A multivariate analysis of breeding bird species of western and central Canadian boreal forests: Stand and spatial effects

We examined the abundance and distribution of 80 bird species in 101 breedi ng bird census plots collared from a wide range of sources in the boreal fo rest of western Canada. In order to examine the relative importance of stan d attributes and geographical location we used Canonical Correspondence Ana lysis (CCA). Two-way indicator species analysis (TWINSPAN) identified seven stand groups ranging from young stands to mature mixed woods to spruce-dom inated stands and finally wet sites. CCA demonstrated that hardwood and sof twood composition of stands, amount of jackpine, stand age, upland versus w etland classification, and geographic location of stands explained signific ant variation in bird species composition and abundance. Using partial CCA analysis, we partitioned out the total variation in the species matrix as f ollows: i) non-spatial environmental variation: 24%; ii) spatially structur ed environmental variation: 3%; iii) spatial species variation that is not shared by the environmental variables: 14%; and iv) unexplained variation a nd stochastic fluctuations: 59%. Stand variables explained 27% of the varia tion in the species matrix; however, the geographical component explained o nly one ninth of this variation. Thus, while geographical location contribu ted significantly to variation in the species matrix, its overall effect wa s small. Our results suggest that for several boreal forest birds, useful h abitat or stand association models can be based on existing forest inventor y and ecosystem site classification data.