| Abstract: | Habitat associations of 40 bird species were analysed using data from a survey of 12892 garden sites throughout the United Kingdom between 1995 and 2002. Gradients in 20 garden habitat variables and 25 variables describing the surrounding local habitat were derived from Detrended Correspondence Analysis. The main garden axis explained 13.3% of variation in the data and described a gradient from sites containing and bounded by high levels of vegetation cover to open sites bounded by walls and fences. The main local axis explained 29.9% of variation in the data and described a gradient from rural to urban habitats. Higher probability of occurrence was detected in larger sites in 25 species and in smaller sites in four (urban-associated) species. Many species (22 out of 40) were significantly associated with the local gradient: seven species showed the highest probability of occurrence in sites within more urbanised habitats. Other species were most likely to occur in sites within rural or suburban habitats. Only five species showed a significant association with garden habitat gradient, four of these being most likely to occur in sites with highest vegetation cover. These associations were similar when controlling for garden size and for food provision in gardens. The results imply that the likelihood of many species occurring in gardens is dependent on the surrounding local habitat rather than the garden habitat. However, survey participants were likely to have "bird-friendly" gardens which may have reduced the variation in garden habitat across the sample. An assessment of the representativeness of survey sites is needed to further understand habitat associations of garden birds. |
