This is a very open ended question. From my experience on streams, this is a two-fold problem.
There is the channel restoration and all this entails, i.e., putting structure in stream (e.g., rocks, logs, etc.), reconnecting channel to flood plain (e.g., difficult or impossible to do in developed areas), and restoring vegetation to stream banks (shade to streams is important). Often in urban areas the channel banks will be hardened (which is not conducive to biological improvement for that particular area) to protect infrastructure and property.
Secondly, particularly in urban areas, storm generated flows to the stream require storage and gradual release to mimic the predevelopment hydrology. Prior to development, which increases impervious cover (e.g., streets, roofs, parking lots, runways, etc.), a large percentage of rainfall infiltrated into the ground and replenished the stream flow between storms. Streams in post-developed areas have flashy stream flows, i.e., large, erosive flows during rain events and little flow, due to decrease pervious cover, between rain events. That is, unless storage for rain fall induced flows has been incorporated into the watershed during development.
In evaluating a stream restoration, we found little or no biological improvement by assessing macroinvertebrates. Macroinvertebrates will be one of the first organisms to return to a stream restoration area and typically are food for higher order organisms or basis of food web. The stream restoration did not include building storage for stormwater in the upper watershed area; therefore, the channel was still experiencing large storm induced flows. We wrote a report and paper. Report available:
"Evaluation of Receiving Water Improvements from Stream Restoration (Accotink Creek, Fairfax City, Virginia)" U.S. Environmental Protection Agency, Office of ReseReport No. EPA/600/R-08/11