Streambank Erosion Associated with Grazing Activities in Kentucky

Final Report for GS02-014

Project Type: Graduate Student
Funds awarded in 2002: $9,836.00
Projected End Date: 12/31/2004
Region: Southern
State: Kentucky
Graduate Student:
Major Professor:
Dwayne Edwards
University of Kentucky
Expand All

Project Information

Summary:

The effects of cattle grazing on stream stability have been well documented for the western portion of the United States, but are lacking for the east. Stream and riparian damage resulting from grazing can include alterations in watershed hydrology, changes to stream morphology, soil compaction and erosion, destruction of vegetation, and water quality impairments. However, few studies have examined the successes of best management practices (BMPs) for mitigating these effects. The objective of this project was to assess the ability of two common BMPs to reduce streambank erosion along a central Kentucky stream. The project site consisted of two replications of three treatments: 1) an alternate water source and a fenced riparian area to exclude cattle from the stream except at a 3.7 m wide stream ford, 2) an alternate water source with free stream access, and 3) free stream access without an alternate water source (i.e. control). Fifty permanent cross sections were established throughout the project site. Each cross section was surveyed monthly from April 2002 until November 2003. Results from the project indicated that the incorporation of an alternate water source and/or fenced riparian area did not significantly alter stream cross sectional area over the treatment reaches. Rather than exhibiting a global effect, cattle activity resulted in streambank erosion in localized areas. As for the riparian exclosures, changes in cross sectional area varied by location indicating that localized site differences influenced the processes of aggradation and/or erosion. Hence, riparian recovery within the exclosures from pre-treatment grazing practices may require decades, or even intervention (i.e. stream restoration), before a substantial reduction in streambank erosion is noted.

Introduction

Over a quarter of the land area within the United States is used for grazing activities supporting nearly 100 million cattle and calves (USDA-NASS, 1997; Vesterby and Krupa, 1997). While these cattle are a major component of the U.S. agricultural trade, improperly managed grazing cattle can contribute significant pollutant loads to the nation's waterways. The U.S. Environmental Protection Agency (U.S. EPA) (2000) identified agriculture as the predominant source of nonpoint source pollution (NPS), impairing 48% of the assessed rivers and streams. This value more than doubled the next leading source of NPS, hydrologic modifications. The two leading pollutants for rivers and streams were pathogens and sediment, constituents linked to cattle production (CAST, 2002; U.S. EPA, 2000; Belsky et al., 1999; Clark 1998). Cattle producers often use rivers and streams as the primary water source for their grazing livestock, resulting in increased activity along the water's edge. Streambank erosion occurs when livestock hooves trample banks and excessive grazing reduces riparian vegetation (Belsky et al., 1999).
Controlling or reducing agricultural NPS is an important step towards improving the quality of our nation's streams. A system of best management practices (BMPs) is the most likely means of achieving this goal in an effective and cost-efficient manner. However, developing a successful NPS pollution control program targeting grazing practices can be difficult, especially in the humid region of the United States. The majority of research on the impacts of cattle grazing and the subsequent effect of BMPs to reduce these impacts has occurred in the western portion of the U.S., a region with a markedly different climate than the eastern U.S. (McInnis and McIver, 2001; Belsky et al., 1999; Clark., 1998; Magilligan and McDowell, 1997). While it is important to examine the individual effects of BMPs for reducing NPS associated with grazing activities, an understanding of the water quality benefits derived from multiple BMPs may provide insights that allow for more informed managerial decisions. Minimizing the impacts of grazing on stream health will likely necessitate the incorporation of both structural (i.e. riparian buffers) and cultural (i.e. managed grazing) BMPs (Logan, 1990).
Few studies in the humid region of the United States have examined the impacts of grazing BMPs on water quality (Line et al., 2000; Sheffield et al., 1997; Owens et al., 1989). Only isolated studies examined streambank erosion associated with the use of a grazing BMP, though they yielded promising results. Sheffield et al. (1997) noted a 77% reduction in streambank erosion along a southwest Virginia stream following implementation of an off-stream water source. At a Tennessee stream, Trimble (1994) measured a six-fold increase in gross bank erosion along uncontrolled grazing sites as compared to reaches with exclusion fencing. While these studies provided useful information, they could not fill all the gaps in knowledge. Notably, these studies examined the effectiveness of a single BMP versus multiple controls more commonly implemented on farms, the erosive forces of cattle grazing and stream flow were not separated, and comparisons were not made between fenced and non-fenced treatments. As evident by these gaps in information, a need exists for obtaining additional information with regard to grazing BMPs, especially multiple BMPs, and their effectiveness at reducing streambank erosion. To fill the void, this project sought to determine the ability of an alternate water source with and without exclusion fencing (consisted of a 9.1 m wide riparian zone equipped with a 3.7 m wide stream ford) to reduce streambank erosion along a central Kentucky stream. Results from this project will provide stakeholders with necessary information regarding the effectiveness of these BMPs for reducing streambank erosion in central Kentucky and possibly within the humid region of the United States.

Project Objectives:

The goal of this project is the provide the agricultural community with a better understanding of the impacts of cattle grazing on stream bank erosion, thus enhancing current cattle production methods on small farms in Kentucky and possibly the eastern United States. Important to note is that this project is a subset of a larger research endeavor into the grazing impacts of cattle and BMPs on the water quality of a Kentucky stream whose objectives are to 1) determine whether the provision of cultural forms of BMPs such as an alternate water source, an alternate shade source, and/or the placement of supplemental feeding areas alter cattle behavior, 2) determine if the above BMPs collectively improve water quality, 3) determine if the inclusion of a fenced riparian zone with the BMP package significantly improves water quality over the BMP package alone, and 4) educate the agricultural community, especially livestock producers, about management systems that minimize the adverse environmental effects of grazing while maintaining production levels. This particular project, in combination with the larger research endeavor of which it is a part, address two main goals of the Sustainable Agriculture Research and Education (SARE) program by providing an assessment and evaluation of animal production practices to strengthen the agricultural competitiveness of Kentucky's cattle producers while seeking to conserve soil, water and stream habitats.

Cooperators

Click linked name(s) to expand/collapse or show everyone's info
  • Dwayne Edwards

Research

Materials and methods:

Study Area
The study area is located on the University of Kentucky's Animal Research Center (ARC) in Woodford County, Kentucky, approximately 15 miles northwest of Lexington, Kentucky (38°02' N, 84°36' W). The climate is humid and temperate with a mean monthly rainfall ranging from 66 mm in October to 118 mm in July with a mean annual rainfall of 1150 mm. Temperatures range from a mean monthly average of 0.3°C to 24.3°C with a mean annual temperature of 12.6°C (University of Kentucky Agricultural Weather Center, 2004). The ARC is characterized by gently rolling hills with elevations ranging from approximately 240 to 260 meters above mean sea level. The valleys within the study area are narrow with a limited area for floodplain development. Valley slopes within the riparian area are approximately 5% while valley slopes outside of the riparian area range from 12 to 17%. One stream drains much of the ARC through two second-order tributaries, Camden Creek and Pin Oak, whose confluence is near the property boundary of the ARC. Camden Creek originates outside of the boundaries of the ARC in horse-grazed pastures while Pin Oak, and its headwaters, begin on the ARC. These streams flow over or near bedrock each with a slope of about 0.5%. The entire study reach along both streams is characterized as a long run, devoid of perceivable riffles and pools. Camden Creek flows in a southwesterly direction, and Pin Oak flows in a northwesterly direction (fig. 1). {See hard copy report on file at Southern SARE Office.} The ARC is located in a significant karst terrain characterized by numerous shallow sinkholes. Previous surveys revealed that approximately 30% of the ARC drains to sinks (Fogle, 1998). Soils at the study site are derived from limestone and consist of the Hagerstown (fine, mixed, mesic Typic Hapludalf) and McAfee (fine, mixed, mesic Mollic Hapludalf) soil series along Pin Oak and the Hagerstown and Woolper (fine, mixed, mesic Typic Argiudoll) soils series along Camden Creek (Jacobs et al., 1994). The land use along the lowermost reaches of these tributaries is pasture. Land use within the subwatershed of Camden Creek consists primarily of pasture (hay and grazing) and crop (corn, wheat and tobacco) while land use in the subwatershed of Pin Oak is predominately crop (corn, wheat and tobacco). The pastures at the ARC, including those in the study area, are dominated by endophyte (Neotyphodium coenophialum) infected tall fescue (Festuca arundinacea).

Treatments
Data collection involved two replications (one replicate was located on Camden Creek and the other on Pin Oak) of three treatments (i.e. pasture plots), listed in downstream order as 1) an alternate water source and a fenced 9.1 m wide grassed riparian area to exclude cattle from the stream (equipped with a 3.7 m wide stream ford, which was constructed in accordance with United States Department of Agriculture-Natural Resource Conservation Service (USDA-NRCS) specifications as outlined in Code 576), 2) an alternate water source with free stream access, and 3) free access with out an alternate water source (i.e. control) (USDA-NRCS, 2004) (fig. 1). Treatments were ordered such that the anticipated severity of the treatment increased in the downstream direction. Selection of the treatment order followed work by Trimble (1994) who implemented a similar experimental design to assess the impacts of uncontrolled grazing on streambank erosion rates. Trimble (1994) reasoned that this treatment order would reduce the possibility of eroded bank sediments destabilizing down-gradient channel reaches through the formation of channel bars, which might alter flow patterns such that stresses are increased along stable banks. The pasture plots used for each treatment within a replication spanned the stream with similar stream frontage (Table 1). The replication, along Camden Creek, contained pasture plots with an area of approximately 2 ha while the other pasture plots, located along Pin Oak, were nearly 3 ha. The difference in plot size for the replications resulted from the amount of land available for the study. Every attempt was made to ensure that other plot characteristics such as topographical features, soils, existing shade, and riparian characteristics (if applicable) were as consistent as possible among the treatments.
The alternate water source consisted of insulated waterers installed on a concrete base with a geotextile-gravel pad. Located in the upland areas of the pasture plots, these waterers were between 87 to 121 m from the streams (Table 1). {See hard copy report on file at Southern SARE Office.} High tensile electrical fence was used to separate the pasture plots and to exclude cattle from the riparian areas. The excluded riparian areas consisted primarily of ungrazed tall fescue, as additional vegetation such as trees and shrubs were not planted. Stream crossings (i.e. fords) were constructed using geotextile, a 15-cm layer of riprap (approximately 10-cm diameter), and a 10-cm compacted layer of crushed rock as outlined in Code-576 (USDA-NRCS, 2004). Fertilizer (ammonia-nitrate) was applied annually to all pasture plots at a rate of 45 kg/ha prior to the start of the grazing season. Cattle stocking densities were varied throughout the grazing seasons based on the amount of available forage (Table 2). However, the maximum practical level was used with the goal of maintaining the same stocking densities for all treatments within a replication. Initial stocking densities were set at 1,300 kg/ha and were adjusted based on visual assessment of available forage. Cattle were weighed on a monthly basis during the grazing season (typically mid-April until late October). During the 2002 and 2003 grazing seasons, steers were grazed on the pastures except during November and December 2002 when cows were placed on the pastures. During August and September of 2002, the stocking density was maintained by feeding hay (ad libitum) when forage supply became limited due to drought conditions. Supplemental hay was located in the upland areas away from the streams and on the opposite side of the pasture plots in comparison to the alternate water sources. All animal protocols were approved by the University of Kentucky Institutional Animal Care and Use Committee (protocol # 00245A2001).

Cross Sectional Areas
Using guidelines set forth by Harrelson et al. (1994), fifty permanent cross sections (23 along Camden Creek and 27 along Pin Oak) were established for surveying erosion levels. Cross sections were erected at both random locations and near areas anticipated as frequent travel paths for cattle. Along Pin Oak, one cross section was established at the confluence of Pin Oak with an unnamed ephemeral channel. Each cross section was established perpendicular to stream flow. Cross-sectional surveys were conducted monthly from April 2002 until November 2003 resulting in a total of 18 surveys of each cross section. Conventional leveling techniques, as outlined in Harrelson et al. (1994), were employed, and elevation measurements were taken at 0.3 m horizontal intervals. Data from all of the cross-sectional surveys were used in the analysis except for those collected during January 2003 (when the alluvial material was frozen).
Computation of changes in cross-sectional area with subsequent surveys was a multi-step process. Each recorded elevation within a cross section was subtracted from an upper plane (i.e. constant elevation) that was greater than all of the measured elevations for that cross section to obtain a cross-sectional area at that location and time period . This step was performed for all 50 cross sections and for all 17 periods. The cross sectional areas (AT,j,k) for each cross section and for each sample period were computed using
(1)
where x is the lateral station along the cross section and y is the elevation at the corresponding lateral station. Lateral station is represented by the subscript i, sample period is represented by the subscript j, and cross section location is represented by the subscript k. Differences in cross-sectional area were computed for each cross section by subtracting the cross-sectional area for the previous period (j-1) from the cross-sectional area for the current period (j). At each individual cross section, increased values of AT,j,k from one cross-sectional survey period to the next indicated alluvium loss or erosion while decreased values of AT,j,k from one cross-sectional survey period to the next indicated alluvium gain or aggradation. A total of 16 cross-sectional area differences were computed for each cross section. For a given sampling period, all cross-sectional area differences computed for the cross sections within a treatment and replication were averaged to obtain an overall mean for the respective pasture plot (Table 3). {See hard copy report on file at Southern SARE Office.}

Cattle Position
Understanding the impact of grazing activity on streambank erosion required knowledge of animal position. Global positioning system (GPS) collars, GPS_2200 Small Animal GPS Location Systems (Lotek Engineering, Inc., Newmarket, ON)*, were used to collect position information on a sample of cattle from each pasture plot (Table 4) Detailed descriptions of the GPS collars were presented in Turner et al. (2000) and Agouridis et al. (2004). Position information was collected over seven, 18-day periods during May, August and November 2002 as well as April, June, July and October 2003. A five-minute sample interval, the smallest permitted with the GPS collars, was selected. Data from the GPS collars were filtered and differentially corrected allowing use of the most accurate position points in the analysis (Agouridis et al., 2004). To ensure that the filtering process did not skew the data, one-way repeated measures analysis of variance (ANOVA) tests (a=0.05) were conducted to compare the percentage of usable (i.e. non-filtered) position points 1) between each pasture plot in its entirety and 2) between each pasture plot along the streambanks (Table 5). No statistical difference was detected between the usable position points collected in the pasture plot and those along the streambanks. Additionally, no statistical differences were noted between pasture plots for the percentage of usable points along the streambanks (i.e. the position points used in the streambank erosion analysis).
Prior to the start of the project, a base map identifying key pasture features was created using a Real Time Kinematic Global Positioning System (RTK-GPS) with a published horizontal accuracy of 20 mm. Key pasture features included the streambanks of Camden Creek and Pin Oak, fences, trees, alternate water sources, fenced riparian areas, and stream crossings. The base map was used in conjunction with data collected from the GPS collars during the seven cattle-monitoring periods to characterize cattle position along the streambanks. A 5-m buffer from the edge of the streambanks was created in ArcView® for each pasture plot. The five-meter buffer was selected because it represents the estimated maximum horizontal error associated with the GPS collars in an open field environment (Agouridis et al., 2004). Separation of instream and near bank cattle positioning was not performed because of technological limitations with the GPS collars (i.e. accuracy level) prohibited specific identification of instream versus near stream cattle position. For each GPS collar-monitoring period, all GPS collar data points that fell within this buffer were totaled (GPSs,p,j). A five-meter buffer around each cross section was overlain on the five-meter stream buffer, and all of the GPS collar data points that fell within this overlay were totaled (GPSov,p,j). Finally, the percentage of cattle activity within five meters of the stream associated with each cross section and each GPS monitoring period was computed using
(2)
where the subscript s denotes the five-meter stream buffer around the stream; the subscript ov denotes the overlay of the five-meter stream buffer around the stream and five-meter cross section buffer around the cross section; and the subscript p denotes the pasture plot that contained the cross section. As with differences in cross-sectional areas, all cross-sectional cattle position values within each treatment and replication were averaged to obtain an overall mean for the respective pasture plots for each sampled period (Table 6). {See hard copy report on file at Southern SARE Office.}

Stocking Densities
All cattle on each of the pasture plots were weighed at 28-day intervals during the grazing season for both years of the project. The final weights and cattle numbers for each pasture plot and for each period were used to compute stocking densities (Table 2). Every attempt was made to maintain equivalent stocking densities across the pasture plots within a replicate for a given period. Stocking densities varied with available forage, ranging from 1670 kg/ha at the early stages of the grazing seasons to 720 kg/ha during the latter part of the grazing seasons.

Stream Discharges
Stream discharge data were collected at the most downstream edge of each replication (i.e. Camden Creek and Pin Oak) using compound 90° V-notch weirs and ISCO 4220 flow meters (pressure transducers) (fig. 1). Discharge data were collected at 10-minute intervals at the two weirs for the duration of the study. Each weir was located approximately 5 m downstream from the respective most downstream treatments. Average discharges were computed from flow values collected during the period prior to each cross-sectional survey. For example, if a cross-sectional survey was performed on September 3, 2002 and the subsequent survey was conducted on October 2, 2002, then the average discharge for the period was assigned to the October survey. Since flow data were not available at each cross section, ArcView® was used to prorate the outlet flow contributions to each cross section based on the cross section's watershed area (eqn. 3).
(3)
Q represents discharge (m3/s), WS represents watershed area, and the subscript w represents the weir. Flow data was not available for the ephemeral channel that flows into Pin Oak. As with differences in cross-sectional areas, all cross-sectional stream discharge values within a treatment and replication were averaged to obtain an overall mean for the respective pasture plots for each sampled period (Table 7). {See hard copy report on file at Southern SARE Office.}

Time
The parameter time was defined as the time lapse or interval from the start of cross-sectional data collection (i.e. prior to cattle introduction in April 2002) until the end of the project (i.e. following cattle removal in November 2003). A time value in relation to the original cross-sectional survey was computed for each subsequent cross-sectional survey, and ranged from 59 days in June 2002 to 571 days in November 2003.

Statistical Analysis
In this study, the experimental unit and blocking unit were stream (i.e. Camden Creek and Pin Oak) and cross sections were subsamples within the experimental unit. For each sampled period, the cross-sectional area differences within each treatment and replication were averaged to obtain an overall mean. Multivariate repeated measures analysis techniques were conducted using the Mixed model in SAS® (PROC MIXED) to determine the effects treatment, time, stream, stocking density, cross-sectional cattle positions, and cross-sectional streamflow had on changes in cross-sectional area (SAS, 1985). Because the number of cross sections or subsamples differed amongst the pasture plots, the subsamples were weighted for unequal sizes in the Mixed model. The Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC) were used to evaluate the covariate structure of the model. Construction of the model occurred by first eliminating nonsignificant continuous variables (i.e. stocking density, cross-sectional cattle positions, and cross-sectional streamflow) at a=0.05 followed by categorical variables without interaction (i.e. stream and time). Since the objective of the study was to assess the ability of treatment (i.e. BMPs) to reduce streambank erosion, treatment and variables interacting with treatment remained in the model during the model reduction process.

Research results and discussion:

Three iterations of the mixed model were conducted to assess the effects of the categorical and continuous variables on changes in cross-sectional areas. Results of these iterations indicated that none of the examined variables (stream, treatment, time, stocking density, cross-sectional cattle position, and cross-sectional stream flow) were significant predictors of changes in cross-sectional area at a=0.05 level of significance. Only one variable, cross-sectional cattle position, was significant at a=0.10 level of significance (P=0.06).

Cattle Position and Activity
Based on previous research into the impacts of cattle grazing on streambanks, it was expected that cross-sectional cattle positions would be a stronger predictor of change in cross-sectional area (Belsky et al., 1999; Sheffield et al., 1997; Trimble, 1994; Kauffman et al., 1983). The moderate nature of the trend was likely due to the position accuracy limitations of the GPS collars coupled with a lack of information regarding actual cattle activity associated with the position points. While the GPS collars were equipped with two-axis activity sensors, these sensors provided a poor sense of animal activity as 1) collar fit, which differed for each animal, impacted sensitivity, 2) orientation of the collar on the animal impacted sensor performance (i.e. which side of the collar was facing outward), and 3) movement of the animal’s head alone could not be readily differentiated from movement of the entire animal (i.e. drinking or grazing versus walking). As such, the nature of the position points (i.e. active or static) was not differentiated meaning that all cross-sectional cattle positions were treated equally in the model, regardless of the level of activity. In actuality, a large number of position points may have been associated with low levels of cattle movement (i.e. cattle loitering in the stream) and relatively small changes in cross-sectional area while a small number of position points with high levels of cattle movement (i.e. cattle walking along the streambanks) may have been linked to relatively large changes in cross-sectional area (Table 8).
Examination of changes in cross-sectional areas over the course of the study, with respect to cross section location in the pasture plots, revealed interesting aspects. Cross sections that demonstrated some of the strongest increasing trends with respect to cross-sectional area 1) did not necessarily have the greatest percentage of cattle position points and 2) were often located along or near preferred stream crossing points (i.e. locations of active movement). For example, cross sections C6, C7, C13, C20, C21, P14, P18, and P26 are all located near fence-lines (i.e. newer cattle stream crossings or paths) or former cattle guards marking former fence-lines (i.e. old cattle stream crossings or paths) (fig. 1). Similarly, not all cross sections that demonstrated a strong decreasing trend in cross-sectional area had a small percentage of cross-sectional cattle position points (i.e. C5). Higher levels of cattle position points were associated with cross sections adjacent to the stream fords due, in part for the non-fenced riparian area pasture cross sections and entirely for the fence riparian area pasture cross sections, to the accuracy level of the collars. For the pasture plots with fenced riparian areas, cattle were excluded from the riparian areas, and as such cattle position points should not have been recorded in relation to cross sections located in these excluded areas. Furthermore, the two cross sections with the highest number of recorded cattle positions, P8 and P9, demonstrated a weak to moderate decreasing cross-sectional area trends (R2=0.24 and R2=0.43, respectively). These cross sections were located in areas where cattle frequently loitered as determined by numerous visual observations, thus highlighting the greater importance of cattle activity rather than cattle position in erosion processes.
The incorporation of a fenced riparian zone into a grazing BMP program may prove beneficial in the reduction of erosion rates, but a significant period of time may be required if these benefits are to be realized. The lack of significance of the variable treatment is likely the result of the geomorphologic nature of the stream and the length of the study. Throughout the study reach, as well as the reaches upstream of the study area, the discharge from both Camden Creek and Pin Oak flows over or near bedrock. Therefore, sediment supply to the streams is limited to contributions from runoff and the streambanks. Previous geological research at the ARC revealed the importance of groundwater as Gremos (1994) discovered nearly 80 sinkholes and sinks (rounded depressions) on the ARC. Maury and McAfee, the dominant soil series at the ARC, are in the hydrologic soil group B, which is characterized by moderate infiltration rates (Haan et al., 1994; Jacobs et al., 1994). With limited amounts of runoff contributing to the flow in Camden Creek and Pin Oak, the majority of the instream sediments are produced from streambank erosion processes (i.e. cattle activity and discharge). The lack of significance for the variable discharge is indicative of the potentially lengthy period of time required for notable changes in the cross-sectional areas (i.e. channel narrowing) to occur under the influence of discharge alone. Furthermore, the sampling period to sampling period oscillation between increases in cross-sectional area and decreases in cross-sectional area for certain cross sections was likely the result of varying sediment supplies (i.e. degree of streambank trampling by cattle) and transport rates (i.e. stream discharge). With regard to sediment supply, no notable degradation of the stream fords was perceived over the course of the study meaning that no rills formed along the stream fords and loss of rock along the top layer was minimal.

Eliminated Variables
The variables identified as nonsignificant in the mixed model (i.e. stream, treatment, time, stocking density, and discharge) were as interesting as the significance of cattle position (P=0.06). The nonsignificant nature of the variable stream indicates that the results of this study are applicable to other areas within central Kentucky and possibly similar bedrock streams in the humid region of the U.S. With regard to treatment, the flawed assumption made was that the rate of both streambank erosion for the unfenced treatments and the recovery phase for the fenced treatments from pre-study grazing pressures would be similar. In actuality, Camden Creek and Pin Oak resembled a broken leg model in which recovery occurs but at a slow rate following the removal of grazing pressures, as it takes a much longer time for the stream to reach a state of equilibrium (Sarr, 2002). The low potential for upstream sediment influxes, due to the nature of the streams and watersheds, coupled with the low potential for sediment deposition, due to the straightness of the channels, indicates that grazing recovery within the study reach may require several years to be realized, and may in fact require intervention (i.e. stream restoration). Costa (1974) noted that straight channel reaches are slow to recover because of the minimal occurrence of flow restrictions, which promote sediment deposition. Magilligan and McDowell (1997) noted that channel adjustment on four alluvial streams in Oregon occurred in part by sediment deposition following the exclusion of cattle. Trends of decreasing cross-sectional area were noted for some cross sections (i.e. C18, C19, P26) in close proximity to the backwaters created by the weirs. The weirs can be characterized as small dams that promoted the settling of suspended sediments, which probably originated from the streambanks. As for the alternate water source, its presence alone did not reduce streambank erosion. Although not directly measuring streambank erosion rates, Line (2003) noted significant decreases in suspended sediment concentrations did not occur with an alternate water source alone but did occur following exclusion fencing.
A basic assumption made with regards to streambank erosion, over the course of the study, at Camden Creek and Pin Oak that helped to explain the insignificance of time was that following the introduction of cattle, erosion would occur steadily over time until cattle were removed, and little recovery would take place after their removal during the off-grazing season. Following cattle removal, erosion rates would continue slowly and possibly plateau, then again increase following the re-introduction of cattle. The basic flaw with this assumption was that cattle position, and hence cattle activity, within a cross section would be constant throughout the time the animals were in the pasture. Plots of GPS collar data revealed that while cattle favored certain sections of the stream, the rates at which they frequented these sections varied throughout the study (fig. 2). Furthermore, the level of activity within a cross section was a more important indicator of erosion than merely cattle presence in the cross section (Table 8). The pressure exerted on streambanks by cattle hooves can result in bank shear and bank sloughing (CAST, 2002; Warren, 1986). Cattle frequently standing on the bedrock bottom of the stream, as visually confirmed with cross section P8 and P9, had a lesser impact on cross-sectional areas (i.e. decreasing trends) when compared to cross sections located near new or established cattle stream crossings or paths near fence lines. This observation indicates that erosion can happen quickly in areas associated with frequent cattle movement but may occur more slowly in reaches noted for cattle loitering.
In addition to time and treatment, stocking density was not found to be a significant predictor of change in cross-sectional area. Stocking density was greatest during the early cooler months of the grazing season when forage was plentiful. As the grazing season continued, stocking densities in the pasture plots generally decreased as forage availability decreased. By fluctuating stocking densities based on available forage, as determined by visual observation, and supplementing forage (i.e. hay) during August and September 2002 during drought conditions, stocking densities were essentially a managerial BMP. Zobisch (1993) noted that soil loss increased with stocking density, but that reductions in soil loss occurred when stocking densities were managed to maintain at least a 40% foliage cover level. Furthermore, these managed stocking densities did not affect increasing trends in cross-sectional area (i.e. erosion) for high traffic locations such as along fence lines, indicating that erosion can occur with a small number of animals. If stocking density had remained constant throughout the grazing seasons, particularly the 2002 grazing season, the level of erosion may have been greater in these high traffic areas.
Surprisingly, cross-sectional discharge was not significantly related to change in cross-sectional area. Lane (1955) indicated that increases in flow produce increases in sediment, assuming median particle size and slope remain constant. As such, the expectation was that increased discharge levels would result in increased rates of streambank erosion, especially in the unfenced treatments that lacked substantial riparian vegetation. The lack of a significant relationship between discharge and change in cross-sectional area is likely due to 1) the nature of the streambank materials and 2) the overriding influence of cattle position (i.e. activity) on change in cross sectional area. Soils along the study reaches are from the Maury and McAfee soil series, which are silt loams. Cohesive banks, such as those in the study reaches, are generally resistant to high fluid shear that is present with elevated discharges. Additionally, these soils have low shear strength making them more susceptible to mass failure, which may be influenced by cattle activity (Lawler et al., 1997). Lawler et al. (1997) noted that cohesive banks rarely experience mass failures during periods of high discharge, but that these banks are more susceptible to such events within hours to days after the recession of higher flows. While these banks may have been more susceptible to erosion as a result of greater discharges, a time-lag likely preceded any mass failures, provided that these mass failures occurred at the study site.

Conclusions
Streambank erosion along two bedrock bottom second-order perennial streams was positively correlated with cattle position within the cross section (P=0.06). While knowledge regarding cattle position was useful, information pertaining to specific cattle activity (i.e. walking versus loitering) would have been of greater assistance as the number of position points within a cross section did not readily relate to high levels of cattle movement (i.e. streambank trampling). Cross sections that demonstrated the highest erosional trends did not have the highest number of position points, but rather were located in areas typical of greater cattle movement, such as preferred stream crossings or paths. Cattle continued to impact the stream if allowed access indicating that the use of an alternate water source alone did not reduce streambank erosion rates. The trend of decreasing cross sectional area for locations within the exclusion areas, while not significant, suggests that the inclusion of a fenced riparian zone into a grazing BMP program may prove beneficial, but years may be required before any morphological benefits are apparent. While some systems may recover quickly, the low supply of upstream sediments to the study reaches on Camden Creek and Pin Oak likely means that several years of rest or even intervention (i.e. stream restoration) may be required before pre-grazed conditions are realized.

Participation Summary

Educational & Outreach Activities

Participation Summary:

Education/outreach description:

Agouridis, C.T., D.R. Edwards, S.R. Workman, J.R. Bicudo, J.L. Taraba, and E.S. Vanzant. 2005. Streambank Erosion Associated with Grazing Practices in the Humid Region. Transactions of the ASAE 48(1): 181-190

Agouridis, C.T. and D.R. Edwards. BAE Study Evaluates Impacts of Cattle Grazing on Streambank Erosion. UK BAE SARE Project Newsletter, Spring 2005. www.bae.uky.edu/WQ406.

Agouridis, C.T., D.R. Edwards, S.R. Workman, J.R. Bicudo, J.L. Taraba, E.S. Vanzant, and R.S. Gates. 2004. Streambank Erosion Associated with Grazing Practices in the Central Kentucky. Paper for the 2004 ASAE International Meeting, Paper #042227, Ottawa, Canada, August 2-4, 26 p.

Agouridis, C.T. 2004. Cattle Production Practices in a Small Grazed Watershed of Central Kentucky. Ph.D. Dissertation, University of Kentucky.

Bicudo, J.R., C.T. Agouridis, B.K. Koostra, and S.R. Workman. 2003. Cattle Production Practices in Grazed Watersheds. Article for Environmental & Natural Resource Issues, ENRI Task Force Newsletter, Fall Issue.

Agouridis, C.T. and D.R. Edwards. BAE Study Evaluates Impacts of Cattle Grazing on Stream Bank Erosion. UK BAE SARE Project Newsletter, Spring 2003. www.bae.uky.edu/WQ406.

Workman, S.R., J.R. Bicudo, E.S. Vanzant, D.R. Edwards, and C.T. Agouridis. 2004. Cattle Production Practices in Grazed Watersheds of the Humid Region. Presentation for the USDA-CSREES National Water Quality Conference in Clearwater, Florida, January 11-14.

Agouridis, C.T., B.K. Koostra, S.R. Workman, J.R. Bicudo, E.S. Vanzant, and D.R. Edwards. 2003. Cattle Production Practices in Grazed Watersheds of the Humid Region. Presented at the University of Kentucky Animal Research Center in Woodford County, Kentucky for the Woodford County Farm Bureau, November 18. 20 attendees.

Kentucky Master Cattleman Program. Environmental Stewardship and Industry Issues Sessions.

Additional tours included: Nelson County, Kentucky Farmers (55 attendees); Kentucky Association of Conservations Districts (50 attendees); Oregon Seed Industry (20 attendees); Dr. Joseph Jen, Under Secretary for Research, Education, and Economics, USDA; and Dr. Colien Hefferan, Administrator USDA-CSREES.

Project Outcomes

Project outcomes:

This project provides information regarding the impacts of cattle grazing on stream bank erosion on two central Kentucky streams. Results from this project have been incorporated into the Kentucky Master Cattleman Program, in which more than 1000 producers from 101 Kentucky counties representing 51,000 cattle have participated since 2000.

Recommendations:

Areas needing additional study

Since the project site was located along two second-order perennial bedrock bottom streams, future effort should focus on examining streambank erosion associated with cattle grazing along streams with differing geomorphologic characteristics.

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.