Table 2

Linear regressions exploring relationships between biodiversity value (habitat heterogeneity, tree cover and species richness), area and psychological well-being. Each pairwise combination of variables was tested separately and the resulting model, where significant, is shown. Then, area was entered into each significant model as a second term and backward stepwise selection was used to eliminate any term satisfying the removal criterion (p>0.05) with the smaller partial R2 at each stage. In two cases, shown in bold, the biodiversity measure was eliminated from the multivariate model, indicating that the effect of area was stronger than that of the biodiversity measure. In all other cases, the area term was excluded. b=model coefficient. In all cases, model significance 0.01<p<0.05.

reflectiondistinct identitycontinuity with pastattachment
no. of habitatsb=0.131b=0.139b=0.099
F=16.03F=5.267F=4.697
R2=0.552R2=0.288R2=0.265
tree cover
plant richnessb=0.747b=1.012
F=6.336F=5.034
R2=0.328R2=0.279
butterfly richness
bird richnessb=0.033b=0.024
F=5.538F=6.278
R2=0.299R2=0.341
greenspace areab=0.024b=0.035b=0.028
F=5.163F=4.842F=6.096
R2=0.284R2=0.271R2=0.319