Dairy farming in the United States has become increasingly capital-intensive, using management schemes that confine large herds of highly productive dairy cows on a small part of the farm, while practicing high-input crop production on most of the land. Large machines harvest the crops and bring them to the cows while other machines haul the cow manure out to the fields to fertilize the soil. This “confinement feeding” system has become the norm on dairy farms in the United States, requiring farmers to invest heavily in machinery and facilities needed for high-input crop production, animal housing, and for manure handling and storage. Herd size increases to support the capital expenses involved with confinement feeding management, and when farmers milk more cows than can be maintained with the feed produced on the farm itself, additional feed must be purchased and imported onto the farm to maintain the herd. The imported feed brings with it large amounts of nitrogen (N) and phosphorus (P). Cows excreted 60-75% of the nutrients ingested, although efforts to fine-tune animal diets may somewhat reduce the excretion of N and P. Eventually the herd supplies more manure N and P than the farm fields can properly assimilate. The resulting accumulation of N and P increases the potential for nutrient loss that can lead to pollution of groundwater, nearby streams and eventually the Chesapeake Bay.
Some farmers are now using a very different approach to dairy production called management intensive grazing (MIG), a grass-based farming system. Under MIG, the farm’s fields are primarily grassed pastures subdivided into paddocks. To best utilize available forage, the herd is moved to a fresh paddock once or twice a day. Grazed forage is the primary source of protein and energy for the cows, eliminating the need for feed crop production and its expensive, energy-demanding infrastructure. Because MIG farms do not require the expensive infrastructure of confined feeding, herd size can remain closer to that which can be sustained by the farm itself, while still yielding profits that support the farmer and family. The smaller herd size means much less feed is imported onto MIG farms, which can reduce the potential for pollution from excess nutrient accumulation
Previous research on intensive grazing in the NE USA and elsewhere has used high levels of N fertilizer on ryegrass paddocks and soil column lysimeters or shallow drain tiles to measure nitrate leaching to groundwater. This research led policy makers to have concerns that intensive grazing would result in unacceptably high nitrate leaching, mainly because of uneven urine distribution by cows. Such excessive N leaching would make MIG environmentally untenable. However, we hypothesized that the lysimeters did not accurately represent the hydrology of the natural pasture system. We also hypothesized that the low fertilizer approach to MIG used by graziers in the mid-Atlantic region would be less subject to N leaching that the high fertilizer approach studied in Europe and New Zealand where excessive groundwater nitrates were reported. We developed this project to determine the actual environmental effects of MIG in actively-grazed pastures.
The project set out to measure the water quality effects of dairy farming on three well-managed farms in central Maryland – two MIG farms, and one confined feeding farm. A specific objective was to determine if MIG farms actually produced the high nitrate levels (averaging 15 to 35 mg/L) in groundwater predicted by earlier studies. To avoid the unrealistic conditions and methodological artifacts that we believe exaggerated N leaching in previous studies, we used piezometers (monitoring wells) to study N and P in actual groundwater under six dairy farm watersheds. Four of the watersheds were characterized by management intensive grazing (MIG) pastures and two by manured cropland associated confinement-feeding dairying. Because in two watersheds, a perennial streams bisected the grazed land, we were also able to collect data on stream water contamination on one grazing-based dairy farm during both base flow and storm flow events. In addition, we studied the farm scale nutrient balance and economic costs and returns for the three commercial dairy farms on which these watersheds were located.
Through workshop sessions, conferences, pasture walks and publications, we attempted to inform extension personnel, policy makers and regulators, as well as dairy farmers, of the benefits of MIG systems and the water quality found on the study farms. As a result, regulators and farmers in the mid-Atlantic are interested in promoting MIG as an alternative to confined dairy farm management and as environmental Best Mangement Practice.
Project LNE-04-213 was a continuation of project LNE- 01-152. This final report summarizes both of these projects as they were really two phases of the same endeavor. Project LNE-04-213 made possible the collection of a critical third year of data on the environmental performance of the three study farms in central Maryland (one confined feeding-based and two grazing-based farms). The project(s) used piezometer wells and stream sampling to measure nutrient losses in two watersheds on each of the three farms. The project results were communicated to small and medium-sized dairy farmers through presentations at workshops in Maryland and Pennsylvania and a farmer-authored fact sheet that has been published and is being distributed. The results were presented to regulators (State Depts. of Agriculture and/or Depts. of Environment or Natural Resources) and extension agents with the aim of providing sufficient information on nutrient issues to allow appropriate state programs to be developed for MIG systems.
Management intensive grazing (MIG) is closely associated with an increase in both profitability and life-style quality for dairy farmers who successfully switch from the confined-feeding system to the grass-based production system. Furthermore, because MIG systems improve profitability without increasing milk production either per farm or per cow, they offer an alternative to the higher production-lower price treadmill that dairy farming has been on for decades. The literature also suggests that adoption of the MIG system allows farmers to better conserve their soil resources. Few practices improve soil quality and protect against soil erosion as well as permanent grass vegetation.
However, research in several locations, including the Northeast, has questioned the environmental impact of some MIG systems with regard to their potential losses of nitrogen and phosphorus to groundwater. Such questions about environmental impacts could serve as a major roadblock to acceptance of MIG systems as best management practices (BMPs) by regulatory agencies and thus, a roadblock to the adoption of such systems by more dairy farmers. This project set out to measure the water quality and calculate the nutrient balance on three well-managed dairy farms in central Maryland – two MIG farms, and one confined feeding farm. A specific objective was to determine if MIG farms actually produced the high nitrate levels in groundwater predicted by earlier studies. The project collected data on the environmental and economic performance of these three farms. The project used piezometer wells and stream sampling to measure year-round nutrient losses by leaching in two watersheds on each of the three farms. The project also focused on the economic impacts of MIG by conducting a cost-returns analysis of each farm.
The results of three years of monitoring showed MIG watershed had relatively low levels of nutrient pollution, especially nitrate levels, which had been considered the greatest risk from intensive grazing systems. Nitrate concentrations on the four grazed watersheds averaged from 4-7 mg L-1. Only two of the four grazed watersheds exceeded the EPA maximum contaminant load (10 mg NO3-N L-1) during two of the 13 seasonal sampling periods of the study. Those two seasons were during periods of rapid groundwater recharge and heavier than normal precipitation. In the two Confined watersheds, nitrate concentrations averaged from 6.8-10.9 mg L-1. One of the confined watersheds had nitrate concentrations that were not significantly below the EPA maximum contaminant load during 8 of the 13 seasonal sampling periods, and the second confined watershed failed to stay below this limit during two of the sampling periods. These high nitrate concentrations occurred even though the cropped land on the confined feeding dairy farm was being managed under nutrient management guidelines.
Phosphorus (P), not usually considered at risk for leaching to groundwater, was found in higher than expected levels in all six of the sampled watersheds, but the levels were not related to current management. Concentrations ranged from 0.034-0.233 mg Dissolved Reactive P L-1 under the six watersheds. The most striking difference was between the two Frederick County farms and the Baltimore County farm. Groundwater DRP was similar on the neighboring confined and grazed Frederick County farms, but was much lower on the Baltimore County farm, possibly due to sorption by calcareous parent material. Management did not appear to affect P concentrations, but P was present at very high levels.
Furthermore, two streams were monitored and showed no increase of N or P during base flow after entering and flowing though the MIG watersheds. Nitrate levels were low, and decreased during storm flow. The dissolved reactive P levels were in excess of the EPA limit (0.01 mg DRP L-1) at 0.025-0.029 mg DRP L-1during base flow, but total dissolved P was within the EPA limit. This suggests that the grazed watersheds did not contribute significant quantities of nutrients from subsurface flow. Phosphorus concentrations did increase during storm events with high runoff in one of the two streams after flowing past a cow congregation area in one of the paddocks alongside the stream. Management of this area has since been changed to ameliorate the situation.
An additional observation of the study was that a substantial portion of the nutrient losses were in organic form, meaning that a sizeable portion of the nutrients present in groundwater under animal-based systems may be in the previously unnoticed organic fractions. Dissolved organic N made up between 12-27% of the total dissolved nitrogen in groundwater. Likewise, dissolved organic P, which is less likely than inorganic P to be held in the soil profile, was measured at 25-45% of the total dissolved P present in groundwater. Management did not affect the proportion of organic N or P present in groundwater, with similar proportions under both confined and grass-based systems.
The results of the project have been communicated to small to medium-sized dairy farmers through the use of presentations at workshops and meetings in Maryland, Virginia and Pennsylvania and a farmer-authored booklet that has been published and is being distributed. More than 300 copies have been used by farmers in Vermont alone.
Research results have been presented to the regulators (State Depts. of Agriculture and/or Depts. of Environment or Natural Resources), extension agents and pasture scientists. With the positive environmental findings and outreach efforts, the project has substantially contributed to reducing environmental roadblocks to the development of appropriate state programs to encourage MIG systems. The Maryland Department of Agriculture now is promoting the use of grazing systems. The Center for Sustainable Agriculture and Vermont Pasture Network is also involved in promoting grazing, and is using work from this study as part of its initiative. In this way, the project has contributed to improving the diversity, profitability and environmental impacts of dairy farming in the Northeast.
1. Maryland and Pennsylvania nutrient management regulators (state policy makers, state and private nutrient management advisors, extension agents, and conservation district personnel) that learned about the environmental and economic impacts of grazing from this project will promote grazing under certain conditions as a sustainable agricultural practice that will contribute to their state’s nutrient management goals.
This performance target has been substantially met. We have obtained credible environmental data showing that intensive grazing dairy systems do not pose greater water quality threats than conventional confined feeding, manure mangement systems. In fact the results suggest that grazing will have a lower negative impact than manured fields. These results have been presented and discussed in numerous forums and are being used by regulators to justify promotion of grazing and by modelers to improve the Chesapeake Bay water pollution models which guide environmental policies.
2. Forty of the confinement-feeding dairy farmers in Maryland and Pennsylvania who learned about the environmental and economic impacts of grazing from this project will take steps to switch to grazing
This performance target was, in hindsight, probably unrealistic for this particular project. To be sure, increasing numbers of dairy farmers are seriously considering the grazing alternative, and our project activities probably contributed to that trend. Because the project needed three years of difficult and intensive field sampling and another year of lab analyses before it could offer environmental assurances, the direct impact on dairy farmers was limited and is just now being felt. However, our economic analysis of the profitability of the three farms came to conclusions similar to those from other more extensive economic studies and added evidence for the desirability of considering the grazing option.
A major and unique product of this project, the farmer-authored, highly readable and attractive booklet “Making the Switch” was published in the last 6 months of the project. It has proved very popular, and its influence is being felt. Requests for many hundreds of copies are still coming in from extension agents, farm groups. and private concerns in the region.
Three dairy farms in Maryland were selected for this study. All three farms, herein designated Grazed 1, Grazed 2 and Confined, have been specialized dairy operations for at least 30 years and have included livestock for at least 100 years prior to the study. Grazed 1 (83 ha, 105 milkers) and Grazed 2 (71 ha, 150 milkers) began using a MIG system in 1995 and 1994, respectively. At the commencement of this study, both Grazed 1 and Grazed 2 had been managed as MIG farms for six and seven years, respectively. The Confined farm (245 ha, 400 milkers) uses conventional feeding system and has used no-till management as a soil conservation practice on its cropland since 1962. The 2002-2003 average animal units (454 kg or 1000 lbs) per ha on the three farms were 0.95, 2.2, 2.1.
The grazing farms were selected from a very small pool of existing MIG farmers in Maryland. The farms chosen each had two pasture watersheds suitable for groundwater monitoring and were two of the earliest adopters of MIG in Maryland. They had been using MIG for long enough for their soils to have approached steady state conditions and for the farmers to have developed successful MIG management systems. The confined farm was chosen because of its location adjacent to one of the MIG farms, with similar soils and topography. The Confined farm has a long history of collaborating with Maryland Cooperative Extension and USDA Natural Resource Conservation Service personnel and is an early adopter of conservation tillage practices.
The farms were selected with the help of local agriculture extension agents who identified farmers with reputations as good managers and good land stewards and who would be willing to participate in the study. The farmers joining the study allowed us to install nests of piezometers in their fields or pastures, shared their financial and nutrient management records for economic and nutrient balance analyses. For twelve months of the study, the three farmers collected bulk-tank milk samples which were analyzed for milk urea nitrogen (MUN), a predictor of cow N excretion (Jonker et al., 1998). Levels on all three farms averaged 13.6 to 14. 6 mg MUN/dl, within range of 10-16 mg /dl considered optimal for efficient N utilization (Jonker et al., 1998).
Grazed 1 and Confined occupy adjacent
tracts of land in Frederick County, Maryland, in the lowland section of the Piedmont Plateau physiographic province, where the average precipitation is 1026 mm, and average annual temperature is 13°C. Soils on Grazed 1 are primarily Fauquier silt loams and Myersville silt loams. The soils on Confined are Fauquier silt loams and Highland silt loams. Grazed 2 occupies land in eastern Baltimore County, Maryland, in the upland section of the Piedmont Plateau physiographic province, where the average precipitation is 1039 mm, and average annual temperature is 13°C. Its soils are mainly Glenville loams, overlaying a Cockeysville marble parent material (personal communication, Robert Shedlock, U.S. Geological Survey, June 3, 2003).
Grazed 1 and Grazed 2 provide the dietary energy needs of the herd primarily through grazed forage. On these farms, hay is made to store excess forage for winter feed. Additional hay may be purchased when on-farm hay production is insufficient to support the herd through the winter. Relatively small quantities of purchased supplemental grain are imported to the farm. In 2001, the Grazed 2 lactating cows were fed. Grazed 1 cows have been supplemented at 3.6 kg cow-1 day-1 (roughly 1% of body mass) since 1999. Grazed 2 cows were fed supplemental grain at 3.6 – 6.8 kg cow-1 day-1 during 2001 -2002 and 3.6 kg cow-1 day-1 from 2003 onward.
Confined produces crops in a six-year corn-corn-oats-alfalfa- alfalfa- alfalfa rotation. Additional feed and bedding is purchased to support the herd. Manure from the herd is applied in liquid form to cropland on the farm. Supplemental fertilizer, at 56 kg N ha-1, is used when needed, mostly on the corn fields.
Within each farm, two watersheds, identified as A and B, were selected for groundwater monitoring using piezometers. The watersheds were chosen because the majority of the land within each watershed was under the management of the farmer, and topography suggested that the groundwater would be within the reach (8 m) of the drilling equipment available to the project. In each pair of watersheds from a given farm, one watershed was determined to be the ‘homestead’ watershed, which historically (100+ years) received a greater proportion of nutrients because of its close proximity or convenience to the barn and homestead. The three homestead watersheds were Confined A, Grazed 1A and Grazed 2B. In four of the watersheds, 100% of the land is under the management of the farmer and surrounding land use is primarily low-density single-family houses, with less than 1 house per hectare. One control piezometer was installed on each farm upslope of the farm management activities, to measure the baseline concentration levels coming onto the farm.
A tipping-bucket rain gauge (Spectrum Technologies; Plainfield, IL) was installed within or near each watershed. The rain gauge closest to the farmer’s house or barn was supplied with a digital display that could be viewed and recorded by the farmer, and the rain gauge in the watershed farther from the homestead was connected to a downloadable datalogger (HOBO® Shuttle and event logger; Onset Computer Corporation; Pocasset, MA). Recording by the farmers was not consistent, and was influenced by their work schedules. If rain data was available from the neighboring farm, it was used if one farm did not have rain data for a period of time. In the rare cases that the recording monitor or rain gauge was not operative and the farmer’s records were not complete, National Oceanic and Atmospheric Administration records from the nearest station were used.
In each watershed, a transect of nested piezometers was installed consisting of three nests spaced 18 m apart, starting at the watershed discharge point and following the flow line upslope. A hinged, slatted wood box constructed from 0.9m x 0.9mx 15 cm high was installed over each nest to protect the tops of the piezometers from equipment and grazing cows.
Three piezometers initially were installed within each nest. The piezometers were made of 5 cm inner diameter polyvinylchloride pipe. The deepest 1 m of each piezometer was slotted. The piezometers were installed in the spring of 2001, when groundwater levels were beginning to drop. The shallowest of the three piezometers was installed to a depth where it could just reach the groundwater at the time of installation, with the next two piezometers installed 1 and 2 m deeper. A fabric filter sock (Drain-Sleeve ® Fabric Sock; Carriff Corporation, Inc.; Midland, NC) was fitted over the 1-m of slotting at the bottom of each piezometer and taped in place. Clean sand was poured into the installation hole around the piezometer. A plug of bentonite powder (Wyoming Bentonite for Water Well & Geotechnical Sealing; Drillers Service, Inc.; Hickory, NC) was used around the upper 30 cm of each piezometer to prevent flow around the wall of the piezometer or down the installation hole.
Extreme drought conditions in 2001 caused groundwater levels to drop below the reach of the deepest piezometers in two of the watersheds. Because this lack of recharge during winter and spring 2002 suggested that future sampling might also be limited, a fourth, 1 m deeper piezometer was added in each of the affected nests in October 2002. Piezometers ranged 1.07 to 6.69 meters deep.
Groundwater samples were collected biweekly, beginning in May 2001. Prior to sampling, the depth to groundwater in each piezometer was measured using a water depth indicator. The shallowest piezometer in each nest containing at least one meter of groundwater was bailed and allowed to recharge for approximately two hours. Samples of 120-150 mL were then taken using a ball valve bailer. The pH of each sample was measured, and the sample was acidified to pH<3 with 2-3 drops of 4M H2SO4. The samples were returned to the lab on ice, where they were stored under refrigeration at 4°C until analysis.
Soil pore water sampling
Within each nest box, two ceramic-tipped suction cup (100 kPa bubbling pressure) lysimeters were installed at a 45° angle to the surface of the ground, using a drop-hammer device to make the pilot hole in soil. The lysimeters, one 90-cm and the other 120-cm long, were installed with the tops of the lysimeters within the wooden nest box, and the ceramic suction cup end extending below ground into surrounding pasture or field, so as to place the ceramic tip 60 or 90 cm below the soil surface and at least 30 cm outside of the box.
Whenever soil moisture allowed, soil macropore water samples were collected using ceramic-tipped suction-cup lysimeters at the time of groundwater sampling. This occurred eleven times on each farm between June 2002 to June 2004, mainly in the fall and spring months for a total of 26 sampling dates. To collect the samples, a suction of 70 to 80 kPa was pulled on the lysimeters, using a hand vacuum pump (Irrometer, Inc.; Riverside, CA), and the lysimeter tubes clamped off to hold this vacuum. The sample was collected two to four hours later by drawing it into an Erlenmeyer flask and then pouring into a sealable plastic vial. The Erlenmeyer flask was rinsed with distilled water in between the collection of samples. The lysimeter water samples were acidified with of 4M H2SO4, with one drop added for every 30-50 ml of sample. They were transported on ice to the lab, where they were refrigerated until analysis.
A second order stream flows through each of the two pasture watersheds on Grazer 2. Five sites along each stream were identified for regular sampling, with the site 1 located at the point of the stream’s entrance onto the farm, and each consecutive site being approximately 100 m downstream. Stream A meanders through the watershed, has grassy riparian vegetation and two stone-stabilized stream crossings for the herd. Before Stream A enters the watershed, it flows through a wooded area. Stream B was channelized more than 50 years ago, and has brushy riparian vegetation with a few trees. The herd typically has no access to Stream B, with the exception of once or twice a year along sampling sites 4 and 5. Upstream from sampling site 1, before Stream B enters the farm, it flows through a conventionally tilled confined feeding dairy farm which was not part of this study.
Grab samples were collected biweekly from May 2001 through June 2004. On 11 occasions the stream was in storm flow during and immediately following heavy rainfall. Sample pH was measured in the field immediately upon collection, and, beginning in January 2002, each sample was acidified to pH<3 with 2-3 drops of 4 M H2SO4. The samples were transported on ice to the lab where they were stored under refrigeration at 4°C until analysis.
Soil profiles from distinct topographic areas (landscape units) within each watershed were collected with a bucket augur in the spring of 2002 and 2003. Two to four profiles per watershed were laid out in a 10-cm diameter trough, divided into horizons, described with regard to color, texture and other morphological features. Composite samples of the upper 15 cm of each area were collected with a hand-held corer. The composite samples and samples from each horizon were taken to the lab in sealed plastic bags on ice, where they were rapidly air-dried under a fan at room temperature.
Sub-samples of soil were ground and analyzed for total C, H, and N by high temperature combustion using the CHN 2000 (LECO Corporation; St. Joseph, Michigan) (Campbell, 1992). Samples were also analyzed for pH (1:1 in water), percent organic matter (loss on ignition) and Melich 1 extractable, Mg, P, K, and Ca. Soluble soil N and P in the soil profile samples were estimated by shaking 3 g soil with 30 mL of 0.5M K2SO4 for 30 minutes at 100 rpm, then centrifuging for 10 minutes at 3000 rpm and then filtering (No. 42 Whatman filter papers).
Water samples were filtered under vacuum through 0.2μm filters (polycarbonate membrane, Nuclepore ® Corporation Filtration Products; Pleasanton, CA). Filtered samples were analyzed for NO3-N using a Technicon Autoanalyzer II flow injection analyzer with a cadmium reduction column and a 2:1 distilled water dilution loop at a rate of 30 samples h-1 (Technicon Industrial Method No. 487-77A, 1977). To bring the samples to within the necessary pH range for use in the Autoanalyzer (5-9), 0.5 ml of 0.1M NaOH in 10% NaAc was added to each.
Ammonium concentration in water samples and soil extracts was determined using an Orion 9512 ammonia specific gas-sensitive electrode (Banwart et al., 1972). One ml of 5M NaOH ionic strength adjusting (ISA) solution with pH color indicator was added to 10.0 ml of sample to bring the sample pH to >13. A Teflon-coated stir bar was added, the sample vial placed on a magnetic stirrer, and the electrode was then lowered into the sample for a reading. When the change in mV slowed to <1mV s-1, the mV reading was recorded for samples and ammonium-N standards (0, 0.1. 1. 10 mg L-1) and a logarithmic standard curve constructed.
Filtered water samples and soil extracts were analyzed for dissolved reactive P (DRP) using the ascorbic acid method (A.P.H.A., 1992). Ten mL of sample was used, with 1.6 mL of the reagent containing 5 N sulfuric acid, ammonia molybdate solution, 0.1 M ascorbic acid and potassium antimonyl tartrate solution. Absorbance was measured at 880 nm.
Total dissolved nitrogen (TDN) and phosphorus (TDP) were determined by a modified alkaline persulfate microwave digestion (Cabrera and Beare, 1993; Hosomi and Sudo, 1986; Johnes and Heathwaite, 1992; Littau and Englehart, 1990). Samples were digested using an alkaline persulfate reagent (45g K2SO8 and 9.5g NaOH per 1.0L) heated in a microwave in 120 mL pressurized Teflon vessels at full power (actual output 675 W) for 750 seconds. Dissolved organic N (DON) was calculated as:
DON = TDN- Total dissolved mineral N (NH4-N + NO3-N)
Dissolved organic P (DOP) was calculated as:
DOP = TDP – DRP
The amount of total suspended solids (TSS) was determined by drying and weighing each filter after filtration. Each filter was then microwave digested to determine total N and P as describe above.
Potentially mineralizable soil nitrogen
Composite soil samples of the surface horizon from each watershed were incubated at 60% water-filled pore space (Linn and Doran, 1984) for 16-days at 30 + 1 °C (Drinkwater et al., 1996) as described by (Sainju et al., 2002) Each 10.0 g soil sample was incubated in a gas-tight 1-L chamber with 3.0 ml of 0.50 M NaOH in a plastic vial and a vial of distilled water to maintain soil humidity. At the end of the incubation period, the soils were extracted with 0.1 M K2SO4, filtered and analyzed for nitrate-N with a salicylic colorimetric method modified from (Cataldo et al., 1975). Ammonia in the soil extracts was measured using an ammonia-gas sensitive electrode described above for groundwater samples. Initial nitrate and ammonium-N in soil samples was measured on subsamples that had not been incubated, but stored dry at room temperature. The difference in inorganic N extracted from incubated and non-incubated soil was considered to be mineralizable N.
Drainage volume was calculated using the WATBAL model (Vinten, 1999), a monthly water balance model which uses inputs of temperature, slope, vegetative cover, rainfall and cloud cover to estimate evapotranspiration, changes in soil moisture, runoff, and finally, drainage (Appendix B). Large-scale (1:1200) topographic maps were made to determine the boundaries, slope, slope aspect and area of each watershed.
To calculate the rate of groundwater flow through the six watersheds, hydraulic conductivity was calculated using slug tests performed on each watershed. Rising-head slug tests were carried out for 62 of the 65 piezometers, using standard methods (American Society for Testing Materials, 1997). Hydraulic conductivity for each piezometer was calculated using the(Bouwer and Rice, 1964) methodology, with adjustments for vertical contrasts in substratum texture.
Statistical analyses for groundwater variables were conducted on samples collected from June 2001 through June 2004. Due to the dry conditions that occurred during the first leaching period, groundwater was not available in two of the grazed watersheds until October 2002 (Grazed 1 B and Grazed 2 B), and analyses including those watersheds cover the period from October 2002-June 2004.
Therefore, for the purpose of statistics, data were grouped in three ways:
1) beginning in October 2002, when piezometers in all six watersheds provided groundwater;
2) beginning in July 2001, when piezometers in four of the watersheds provided groundwater; and,
3) beginning in October 2002, a comparison of groundwater from the six watersheds based on proximity to the homestead or barnyard.
One of the two watersheds on each farm was determined to be the ‘homestead’ watershed, historically receiving a greater proportion of nutrients because of its close proximity or access to the barn and homestead for the past one hundred years or more.
The nutrient concentration data were subjected to analysis using repeated measures GLM. The seasonal average for each sampling period (e.g. Winter 2002, Spring 2003, etc.) was used for each sampling nest to avoid pseudoreplication over time.
The model included the effects of farm, piezometer nest, and watershed nested within farm. Rainfall and sample pH were originally considered as covariates, but were removed because they were not significant. A GLM was also run for these parameters in macropore water taken via ceramic-tipped suction lysimeters. One-tailed t-tests were run to compare averages to the EPA maximum contaminant loads and limits.
The nutrient and sediment concentrations for the stream samples were analyzed using repeated measures GLM, with covariates of season and sample pH. Analyses were done fore both streams together, using effects of sampling site and flow, and for interactions of stream, flow, and sampling site. Separate analyses were done for base and storm flow to take advantage of the much larger data set for base flow samples.
We did an analysis of the nitrogen and phosphorus balance on the three farms for the years 2003-2004, using mainly farmer records and literature values, but also some of our data on pasture composition. The nutrient balance for N and P was calculated from inputs of feed, hay, fertilizer, legume nitrogen fixation, and atmospheric deposition minus outputs as milk, hay and animals sold. The nutrient balance was calculated for the whole farm and on a per land area basis. Each farmer completed several questionnaires detailing nutrient imports and exports. Questions included: the amount and type of fertilizer purchased and applied, manure application and test results, hay and bedding material purchased or sold, the amount and type of feed purchased, daily ration information (including net energy intake, % crude protein, % phosphorus), herd size and breeds, and milk and animals sold. The questionnaires were followed with interviews with the individual farmers for clarification and further information on farm history. Each interview was at least one hour in length. Information on crops grown and acreage (including leguminous crops) was provided by the Confined farmer, and forage quality information based on samples from the study watersheds was used to determine the proportion of nitrogen-fixing species in grazed pastures.
The nitrogen fixation rates were estimated as 125 kg ha-1 for pure clover and 200 kg ha-1 for pure alfalfa stands. Milk nitrogen was calculated using milk urea nitrogen data from the three farms. Nitrogen in rations was calculated from crude protein content (protein = N*6.25). Exports in animals, hay and silage sold off farm were calculated from data on weight and estimated nutrient contents from the literature. Exports were subtracted from imports to give surplus. The total was divided by the area of the farm for a per hectare value
The three project farms were included in the University of Maryland Agriculture and Natural Resource Economics Department’s Maryland Dairy Farm Business Summary (Johnson et al., 2004), an extension effort that conducts a detailed economic audit and allows farmers to compare their economic strengths and weaknesses to a group of about 30 dairy farms.
The research results were disseminated in multiple ways. Project collaborator and extension educator, Stan Fultz, conducted many pasture walks on various grazing farms each year of the project. Project personnel presented research results to farmer and environmental professional audiences at numerous forums, including farmer meetings, environmental conferences, workshops and committee meetings. A fact sheet was prepared and printed and a second edition of the fact sheet was distributed via the internet. A farmer-authored booklet was produced providing information to fellow dairy farmers on how to make the switch to grazing.
Based on the three leaching seasons sampled, including the first year which was exceptionally dry with little normal leaching, and the third year which we completed under phase II grant funding, the average nitrate-N in the groundwater was 4.8 ppm in the four grazed watersheds and 8.9 ppm in the two watersheds on the confined-feeding farm. The nitrate-N was well below the EPA groundwater standard, and far lower that the 15.5 and 32 ppm average nitrate-N predicted by the models based on previous (faulty) research. Groundwater nitrate concentrations in the two Confined watersheds averaged 6.8 and10.9 mg L-1 and exceeded the EPA limit regularly over the sampling period.
Groundwater and pore water nitrate
From May 2001 to March 2002, Maryland experienced a drought, with little to no groundwater recharge. For almost a year following these extremely dry conditions, there was heavier than normal rainfall. During dry periods and through most of the growing season, plants take up both water and nitrogen, allowing little opportunity for nitrate leaching. It is during periods when precipitation exceeds evapotranspiration and groundwater recharges that nitrate leaching is most likely (Staver and Brinsfield, 1998).
Under the study watersheds, groundwater nitrate concentrations were especially high when leaching followed dry periods, apparently flushing into the groundwater nitrate retained in the soil profile. Similar nitrate flushing was reported by Tyson et al. (1997) and Unwin (1986). The most notable nitrate concentration peaks occurred in Fall 2002 and Winter 2003. Other observed fluctuations in nitrate concentrations were similar to seasonal variations reported in other studies, with greater nitrate levels in wetter and colder months (Hack-ten Broeke et al., 1996; Kolenbrander, 1981; Owens et al., 1992; Saarijarvi et al., 2004; Stout et al., 1997).
Even with elevated nitrate concentrations following drought conditions, groundwater under grazed pastures did not reach the excessively high levels previous research had predicted. Predictive models developed by Stout et al (2000) predicted mean annual groundwater nitrate concentrations of 15 and 32 mg NO3-N L-1 for the grazed watersheds with the stocking rates found on the two MIG farms in the present study. Instead, the mean annual nitrate-N concentrations actually observed on the four MIG watersheds were between 4 and 7 mg L-1.
With all six watersheds included in the statistical model that evaluated seasonal averages for 12 seasons (summer 2001 through spring 2004), the main effect of watershed on nitrate concentration was significant, with the highest concentrations in the two watersheds on the Confined farm. Seasonal average groundwater nitrate concentrations in the two Confined watersheds ranged from 6.8 to 10.9 mg L-1 and exceeded the EPA maximum contaminant load during eight seasons in the study period. The high nitrate-N levels in groundwater under manured cropland, even with the implementation of an approved nutrient management plan is in keeping with the observation of (Angle, 1990) that nitrate-N concentrations are expected to exceed 10 mg L-1 with high yielding row crops under in Maryland conditions.
In contrast, seasonal average groundwater nitrate concentrations ranged from 3.4 to 8.2 mg L-1 on the four grazed watersheds, with two exceptions. The 95% confidence interval of groundwater nitrate concentrations under grazed watersheds were significantly below the EPA maximum contaminant load except during two seasons of rapid groundwater recharge and heavier than normal precipitation levels: Winter 2003, when Grazed 1A averaged 9.6 mg NO3-N L-1, and Spring 2003, when Grazed 2B averaged 11.6 mg NO3-N L-1.
In all three statistical models, depth to groundwater and season were highly significant effects, either as main effects or within interactions. Nitrate concentrations increased as groundwater recharged in all six watersheds, with the most notable increases in the Fall 2002 and Winter 2003.
Using rainfall and climate data and the WATBAL model, drainage was calculated as 268 mm yr-1 in the Frederick County watersheds and 302 mm yr-1 in the Baltimore County watersheds. Based on average nitrate concentrations in the groundwater and calculated drainage, the N exported in groundwater was 173 kg/ha for Confined, 54 kg/ha for Grazed1 and 71 kg/ha for Grazed 2. Extrapolated to the whole farm, this would yield a total yearly export of 4459-7154 kg NO3-N from the 245 ha Confined and yearly exports of 780-1137 kg from Grazed 1 (83 ha) and 710-1583 kg from Grazed 2 (71 ha). The estimates of groundwater loading by the three farms were significantly predicted by the N surpluses calculated by the nutrient balance accounting according to the linear regression equation (N=3, R2 = 0.91**):
Nitrate-N leaching (kg N ha-1 y-1) = 7.7 + 0.11 * Annual N surplus (kg N ha-1 y-1)
Phosphate levels in groundwater under all watersheds was generally above 20 ppb, often cited (Franklin et al., 2002; Heathwaite and Dils, 2000) as critical for causing eutrophication in surface waters. Dissolved reactive phosphorus in groundwater was much lower in both Baltimore County grazed watersheds, at an average of 0.04 ppm, as compared to 0.19-0.20 ppm in the four Frederick County watersheds. The lower P levels in the Baltimore County watersheds were almost certainly due to the presence of calcareous parent material underlying those watersheds, rather than a difference in management. For N, but not for P, the watershed nearest the barns (homestead) that historically received the most manure had significantly higher groundwater nutrient levels than the watershed farther from the barns (away).
Stream water in both streams on the Baltimore County grazing dairy farm showed little change in N concentration from where the stream enters to where it leaves the farm during base or storm flow, indicating minimal impact of the grazed pastures on the surface water quality. There were higher levels of total P and lower levels of total N during storm events than during base flow. These trends were as expected because much P is carried with soil particle in surface runoff during storm events, while most N reaches stream by leaching to groundwater, so storm generated surface flow tends to dilute the N in streams (even if it increases the total load).
A winter “camping area” or congregation area of highly trampled and nutrient enriched soil where the dried-off herd was kept and fed hay during the winter resulted in increased P concentrations in samples taken downstream from this area in one of the streams early in the study. The farmer changed his overwintering practices in response to this data.
Calculated P surpluses on the three farms were 9.9 kg P/ha for Confined, 1.5 kg P/ha for Grazed 1 and 7.8 kg P/ha for Grazed 2. Unlike for nitrogen, there was no relationship between P surpluses and P loading to groundwater. Rather, total dissolve P loading to groundwater seemed mainly related to underlying geologic material, with the acidic rock Frederick County farms losing 0.60 to 0.62 kg P/ha and the calcareous rock Baltimore County farm losing only 0.26 kg P/ha.
The total N and P determined after microwave persulfate digestion in ground and stream water samples showed that significant quantities of organic N and P containing compounds were present in addition to the customary nitrate-N and dissolve reactive phosphate-P reported above. There were no significant watershed or seasonal effects on the proportion of total N that was organic; the organic N percentage fluctuated, but averaged about 20%. The percentage of the total P that was organic was significantly different for among watersheds, being about 20% in the Frederick County watersheds, but averaging 43% in the two Baltimore County watersheds. We attribute this difference to the more effective removal of inorganic P than organic P from the water by the calcareous parent materials in the Baltimore County watersheds. These results indicate that the nitrate and dissolve reactive phosphate parameters traditionally reported in groundwater pollution studies may not be adequate to assess the total loads of these two nutrients.
In January of 2003 and 2004, we presented results of the water quality and economic analyses at sessions in the annual Farming for Profit and Stewardship Conference sponsored by Future Harvest CASA and Maryland Cooperative Extension. We reached a combined audience of 90, made up of farmers, extensionists and regulators. In March 2004, we were invited to present our environmental impact information to the Scientific and Technical Advisory Committee of the Chesapeake Bay Program, a group of 15 high-level scientists who advise the state and federal agencies involved with the Chesapeake Bay restoration efforts. In June 2004, we presented the results of the full leaching season at the annual meeting of the American Forage Grassland Conference in Virginia, and reached 75 farmers, researchers and policy makers.
In September 2004, we were able to take advantage of an existing conference, the 12th National Nonpoint Source Monitoring Workshop, held in Ocean City, Md, that brought together many regulators involved with water quality in mid-Atlantic and other states. At this conference we addressed a session with approximately 50 policy makers, government officials and environmental managers and discussed our data showing no increase in nitrate leaching from grazing dairies.
In January 2006, the project director presented our results to an audience of 60 dairy and beef cattle farmers with an interest in grazing practices. This presentation resulted in an active discussion of and interest in the water quality impacts of grazing practices.
Project collaborator and extension educator, Stan Fultz, conducted a total of 33 pasture walks during the project period, 2003-2004, including many on the project grazing farms. These pasture walks spread the story of grazing, our research results and other environmental impacts to 871 attendees, mostly dairy farmers. The project PI and project research assistant also made presentations to farmers at several of these pasture walks.
We created a 14-page farmer-authored booklet entitled Making the Switch: Two Successful Dairy Graziers Tell Their Stories. Seven to eight years after they began to use grazing instead of confined feeding for their milk cows, two successful Maryland dairymen sat down for leisurely interviews on how they changed to grazing and how grazing changed their lives. The highly readable, intimate and well illustrated booklet contains their stories in their own words, edited only minimally for readability and brevity. It is obvious that by the time project research assistant, Rachel Gilker, conducted the interviews for this booklet, she had developed a close rapport with both farmers during the three years she had been studying the groundwater quality under their pastures.
Finally, a PhD dissertation was submitted in July 2005. From that dissertation, several scientific papers are under various stages of review and submission.
Additional Project Outcomes
Impacts of Results/Outcomes
The results of the project have been communicated to small to medium-sized dairy farmers through the use of presentations at workshops and meetings in Maryland, Virginia and Pennsylvania and a farmer-authored booklet that has been published and distributed. After seeing the data on stream P pollution, the farmer of Grazer 2 adjusted his management of the winter camping area to reduce its impact on the stream.
The results have been presented to the regulators (State Depts. of Agriculture and/or Depts. of Environment or Natural Resources), extension agents and pasture scientists. With the positive environmental findings and outreach efforts, the project has substantially contributed to reducing environmental roadblocks to the development of appropriate state programs to encourage MIG systems. The Maryland Department of Agriculture now is promoting the use of grazing systems. The Center for Sustainable Agriculture is seeing more and more interest in grass-based farming. In this way, the project has contributed to improving the diversity, profitability and environmental impacts of dairy farming in the Northeast.
The experiences and insights related in Making the Switch: Two Successful Dairy Graziers Tell Their Stories have been widely disseminated are now inspiring others to explore the possibilities of grazing, and helping others to avoid or better deal with some of the struggles these dairymen experienced along the way. More than 1000 copies of the booklet have been distributed to extension personnel, regulators and farm advisors, including 250 copies that were requested by the Vermont Pasture Network which disseminates them directly to farmers. Requests for this unique booklet are still coming in from up and down the East Coast, more than a year after publication.
The only economic analysis included in the project was the economic performance data on the three study farms obtained from their participation in the Maryland Dairy Farm Business Summary (Johnson et al., 2004)(as described in the materials and methods and results). The “Making the Switch” booklet includes sidebars with some of the factual information on nutrient balance and some measures of profitability from the Maryland Farm Business Summary. The economic analysis for the three project farms showed that the grazing farms had a lower cost of production for milk and greater profit per unit of milk sold ($6.99/CWT milk for Grazed 1 and $4.34 for Grazed 2) than did Confined ($3.60/CWT milk). However, profit per acre of land was greatest for Confined ($486/acre compared to $437 for Grazed 1 and $352 for Grazed 2).
Although the target audience for our research results is primarily regulators and farm advisors, farmers themselves have also found the environmental data to be of interest. Our work has given a boost to and has been integrated into extension efforts to promote grazing by Maryland farmers. The number of dairy farmers who use grazing has increased substantially during the project, with nearly 30 switching in Washington and Frederick counties. Currently we estimate that 15 to 20% of Maryland’s 600 dairy farms use grazing to some degree.
Areas needing additional study
As is usual with research, our project may have raised nearly as many questions as it answered. A few will be considered here as subjects worthy of further research.
While our results clearly should put to rest the concern that intensive grazing will inevitably result in massive nitrate loss from urine spots (an initial concern that motivated this project), they did show that nitrate concentrations in groundwater under dairy farms are highly variable, with some samples that far exceeded the 10 ppm standard for nitrate –N. While these high N episodes were much more common on the Confined feeding watersheds, the Grazed farms had them as well and they did not always appear correlated with any of the weather, grazing or manure application events monitored in our study. Finding the cause of these N spikes could be very helpful in curbing the overall N loading to groundwater.
Our data clearly shows that, on each farm, N losses were greater in the watershed most convenient to the homestead and therefore with a long history of heavy manure applications. This N leaching difference persisted even though since the advent of Maryland’s Nutrient Managment Program more than a decade prior to the initiation of our project, manure on all three farms had been applied evenly to fields according to soil tests and agronomic N use. Such residual effects did not appear to hold for phosphorus leaching, however. The issue of long term residual effects is a very important one for environmental policy as such effects may make it take a very long time before the benefits of nutrient managment practices are seen in the either the local groundwater or the Chesapeake Bay. The nutrient accumulation and release mechanisms and dynamics and possible approaches remediation in long term over-manured soils is an area that needs further study.
Third, our results clearly show that measuring the usual, ammonium and ortho-phosphate in groundwater or stream water will not suffice to tell the nutrient balance or water quality story. We found that dissolved organic forms of both N and P were present in significant amounts (up to 40% of the total nutrient leaching). Future nutrient pollution research should measure and take these forms into account, especially for animal-based agriculture.
Finally, there remains much to learn about the grazing itself and how soils, water, animals and grass all interact on intensively managed pastures.
A.P.H.A. 1992. Standard Methods for the Examination of Water and Wastewater. 18 ed. American Public Health Association, Washington, D. C.
American Society for Testing Materials. 1997. ASTM standards on environmental sampling. ASTM, West Conshohocken, PA.
Angle, J.S. 1990. Nitrate leaching losses from soybeans (Glycine max L. Merr.). Agric. Ecosystems and Environ. 31:91-97.
Banwart, W.L., M.A. Tabatabai, and J.M. Bremner. 1972. Determination of ammonium in soil extracts and water samples by an ammonia electrode. Communications in Soil Sci. Plant Anal. 3:449-458.
Bouwer, H., and R.C. Rice. 1964. Basic instructions for falling-head seepage meter technique. Water Conservation Laboratory, Soil and Water Conservation Research Division, ARS-USDA, Tempe, AZ.
Cabrera, M.L., and M. Beare. 1993. Alkaline persulfate oxidation for determining total nitrogen in microbial biomass extracts. Soil Sci Soc Am J 57:1007-1012.
Campbell, C.R. 1992. Determination of Total Nitrogen in Plant Tissue by Combustion., p. 21-23 Plant analysis reference procedures for the Southern Region of the U.S. , Bulletin 368. USDA Southern Coop. Res. Ser., Washington, D.C.
Cataldo, D.A., M. Haroon, L.E. Schrader, and V.L. Youngs. 1975. Rapid colorimetric determination of nitrate in plant tissue by nitration of salicylic acid. Comm. Soil Sci. Plant Anal. 6:71-80.
Drinkwater, L.E., C.A. Cambardella, J.D. Reeder, and C.W. Rice. 1996. Potentially mineralizable nitrogen as an indicator of biologically active soil nitrogen, p. 217-229, In J. W. Doran and A.
J. Jones, eds. Methods for assessing soil quality. Soil Science Society of America, Inc., Madison, WI.
Franklin, D.H., J.L. Steiner, M.L. Cabrera, and E.L. Usery. 2002. Distribution of Inorganic Nitrogen and Phosphorus Concentrations in Stream Flow of Two Southern Piedmont Watersheds. J Environ Qual 31:1910-1917.
Heathwaite, A.L., and R.M. Dils. 2000. Characterising phosphorus loss in surface and subsurface hydrological pathways. The Science of The Total Environment 251-252:523-538.
Hosomi, M., and R. Sudo. 1986. Simultaneous determination of total nitrogen and total phosphorus in freshwater samples using persulfate digestion. International Journal of Environmental Studies 27:267-275.
Johnes, P.J., and A.L. Heathwaite. 1992. A procedure for the simultaneous determination of total nitrogen and total phosphorus in freshwater samples using persulphate microwave digestion. Water Research (Oxford) 26:1281-1287.
Johnson, D., D. Schwartz, S. Fultz, and M. Bell. 2004. Maryland Dairy Farm Business Summary. Information Series 200403. Department of Agriculture and Resource Economics, University of Maryland, College Park, MD.
Jonker, J.S., R.A. Kohn, and R.A. Erdman. 1998. Using milk urea nitrogen to predict nitrogen excretion and utilization efficiency in lactating dairy cows. Journal of Dairy Science 81:2681-2692.
Linn, D.M., and J.W. Doran. 1984. Effect of water-filled pore space on carbon dioxide and nitrous oxide production in tilled and untilled soils. Soil Sci Soc Am J 48:1267-1272.
Littau, S.E., and W.G. Englehart. 1990. Microwave sample preparation for total nitrogen and phosphorous determination. CEM Corporation, Matthews, NC.
Sainju, U.M., B.P. Singh, and S. Yaffa. 2002. Soil Organic Matter and Tomato Yield following Tillage, Cover Cropping, and Nitrogen Fertilization. Agron J 94:594-602.
Vinten, A.J.A. 1999. Predicting water and chloride transport in drained soils derived from glacial till. J. Environmental Quality 28:980-987.