Nutrients in the Swan-Canning Estuary  ||  Stream restoration of Wilson Creek
Ohio River ecosystem study  ||  James River ecosystem study

Stream Restoration of Wilson Creek, Ky.

While at the University of Louisville, I collaborated on a stream restoration project with fellow faculty members Jeff Jack (biology) and Art Parola (engineering). The project team also included Chuck Rhodes and Adam Datillo of the University of Kentucky School of Forestry. The project was organized by Margaret Shea, natural areas manager, Bernheim Arboretum and Research Forest, and funded by the Environmental Protection Agency.

Wilson Creek prior to restoration. Note lack of natural meanders and predominance of bedrock (erosion of streambed).

Background
Streams and their associated riparian ecotones provide important goods and services through their role in the cycles of water, energy and materials. Vital services include water storage, maintenance of biodiversity and mitigation of downstream nutrient transport. Recent interest in nutrient retention in streams and rivers reflects concerns about eutrophication in downstream and coastal ecosystems. Nutrients are retained in streams through biological uptake by algae and bacteria. The efficiency with which biota remove nutrients from stream water depends on water transit time (how long water is resident within the stream) and by rates of metabolism (biological activity). Channelization of streams increases the velocity with which water moves through the channel and thereby diminishes opportunities for biotic uptake. As part of a stream restoration project, nutrient uptake rates were measured before and after restoration to determine whether naturalizing the channel resulted in increased nutrient retention.

Site description
The restoration site, Wilson Creek, is a third-order stream located approximately 60 km south of Louisville, Ky. The 1-km reach selected for restoration included 6.5 ha of floodplain and was located within the Bernheim Arboretum and Research Forest. The reference site, Harts Run is a third-order tributary that flows into Wilson Creek approximately 1 km below the site of restoration. Like many streams in this region, Wilson Creek was channelized and relocated to the margin of its floodplain, adjacent to valley hill slope, for bottomland agriculture. This alteration resulted in a stream channel that was incised, entrenched and confined (bank-full capacity comparable to a 10-year event). Runs comprise nearly all of the study reach while pools and riffles represent less than 10 percent of the total length. Substrate is exposed bedrock (dolomitic limestone) with isolated patches of gravel and siltstone cobble. The catchment of Harts Run (7.5 km²) is located within the Bernheim Arboretum and Research Forest and is 95 percent forested. Like Wilson Creek, its channel has been impacted by agricultural activities in the floodplain. An important distinction is that the upper section of Harts Run was not relocated to the margin of the floodplain. Secondary re-sorting of bank and floodplain materials during the past 60-100 years have resulted in a meandering channel that is dominated by riffles and pools with gravel and cobble substrate (little exposed bedrock). Its selection as a reference site serves to quantify the condition attained by a stream that has recovered to a more natural state rather than to typify pristine conditions occurring prior to European settlement. Both Wilson Creek and Harts Run are shaded by a relatively mature riparian canopy (50+ years) dominated by sycamores (Platanus occidentalis) and white oaks (Quercus alba).

Stream restoration
The purpose of the restoration was to reconnect Wilson Creek with its floodplain and to establish native riparian, bottomland forest and wet meadow communities in the floodplain. Stream-floodplain connectivity was re-established by relocating the channel and reducing its bank-full capacity. A detailed description of channel design and construction will be provided elsewhere and briefly summarized here. The design of the restored channel followed Rosgen (1996) with parameter ranges for bank-full dimensions, meander belt width, meander radius and channel slope obtained from reference streams. The morphometry and location of the designed channel was determined in part by historical considerations, as revealed by underlying deposits and microtopography of the floodplain, and the desire to achieve a profile that would sustain long riffles with short runs into deep pools. Floodplain terracing was completed with a bulldozer while pools were excavated with a backhoe to avoid compaction of alluvium. Riffles were lined with gravel taken from the old channel, and jute or burlap fabric was used to stabilize banks. The meandering form of the designed channel resulted in a total stream length of 944 m (vs. 823 m prior to restoration). Shrubs and trees adapted to occasional flooding were planted proximal to the designed channel: native giant cane (Arundinaria gigantean), American sycamore (Platanus occidentalis), boxelder (Acer negundo), black willow (Salix nigra), dogwood (Cornus sp.) and northern spicebush (Lindera benzoin).

Measuring stream nutrient retention
Retention can be quantified by injecting nutrients and a conservative tracer, such as salt or dye, and measuring their disappearance downstream (injection experiments). Two sub reaches were selected in each of the reference (Harts Run), channelized (Wilson Creek, prerestoration) and designed (Wilson Creek, post-restoration) streams. Sub-reach lengths were fixed despite changing discharge and transit time in order to obtain site-specific measurements of nutrient and water dynamics through time. Experiments were performed during a spring index period (mid-April to mid-June) to characterize stream properties over a range of discharge, canopy and temperature conditions. Work in 2002 focused on hydrodynamics (conservative tracer only), whereas experiments in 2003 and 2004 entailed co-injection of conservative and nonconservative solutes. A total of 39 experiments were performed at the reference (N = 12), channelized (N = 15) and designed (N = 12) streams.

Result
Transient storage of water within the Wilson Creek channel prior to restoration was consistently lower than that observed in a nearby reference stream (Hart’s Run). Naturalization of the channel did not substantially improve transient storage over most of the restored section since the new channel was deeper and narrower. However, in sections where remnants of the old channel connected to the restored channel (serving as backwater areas), transient storage was much higher and exceeded that of the reference stream. We also observed high nitrogen retention in these reaches. The presence of recirculating zones, which connected the active (designed) channel with the backwater (remnant) channel, may facilitate retention through denitrification occurring within the latter. Retention of N and P in the new channel was otherwise similar to rates observed prior to restoration and lower than that measured in the reference stream. Further work in 2005 will focus on N processes in backwater zones.

Acknowledgments
I am indebted to Randall Kelly for his assistance in the field and to Rich Schultz for sample analyses. Robert Hall and Pat Mulholland gave useful advice on the execution and interpretation of solute injection experiments.