Top Navigation Bar

Effects of Cattle Grazing and BMPs on Stream Water Quality

Carmen T Agouridis , Engineer Associate IV, ASAE Member

University of Kentucky , 128 C.E. Barnhart Bldg., Lexington, KY 40546 , USA

Dwayne R Edwards , Professor, ASAE Member

University of Kentucky , 128 C.E. Barnhart Bldg., Lexington, KY 40546 , USA

Steve R Workman , Associate Professor, ASAE Member

University of Kentucky , 128 C.E. Barnhart Bldg., Lexington, KY 40546 , USA

Jose R Bicudo , Assistant Professor, ASAE Member

University of Kentucky , 128 C.E. Barnhart Bldg., Lexington, KY 40546 , USA

Joseph L Taraba , Extension Professor, ASAE Member

University of Kentucky , 128 C.E. Barnhart Bldg., Lexington, KY 40546 , USA

Eric S Vanzant , Associate Professor

University of Kentucky , 907 T.P. Cooper Bldg., Lexington, KY 40546 , USA

Richard S Gates , Professor and Chair, ASAE Member

University of Kentucky , 128 C.E. Barnhart Bldg., Lexington, KY 40546 , USA

This is not a peer-reviewed article.

Paper No: 042131
An ASAE Meeting Presentation

Written for presentation at the 2004 ASAE/CSAE Annual International Meeting

Sponsored by ASAE/CSAE
Fairmont Chateau Laurier, The Westin, Government Centre
Ottawa, Ontario, Canada
1 - 4 August 2004

Abstract. Cattle production is a major component of Kentucky’s agricultural economy, accounting for approximately 15% of the total agricultural sales in 2000. There are over 2.2 million beef cattle and calves in the state making Kentucky the number one beef producer east of the Mississippi River. Research into the effects of cattle grazing on stream water quality has been well documented in the western portion of the United States with some estimates indicating that 80% of the damage to riparian areas was caused by grazing livestock. However, the impacts of grazing cattle in a humid environment may differ significantly from those witnessed in the arid West. Furthermore, relatively little information exists regarding the effectiveness of grazing best management practices (BMPs), such as alternate water sources, alternate shade sources, supplemental feeding, and riparian buffers, for improving the water quality of streams in grazed watersheds of the humid region. As part of a larger research endeavor into cattle production practices in the humid region, water samples were collected over a two year period at the project site located on the University of Kentucky’s Animal Research Center. The project sites consisted of two replications of three treatments: control, selected BMPs with free access to the stream, and selected BMPs with limited access to the stream. Grab samples were collected at the upstream and downstream pasture edges. Samples were analyzed for nitrate-nitrogen, ammonium-nitrogen, total Kjeldahl nitrogen, dissolved orthophosphate, total phosphorus, total suspended solids, pH, chemical oxygen demand, five-day biochemical oxygen demand, fecal coliforms, and fecal streptococci. Results indicated that minimal water quality benefits were incurred by implementing the BMP systems (i.e. treatments). One of the most substantial understandings gleaned from the project was the importance of upstream land use, and to some degree soils, when attempting to identify significant treatment effects within a small reach. Additionally, the karst geology, which is characteristic of the Bluegrass Region of Kentucky, influenced the rate of transport (i.e. flashy system with quick response time to rainfall) of nutrients from upland areas (i.e. row crops), especially along Pin Oak. These external factors may have resulted in the lack of uniformity in significant constituent concentration differences between the two streams when cattle were present. Furthermore, the background constituent concentration levels may have prohibited the identification of treatment effects. Results from this project indicated that minimal water quality benefits were incurred by implementing a BMP system (with or without a partially excluded riparian zone). However, these results may differ if cattle were completely excluded from the stream or if the BMP system was implemented at a site with larger pastures, different geology (nonkarst), soils (low in phosphorus), or stream morphology (nonbedrock bottom channel).

Keywords. Best management practices, nutrients, bacteria, stream.


Introduction

Cattle production is a major component of Kentucky's agricultural economy, accounting for approximately 13% of the total agricultural sales in 2002 making Kentucky the largest beef producing state east of the Mississippi River (state rank: 8th) with over 2.4 million cattle (KASS, 2003). Additionally, Kentucky's pastures are characterized as rolling-to-steep, frequently permeated with streams and dry runs. Research indicates that uncontrolled livestock access to streams can be greatly damaging to the ecological integrity and sustainability of riparian ecosystems. Considering the number of pasture grazed beef cattle in the state, the expected upward trend in beef production, and Kentucky's 143,200 kilometers of rivers and streams, understanding and assessing the potential for livestock damage to riparian ecosystems in Kentucky's humid environment is quite important.

The U.S. Environmental Protection Agency (EPA) Office of Water (2000) noted that fecal bacteria and siltation were among the most common pollutants to Kentucky's rivers and streams. In Kentucky, 73% of the river miles surveyed for bacteria do not fully support swimming because of high fecal coliform levels. In their 2000 Water Quality Report, the EPA's Office of Water noted that the nation's leading source of river and stream impairment was agricultural activity with the states reporting that agriculture was a source of contamination for 48% of the documented impaired river and stream miles. Approximately 41% of the continental United States (364 million hectares) is dedicated to agricultural production of which 43% is pasture and rangeland. With such a large amount of land dedicated to a practice identified as one of the main contributors to stream impairment, methods for determining and managing the effects of livestock on the environment must undergo careful consideration (1997 Census of Agriculture as reported by the EPA Office of Water, 2000).

Research into the effects of cattle grazing on water quality has been well documented in the western United States. Researchers estimated that 80% of the damage incurred by streams and riparian systems in these arid environments was from grazing livestock (U.S. Department of the Interior, 1994 as reported in Belsky et al., 1999). Stream and riparian damage resulting from livestock grazing includes alterations in watershed hydrology, changes to stream channel morphology, soil compaction and erosion, vegetation destruction and water quality impairments (Belsky et al., 1999; CAST, 2000). Only a small amount of research has been conducted with regards to best management (BMP) impacts on water quality, and this research predominately focused on the effects of a single BMP, namely an off stream water source. Sheffield et al. (1997) noted a 90% reduction in total suspended solids (TSS), a 54% reduction in total nitrogen (TN), an 81% reduction in total phosphorus (TP), and a 75% reduction in sediment bound phosphorus (SBP) following the installation of an offstream water source along a southwest Virginia stream. Line et al. (2000) also examined the effectiveness of an offstream water source, but concluded that this BMP alone was not successful at reducing pollutant loads to a North Carolina stream.

One approach to minimizing the impacts of cattle grazing on stream quality is through the use of a program of best management practices (BMPs). Both structural and cultural control BMPs are well suited for reducing the impacts of cattle grazing on the health of a stream. Structural control BMPs modify the transport of pollutants to waterways (i.e. riparian buffers). Cultural control BMPs are designed to minimize pollutant inputs to waterways through land management practices (i.e. managed grazing) (Logan, 1990). By using a program of BMPs (i.e. multiple BMPs) comprised of structural and cultural BMPs, the achievement of water quality goals is more likely, since the effectiveness of each BMP will vary depending upon the pollutant(s) of concern. As evident by the lack of literature focusing on the humid, temperate region of the United States, a need exists for detailing the ability of a program of BMPs to reduce nonpoint source (NPS) loads from grazed pastures to streams. To fill the gap in the literature, this project was designed to assess the effects of two BMP systems (each consisting of an alternate water source, alternate shade sources, and pasture improvements with one system also having a 9.1 m wide riparian zone equipped with a 3.7 m wide stream crossing) on water quality. Constituents examined in the project were those indicative of manure contamination and include nitrate-nitrogen (NO 3 -N), nitrite-nitrogen (NO 2 -N), ammonia-nitrogen (NH 3 -N), total Kjeldahl nitrogen (TKN), organic nitrogen (Org-N), total nitrogen (TN), dissolved orthophosphate (PO 4 -P), total phosphorus (TP), pH, chemical oxygen demand (COD), five-day biochemical oxygen demand (BOD 5 ), total suspended solids (TSS), fecal coliforms (FC), and fecal streptococci (FS). Results from this project will provide stakeholders with information regarding the impacts of these BMP systems on the water quality of two central Kentucky streams.

Methods

Study Area

The study area is located on the University of Kentucky's Animal Research Center (ARC) in Woodford County, Kentucky. 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 (University of Kentucky Agricultural Weather Center, 2004). The ARC is characterized by gently rolling hills with elevations ranging from approximately 240 to 260 meters. One stream drains much of the ARC through two bedrock bottom second-order tributaries, Camden Creek and Pin Oak, whose confluence is near the property boundary of the ARC. The total drainage area for Camden Creek is 465 ha, and the total drainage area for Pin Oak is 337 ha (fig. 1). The ARC is located in a significant karst area with approximately 30% of the farm draining to sinks (Fogle, 1998) (fig. 2). Soils at the study site are derived from limestone and consist of the Hagerstown and McAfee soil series along Pin Oak and the Hagerstown and Woolper along Camden Creek (Jacobs et al., 1994). Land use along the lowermost reaches of these tributaries is pasture, which is dominated by endophyte ( Neotyphodim coenophialum ) infected tall fescue ( Festuca arundinacea ). For Camden Creek, the primary agricultural activity within the watershed is grazing (beef, sheep and equine) with less than 10% of the area of the ARC above the study site dedicated to row crop production. Activities in the Pin Oak watershed are similar to those in the Camden Creek watershed, but with increased levels of row crop production (20% of the area of the ARC above the study site).

Treatments

Data collection involved two replications (Camden Creek and Pin Oak) of three treatments (i.e. pasture plots), listed in downstream order as 1) BMPs and a fenced 9.1 m wide riparian area to exclude cattle from the stream and equipped with a 3.7 m wide stream crossing (Riparian)), 2) BMPs with free stream access (BMP only), and 3) free stream access with limited BMPs (Control) (fig.1). The limited BMPs included in the control treatment predominately consisted of natural shade, since the decision was made not to remove any old growth trees from the site for the purpose of this study. Treatments were ordered such that the anticipated severity (i.e. water quality impact) of the treatment increased in the downstream direction. The implemented BMPs included an alternate water source, alternate shade (started in July 2003), and pasture improvements consisting of fertilizer plots and herbicide plots, each 30.5 m × 30.5 m (started in April 2003). 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. The fertilizer plots received an additional 11 kg/ha of fertilizer and a herbicide was applied for inhibition on reproductive growth of fescue to the herbicide plots . The pasture plots used for each treatment within a replication spanned the stream with approximately equal stream frontage. One 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. Every attempt was made to ensure that plot characteristics such as topographical features, soil, existing shade, riparian characteristics (if applicable), and linear feet of stream frontage was as consistent as possible among the treatments.

Water Quality Monitoring and Analysis

Selection of the sampling method (i.e. grab, storm event or both) was based on a number of factors such as probability of success in detecting treatment differences, budgetary constraints and logistical constraints. To better assess the probability of successfully detecting treatment differences, a five-year data set (1996 to 2001) from a separate monitoring project, compiled from grab samples that were collected from monitoring stations both upstream and downstream of the research site (along Camden Creek and Pin Oak) was analyzed using nonparametric one-way ANOVA (Kruskal-Wallis one way analysis of variance on ranks). The data set consisted of bi-weekly grab samples from the sites upstream and downstream along Camden and Pin Oak Creeks and a tributary entering Pin Oak. The analysis was conducted on the constituents pH, temperature (T), dissolved oxygen (DO), PO 4 -P, TP, total solids (TS), TSS, total coliforms (TC), FC, FS, NH 3 -N, NO 3 -N and total organic carbon (TOC). Results from this preliminary analysis indicated that differences existed between five of the constituents monitored at the upstream and downstream stations of Camden Creek (pH, PO 4 -P, TS, FC and TOC), three of the constituents monitored at the upstream and downstream stations of Pin Oak (PO 4 -P, TS and FC), and five of the constituents monitored at the downstream station and the upstream tributary of Pin Oak (PO 4 -P, TP, FC, NO 3 -N and TOC). These results indicated that it was probable to detect water quality differences for nutrient, sediment and bacterial constituents between monitoring locations along Camden Creek and Pin Oak. Based on this information coupled with the expense of establishing storm event monitoring stations (estimated at $5,000 per station with the need for four stations), the expense of analyzing a much larger number of storm event samples, and logistical/laboratory constraints, the decision was made to collect bi-weekly grab samples.

Bi-weekly stream flow samples were collected at the treatment boundaries for each replication, totaling four sampling sites per replication per stream (fig. 1). The upstream sampling location for the Riparian treatment was identified as G1 (G indicating grab sample and 1 indicating location) while the respective downstream sample location was identified as G2. For the BMP Only treatment, upstream sample location was identified as G2 and the downstream sample location was identified as G3. With the Control treatment, the upstream sample location was identified as G3 and the downstream sample location was identified as G4. Differences between the upstream and downstream locations were indicative of the impact of the treatment (Riparian, BMP and Control) on stream water quality. For example, the impact of the Riparian treatment was determined by subtracting each respective constituent concentration at sample location G1 from sample location G2 (i.e. downstream minus upstream). A positive difference indicated that the respective constituent concentration increased across the treatment. Hence, references in this article to treatment are indicative of differences between the treatment upstream and downstream locations while references to location are indicative of concentrations or levels. Samples were collected from January 2002 until November 2003 for a total of 47 sampling events. On Camden Creek, samples were collected almost simultaneously at locations G1 and G3 followed by the near simultaneous collection of samples at locations G2 and G4. For Pin Oak, samples were collected almost simultaneously for locations G2 and G3 followed by the near simultaneous collection of samples at locations G1 and G4. The average time between grab sample collections along the same stream was approximately two minutes. The samples predominantly represented base flow conditions, although sampling occasionally occurred during storm events (<5 % of the samples). Samples were immediately placed on ice and transported to the Biosystems and Agricultural Engineering Water Chemistry Laboratory for immediate analysis or storage in accordance with standard methods (Greenberg et al., 1992). The constituents NO 3 -N, NO 2 -N, NH 3 -N, TKN, PO 4 -P, and TP were analyzed using an IRAMA RAS-1000 auto analyzer.

Cattle Positions

Understanding the impact of grazing activity on water quality required knowledge of animal position, especially in relation to streamside areas. 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.

(*Use of trademark does not imply endorsement by the University of Kentucky.)

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 November 2003. A five-minute sample interval, the smallest permitted with the GPS collars, was selected. Data from all of the GPS collars were filtered and differentially corrected allowing use of only the highest quality position points in the analysis (Agouridis et al., 2004).

Prior to the start of the project, a base map identifying key pasture features was created using a Trimble Real Time Kinematic Global Positioning System (RTK-GPS) (5800 RTK rover, MS750 base station) with a published horizontal accuracy of 2 cm. Key pasture features included the streambanks, fences, trees, and all BMPs . This base map was used in conjunction with the data collected from the GPS collars during the seven cattle-monitoring periods to characterize cattle activity along the streambanks. A five-meter buffer from the edge of the streambanks was created in ArcView® for each pasture plot. The fraction of time the cattle spent in or beside each stream (for each treatment) was determined by dividing the total number of GPS points collected (summed for all collars) within five meters of the stream for each treatment (Riparian, BMP Only and Control) and each replication (Camden Creek and Pin Oak) by the respective total number of GPS points (summed for all collars) collected within each treatment and within each replication.

In addition to differences in the number of usable points for each pasture and sampling event, the amount of stream length available to the cattle also differed (Table 1). Therefore, the stream length of each pasture plot was normalized. Finally, the normalized stream length was incorporated into the normalized streamside position data for each stream, respectively. Values greater than one indicated that cattle demonstrated an affinity for the streamside while values less than one indicated that cattle tended to avoid the streamside.

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. 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 90º 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 grab sample collection period (9:00 am until 10:00 am). Since all grab samples along a stream were collected within a five-minute interval, the assumption was made that the flow at each grab sample location was the same for that stream for the sampled period (i.e. flow was not adjusted for the sampling locations upstream of the weir) (Tables 2 and 3).

Statistical Analysis

Nonparametric repeated measures analysis of variance (Friedman one-way repeated measures analysis of variance on ranks) were conducted, using SigmaStat ® , to determine whether treatment significantly affected the concentration differences (i.e. downstream concentration minus upstream concentration) of the examined constituents ( a = 0.05, n=47) (SPSS, 1997). Nonparametric procedures were used instead of parametric procedures because the data were not normally distributed as indicated by the Kolmogorov-Smirnov (with Lilliefor’s correction) test for normality ( a =0.05). Dunn’s method was used for multiple comparisons when the ANOVA indicated a significant treatment effect on constituent differences. This method was selected because the sample sizes in the different treatment groups differed for certain constituent differences. Because no two-factor test exists for non-normally distributed data and the data were not readily transformable (i.e. log 10 , natural log, reciprocal, exponential, square, square root, and arcsin square root), each stream was analyzed separately. Treatment (i.e. pasture) was selected as the factor and each sampling date was treated as a replication. Using a Mann-Whitney rank sum test, the constituent concentrations (i.e. sample locations) and constituent concentration differences (i.e. treatments) between Camden Creek and Pin Oak were examined to better assess the degree of similarity (i.e. effect of differing watersheds) between the two streams over all treatments (n=3) and dates (n=47).

Following the initial assessment, the data set was separated into sampling events that occurred when cattle were present on the pasture (n=27) and sampling events that occurred when cattle were absent from the pastures (n=20). Using SigmaStat ® , nonparametric repeated measures analysis of variance (Friedman one-way repeated measures analysis of variance on ranks) were conducted on each data-subset (i.e. cattle present and cattle absent) to determine 1) whether treatment significantly affected the concentration differences of the examined constituents and 2) whether any significant differences in constituent concentration differences were the likely result of cattle activity. Additionally, the effects of treatment on cattle positions were examined using one-way repeated measures ANOVA in SigmaStat ® (each stream separately examined), and the water quality data from each sample location and for each sampling period were examined for seasonality using Kruskall-Wallis one-way ANOVA (nonparametric) in WQStatPlus ® (NIC, 2003). Seasonality was examined at the a =0.10 level of significance to ensure that weak seasonal trends were identified. However, the effects of seasonality and/or flow were not removed from the data (i.e. all, cattle present and cattle absent).

Results and Discussion

Camden Creek

Results from the nonparametric repeated measures ANOVA conducted on the Camden Creek data indicated that the concentration differences or level differences of four constituents (NH 3 -N, pH, TP, and PO 4 -P) differed significantly with respect to treatment . Subsequent analysis of the data in accordance to the presence (n=27) or absence (n=20) of cattle revealed that NH 3 -N, pH and TP differed significantly with respect to treatment when cattle were present on the pasture plots at the time of sampling while only pH differed significantly with respect to treatment when cattle were absent from the pasture plots at the time of sampling. Seasonal trends were detected for TP and PO 4 -P.

Ammonia Nitrogen

Concentration differences of NH 3 -N were significantly greater for the Control treatment when compared to the Riparian and BMP treatments for the data set comprised of samples collected when cattle were present on the pasture plots. These results seem to indicate that the presence of cattle resulted in the increase in NH 3 -N levels for the Control treatment since this parameter was not significant for the data set comprised of samples collected when the cattle were absent from the pasture plots. Results from the examination of the cattle position data revealed that cattle spent over 8% of the time (approximately 434 hours) either in the stream or within 5 m of the streambanks in the Control treatment (Table 1). This value did not differ significantly from the BMP treatment (8%), but was significantly greater than the Riparian treatment where the cattle spent approximately 2% of the time (approximately 117 hours) either in the stream crossing or within 5 m of the streambanks. However, normalization of the cattle position data by linear foot of available stream length revealed an interesting aspect. When the amount of time that cattle spent within 5 m of the streambanks was adjusted based on the amount of stream length that was accessible to the cattle in each treatment, a significantly greater percentage of the time was spent in the stream crossing of the Riparian treatment than along the streambanks of either the BMP or Riparian treatments. Even so, the cattle did not demonstrate a strong affinity towards Camden Creek. CAST (2002) noted that cattle defecate and urinate more frequently near loafing and watering areas. As such, it would appear that the cattle would contribute a greater localized concentration of waste products to the stream in the Riparian treatment as compared to the BMP and Control treatments, as a result of loitering in the stream crossing. Unfortunately, knowledge regarding the locations and amounts of fecal and urine deposition cannot be gleaned from the cattle position data.

Bicudo et al. (2003) observed that cattle in the Camden Creek Riparian treatment consumed a larger amount of water from the offstream watering source (as much as 87 L/1000 kg LAW-hr) during the morning hours of the months of August and September 2002 when compared to the BMP treatment. Peak rates of water consumption for the Riparian treatment were between 8:00 am and 6:00 pm. The authors hypothesized that the difference in water consumption rates was likely due in part to the presence of shade near the water area. McIlvain and Shoop (1971) also noted the importance of shade in relation to cattle grazing patterns. Following water, these researchers noted that shade was the next primary driving factor for grazing distribution. Examination of the amount of shade available to the cattle revealed that 8.4% of the Riparian treatment pasture area had shade, 7.8% of the BMP treatment pasture area had shade, and only 4.5% of the Control pasture area had shade. Considering the higher levels of activity exhibited by cattle at sunrise and sunset, increased water intake rates associated with grazing, the greater levels of internal heat associated with feed intake, especially on endophyte infected fescue, and the backwater pool created by the weir, the collection of the grab samples may have closely coincided with unseen streamside cattle activity (i.e. cattle were not continuously monitored throughout the study) (Al-Haidary et al., 2001; Albright and Arave, 1997; Blackshaw and Blackshaw, 1994). With less available shade, cattle in the Control treatment may have spent more time cooling in Camden Creek. The increase in NH 3 -N levels across the Control treatment (sample location G4 versus G3) may be indicative of higher levels of ammonification or urease hydrolyzation resulting from an increased in stream or streamside waste supply.

Another potential source of NH 3 -N to the stream was from runoff from the fertilizer plots. Examination of the fertilizer plot areas as a percentage of pasture plot area revealed that similar percentages of pasture plot coverage were present for all three treatments . Additionally, the soils on which the fertilizer plots were established are predominately from the same hydrologic group (C) (Jacobs et al., 1994). Average slopes within these pasture plots are also similar, indicating that the difference in NH 3 -N concentration differences between the Control treatment and Riparian and BMP treatments was probably not due to differences in runoff volumes. Examination of the NH 3 -N sample data for seasonality revealed no trends. Based on these results, it seems likely that the significant increase in NH 3 -N concentration difference for the Control treatment was linked to the increased presence of cattle along the streamside.

pH

The level (i.e. difference) of pH differed significantly between the Control treatment and the BMP treatment (Control>BMP) when cattle were present and it differed significantly between the Control treatment and the Riparian and BMP treatments (Control>BMP, Control>Riparian, BMP=Riparian) when cattle were absent. The significance of pH regardless of cattle presence on the pastures suggests that the controlling factors for pH were external (i.e. groundwater, soils and/or pool by weir). Examination of the soils maps generated by Jacobs et al. (1994) revealed that the predominate soil series in the Riparian treatment is Hagerstown B; Hagerstown B and Woolper B are the predominate soil series for the BMP treatment; and Woolper B is the predominate soil series for the Control treatment. Hagerstown soils are typically neutral (7.3<pH>6.6) to slightly acidic (6.5<pH>6.1) while Woolper soils are typically neutral to mildly alkaline (7.8<pH>7.4). The influence of soils on pH was reflected in the general increase in pH in the downstream direction from sample location G1 to G4. The presence of the backwater pool created by the weir at the downstream edge of the Control pasture plot may also have influenced pH. It is possible that waters in this pooled area exhibited a greater connection to the hyporheic zone, thus impacting pH. However, no data regarding hyporheic flow paths were collected. Tests for seasonality revealed no trends associated with pH.

Total Phosphorus

The concentration difference of TP differed significantly between the BMP and the Control treatments when cattle were present in the pastures but did not differ significantly among the treatments when cattle were absent from the pastures Median concentration differences revealed an increase in TP for the BMP treatment and a decrease in TP for the Control treatment. The geology of the soils coupled with the seasonality of the data prohibited the clear identification of cattle activity as the primary source of TP between the two treatments. For example, the Hagertown soils were derived from the Granular Phosphatic Limestone Member of the Cynthiana Formation, which is noted for its higher phosphorus content. The Woolper soil series along Camden Creek was predominately formed from the Grier Limestone Member, and this soil series is closely associated with the Hagertown soil series. While seasonal trends were noted for three of the sampled locations along Camden Creek (G1, G2 and G3), it was not possible to distinguish if the seasonality was related to flow or to the grazing season (i.e. presence of cattle). The lack of seasonality for sample location G4 seemed to indicate that one of these two factors differed at the location (i.e. change in flow or change in cattle activity in comparison to the other locations). One explanation is that the decrease in median TP concentration differences for the Control treatment as compared to the BMP treatment may have been related to the presence of the weir located at the downstream sampling location, G4, along Camden Creek. The weir created areas of lower velocities, which were conducive to the settling of sediments and particulate matter. Therefore, it is difficult to attribute the relatively higher median TP concentration difference in the BMP treatment (or the relatively lower TP concentration difference in the Control treatment) to cattle activity when all three treatments may have exhibited statistically equivalent medians in the absence of the weir.

Orthophosphate

Median concentration differences of PO 4 -P differed significantly between the BMP and Riparian treatments when all 47 sampling events were examined, but did not display significance when the data set was examined in relation to cattle presence or absence. Orthophosphate seemed to be more strongly linked to flow as high flow events such as those for February 6, 2002, November 13, 2002, February 19, 2003, and September 3, 2003 resulted in higher PO 4 -P levels for all sample locations along Camden Creek. Furthermore, the cyclic decrease in PO 4 -P levels during the later spring to early fall months even in the presence of higher flows and cattle suggests uptake mechanisms were occurring (i.e. plant and/or microbial uptake).

Pin Oak

Results from the nonparametric repeated measures ANOVA conducted on the grab sample data collected from Pin Oak indicated that the concentration differences or level differences of two constituents (NO 3 -N and pH) differed significantly with respect to treatment. Subsequent analysis of the data in accordance to the presence (n=27) or absence (n=20) of cattle revealed that NO 3 -N, pH, TN and FC differed significantly with respect to treatment when cattle were present on the pasture plots at the time of sampling while NO 3 -N and pH differed significantly with respect to treatment when cattle were absent from the pasture plots at the time of sampling. Seasonality was present for NO 3 -N, pH, TN and FC.

Nitrate Nitrogen

The median concentration differences of NO 3 -N differed significantly between the BMP treatment and the Riparian and Control treatments when cattle were both present and absent from the pasture plots with median NO 3 -N levels significantly lower for the BMP treatment. The significance of NO 3 -N regardless of the presence or absence of cattle suggests that factors external to cattle were influential. Reconnaissance of the project area along Pin Oak revealed several interesting points. First, the watershed for Pin Oak receives a significant contribution (20% of the ARC) from row crop production areas. Additionally, dye trace studies conducted by the Kentucky Geologic Survey revealed that sink holes present in the row crop areas are connected to a small tributary that flows into Pin Oak upstream of the study site (fig. 2). Analysis of groundwater data collected quarterly from two groundwater monitoring wells located in the crop production areas revealed that NO 3 -N concentrations ranged from 2.03 mg/L to 9.49 mg/L with a mean of 6.04 mg/L. Further examination of the grab sample data revealed that all sample locations on Pin Oak exhibited seasonality with respect to NO 3 -N. The largest NO 3 -N concentrations occurred during the winter months with steady decreases in concentrations occurring throughout the year. Short periods of increased NO 3 -N levels appeared to be linked to increased flow levels. These increases were likely the result of runoff and increased groundwater contribution from crop land areas. Plant uptake likely resulted in the decline in NO 3 -N concentrations at the sample locations throughout the year.

Regardless of cattle presence or absence, the BMP treatment had the largest decrease in median NO 3 -N concentration difference relative to the Riparian and Control treatments. The reason for this treatment effect was not clear, but a subsurface linkage may have existed. Along the downstream edge of the BMP treatment and within the Control treatment, several groundwater seeps exit directly into Pin Oak via pipes.

pH

The level of pH (i.e. difference) was significantly greater for Control treatment than the Riparian and BMP treatments when cattle were both present and absent. The significance of pH regardless of cattle presence on the pastures suggests that, similar to NO 3 -N, the controlling factors for pH were external (i.e. groundwater, soils and/or weir). Examination of the soils maps generated by Jacobs et al. (1994) revealed that the predominate soil series in the Riparian treatment is McAfee D with some McAfee C, Maury B and Maury C, McAfee D is the predominate soil series for the BMP treatment, and McAfee C and McAfee D are the predominate soil series for the Control treatment. McAfee soils are soils are typically neutral (7.3<pH>6.6) to slightly acidic (6.5<pH>6.1) while Maury soils are characterized as neutral for the top 60 cm, progressing to strongly acidic (5.5<pH<5.1) at a depth of 250 cm. For the locations near Pin Oak, depths to bedrock typically remained less than 60 cm.

The influence of soils on pH was reflected in the general increase in pH in the downstream direction from sample location G1 to G4. Overall, the median differences in pH levels increased in the downstream direction from the Riparian treatment to the Control treatment. Similarly, the general soil types changed from those with a more acidic reaction to those with a more neutral reaction. Tests for seasonality revealed seasonal trends associated with pH levels at locations G1 and G2. Further examination revealed that increases in stream flow were linked to reductions in pH levels at these two sampling locations. It is possible that waters in the pooled area (behind the weir located between sample locations G1 and G2) exhibited a greater connection to the hyporheic zone, thus impacting pH.

Total Nitrogen

The concentration differences of TN were significantly greater for the Control treatment than the Riparian treatment when cattle were present in the pasture plots. No significant differences were detected between the treatments in the absence of cattle. A reduction in the median TN concentration difference occurred in the Riparian treatment while an increase in the median TN level difference was noted for the Control treatment. Tests for seasonality indicated that seasonal trends were present for TN levels at all four sampling locations along Pin Oak. Concentration versus time plots of TN closely resembled those of NO 3 -N indicating that crop production significantly impacted nitrogen levels in Pin Oak.

Fecal Coliform Bacteria

The median concentration difference of FC differed significantly between the Riparian and the Control treatments when cattle were present in the pastures, but no significant differences were detected when cattle were absent from the pastures. Surprisingly, the Riparian treatment had an increase in FC level differences while the Control treatment had a decrease in FC level differences. Examination of the cattle position data revealed that for the seven GPS monitored periods, cattle spent significantly less time (9% of the total time) in the stream or within 5 m of the streamside in the Riparian treatment as compared to the BMP and Control treatments (Table 1). However, when the position data was normalized by linear feet of available stream length, the opposite trend was apparent. Within the Riparian treatment, cattle demonstrated a strong affinity towards the loitering in or drinking from the stream at the crossing. The stream crossing, as constructed on the bedrock bottom channel, provided the cattle with an ideal location for seeking relief from the heat and humidity. Cattle were observed loitering in the stream crossing for hours at a time. Photographic documentation of the stream crossing indicated that cattle frequently defecated in the stream crossing, thus readily introducing fecal coliforms into the stream. According to Soto-Grajales (2002), nearly 95% of microbes entering streams settle in the bottom sediments within 50 m of their entry point.

The decrease in FC level differences for the Control treatment was likely the result of settling due to the backwaters of the weir rather than decreased FC inputs into the stream. Analysis of the cattle position data demonstrated no significant difference between the BMP and Control treatments with regards to the time the cattle spent in the stream or within 5 m of the streamside. Even though a reduction in the FC difference for the Control treatment was the likely result of the pooled area created by the weir, it is not recommended to install similar structures (i.e. check dams) in streams. The installation of such structures impacts other aspects of the stream such as its ability to transport sediment (i.e. stability) and its function as habitat. Reducing the streams ability to transport sediment while maintaining flow conditions can result in channel incision (Werritty, 1997) and habitat degradation (Giller and Malmqvist, 1998).

Comparison of Streams

A comparison of median constituent concentrations (i.e. locations) for Camden Creek and Pin Oak revealed that the streams differed significantly for 10 constituents, accounting for over 70% of the constituents sampled. For nine constituents (approximately 64%), Pin Oak had higher median concentrations, mainly with nitrogen and phosphorus species. Elevated levels of nutrients in Pin Oak were a reflection of the watershed land uses (i.e. row crop production). Along Camden Creek, higher levels of fecal coliform contamination, organics and solids were a reflection of the upstream land use within that watershed (i.e. grazing). However, when the median constituent concentration differences were examined (i.e. treatments), only pH differed significantly between the streams. These results indicated that, although many constituent concentrations were significantly different between Camden Creek and Pin Oak with regard to sample location, changes in the constituent concentrations (i.e. differences) between the respective upstream and downstream sample locations (i.e. treatment effects) were quite similar between the two streams.

Conclusions

Biweekly grab samples were collected from two bedrock, second order perennial streams to assess the effects of two BMP systems (each consisting of an alternate water source, alternate shade sources, and pasture improvements while one system also had a 9.1 m wide riparian zone equipped with a 3.7 m wide stream crossing) on water quality. Grab samples were analyzed for NO 3 -N, NO 2 -N, NH 3 -N, TKN, Org-N, TN, PO 4 -P, TP, pH, COD, BOD 5 , TSS, FC and FS. Using GPS collars, streamside cattle activity was recorded for seven, 18-day periods with a five-minute sample interval and was normalized based on linear feet of available stream length. Stream discharge data were collected at the most downstream edge of each replication (i.e. Camden Creek and Pin Oak) at 10-minute intervals.

Results from the analyses of the Camden Creek dataset indicated that treatment significantly affected concentration differences or level differences of NH 3 -N, TP and pH when cattle were present and only pH levels when cattle were absent. While cattle activity seemed to influence NH 3 -N concentration differences, it was not possible to directly link TP responses to cattle activity alone as these values demonstrated some dependence on flow. Interestingly, these results (i.e. significance when cattle present but not when cattle absent) highlight the fairly rapid response and recovery time exhibited by Camden Creek. Sarr (2002) identified such streams as following a “rubber band model” characterized by quick recovery time with minimal hysteresis. As for Pin Oak, treatment significantly affected concentration differences or level differences for NO 3 -N, pH, TN and FC. Similar to Camden Creek, the majority of these significant constituents were linked to seasonality (i.e. stream discharge and plant uptake mechanisms) and upstream land uses (i.e. NO 3 -N and row crop production). Cattle activity seemed to cause the increase in FC differences across the Riparian treatment as cattle demonstrated a strong affinity towards loitering in or drinking from the stream at the crossing. And while FC decreases were seen across the Control treatment likely as a result of the weir, the installation of check dams is not recommended.

One of the most substantial understandings gleaned from the project was the importance of upstream land use, and to some degree soils, when attempting to identify significant treatment effects within a small reach. Additionally, the karst geology, which is characteristic of the Bluegrass Region of Kentucky, influenced the rate of transport (i.e. flashy system with quick response time to rainfall) of nutrients from upland land areas (i.e. row crops), especially along Pin Oak. These external factors may have resulted in the lack of uniformity in significant constituent concentration differences between the two streams when cattle were present. Furthermore, the background constituent concentration levels may have prohibited the identification of treatment effects. Results from this project indicated that minimal water quality benefits were incurred by implementing a BMP system (with or without a partially excluded riparian zone). However, these results may differ if cattle were completely excluded from the stream (i.e. no stream crossing) or if this project were conducted on a site with larger pastures, different geology (non karst), soils (not high in P) or stream geomorphology (non bedrock bottom channel).

References

Agouridis, C.T., T.S. Stombaugh, S.R. Workman, B.K. Koostra, and D.R. Edwards. 2004. Suitability of GPS for Grazing Studies. Trans ASAE 47(4): 1-9.

Albright, J.L. and C.W. Arave. 1997. The Behavior of Cattle. Chapter 5: Feeding Behavior. CAB International, New York.

Al-Haidary, A., D.E. Spiers, G.E. Rottinghaus, G.B. Garner, and M.R. Ellersieck. 2001. Thermoregulatory Ability of Beef Heifers Following Intake of Endophyte-Infected Tall Fescue During Controlled Heat Challenge. J. Anim. Sci. 79(7): 1780-1788.

Belsky, A.J., A. Matzke, and S. Uselman. 1999. Survey of Livestock Influences on Stream and Riparian Ecosystems in the Western United States. J. Soil and Water Conserv.. 54(1): 419-431.

Bicudo, J.R., C.T. Agouridis, R.S. Gates, S.R. Workman and E.S. Vanzant. 2003. Effects of Air and Water Temperature and Stream Access on Grazing Cattle Water Intake Rates. Paper No. 03-4034. ASAE International Meeting, Las Vegas, NV.

Blackshaw, J.K. and A.W. Blackshaw. 1994. Heat Stress in Cattle and the Effect of Shade on Production and Behaviour: A Review. Australian J. Exp. Agric. 34(2): 285-295.

Clark, E.A. 1998. Landscape Variables Affecting Livestock Impacts on Water Quality in the Humid Temperate Zone. Canadian J. Plant Science 78(2): 181-190.

Council for Agricultural Science and Technology (CAST). 2002. Environmental Impacts of Livestock on U.S. Grazing Lands. Issue Paper 22. Ames, Iowa.

Daniel, T.C., A.N. Sharpley and J.L. Lemunyon. 1998. Agricultural Phosphorus and Eutrophication: A Symposium Overview. J. Environ. Qual. 27: 251-257.

Fogle, A.W. 1998. Impact of Topographic Data Resolution on Hydrologic and Nonpoint-Source Pollution Modeling in a Karst Terrain. Report of Investigations 13. Kentucky Geological Survey, Lexington, Kentucky.

Giller, P.S. and B. Malmqvist. 1998. The Biology of Streams and Rivers. Oxford University Press, New York.

Jacobs, S., R.D. Jones, and A.D. Karathanasis. 1994. Soil Survey of (Pin Oak) University of Kentucky Woodford County Research Farm.

Kentucky Agricultural Statistics Service (KASS). 2003. Kentucky Agricultural Statistics 2002-2003. [On-line]. Available at http://www.kyagr.com

Line, D.E., W.A. Harman, G.D. Jennings, E.J. Thompson, and D.L. Osmond. 2000. Nonpoint-Source Pollutant Load Reductions Associated with Livestock Exclusion. J. Environ. Quality 29(6): 1882-1890.

Logan, T.J. 1990. Agricultural Best Management Practices and Groundwater Protection. J. Soil and Water Conserv. 45(2): 201-206.

McIlvain, E.H. and M.C. Shoop. 1971. Shade for Improving Cattle Gains and Rangeland Use. J. Range Management 24(3): 181-184.

NIC Environmental Division. 2003. WQStatPlus. Indianapolis, IN.

Sarr, D.A. 2002. Riparian Livestock Exclosure Research in the Western United States: A Critique and Some Recommendations. Environ. Manage. 30(4): 516-526.

Sheffield, R.E., S. Monstaghimi, D.H. Vaughan, E.R. Collins, Jr., and V.G. Allen. 1997. Off Stream Water Sources for Grazing Cattle as a Stream Bank Stabilization and Water Quality BMP. Trans. ASAE 40(3): 595-604.

Soto-Grajales, N. 2002. Livestock Grazing and Riparian Areas in the Northeast. USDA.

SPSS. 1997. SigmaStat® Statistical Software User’s Manual. SPSS, Inc., Chicago, IL.

Turner, L.W., M.C. Udal, B.T. Larson, and S.A. Shearer. 2000. Monitoing Cattle Behavior and Pasture Use with GPS and GIS. Canadian Journal of Animal Science . 80(3): 405-413.

United States Department of Agriculture-National Agricultural Statistics Service (USDA-NASS ). 1997. Census of Agriculture. Washington, D.C.

United States Environmental Protection Agency (U.S. EPA). 2000. National Water Quality Inventory Report to Congress (305b Report): 2000 Water Quality Report. [On-line]. Available at http://www.epa.gov.

University of Kentucky Agricultural Weather Center. 2004. Climatology. [On-line]. Available at http://wwwagwx.ca.uky.edu/climdata.html.

Werritty, A. 1997. Short Term Changes in Channel Stability. Applied Fluvial Geomorphology for River Engineering and Management. Eds. C.R. Thorne, R.D. Hey, and M.D. Newsom. John Wiley & Sons, West Sussex.

Table 1. Streamside GPS Collar Points Collected Over Seven, 18-Day Periods.

Treatment

Camden Creek

Pin Oak

%Streamside*

GPS Collar Points

Normalized Streamside GPS Collar Points

%Streamside GPS Collar Points

Normalized Streamside GPS Collar Points

Riparian

2.0

0.41**

9.0

4.08

BMP Only

8.0

0.80

15.2

0.15

Control

8.3

0.83

13.7

0.14

* GPS collar points within five meters of the streambanks.

** Values greater than one indicate that the cattle demonstrated an affinity for the

streamside.

Table 2. Stream Discharge (m 3 /s) for Camden Creek for the Water Quality Sampling Periods.

Date

Stream Discharge *

January 9, 2002

0.0362

January 23, 2002

0.0835

February 6, 2002

0.1250

February 20, 2002

0.0498

March 6, 2202

0.0360

March 20, 2002

-- **

April 3, 2002

0.3110

April 17, 2002

-- **

May 1, 2002

-- **

May 15, 2002

0.2800

May 29, 2002

0.0599

June 12, 2002

0.0416

June 26, 2002

0.0156

July 10, 2002

0.0060

July 24, 2002

0.0014

August 7, 2002

0.0009

August 21, 2002

0.0009

September 4, 2002

0.0003

September 18, 2002

0.0003

October 2, 2002

0.0057

October 16, 2002

0.0697

October 30, 2002

0.4090

November 13, 2002

0.2450

December 4, 2002

0.0320

December 11, 2002

0.1170

January 8, 2003

0.1198

January 22, 2003

0.0348

February 5, 2003

0.0807

February 19, 2003

0.1719

March 5, 2003

0.1127

March 19, 2003

0.0515

April 2, 2003

0.0807

April 16, 2003

0.0705

April 30, 2003

0.0501

May 14, 2003

0.2169

May 28, 2003

0.1062

June 11, 2003

0.1464

June 25, 2003

0.0651

July 9, 2003

0.0204

Table 2. (continued).

Date

Stream Discharge *

July 23, 2003

0.0190

August 6, 2003

0.0127

August 20, 2003

0.0068

September 3, 2003

0.2611

September 17, 2003

0.0062

October 1, 2003

0.0379

October 15, 2003

0.0136

October 29, 2003

0.0227

* Average values computed from 10-minute interval discharge data collected on the indicated days during the grab sample collection periods (9:00 am until 10:00 am).

** No data recorded. Instrumentation malfunctioned.

Table 3. Stream Discharge (m 3 /s) for Pin Oak for the Water Quality Sampling Periods.

Date

Stream Discharge *

January 9, 2002

0.0286

January 23, 2002

0.0340

February 6, 2002

0.0929

February 20, 2002

0.0286

March 6, 2202

0.0144

March 20, 2002

1.0100

April 3, 2002

0.3210

April 17, 2002

0.1080

May 1, 2002

0.3260

May 15, 2002

0.2370

May 29, 2002

0.0515

June 12, 2002

0.0442

June 26, 2002

0.0122

July 10, 2002

0.0119

July 24, 2002

0.0060

August 7, 2002

0.0031

August 21, 2002

0.5660

September 4, 2002

0.0031

September 18, 2002

0.0060

October 2, 2002

0.0142

October 16, 2002

0.0453

October 30, 2002

0.1420

November 13, 2002

0.2200

December 4, 2002

0.0227

December 11, 2002

0.0382

January 8, 2003

0.0824

January 22, 2003

0.0289

February 5, 2003

0.0453

February 19, 2003

0.4378

March 5, 2003

0.2138

March 19, 2003

0.0278

April 2, 2003

0.0513

April 16, 2003

0.0447

April 30, 2003

0.0351

May 14, 2003

0.2039

May 28, 2003

0.1311

June 11, 2003

0.1110

June 25, 2003

0.0470

July 9, 2003

0.0289

July 23, 2003

0.0244

August 6, 2003

0.0224

Table 3. (continued).

Date

Stream Discharge *

August 20, 2003

0.0096

September 3, 2003

0.0779

September 17, 2003

0.0068

October 1, 2003

0.0198

October 15, 2003

0.0176

October 29, 2003

0.0286

* Average values computed from 10-minute interval discharge data collected on the indicated days during the grab sample collection periods (9:00 am until 10:00 am).

Figure 1 is not available in html, see pdf for complete version.

Figure 1. Base Map of Pasture Plots. Plot 1 (Riparian) treatment is the BMPs and a fenced riparian area to exclude cattle from the stream except at a 3.7 m wide crossing; plot 2 (BMP Only) treatment is BMPs with free stream access; and plots3 (Control) are free access with no BMPs (control) except for herbicide, fertilizer, and alternate shade that was added in the last few months of the study.

Figure 2 is not available in html, see pdf for complete version.

Figure 2. Karst Geology of ARC. Plot 1 is the Riparian treatment, plot 2 is the BMP treatment, and plot 3 is the Control treatment.


Copyright © 2008 American Society of Agricultural and Biological Engineers. All rights reserved.