. Alternatively, a simple-effect analysis was used to evaluate the difference of levels of one factor under a certain condition of another factor when significant interaction were found. Linear model was applied to test the relationship between phylogenetic diversity and species richness.