Maintenance of Natural Sustainable Riparian Communities Located Within the Middle Sheyenne Watershed

Final Report for GNC09-113

Project Type: Graduate Student
Funds awarded in 2009: $9,528.00
Projected End Date: 12/31/2011
Grant Recipient: North Dakota State University
Region: North Central
State: North Dakota
Graduate Student:
Faculty Advisor:
Edward DeKeyser
North Dakota State University
Expand All

Project Information

Summary:

Geologic features of the Middle Sheyenne River Valley located within Eddy and Nelson counties in eastern North Dakota are dominated by alluvial terraces and flood plains, characteristic of valley type VIII as described by Rosgen. Valley type VIII supports stream types C, D, E, F, and G. Nine cross-sections were sampled to identify stream type using Rosgen’s classification of natural rivers. Classification of these cross-sections resulted in seven C5/6 and two E5/6 channels. The results of our stream classification lead us to conclude that stream type succession scenario 1 is most appropriate for the Middle Sheyenne River. The stable reference reach is an E channel; when stability is lost it first becomes a C channel, which is transformed into a Gc channel, followed by a F channel. As it begins to stabilize and a new floodplain is established a Bc channel forms, then an entrenched C channel, and finally it becomes an entrenched E channel. In the state-and-transition model the E and C channels, the potential natural channels, form state one. State two is comprised of the unstable Gc and F channels. State three is comprised of the Bc and the entrenched C and E channels that have restabilized the system.

Plant communities associated with the riparian ecosystem of the Middle Sheyenne River include: 1) the greenline, 2) woodlands, 3) shrublands, 4) grasslands and 5) wetlands. Plant communities were evaluated using a modified Whittaker plot to determine species composition and cover. The greenline community is a water sedge community that is subject to invasion by reed canarygrass. The woodland communities found within the floodplain are American elm/Sprengel’s sedge, green ash/Sprengel’s sedge, and bur oak/Spengel’s sedge. When changes occur in the canopy of woodlands the understory is vulnerable to invasion by Kentucky bluegrass or smooth bromegrass. Due to Dutch elm disease the American elm/Sprengel’s sedge community has been lost or compromised. The shrublands are dominated by northern hawthorn and Sprengel’s sedge with canopies subject to invasion by buckthorn and understory subject to invasion by Kentucky bluegrass. Prairie communities within the floodplain were warm season graminoid/western snowberry, now dominated by Kentucky bluegrass or smooth bromegrass. Wetland communities are characterized by a native sedge community; however, when compromised are dominated by reed canarygrass.

Introduction:

Ecological site descriptions (ESDs) are reports that characterize a site by documenting the site’s resources (USDA, NRCS 2003; Bestelmeyer and Brown 2010). Resources that are required to be documented within an ESD include physiographic features, climate, water features that are influencing the management of the site, soil features that are representative of the site, ecological dynamics of the site in the form of a state-and-transition model (STM), vegetation dynamics, and supporting information (USDA, NRCS 2010a). According to the USDA, NRCS (2003), an ecological site is “a distinctive kind of land with specific physical characteristics that differs from other kinds of land in its ability to produce a distinctive kind and amount of vegetation.” STMs are developed in conjunction with ecological site descriptions, defining alternative plant communities occurring within a site and possible drivers of transitions between both states and communities (Bestelmeyer et al. 2003; Bestelmeyer et al. 2004; Bestelmeyer et al. 2009; Bestelmeyer and Brown 2010; USDA, NRCS 2010a).

Riparian ecosystems are transitional ecosystems occurring between terrestrial ecosystems, where hydrology has little influence, and aquatic ecosystems where hydrology has a significant impact on ecosystem function and formation (Gregory et al. 1991; Naiman et al. 1993; Svejcar 1997). The stream channel and its associated floodplain that is influenced by the stream’s hydrology, specifically the water table and high flow events, are the primary components comprising a riparian ecosystem (Naiman et al. 1993). In addition to hydrology, riparian ecosystems are also influenced by geomorphology, climate, soils, vegetation, ecological processes, and management (Kovalchik and Chitwood 1990; Gregory et al. 1991; Niaman et al. 1993; Svejcar 1997; Lytle and Poff 2004).

Standard ESDs and STMs fail to adequately describe the complex ecological processes and vegetative dynamics of riparian ecosystems (Leonard et al. 1992). In order for ESDs to adequately describe riparian system’s geomorphology, specifically valley type and stream channel classification (Leonard et al. 1992; Stringham and Repp 2010), and the riparian complex need to be incorporated (Winward 2000; Stringham and Repp 2010). Similar to standard STMs, riparian STMs are used to describe ecosystem dynamics (Stringham and Repp 2010). However, due to the influence of hydrology in riparian ecosystems some adaptations must be made to adequately describe ecological processes and predict responses to disturbance. According to Zweig and Kitchens (2009) hydrology is the major factor driving transitions between states within riparian ecosystems. Riparian STMs need to include channel classification, channel evolution models, description of fluvial landforms, plant community phases comprised of multiple community components, and soil-water-vegetation dynamics taking place within the ecosystem.

The development of ESDs and STMs for riparian ecosystems is a relatively new endeavor, with a meager three drafts being completed by the NRCS for the entire nation that still need to be approved (Repp et al. 2011a; Repp and Boyer 2011; Repp et al. 2011b). Within North Dakota a riparian ESD and associated STM is currently being drafted for Magpie Creek in the western portion of the state (Repp 2010); however, no work has been conducted on perennial systems in the state.

Project Objectives:

The objectives of this project are: 1) identify the natural sustainable plant communities and best management practices of the above watershed through Ecological Site Description development, and 2) provide rangeland technical assistance through media development and consultation with relevant land managers. The ESD and STM will describe channel morphology, community phases, and plant community components associated with each state. In addition a description of environmental and anthropogenic disturbances triggering transitions between states and community phases within the riparian ecosystem of the Middle Shyenne River in eastern North Dakota, aiding in the establishment of realistic goals for restoration and maintenance of natural sustainable communities. Development of riparian range and forestry management recommendations targeted at restoring proper ecosystem function. Development of educational media including brochures, pamphlets, extension manuals, and meetings/ field days/ workshops with ranchers to review grazing recommendations for riparian ecosystems.

Cooperators

Click linked name(s) to expand
  • Edward DeKeyser

Research

Materials and methods:
Study Area

This study was conducted along the Middle Sheyenne River in eastern North Dakota, stretching from near Warwick in northeastern Eddy County to near McVille in southwestern Nelson County. Specifically the study sites were located within the riparian ecosystem associated with the Middle Sheyenne River Watershed.

Climate

The climate of the study area is characterized by marked variations in both temperature and precipitation. Average annual precipitation is approximately 50 cm/year near McHenry, North Dakota, of which over 80% occurs during the growing season April through October (NDAWN 2011). The study location averages 124 frost-free days annually with monthly mean, average daily temperatures ranging from -15°C in January to 20.6°C in July (NDAWN 2011).
Growing season precipitation was 48.8 cm in 2008, exceeding the 30-year average by 8.8 cm (NDAWN 2011). Throughout 2008 all months received above average precipitation except January, March, April, May, and July. The 2009 growing season was characterized as average with a total growing season precipitation of 43.2 cm, or 3.2 cm above the 30-year average. However, nearly half of the growing season months received below average precipitation in 2009. In 2010 growing season precipitation exceeded the 30-year average by 12.3 cm, totaling 52.3 cm from April through October. Precipitation during the study period was above average, particularly in September and October of all years, June of 2008, and May of 2010.
The length of the growing season exceeded the 100-year average throughout the study period. Both the first and last frost free dates were later than the 100-year average in 2008, extending the growing season by fifteen days. However, the first frost free day was earlier than average in 2009 and 2010. The last frost free day was later than average in 2009, whereas the last frost free day was slightly earlier than average in 2010. Despite this, the total frost free period was longer than the long-term average by 26 and 7 days in 2009 and 2010, respectively.
Monthly mean daily average temperatures were near the 30-year average in 2008, with the exception of February and December which had lower than average temperatures. Monthly average temperatures were below the 30-year average in 2009 except in September and November. In 2010 monthly mean daily average temperatures showed little variation from the 30-year average, with the exclusion of March and April which received above average temperatures.

Geology

The Sheyenne River is the longest tributary of the Red River of the North and flows through the Glaciated Plains Region of North Dakota (Severson and Sieg 2006). The landscape of the Glaciated Plains is classified by undulating collapsed topography located within glacial till. The Sheyenne River Valley is a meltwater trench formed by glacial meltwater flowing south from glacial Lake Souris into Lake Agassiz during the Wisconsin glaciation (Bluemle 1973; Severson and Sieg 2006).
The Sheyenne River Valley has depth measuring between 30.5 to 45.7 meters and a width ranging from 0.4 to 1.6 kilometers in Eddy County (Bluemle 1965). The valley’s dimensions are a function of geologic substrate; the valley is deep and constrained in end moraine, and is shallow and wide as it cuts through outwash and ground moraine. In Eddy County the floodplain of the Sheyenne River Valley is composed primarily of very deep alluvial silts and sands with high organic matter content; however, in some areas of the valley there are shale outcrops. There are four historic terraces associated with the Sheyenne River ranging from heights of 12.2 to 30.5 meters. There are a number of low terraces found within the valley developed from recent events, none of which can be traced for a significant distance along the Sheyenne River.
The Sheyenne River Valley is 30.5 meters deep and 1.2 kilometers wide on average as it cuts through Nelson County (Bluemle 1973). The valley’s dimensions are a function of geologic substrate. The valley is deep and constrained in end moraine, and shallow and wide as it cuts through outwash and ground moraine. The floodplain of the Sheyenne River Valley in Nelson County is composed of sand, silt, and clay alluvium overlaying glacial outwash. There are three historic terraces associated with the Sheyenne River in Nelson County ranging from heights of 7.6 to 33.5 meters above the current floodplain. These historic terraces are comprised mainly of a gravel substrate. There are a number of low terraces found within the valley developed from recent events, none of which can be traced for a significant distance along the Sheyenne River.

Vegetation

The woodlands found in association with the riparian ecosystems of the Middle Sheyenne River have canopies comprised of green ash (Fraxinus pennsylvanica), American elm (Ulmus americana), boxelder (Acer negundo), and bur oak (Quercus macrocarpa) (Nelson 1964; Dix and Smeins 1967; Meinke 1991; Stroh 2002). Prior to Dutch elm disease, American elm comprised approximately 25% of riparian forests associated with the Sheyenne River, but currently only account for approximately 2% (Stroh 2002). Green ash and American elm are located on lower terraces, whereas boxelder and oak dominate higher more well developed terraces. The dominant understory species in riparian woodlands is Spengel’s sedge (Carex spengelii) (Nelson 1964; Meinke 1991). Forbs that are common in the understory of riparian woodlands include meadow anemone (Anemone canadensis), catchweed bedstraw (Galium aparine), tall white violet (Viola canadensis), Pennsylvania pellitory (Parietaria pensylvanica), early meadow rue (Thalictrum venulosum), common burdock (Arctium minus), catnip (Nepeta cataria), yellow wood sorrel (Oxalis stricta), and spotted touch-me-not (Impatiens capensis) (Meinke 1991).
Shrub communities are found on the edges of woodlands within the Middle Sheyenne River Valley (Nelson 1964; Meinke 1991). Common shrub species of this ecosystem include northern hawthorn (Crataegus rotundifolia), chokecherry (Prunus virginiana), western snowberry (Symphoricarpos occidentalis) and juneberry (Amelanchier alnifolia).
The composition of prairie communities located within the Middle Sheyenne River Valley is a function of water availability (Dix and Smeins 1967; Meinke 1991). Communities located on fluvial features with a lower water table are dominated by little bluestem (Schizachyrium scoparium), switchgrass (Panicum virgatum), porcupine grass (Heterostipa spartea), prairie dropseed (Sporobolus heterolepis), big bluestem (Andropogon gerardii), and Indian grass (Sorghastrum nutans). In areas with a higher water table the communities are comprised primarily of big bluestem, switchgrass, little bluestem, Indian grass, northern reedgrass (Calamagrostis stricta), prairie cordgrass (Spartina pectinata), baltic rush (Juncus balticus), and fowl bluegrass (Poa palustris). Prairie communities that have been disturbed are subject to invasion by Kentucky bluegrass (Poa pratensis) and increases in the shrub species western snowberry (Symphoricarpos occidentalis) (Meinke 1991). In addition communities in close proximity to cropland are subject to invasion by quackgrass (Agropyron repens), Kentucky bluegrass, and smooth bromegrass (Bromus inermis).
The greenline plant communities adjacent to the Middle Sheyenne River are comprised primarily of water sedge (Carex aquatilis), slough sedge (Carex atherodes), and reed canarygrass (Phalaris arundinacea) (Dix and Smeins 1967).

Sampling Procedures

Valley type delineation
Valley type was delineated from site geology, which was determined from topographic maps, soils maps, and aerial photography using methods described by Rosgen (1996) and within the National Engineering Handbook Part 654 (USDA, NRCS 2007). Valley morphology, basin relief, geologic substrate, and fluvial features were used to determine valley type.

Stream type delineation
Stream type was delineated using Rosgen’s stream classification system (Figure 2.3) (Rosgen 1985, 1994; Harrelson et al. 1994; USDA, NRCS 2007; USDA, NRCS 2010b). Nine potential reference reaches were stratified for collection of cross-section and longitudinal data (Table 2.4). Reaches sampled consisted of a complete meander (Figure 2.2). Within the cross-section elevation of important features including, the water’s edge, bankfull discharge, and flood plain features such as terraces were collected using a real-time kinematic (RTK) device and analyzed with ArcMap. Elevation changes within the stream were collected using a weighted tape, due to channel depth, and were also analyzed with ArcMap. Longitudinal data were collected using the RTK device to determine the elevation of the water surface at the cross-section and one full meander downstream. Cross-sectional data were used to determine channel form, whereas longitudinal data were used to measure the slope of the water surface. The measurements collected in the field from the cross-section were used to calculate the entrenchment ratio, width to depth ratio, and channel sinuosity which determine the level one stream classification. The slope collected from the longitudinal profile is a component of stream classification level two. To complete level two channel material was determined using the Wolman Pebble Count (Wolman 1954) which determined the D50 size of the dominant bed material and classify bed material according to size. Pebble Counts were conducted in pools and riffles near each of the nine cross-sections surveyed.

Vegetation Communities
Vegetation communities associated with the nine potential reference reaches of the Middle Sheyenne River were analyzed. Communities were stratified based on plant community type, location within the floodplain, historic use, and grazing management (grazed versus ungrazed). Five plant community types were evaluated including greenline, woodlands, shrublands, prairies, and wetlands. A total of 41 plant communities were sampled; consisting of 9 greenline, 13 woodlands, 7 grasslands, 4 shrublands, and 8 wetlands. Plant communities were named based on the dominate plant species in the canopy and herbaceous layers.
A multi-scaled Modified-Whittaker plot 20 x 50 m (1,000m2) (Figure 2.4) was established at each study site (Stohlgren et al. 1998). Within the 1,000m2 plot a complete list of all the species present was compiled, plant nomenclature is in accordance with Flora of the Great Plains (The Great Plains Flora Association 1986) and the Plants Database (USDA, NRCS 2011). Systematically placed in each Modified-Whittaker plot were ten 1m2 Daubenmire quadrats. Percent cover was assigned to each species found within the Daubenmire quadrats, along with percent of litter and bareground. Additional modifications were made to the protocol established by Stohlgren et al. (1998) to assess shrub and canopy layers of riparian plant communities. Eight 4m2 (1.1m radius) plots were systematically placed within each Modified-Whittaker plot to observe regeneration and richness of shrub species and seedlings (Avery and Burkhart 2002). A 40m2 (11.3m radius) was placed in the center of the Modified-Whittaker plot at woodland sites to evaluate canopy composition (Avery and Burkhart 2002); the number of dead and harvested trees were also collected with this plot. Diameter at breast height (DBH) of tree species was used to estimate age and cover of species. Within the canopy layer crown class of canopy species was measured using the following classes: dominant, codominant, intermediate, and overtopped. Cover of dominant species was estimated using a Forest Densiometers spherical densitometer concave. In woodland communities the amount of light available for plant growth at the herbaceous level was measured as photosynthetically active radiation (PAR) using an ACCUPAR LP -80 ceptometer manufactured by Decagon Devices, Inc.
A line transect was run in the greenline plant community due to the linear shape and ten 1m2 Daubenmire quadrats were used to determine percent cover of each species, litter, and bareground.
Wetlands were sampled using a quadrat method similar to the one employed by Kantrud and Nelson (1996). Ten 1m2 Daubenmire quadrats were used to determine percent cover of each species, litter, bareground, and open water in the wet meadow and shallow marsh zones of wetlands.
A complete species list was compiled within each 20 x 50 m Modified-Whittaker plot of the woodland, shrub, and prairie communities; for the entire greenline transect, and within each wetland zone (Appendix A). Plant nomenclature is in accordance with Flora of the Great Plains (The Great Plains Flora Association 1986) and the Plants Database (USDA, NRCS 2011).

Herbage production
Herbage production clippings was collected for the herbaceous layer of the plant communities that occurred most frequently within the riparian ecosystem of the Middle Sheyenne River. Six plots were randomly selected within four greenline, three shrub, four prairie, and six woodland communities. A 0.18m2 quadrat was clipped in early August to determine total anuual production (USDA, NRCS 2003; Sedivec et al. 2007). Herbacous species were clipped to ground level, whereas only the current year’s growth was collected from woody species such as shrubs as specified by the USDA, NRCS (2003). Clipped herbage was separated by species, dried at 55°C for 48 hr, weights determined, and plant biomass calculated for each plant community.

Statistical analysis
Hierarchial clustering was used to group similar sites for stream type, herbaceous data for all plant communities, shrub and regeneration data for plant communities in which they were present, and canopy data of woodland plant communities. Cluster analysis was performed in PC-Ord (McCune and Grace 2002) with the flexible beta method using a beta value of -0.25 (Lance and Williams 1967). Two groupings for each parameter clustered were analyzed, 50% and 25% data remaining were used to determine potential groupings.
Nonmetric Multidimensional Scaling (NMS) ordination (McCune and Grace 2002) was used to examine differences between stream classification parameters measured among cross-sections sampled. The NMS was performed in PC-Ord using the Correlation distance measure and the slow and thorough autopilot option. Final solutions were selected for low final stress (based on Clarke’s rules of thumb) and final instability (less than 0.00001) (McCune and Grace 2002). Species possessing a Pearson correlation coefficient (r) of 0.40 (absolute value) or greater were selected for discussion.
Nonmetric Multidimensional Scaling (NMS) ordination (McCune and Grace 2002) was used to examine differences in community composition among vegetation data subsets. The NMS analysis was performed in PC-Ord, using the Sorensen (Bray-Curtis) distance measure and the slow and thorough autopilot option. Final solutions were selected for low final stress (based on Clarke’s rules of thumb) and final instability (less than 0.00001) (McCune and Grace 2002). Species possessing a Pearson correlation coefficient (r) of 0.40 (absolute value) or greater were selected for discussion.
Further statistical analyses were conducted using the Multi-response Permutation Procedure (MRPP) (Mielke and Berry 2001). The MRPP analyses were conducted in PC-Ord, using the Correlation (for stream classification data) and Sorenson (for vegetation data) distance measures at a significance level of P < 0.05. MRPP was used to determine if groupings were significant. Groupings tested included stream type, grazing, and the potential groupings from the hierarchical clustering.
Regression analysis was conducted to determine if axes were correlated with environmental factors in woodland communities; specifically, between canopy and herbaceous cover and canopy density and PAR.

Research results and discussion:
Valley Type

The Middle Sheyenne River Valley is a wide valley with an elevation of 130.2 m in the upper portion of the reach and 125.6 m in the lower portion, resulting in an elevation change of 4.6 m and an elevation relief of 3.5 percent. The geology of the valley is characterized by historic terraces formed by glacial activity and a number of low terraces formed more recently within glacial till. These features lead to the classification of the Middle Sheyenne River Valley as Valley Type VIII, which is described as a broad valley with gentle down-valley relief that has a well developed floodplain and numerous glacial and alluvial terraces. Valley Type VIII is able to support stream types C, D, E, F, and G.

The Middle Sheyenne River Valley is classified as Valley Type VIII. According to Rosgen (1996), Valley Type VIII is a glacial valley type with gentle relief down-valley that had a well developed floodplain and multiple glacial and alluvial terraces. The geology of Valley Type VIII is characterized by historic terraces formed by glacial activity and a number of low terraces formed more recently within glacial till. The Middle Sheyenne River Valley lies within a glacial trough, as does the valley associated with the Uncompahgre River in Colorado, which was also classified as Valley Type VIII (Repp et al. 2011). Similarly, the valleys of Magpie Creek in western North Dakota (Repp 2010) and of the Arikaree and Republican Rivers in Nebraska (Repp and Boyer 2011) were also glacial valley types classified as Valley Type VIII. The relief of the Middle Sheyenne River Valley was calculated to be 3.5%; however, the relief of the entire Sheyenne River Valley was estimated to be 1.6% (USACE 2001).

Stream Type

Classification of the nine cross-sections sampled resulted in two cross-sections (3 and 4) being classified as stream E and seven cross-sections (1, 2, 5, 6, 7, 8, and 9) being classified as stream type C (Table 2.4). E channels are slightly entrenched, having an entrenchment ratio of 2.2 or greater, are narrow and deep with a width to depth ratio that is less than 12, and have high sinuosity (greater than 1.5). C channels are also considered to be slightly entrenched; however, these channels are wider and shallower with a width to depth ratio that is greater than 12 and are highly to moderately sinuous, having a sinuosity of 1.2 or greater. All the channels sampled have a normal slope between 0.001 and 0.008; normal slopes are less than 0.02 and between 0.001 and 0.02 for E and C channels, respectively. The channel material at cross-sections 1, 5, 7, 8, and 9 was dominated by silt/clay particles, which have a D50 score of less than 0.062. The remaining cross-sections (2, 3, 4, and 6) had bed materials that were comprised mainly of sands, which have a D50 score ranging between 0.062 and 2.

The NMS ordination of the stream classification parameters produced a three dimensional final solution with a stress of 0.00982, indicating an excellent representation with no prospect of misinterpretation. The final instability was 0 with 43 iterations. The three axes of the final solution accounted for 96.3% of the variation within the dataset, with axes 1, 2, and 3 accounting for 30.7%, 57.6%, and 8.1%, respectively.

Bankfull width and width to depth ratio were positively correlated with axis one, whereas entrenchment ratio was negatively correlated with axis one. Stream classification was not influenced (r < ±0.4) by bankfull depth, maximum depth of the channel, width of the flood prone area, channel material, channel slope, and sinuosity. Axis two was strongly influenced by channel material and width of the floodplain. Axis three was negatively driven by channel slope, and positively driven by bankfull depth.

MRPP analysis indicated that stream channel parameters were not influenced (P > 0.05) by stream channel type or grazing. Analysis of the potential groupings from the hierarchical clustering indicated that the 50% grouping was significant (P ? 0.05); whereas the 25% grouping was not significant (P > 0.05). The 50% cluster was comprised of three groups; group one was comprised of cross-sections 1 and 6, group two was comprised of cross-sections 2 and 3, and group three was comprised of cross-sections 4, 5, 7, 8, and 9. Axes one and two have the greatest influence on the grouping of cross-sections. The cross-sections in group one are positively associated with axis one, but do not appear to be influenced by axis two, indicating that bankfull depth and the width to depth ratio of the channel are influencing channel classification at these cross-sections. This grouping is supported by the fact that cross-sections 1 and 6 are wide having bankfull widths that are greater than 40 m and having depths of 1.44 and 1.97 m, respectively. The cross-sections that make up group two are characterized as having sand as the dominant bed material and having moderate entrenchment ratios. Cross-sections in this group are negatively associated with both axis one and two. Group three is negatively associated with axes one and two. The cross-sections in this group are characterized by entrenchment ratios that are greater than 40 and broad floodplains.

Classification of the nine cross-sections sampled resulted in two cross-sections being classified as stream type E and seven cross-sections being classified as stream type C. In 1998 sampling of 22 cross-sections along the Sheyenne River identified 10 C channels, 8 Bc channels, 2 E channels, 1 G channel, and 1 F channel using Rosgen’s classification system (USACE 2001). Riparian ESDs being developed for other riparian systems characterized by Valley Type VIII have reference channels classified as E or C channels (Repp 2010; Repp and Boyer 2011; Repp et al. 2011). C channels evaluated within the Middle Sheyenne River had width to depth ratios ranging from 15.4 to 28.1 with an average of 19.7, whereas C channels sampled along the entire Sheyenne River had width to depth ratios ranging from 12.2 to 74.7 with an average of 27.8. E channels evaluated within the Middle Sheyenne River had width to depth ratios of 8.3 and 11.6, whereas E channels sampled along the entire Sheyenne River had width to depth ratios of 8.3 and 9.8. All the channels sampled had slopes measuring between 0.001 and 0.008; normal slopes are less than 0.02 and between 0.001 and 0.02 for E and C channels, respectively. The USACE (2001) reported multiple channels with low gradient slopes, slopes of less than 0.001, which are denoted by c. The channel material at four cross-sections was dominated by silt/clay particles, which have a D50 score of less than 0.062 and are denoted by a 6. The remaining five cross-sections had bed materials that were comprised mainly of sands, which have a D50 score ranging between 0.062 and 2 and are denoted by a 5. This is similar to the USCOE’s findings of silt/clay channel materials for eleven cross-sections, sands for ten cross-sections, and fine gravel for one cross-section (USACE 2001).

Evaluation of the different components of Rosgen’s classification of natural rivers (Rosgen 1985; 1994) revealed that the factor separating E and C channels within the Middle Sheyenne River was the width to depth ratio, thus the transition between these stream types is a function of changes to the width to depth ratio. This is supported by evaluation of other stream systems containing both E and C channels (Rosgen 2006; Repp 2010; Repp and Boyer 2011; Repp et al. 2011a).

Stream Evolution Model

The stream evolution models that could potentially apply to the Middle Sheyenne River include models one and five from Rosgen’s classification of natural rivers (Rosgen 2006) (Figure 2.3). These are the evolution models that would be applicable to streams having Valley Type VIII and include both E and C stream types. While evolution model ten possesses both E and C stream types, this model also includes stream type A. Valley Type VIII is unable to support stream type A. Due to the identification of Bc, G, and F channels in addition to E and C channels by the USACE (2001), stream evolution model one with the addition of a Bc channel is being carried out along the Sheyenne River. The stream evolution model is as follows: E ? C ? Gc ? F ? Bc ? C ? E. The E channel’s transition to a C channel is characterized by an increase in the channel’s width to depth ratio; however, both stream types are able to access the associated floodplain in high flow events (Rosgen 1996, 2006). When the C channel becomes unstable through increased erosion and decreased bank stability, down cutting occurs resulting in the formation of a G channel. The G channel has lost connectivity with the floodplain, resulting in the loss of riparian vegetation and increased bank erosion, as stream banks slump the channel widens forming a F channel. As sediment loads diminish and the stream begins to stabilize a Bc channel forms within the F channel; however, a new floodplain has not yet been developed. As stability increases within the system, a C channel that has a new constrained floodplain is formed within the bed of the F channel. As riparian vegetation colonizes the new floodplain the width to depth ratio of the stream decreases and the system becomes more stable, an entrenched E channel is formed. Similar evolution models were found to be carried out along Magpie Creek in western North Dakota (Repp 2010) and the Arikaree and Republican Rivers in Nebraska (Repp and Boyer 2011) with the exclusion of the initial C channel and the Bc channel, respectively.
The hydrologic processes of riparian ecosystems, which are closely linked to stream type, are responsible for the development of the plant communities within riparian ecosystems (Stringham and Repp 2010). The increase in the channel’s width to depth ratio associated with the transition from an E to C channel results in a reduction in floodplain size and a subsequent decline in the water table and soil moisture, altering riparian plant communities (Leonard et al. 1992). Whereas, a reduction in the width to depth ratio has the opposite effect and enhances riparian vegetation. Channel entrenchment associated with G and F channels results in the stream losing connectivity with the floodplain (Rosgen 1985; 1994) and a subsequent loss of the greenline community component (Leonard et al. 1992; Winward 2000).

Vegetation Communities

The geomorphic features of a riparian ecosystem, specifically valley and stream type in combination with the complex vegetation, make up a riparian complex (Winward 2000; Stringham and Repp 2010). The multiple plant communities that occur simultaneously within a riparian ecosystem are plant community components (Stringham and Repp 2010). Plant community components within riparian ecosystems are structural components of the ecosystem and are not indicative of successional processes taking place.
Five primary communities or plant community components were found to be associated with riparian ecosystems of the Middle Sheyenne River: 1) the greenline, which is the line of vegetation on or at the water’s edge, 2) woodlands, 3) shrublands, 4) grasslands and 5) wetlands. The USCOE (1993) identified four primary community types within the Sheyenne River Valley: 1) woodlands, 2) shrublands, 3) grasslands, and 4) wetlands.

Greenline
NMS analysis of the greenline composition resulted in a one dimensional final solution with a stress of 7.13531, indicating a good ordination with little risk of drawing false inferences. The final solution was stable with an instability of 0 for 46 iterations. The final solution accounted for 90.7% of the variation within the dataset.
Axis one was positively correlated with native riparian vegetation with water sedge (Carex aquatilis) exerting the strongest influence. Axis one was negatively driven by invasive species, specifically reed canarygrass (Phalaris arundinacea).
MRPP analysis indicated that greenline composition was not influenced ( P > 0.05) by grazing. However, stream channel type and both the 50 and 25% groupings from the hierarchical clustering significantly influenced (P ? 0.05) greenline composition. The 50% grouping was comprised of three groups: group one included communities associated with cross-sections 7, 8, and 9; group two included communities associated with cross-sections 1, 2, 5, and 6; and group three included communities associated with cross-sections 3 and 4. Further clustering resulted in the 25% grouping being made up of two groups, in which group one was comprised of groups 1 and 2 from the 50% grouping and group 3 became group 2. The 25% grouping matched the stream channel classification grouping with group 1 being made up of cross-sections having a C stream type and group 2 made up those with E stream types. The greenline plant communities associated with the E stream types were correlated with the negative reed canarygrass end of the axis, whereas C stream types were correlated with the positive water sedge end of the axis.
The results of the NMS analysis and MRPP indicates that there are two primary greenline community components associated with the riparian ecosystems of the Middle Sheyenne River, a native community dominated by water sedge and an invaded community dominated by reed canarygrass.
Greenline communities that were dominated by water sedge had a total annual production of approximately 8,000 kg/ha. Water sedge (Carex aquatilis) accounted for 63% of this community’s production. Other native graminoid species included prairie cordgrass (Spartina pectinata), three square (Schoenoplectus pungens), and fowl bluegrass (Poa palustris), which comprised 25%, 8%, and 1% of the community’s production, respectively. Traces of the introduced graminoid species reed canarygrass also made up a minute portion of the community’s production. Native forb species accounted for 2% of production, specifically swamp milkweed (Asclepias incarnate), smooth scouring rush (Equisetum laevigatum), and field mint (Mentha arvensis). Traces of invasive forb species Canada thistle and leafy spurge were also present. The native shrub species Woods’ rose (Rosa woodsii) composed 1% of the community’s production.
Greenline communities that were dominated by the introduced species reed canarygrass has a similar total annual production as greenline communities dominated by water sedge at approximately 8,000 kg/ha. Reed canarygrass made up the bulk of the sites’ production at 88%. Water sedge was still present in some of these communities and accounted for up to 10% of the total annual production. Native forb species, stinging nettle (Urtica dioica) and field mint, accounted for about 2% of the community’s production.
Two primary greenline communities were identified within the riparian ecosystems of the Middle Sheyenne River; a native water sedge community and an invaded reed canarygrass community. This is supported by Dix and Smeins (1967) who reported that the greenline plant communites adjacent to the Middle Sheyenne River were comprised primarily of water sedge, slough sedge, and reed canarygrass. Greenline communities that were dominated by water sedge had a total annual production of approximately 8,000 kg/ha. Water sedge accounted for 63% of this community’s production while other native graminoid species included prairie cordgrass, three square, and fowl bluegrass. Repp and Boyer (2011) reported greenline communities associated with the Arikaree and Republican Rivers were dominated by graminoid species, which accounted for 80% of the site’s nearly 9,000 kg/ha total annual production. The dominant species included prairie cordgrass, three square, and water sedge (Repp and Boyer 2011). Native forbs and shrubs were both reported to make up about 10% of the sites production of the Arikaree and Republican River’s greenline, whereas forbs and shrubs were only found to comprise 2% and 1% of the Middle Sheyenne River’s greenline production, respectively.

Woodland and Shrublands
NMS analysis of the herbaceous layer of woodland and shrublands produced a final solution with two dimensions and a final stress of 8.12549, which indicates a good ordination with little risk of drawing false inferences. This solution was stable with 44 iterations resulting in an instability of 0. Together, axes one and two accounted for 91.5% of the variation within the dataset. Axis one accounted for the majority of the variation (70.9%), whereas axis two only accounted for 20.7% of the variation observed in the understory of woodland and shrubland communities (Figure 2.6).
The species positively correlated with axis one of the herbaceous component of woodland and shrub communities were native species commonly associated with riparian woodlands with Sprengel’s sedge (Carex sprengelii) exerting the most influence. Whereas, the species negatively correlated with axis one were introduced species found to occur in degraded ecosystems, of which smooth bromegrass had the greatest influence. Axis two had species positively correlated by plant species associated with prairie ecosystems and negatively correlated by shrub and forb species associated with woodland and shrubland communities.
MRPP analysis indicated that the herbaceous layer of woodland and shrubland communities was not influenced (P > 0.05) by stream channel type or grazing. Analysis of the potential groupings from the hierarchical clustering indicated that both the 50 and 25% groupings were significant (P ? 0.05). The 50% grouping was comprised of three groups; group 1 was comprised of six woodland and three shrubland communities that were positively correlated with axis 2, group 2 was comprised of two woodland and one shrubland communities that were negatively correlated with axis one, and group 3 was comprised of five woodland communities that were positively correlated with both axis one and two. The 25% grouping included two groups, group 1 combined the invaded understory communities associated with groups 1 and 2 of the 50% grouping, and the native understory communities associated with group 3 made up the second group.
NMS analysis of the shrubs and woody regeneration within woodland and shrub communities resulted in a one dimensional final solution with a final stress of 0.00014, indicating no chance of misinterpreting the ordination. The final solution had an instability of 0 for 33 iterations. The final solution only accounted for 13.5% of the variation observed within the dataset.
Axis one was negatively influenced by shrub and tree species commonly found within woodland and shrub communities in the region including: chokecherry, northern hawthorn, American elm, and western snowberry. No species were observed to have a positive influence (P ? 0.05) on axis one.
MRPP analysis indicated that the shrub layer of woodland and shrubland communities was not influenced (P > 0.05) by stream channel type or grazing. Analysis of the potential groupings from the hierarchical clustering indicated that both the 50 and 25% groupings were significant (P ? 0.05). The 50% group was comprised of four groups, group 1 included two woodland communities that were more negatively correlated with axis one, group 2 included five woodland communities that were moderately negatively correlated with axis one, group 3 included three woodland and one shrubland communities that were moderately negatively correlated with axis one, and group 4 was comprised of two woodland and three shrubland communities that were moderately negatively associated with axis one. The 25% grouping was comprised of three groups, group 1 remained the same as the 50% grouping, group 2 combined groups 2 and 3 from the 50% grouping, and group 4 of the 50% grouping became group 3 in the 25% grouping.
NMS ordination of the historic canopy associated with woodland communities comprising the riparian woodlands of the Middle Sheyenne River did not produce a useful ordination indicating that the canopy composition of woodlands had no structure and the composition was similar to a random arrangement. However, NMS analysis of the current canopy composition of the riparian woodlands of the Middle Sheyenne River resulted in a three dimensional final solution with a final stress of 1.41414, leaving no prospect of misinterpretation of the ordination. Final instability was 0.00043 with 200 iterations. Cumulatively the three axes accounted for 78.9% of the variation observed in the dataset, individually axis one, two, and three accounted for 29.3%, 28.5%, and 21.1% of the variation, respectively.
Green ash (Fraxinus pennsylvanica) was positively correlated with axis one, whereas bur oak (Quercus macrocarpa) was negatively correlated with axis one. The opposite trend was observed for axis two, with green ash negatively influencing the axis and bur oak positively influencing the axis. Axis three was positively influenced by green ash and bur oak, and negatively influenced by American elm (Ulmus americana).
MRPP indicated that neither stream channel classification, nor grazing significantly influenced (P > 0.05) canopy composition of riparian woodlands. In addition neither of the groupings from the hierarchical clustering influenced (P > 0.05) canopy composition.
Regression analysis of the relationship between density of the canopy and axes of the herbaceous and canopy ordinations showed that canopy density did not influence (P > 0.05) either herbaceous or canopy composition, with the exception of axis two of canopy composition. Similar results were found for the relationship between PAR and the axes of the herbaceous and canopy ordinations, with only axis two of the canopy composition influencing (P ? 0.05) the amount of PAR available for plant growth in woodland communities. The communities that are positively associated with axis two of the canopy ordination are characterized by woodlands that are diverse and have a greater canopy density that restricts the amount of PAR in the understory, whereas those negatively associated with axis two are characterized by numerous dead trees and low density open canopy that allow more PAR into the understory.
NMS ordination of the herbaceous, shrub and canopy layers of woodland shrub communities within the riparian ecosystem of the Middle Sheyenne River resulted in the identification of three woodland and two shrubland communities. Woodlands communities include an American elm/Sprengel’s sedge community in which the elms have died due to Dutch elm disease and the understory is now dominated by either smooth bromegrass or Kentucky bluegrass. The most common woodland community component is a bur oak/Sprengel’s sedge community, when changes in canopy density occur within this community the understory becomes invaded with Kentucky bluegrass. The final woodland community is a green ash/Sprengel sedge community component in which the understory is now dominated by reed canarygrass. Shrubland community components include a native community of northern hawthorn/Sprengel’s sedge and an invaded community of buckthorn/Kentucky bluegrass.
What historically were American elm/Sprengel’s sedge communities have been compromised due to the loss of canopy and subsequent invasion of the understory by either smooth bromegrass or Kentucky bluegrass. The understory of the American elm/smooth bromegrass community had a total annual production of approximately 4,000 kg/ha. Smooth bromegrass dominates the understory production at 55%. The remainder of the graminoid production is composed of the introduced species Kentucky bluegrass and quackgrass, which accounting for 28% and 2%, respectively. The native forb species heath aster (Symphyotrichum ericoided) and early meadow rue (Thalictrum venulosum) comprised around 2% of the community’s production. Twelve percent of the production consisted of the introduced species Canada thistle, common milkweed (Asclepias syriaca), and common dandelion (Taraxacum officinale). Traces of the native shrub species red raspberry and western snowberry also accounted for a minute portion of the community’s annual production.
The American elm/Kentucky bluegrass community had a total annual production of 2,300 kg/ha, of which Kentucky bluegrass comprised 70% of the production. Eight percent of the original dominant understory species Sprengel’s sedge remained a component of the community’s composition. The introduced graminoid species quackgrass and smooth bromegrass accounted for 20% of the community’s production. The introduced forb species absinth wormwood (Artemisia absinthium) and common dandelion made up approximately 1% of the community’s production. Traces of the native forb Canadian white violet (Viola canadensis) comprised a portion of the community’s production.
The understory production of the bur oak/Sprengel’s sedge woodland community had a total annual production of 1,400 kg/ha. Eighty-eight percent of the understory production consisted of Sprengel’s sedge, which accounted for nearly all of the graminoid production with the exception of traces of Virginia wildrye (Elymus virginicus) and Kentucky bluegrass. Native forbs accounted for 11% of the community’s production, including hog peanut (Amphicarpaea bracteata), common burdock (Arctium minus), catchweed bedstraw (Galium aparine), northern bedstraw (Galium boreale), wood nettle (Laportea canadensis), early meadow rue, stinging nettle, Canadian white violet. The introduced forbs Canada thistle and common dandelion comprised less than 1% of the total annual understory production. Seedlings of the native woody species beaked hazelnut (Corylus cornuta), northern hawthorn, boxelder, green ash, and bur oak accounted for less than 1% of the community’s production, as did the introduced shrub species buckthorn (Rhamnus cathartica).
The understory production of the green ash/reed canarygrass plant community had a total annual production of 3,407 kg/ha, of which reed canarygrass accounted for 59%. The native graminoid species fowl bluegrass and owlfruit sedge (Carex stipata) accounted for 4%, and introduced graminoid species smooth bromegrass, quackgrass, and Kentucky bluegrass accounted for an additional 33% of the community’s production. Native forbs comprised 1% of the production, including meadow anemone, stinging nettle, and yellow woodsorrel (Oxalis stricta). The introduced forbs Canada thistle and common dandelion accounted for 1% of the community’s production. Native woody species comprised approximately 2% of the production; western snowberry accounted for most of the woody production and traces of boxelder and green ash accounted for a portion of the woody production.
The northern hawthorn/Sprengel’s sedge community had a total annual production of 900 kg/ha. The dominant understory species was Sprengel’s sedge, which accounted for 73% of the community’s production. Native forb species comprised approximately 5% of the community’s production, including catchweed bedstraw, northern bedstraw, common hops (Humulus lupulus), early meadow rue, and Canadian white violet. Native woody species accounted for 17% of the community’s production with western snowberry comprising over 15% and Virginia creeper (Parthenocissus quinquefolia), woods rose, boxelder, and green ash comprising the rest. The introduced shrub species buckthorn accounted approximately 5% of the community’s production.
The buckthorn/Sprengel’s sedge community had a total annual production of 800 kg/ha. Spengel’s sedge is the dominant understory species comprising 71% of the community’s production. However, unlike the northern hawthorn/Sprengel’s sedge community the graminoid species Kentucky bluegrass and smooth bromegrass made up 6% and 1% of the production, respectively. Native forb species accounted for 12% of the community’s production, 10% of which consisted of Pennsylvania pellitory (Parietaria pensylvanica), and the remaining 2% was comprised of common burdock, catchweed bedstraw, northern bedstraw, common hops, Virginia lionsheart (Physostegia virginiana), false lily of the valley (Smilacina stellata), early meadow rue, stinging nettle, and Canadian white violet. Introduced forbs Canada thistle, dame’s rocket (Hesperis matronalis), and common dandelion made up 1% of the community’s production. The native woody species northern hawthorn, chokecherry (Prunus virginiana), wild black current (Rhus trilobata), boxelder, green ash, and American elm comprised 2% of the community’s production. The introduced woody species buckthorn accounted for 7% of the buckthorn/Sprengel’s sedge community’s understory production.
Three woodland communities were identified within the riparian ecosystem of the Middle Sheyenne River: 1) American elm/Sprengel’s sedge, 2) bur oak/Sprengel’s sedge, and 3) green ash/Sprengel’s sedge. Numerous other studies within the riparian ecosystem of the Sheyenne River identified American elm, green ash, and bur oak made up the canopy of riparian woodlands (Nelson 1964; Dix and Smeins 1967; Meinke 1991; Stroh 2002), and Sprengel’s sedge as the primary understory species (Nelson 1964; Meinke 1991). Green ash and American elm were reported to be associated with lower terraces (Stroh 2002). The green ash/Sprengel sedge community component now has an understory dominated by reed canarygrass. Stroh (2002) identified a similar green ash dominated community along the Lower Sheyenne River, except that the dominant understory species was hackberry (Celtis occidentalis L.). While graminoid speices accounted for 96% of the total annual production of the understory, it comprised only about 40% of the understory of the green ash/hackberry community described by Stroh (2002). The American elm/Sprengel’s sedge community has been compromised by Dutch elm disease and the understory is now dominated by either smooth bromegrass or Kentucky bluegrass. According to Stroh (2002), American elm have suffered extreme losses within riparian ecosystems from comprising 25% to 2% of the canopy due to Dutch elm disease, and communities with canopies dominated by American elm are no longer present in the Lower Sheyenne River Valley. Oak have been shown to dominate higher, more well developed terraces within the floodplain of the Sheyenne River (Stroh 2002). The most common woodland community component is a bur oak/Sprengel’s sedge community, changes in canopy density within this community results in the invasion of the understory by Kentucky bluegrass. A similar community was described by Stroh (2002) in association with the Lower Sheyenne River with an understory dominated by western snowberry instead of Sprengel’s sedge. Where we found graminoids to comprise nearly 90% of the herbaceous layer, Stroh (2002) found graminoids accounted for only 49% of the herbaceous layer.
Two shrubland communities were identified within the riparian ecosystem of the Middle Sheyenne River. Shrubland community components include a native community of northern hawthorn/Sprengel’s sedge and an invaded community of buckthorn/Kentucky bluegrass. According to other research conducted within the Sheyenne River Valley shrublands are located on the outskirts of woodlands and are comprised of the native shrub species: northern hawthorn, chokecherry, western snowberry, and Juneberry (Nelson 1964; Meinke 1991).

Prairie
NMS analysis of the herbaceous layer of prairie communities resulted in a two dimensional final solution with a final stress of 0.07289, indicating an excellent representation with no prospect of misinterpretation. Final instability was 0 for 55 iterations. The two axes accounted for 63.7% of the variation within the dataset: axis one accounted for 36.2% of the variation and axis two accounted for 27.6%.
The species positively correlated with axis one are native perennial grassland species, with the exception of Kentucky bluegrass. Species negatively correlated with axis one are the invasive species Canada thistle, woodland species, and plants species associated with disturbance such as smooth bromegrass and showy milkweed. Axis two was positively influenced by plant species associated with mesic prairie plant communities (big bluestem and meadow sweet) and negatively influenced by plant species associated with drier prairie plant communities (Kentucky bluegrass and green needlegrass).
MRPP analysis of the herbaceous composition of prairie communities could not be performed for the stream classification grouping, as all prairie communities sampled were associated with C stream types. MRPP analysis indicated that grazing did not influence (P > 0.05) herbaceous composition of prairie plant communities. Analysis of the potential groupings from the hierarchical clustering indicated that both the 50 and 25% groupings were significant (P ? 0.05). The 50% grouping consisted of three groups, group 1 included two plant communities that were positively associated with axis one and negatively associated with axis two, group 2 included two plant communities that were positively associated with both axes, and group 3 included three plant communities that were negatively influenced by axis one. The 25% grouping was comprised of two groups, group 1 included group 1 and 2 of the 50% grouping, and group 2 was made up of the same communities of group 3 in the 50% grouping.
NMS analysis of the shrub layer of the prairie communities resulted in a one dimensional solution with a final stress of 0.00212, indicating no risk of drawing false inferences. The ordination was stable over 35 iterations with an instability of 0. The final solution accounted for 97% of the variation within the dataset.
The ordination was positively correlated with shrub species requiring moist habitats, such as willows. The ordination was negatively influenced primarily by western snowberry.
MRPP indicated that neither stream channel classification, nor grazing influenced (P > 0.05) shrub composition of prairie communities. The hierarchical clustering resulted in the same grouping for the 50% and 25% levels, which significantly influenced shrub species associated with the prairie communities (P ? 0.05). The four communities that comprised group 1 were negatively correlated with axis one, whereas the three communities included in group 2 were positively correlated with axis one.
NMS ordination of the herbaceous and shrub components of the prairie communities resulted in the identification of three prairie community components. The primary community is a warm season graminoid/western snowberry community, which is subject to invasion creating either a Kentucky bluegrass/western snowberry community or a smooth bromegrass/western snowberry community.
The warm season graminoid/western snowberry community is the historic prairie plant community associated with the riparian ecosystems of the Middle Sheyenne River. This community has been compromised and the closest remaining community is a Kentucky bluegrass/warm season graminoid/western snowberry community. This compromised community is dominated by Kentucky bluegrass, which accounts for 53% of the community’s total annual production of approximately 3,500 kg/ha. Warm season grasses no longer dominate this community, but do account for 4% of the community’s production. The warm season grasses that still remained in the community were big bluestem and prairie cordgrass. In addition to these native warm season grasses, the native cool season grasses porcupine grass, western wheatgrass (Pascopyrum smithii), and green needlegrass (Nassella viridula) made up 13% of the production. The introduced graminoid species quackgrass accounted for 5% of the community’s production. The native forb species western yarrow (Achillea millefolium), western ragweed (Ambrosia psiolstachya), heath aster, harebell (Campanula rotundifolia), northern bedstraw, torch flower (Geum triflorum), Canada goldenrod (Solidago canadensis), prairie goldenrod (Solidago missouriensis), stiff goldenrod (Oligoneuron rigidum), American vetch (Vicia americana), and prairie violet (Viola pedatifida) comprised 8% of the community’s production. The introduced forb species Canada thistle accounted for 1% of the community’s production. Native woody species comprised 16% of the community’s production, 15% of which was made up of western snowberry and the remaining 1% consisted of meadow-sweet (Spiraea alba), boxelder, and green ash.
The Kentucky bluegrass/western snowberry community was dominated by Kentucky bluegrass and western snowberry which accounted for 42% and 31% of the community’s 3,500 kg/ha total annual production, respectively. The introduced graminoid species quackgrass and smooth brome grass comprised 21% of the community’s production. Forb species accounted for approximately 6% of the community’s production, most of which consisted of the introduced species common milkweed (Asclepias syriaca), Canada thistle, and common dandelion and a trace amount which was made up of the native forb ovalleaf milkweed (Asclepias ovalifolia).
The smooth bromegrass/western snowberry prairie community had a total annual production of approximately 4,000 kg/ha. Seventy-five percent of the community’s production consisted of smooth bromegrass. The introduced graminoid species Kentucky bluegrass accounted for 17% of the community’s production. The invasive forb leafy spurge made up 4% of the community’s production. The native shrub western snowberry accounted for 4% of the community’s production.
Three prairie communities were found in association with the riparian ecosystems of the Middle Sheyenne River including: 1) warm season graminoid/western snowberry community, 2) Kentucky bluegrass/western snowberry community, and 3) smooth bromegrass/western snowberry community. The primary community was a warm season graminoid/western snowberry community, which is subject to invasion creating either a Kentucky bluegrass/western snowberry community or a smooth bromegrass/western snowberry community. Prior studies conducted on the composition of the prairie communities within the Middle Sheyenne River Valley reported communities dominated by the native warm season grasses big bluestem, switchgrass, little bluestem, prairie cordgrass, and Indian grass (Dix and Smeins 1967; Meinke 1991). However, according to Meinke (1991), prairie communities that have been disturbed are subject to invasion of the graminoid species Kentucky bluegrass and increases in the shrub species western snowberry. The extent of Kentucky bluegrass/western snowberry communities have been increasing, especially in areas that are overgrazed. In addition prairie communities in close proximity to cropland are subject to invasion by quackgrass, Kentucky bluegrass, and smooth bromegrass (Meinke 1991). The floodplain of the Arikaree and Republican Rivers were found to be dominated by the warm season grasses switchgrass, Indian grass, and big bluestem, which comprised approximately 25%, 20%, and 10% of the total annual production, respectively (Repp and Boyer 2011). The warm season graminoid/western snowberry community has been compromised and is now dominated by Kentucky bluegrass, which accounts for 53% of the community’s total annual production, and warm season grasses only account for 16% of the community’s production currently. The warm season grasses that still remained in the community were big bluestem and prairie cordgrass.

Wetlands
NMS analysis of the wet meadow zone of wetland communities associated with the riparian ecosystems of the Middle Sheyenne River resulted in a one dimensional final solution with a final stress of 10.81170, indicating a good ordination little risk of drawing false inferences. Final instability was 0 for 43 iterations. The final solution accounted for 84.5% of the variation within the dataset.
Reed canarygrass had the greatest positive influence on axis one, along with species associated with disturbed communities. The introduced species quackgrass had the greatest negative influence on axis one; however, the other species that significantly influenced axis one were native wetland species.
MRPP could not be performed to assess if stream channel classification influenced the composition of the wet meadow zone of wetlands, as there were no wetlands found in association with E stream channel types. MRPP analysis of the wet meadow plant community indicated that grazing did not impact (P > 0.05) the species composition. Analysis of the potential groupings from the hierarchical clustering indicated that both the 50 and 25% groupings were significant (P ? 0.05). The 50% grouping included three groups, group 1 was comprised of three oxbow wetlands that were negatively correlated with axis one, group 2 was comprised of two oxbow wetlands that were neither positively or negatively influenced by axis one, and group 3 was comprised of two seasonal and one oxbow wetland that were positively correlated with axis one. The 25% grouping was comprised of two groups, group 1 was the same as in the 50% grouping and group 2 was a combination of groups 2 and 3 from the 50% grouping.
NMS ordination of the shallow marsh zone of wetland communities associated with the riparian ecosystems of the Middle Sheyenne River did not produce a useful ordination, indicating that the composition of the shallow marsh zone of wetlands had no structure and would be considered a random arrangement.
Overall the results of the NMS ordination of wetland plant communities indicated that there were two plant communities; one that was dominated by the invasive species reed canarygrass and the other comprised of quackgrass and native sedges.
Two wetland plant communities were identified in association with the riparian ecosystem of the Middle Sheyenne River; one in which the wet meadow zone is dominated by the invasive species reed canarygrass and the other is comprised of quackgrass, prairie cordgrass, Baltic rush, and native sedges. Meinke (1991) reported that the wet meadow zones of wetlands in Eddy county were dominated by wooly sedge, clustered field sedge, three square, Baltic rush, prairie cordgrass, and spike rush. While, reed canarygrass was not reported to be present in the wet meadow zones of wetlands, it was found to be dominant species within the shallow marsh zone (Meinke 1991).

Participation Summary

Educational & Outreach Activities

Participation Summary:

Education/outreach description:

A series of five extension fact sheets on riparian ecosystem and grazing management recommendation within riparian ecosystems were developed. A journal article on the nutritional value of Carex sprengelii, a dominate forage within riparian woodlands has been submitted to Rangeland Ecology and Management. The doctoral dissertation “Riparian Ecosystems of the Middle Sheyenne River in Eastern North Dakota,” is currently being reviewed. A detailed ecological site description and additional journal articles still need to be completed.

Project Outcomes

Project outcomes:

Evaluation of the riparian ecosystems of the Middle Sheyenne River and other riparian ecosystems within the Great Plains Region lead to the development of the following state-and-transition model for the Middle Sheyenne River.
The building blocks of the state-and-transition model are the stream types found along the Sheyenne River and the stream evolution model that describes the potential stream channels of the Sheyenne River system. The large boxes depicted in the state-and-transition model represent the different states, and the smaller boxes within the larger boxes represent the various community phases.
State one is the potential natural channels, the E5/6 channel is the reference channel but can become a C5/6 channel as a result of stream widening (Rosgen 1996, 2006; Repp 2010; Repp and Boyer 2011). However, the E channel can be recovered if bank stability is increased. The channels in state one are relatively stable when the appropriate native vegetation is maintained.
State two is the most unstable state. The state one channel becomes entrenched forming a Gc5/6 (Rosgen 1996, 2006; Repp 2010; Repp and Boyer 2011). This channel then widens forming a F5/6 channel. In state two active bank erosion is taking place. As the channel becomes entrenched the original floodplain is abandoned and riparian vegetation is lost and riparian plant communities may also be lost.
State three the system begins to stabilize as the floodplain rebuilds and balance between water, sediment, and energy is achieved. Erosion and sedimentation within the F5/6 channel creates a new constrained floodplain, forming a Bc5/6 channel (Rosgen 1996, 2006; Repp 2010; Repp and Boyer 2011). Further stabilization of the channel results in the C5/6 channel, and eventually the E5/6 channel. As the system stabilizes and the floodplain becomes wider riparian vegetation that was lost in state two returns. The endpoint of state three is the same as the potential natural channels, but has a constrained floodplain within the entrenched valley formed in state two.
The transition from the reference state to the entrenched state, T1A, is a result of a loss in vertical stability leading to downcutting and widening of the channel. As the channel becomes entrenched connectivity to the floodplain is lost. The transition to the re-established floodplain state, state three, from state two (T2A) is characterized by floodplain rebuilding and reestablishment of desirable riparian vegetation. Transition T3A back to the entrenched state from state three is similar to transition T1A.
Within state one the community phase change to 1.2 is characterized by decreased bank stability and increased erosion, which causes the stream to get wider and shallower. Despite the increased width to depth ratio the stream maintains connectivity with the floodplain. These changes are accompanied by a loss in desirable greenline vegetation.
The community phase change back to 1.1 requires the revegetation of the greenline by desirable plant species that aid in sediment trapping and reduction of the width to depth ratio.
Within state two the community phase change 2.1 results from increased instability and entrenchment resulting from greater erosion and losses of deep rooted vegetation.
In state three community phase change 3.1 is characterized by increased greenline vegetation, sediment trapping, and redevelopment of the floodplain. As the channel becomes more stable the width to depth ratio decreases and connectivity to the floodplain increases.
Community phase change 3.2 is accompanied by further increase in the greenline vegetation and widening of the floodplain. As stability increases the channel becomes deeper and narrower.
Community phase change to 3.3 is characterized by decreased bank stability and increased erosion, which causes the stream to get wider and shallower. Despite the increased width to depth ratio the stream maintains connectivity with the floodplain. These changes are accompanied by a loss in desirable greenline vegetation.
The historic climax plant communities associated with the riparian ecosystem of the Middle Sheyenne River include: 1) the greenline, which is the line of vegetation on or at the water’s edge, 2) woodlands, 3) shrublands, 4) grasslands and 5) wetlands. The greenline community is a water sedge community. The woodland communities found within the floodplain are American elm/Sprengel’s sedge, green ash/Sprengel’s sedge, and bur oak/Sprengel’s sedge, with the first two communities generally located adjacent to the river, whereas the later is located on higher floodplain features such as natural levees and terraces. The shrublands that boarder the woodlands are dominated by northern hawthorn and Sprengel’s sedge. Prairie communities within the floodplain are warm season graminoid/western snowberry. Wetland communities are characterized by a native sedge community.
Numerous non-native species are opportunistic and capitalize on changes in the natural disturbance regime of the ecosystem. Greenline communities are subject to increases in reed canarygrass, which practically forms a monoculture within the community. Changes in the canopy of woodlands result in increases of Kentucky bluegrass and smooth bromegrass in the understory. Within shrublands increases of buckthorn and Kentucky bluegrass have been observed. Prairie communities are subject to increases and even monopolization of the graminoid species Kentucky bluegrass and smooth bromegrass, in addition to increases in western snowberry. Similar to the greenline, wetland communities are vulnerable to monopolization by reed canarygrass.

Recommendations:

Areas needing additional study

Further research is needed to determine what is driving the ecological processes taking place within the riparian ecosystems of the Middle Sheyenne River. Additional research on how riparian ecosystems within the Middle Sheyenne River Watershed respond to grazing management needs to be conducted. This additional research would aid in the determination of best management practices for riparian ecosystems.

Information Products

Any opinions, findings, conclusions, or recommendations expressed in this publication are those of the author(s) and do not necessarily reflect the view of the U.S. Department of Agriculture or SARE.