• Lake States;
  • hierarchy theory;
  • large wood;
  • redundancy analysis;
  • regression trees


Large wood (LW; pieces with diameter greater than 10 cm and length greater than 1 m) and large-wood jams (LWJs; two or more pieces of LW in contact with each other) are important components of stream ecosystems that are often distributed along stream channels in response to geomorphic and riparian forest factors that interact hierarchically. As a result, information on these relationships is valuable for predicting patterns of wood accumulation and characteristics of individual pieces of wood. We studied relationships between geomorphic and riparian factors and LW and LWJ structure in different geomorphic settings associated with old-growth and second-growth settings in Upper Michigan. We used redundancy analysis (RDA) and regression tree analysis to evaluate changes in LW and LWJ structural characteristics among geomorphic and riparian forest settings. Geomorphic factors explained 38·5% of the variability in LW and LWJ characteristics, riparian forest factors uniquely explained 18·4% of the variance and the intersection of the two categories of environmental factors (i.e. the redundant portion) was 29·8%. At the landscape scale, our multivariate analyses suggest that the presence of rock-plane bedding was an important predictor of the number of LWJs and the percent of channel spanned by LWJs. Our analyses suggest differences in relationships between geomorphic factors and LW and LWJ structure. Channel width, distance from headwaters, gradient and sinuosity were identified by regression tree analyses as the most important variables for predicting LW characteristics, while channel width and confinement were the most important variables for predicting LWJ characteristics. Old-growth settings generally contained a higher proportion of conifer and LW (both in and out of LWJs) with greater diameter and volume than in second-growth settings. Our study supports the view that restoration of wood to streams will benefit from considering the associations of wood structure with landscape and reach-scale geomorphology. Copyright © 2007 John Wiley & Sons, Ltd.