Summary 1. Channel complexity is an important ecological property of stream systems and is often targeted for restoration in channelised urban streams. However, channel complexity is rarely defined explicitly, and little research on channel complexity has been conducted in streams in urban catchments that have not been directly channelised by human activities. Therefore, it remains unclear whether restoration of non-channelised urban streams has improved complexity.
2. We explicitly define channel complexity and use a multimetric approach to provide a comprehensive assessment of complexity in multiple restored, urban and forested streams on the Maryland Coastal Plain and two streams of differing land use in Colorado. We also expand on the Maryland and Colorado results with a literature survey of channel complexity from diverse geographical regions.
3. Many streams draining urban catchments in Maryland had relatively high values of some complexity metrics compared to forested reference streams in Maryland and compared to the values for pristine streams calculated from the literature. This suggests that streams in urban catchments that are not directly manipulated by human activities (e.g. channelisation or piping) may be able to maintain channel structures beneficial for aquatic organisms even when impervious surfaces are the dominant form of land use in the catchment.
4. Restored streams in Maryland had equal or lower values of many complexity metrics compared to streams draining urban catchments in Maryland. This suggests that restoration of streams draining urban catchments did not improve the overall channel complexity.
5. Our results highlight the need to explicitly define and measure the attributes of channel complexity that are targeted during restoration, to determine whether the streams in urban catchments are truly degraded with respect to channel complexity.
6. Combined with recent synthesis work suggesting that biodiversity may not be improved by increasing the channel complexity, these results indicate that targeting catchment processes may prove a more useful approach to restoration than attempting to move channel complexity in streams draining urban catchments towards conditions in forested reference streams.