Potential and actual geomorphic complexity of restored headwater streams in northern Sweden

Stream restoration usually relies on ecological theories presuming that increased habitat heterogeneity leads to higher biodiversity. However, to test this hypothesis a quantitative metric of overall geomorphic complexity is needed. We quantified geomorphic complexity using 29 metrics over five dimensions (sediment distribution, longitudinal profile, cross section, planform, and instream wood) of headwater streams in northern Sweden. We examined reaches with four different restoration statuses after a century of timber floating (channelized, restored, demonstration restored, and unimpacted) to determine (1) whether restoration increases complexity in all dimensions, (2) whether a complexity gradient can be quantified and which metrics can serve as proxies for the gradient, and (3) levels of potential complexity based on large-scale controls (drainage area, glacial legacy sediment, valley slope, valley confinement, old-growth forest/buffer zone, and beaver activity). We found a significantly higher complexity in unimpacted and demonstration restoration sites than in channelized sites in all five dimensions except the cross section (based on the two metrics quantifying variability in the cross section). Multivariate analyses were able to elucidate an apparent complexity gradient driven by three complexity metrics: longitudinal roughness, sediment sorting, and cross section chain and tape ratio. The large-scale factors of valley and channel gradient as well as median grain size, along with restoration status, drive differences in complexity composition. Restoring a reach to its potential complexity is beneficial in regions without reference systems or sufficient data to model flow and sediment processes. Unimpacted and demonstration restoration reaches displayed not only more intrareach variability than channelized reaches but also greater interreach heterogeneity in complexity composition, which supports a focus on reach-scale controls on potential complexity and a landscape-scale view on restoration.