Abstract
A statistical analysis is presented of the relationships between the bird numbers and the characteristics of 42 hedges on a Dorset [England, UK] farm. Multiple regression, principal components analysis and regression of factors, and canonical correlation analysis were used and compared. Regression equations derived from the differences between hedges were used to predict the effects of elm felling on birds; these predictions were compared with the actual changes observed. Bird-rich hedges were characterized as having a large basal area, many tree species, some dead timber and being near scrub habitats. The multiple regression and regression of factors produced interpretable results with reasonable predictive power. Canonical correlation analysis detected pattern in the data set but it apparently had no meaning.