Loading...
HomeMy WebLinkAboutLong Term Effects of Nitrogen Ferilizer Use_201401221138421203Long-Term Effects of Nitrogen Fertilizer Use on Ground Water Nitrate in Two Small Watersheds M. D. Tomer* and M. R. Burkart ABSTRACT to saturation, the thickness of the saturated media, ve-locity and directions of ground water movement, andChanges in agricultural management can minimize NO3–N leach-the type and spatial extent of management changes thating, but then the time needed to improve ground water quality is uncertain. A study was conducted in two first-order watersheds (30 are implemented. and 34 ha) in Iowa’s Loess Hills. Both were managed in continuous Groundwaterqualityinagricultural areasoftheMid-corn (Zea mays L.) from 1964 through 1995 with similar N fertilizer west has been particularly affected by NO3–N leachingapplications (average 178 kg ha 1 yr 1), except one received applica-(Burkart and Stoner, 2001). Changes in cropping prac-tions averaging 446 kg N ha 1 yr 1 between 1969 and 1974. This study tices including multicrop rotations and N fertilizer man-determined if NO3–N from these large applications could persist in agement may help reduce this contamination, but theground water and baseflow, and affect comparison between new crop changes in ground water quality may substantially lagrotations implemented in 1996. Piezometer nests were installed and changes in management. Underestimating the time re-deep cores collected in 1996, then ground water levels and NO3–N quiredforwaterqualityimprovementcouldcauseincor-concentrations were monitored. Tritium and stable isotopes ( 2H,18O)rect assessments of the effect of new practices on waterwere determined on 33 water samples in 2001. Baseflow from the heavily N-fertilized watershed had larger average NO3–N concentra-quality, if the assessments are based on short-term stud- tions,by8mgL 1.Time-of-travelcalculationsandtritiumdatashowed ies. The potential lag in water quality response also ground water resides in these watersheds for decades. “Bomb-peak”has implications for the setting of realistic timelines toprecipitation(1963–1980)mostinfluencedtritiumconcentrationsnear achieve targeted water quality improvements in water-lower slope positions, while deep ground water was dominantly pre-sheds. Unfortunately, documenting the timing of water1953 precipitation. Near the stream, greater recharge and mixed-quality responses to new management practices has re-age ground water was suggested by stable isotope and tritium data,ceived little attention, and there are few reports of wa-respectively. Using sediment-core data collected from the deep unsat-tershed- or landscape-scale changes in ground waterurated zone between 1972 and 1996, the increasing depth of a NO3–N quality resulting from specific changes in agriculturalpulse was related to cumulative baseflow (r2 0.98), suggesting slow management.downward movement of NO3–N since the first experiment. Manage- ment changes implemented in 1996 will take years to fully influence Long-term studies under documented management ground water NO3–N. Determining ground water quality responses are needed to develop knowledge on the timing ofto new agricultural practices may take decades in some watersheds.ground water quality response to new practices. Oppor-tunities to examine these responses with field datashould be identified and exploited. This paper presentsAgricultural land use has frequently been associ-a comparison between two small watersheds (30 andated with loadings of nutrients to ground and sur-34 ha) for which there is nearly a 40-year record offace waters (Burkart and Stoner, 2001; Castillo et al.,management history and stream flow. The hydrology2000; Sauer et al., 2001; Schilling and Libra, 2000.) It is and geology ofthese two adjacent watershedsare essen-also widely reported that agricultural practices can be tiallyidentical.Agriculturalmanagementpracticesweremodifiedtoreducetheseloadingsthroughnutrientman-also identical, except for two periods between 1969 andagement (Kitchen and Goulding, 2001), crop rotation 1974, and from 1996 until present. Both of these experi- (Bolton et al., 1970; Owens et al., 1995; Randall et al.,mental periods were hypothesized to cause a difference 1997), and the use of biological filters (e.g., Peterjohn in NO3–N concentrations in ground water and stream andCorrell,1983;Nelsonetal.,1995;Spruill,2000;Gold baseflow between the two catchments. Could an experi- et al., 2001). Once a change in an agricultural system is mental effect from the first treatment period possiblyimplemented, there may be rapid improvement in the persist and confound a comparison between the currentquality of runoff waters if those changes increase infil-experimental practices? The objective of this study wastration and reduce sediment transport. But in the humid to determine if any current differences in NO3–N con-Midwestern USA, most streamflow is typically domi-centrations could, at least in part, be attributed to anated by baseflow that originates from ground water.residual effect from the 1969–1974 experiment.Thetimingofgroundwaterqualityresponsestochanges This study included several techniques that includedin agricultural management is not readily predicted and estimating the travel time of ground water through thedepends on the size of the area being monitored, depths watershed with physical and isotopic methods. Isotopes of water ( 18O,2H, and 3H) have often been used to help interpret hydrologic systems, both surface waterNationalSoilTilthLaboratory,USDAAgriculturalResearchService,(GenereuxandHooper,1998)andgroundwater(Gonfi-2150 Pammel Drive, Ames, IA 50011. Received 4 Nov. 2002. *Corre-sponding author (tomer@nstl.gov).antini et al., 1998). Changes in stable isotope ( 18O,2H) composition occur due to changes in water phase (evap-Published in J. Environ. Qual. 32:2158–2171 (2003).ASA, CSSA, SSSA677 S. Segoe Rd., Madison, WI 53711 USA Abbreviations:Ks, saturated hydraulicconductivity; TU, tritium units. 2158 TOMER & BURKART: GROUND WATER NITRATE IN TWO SMALL WATERSHEDS 2159 oration, sublimation, condensation) in the atmosphere and NO3–N concentrations of ground water and unsatu- rated sediments are compared between two adjacent(Ingraham,1998),atthelandsurface(WalkerandKrab- benhoft, 1998), and in the shallow subsurface (Barnes first-orderwatershedsintheLoessHillsofsouthwestern Iowa. Hydraulic and isotopic data are used to evaluateand Turner, 1998). Spatial and temporal changes in iso- topic composition can be used to determine relative thesubsurface flowsystem.It ishypothesized thatisoto-pic signatures will help interpret relative ages and path-water sources and dominant hydrologic processes, andthus aid interpretation of storm flow pathways (Buttle ways of ground water in the two small watersheds atthis southwestern Iowa site. If both the physical and theand Sami, 1990; DeWalle et al., 1997; Genereux andHooper, 1998), ground water flow (Gonfiantini et al., isotopic methods indicate that rainfall predating 1980still resides within ground water of these watersheds,1998; Hendry, 1988; Kehew et al., 1998; Matheney andGerla, 1996), and plant water use (Burgess et al., 2000; then any differences in NO3–N concentrations betweenthese two watersheds could result, at least in part, fromDawson and Ehleringer, 1998).Tritium ( 3H) is an unstable isotope, with a half-life the 1969–1974 experiment. The NO3–N concentrationsthemselves were also evaluated and compared with his-of 12.43 years, that increased in the atmosphere and inprecipitation following aboveground testing of nuclear torical data to provide an additional line of evidence for evaluating the possible persistence of NO3–N fromweapons during the late 1950s and early 1960s, then decreased rapidly after 1963 after testing was banned the first experiment. (Ingraham, 1998). This pulse of tritiated waters has al- lowed relative aging of ground water, essentially into SETTING AND BACKGROUNDthree classes: “pre-bomb” (before 1953), “bomb-peak” (from the mid 1950s to the late 1970s or early 1980s),The study took place within Watersheds 1 and 2 ofand more recent waters (after about 1980). These three the Deep Loess Research Station (DLRS), located 10age classes are relevant to this study given the dating km south of Treynor, Iowa (Fig. 1). The station wasof the two experiments. Discerning these relative ages established to carry out research on agricultural hydrol-isbecomingdifficultasisotopicdecayof“bomb”tritium ogy and has a research history dating back to the mid-proceeds. Further discussion and case studies are pro-1960s, withmuch of theresearch focused onerosion andvided by Bradbury (1991), Daniels et al. (1991), Gonfi-nutrient balances under corn production (e.g., Karlen etantini et al. (1998), and Ingraham (1998).al., 1998, 1999; Logsdon et al., 1999). Soils developedNitrogenisotopes(15N/14N)ofnaturalabundancehave in the deep loess are dominantly mapped as Mononaalso been used in environmental studies; however, diag-(Typic Hapludolls), Ida (Typic Udorthents), Napier,nosticuseistypicallyrestrictedtodistinguishinganimal-and Kennebec (both Cumulic Hapludolls) soils (Soil(and/or human-) waste sources from soil organic matter Survey Staff, 1994), and more than a third of the area(SOM) or fertilizer sources (e.g., Spalding et al., 1982;is highly eroded (Karlen et al., 1999).Kendall,1998).Soilorganicmatterandfertilizersources Both watersheds were under a continuous corn rota-can become difficult to distinguish in ground water be-tion with a conventional tillage system from 1964cause their ranges of isotopic ratios overlap, soil N-cycle through 1995 (Table 1) and have shown similar runoffprocesses (i.e., immobilization, mineralization) in time and baseflow volumes (Kramer et al., 1999). There wasmix soil and fertilizer-applied N, and NO3–N in deep an experimental change in nitrogen fertilizer applica-soils and ground water could be subject to denitrifica-tions between 1969 and 1974, when Watershed 1 re-tion, which results in 15N enrichment of residual NO3–N ceived an average of 446 kg N ha 1 yr 1 and Watershed(Kendall, 1998). The two watersheds in this study never 2 received an average of 172 kg N ha 1 yr 1 (Table 1).received manure applications, and the only known These applications were made in spring mostly as anhy-sources of NO3–N are SOM and applied fertilizer.drous ammonia, but some of the increased N in Water-Therefore, natural abundance N isotopes would not be shed 1 was applied as granular NH4NO3 (Burwell etexpected to answer the question being asked in this al., 1976). This experiment was used to assess NO3–Nstudy.movement in deep soils (Schuman et al., 1975; AlbertsNitrateoriginatingfromgeologicparentmaterialscanet al., 1977) and its export in baseflow (Burwell et al.,occur, but these have only been reported as important 1976, 1977). However, total losses of N via stream dis-sources in more arid areas to the west. Boyce et al.charge during the six-year experiment only showed a(1976)reportedthatNO3–NconcentrationsinNebraska 111 kg ha 1 difference between Watersheds 1 and 2loess were diminished in eastern portions of that state’s (Burwelletal.,1977).AlbertsandSpomer(1985)resam-deep-loess region. Regardless, any geologic source for pled the loess profiles of Watersheds 1 and 2 to a 15-mthe NO3–N should not cause a difference in NO3–N depth in 1984, and identified a prominent increase inconcentrations between these two watersheds, which NO3–N centered at a 10.7-m depth in Watershed 1,borderoneanotherand havevirtuallyidenticalgeology,where a concentration of 16 mg NO3–N kg 1 soil con-pedology, and hydrology. The duration of cultivation trasted the 3 mg kg 1 found at 15 m. The watershedsis also believed to be similar. Both watersheds were received similar N applications from 1964 to 1968 andprobablycultivated sinceabout1880,based onPotawat-from 1975 to 1995 (Table 1). Karlen et al. (1999) listedtamie County records of cropland acreage (Larry Kra-annual N applications from 1964 to 1995. Corn grainmer, personal communication, 2002).harvest rarely accounted for more than 110 kg ha 1 ofN removal (Karlen et al., 1998).In this paper, the hydrology, water-isotope chemistry, 2160 J. ENVIRON. QUAL., VOL. 32, NOVEMBER–DECEMBER 2003 Fig. 1. Maps of Watersheds 1 and 2 of the Deep Loess Research Station (DLRS), showing topography (left) and cropping boundaries (right).Topographic contours (left map) are at a 5-m interval and labeled as meters above sea level. Locations of piezometer nests and weirs areshown in both maps; the right map also identifies the nests and shows piezometer transect locations. Table 1 names the piezometer transectsand gives depths of the installations at each piezometer nest. New crop rotations were established in both water- aged 151 kg N ha 1 in Watershed 1 and 117 kg N ha 1 sheds beginning in 1996 (Table 1). Watershed 1 was in Watershed 2. The fertilizer N applications are deter-placed under a corn–soybean [Glycine max (L.) Merr.]mined using soil testing; larger carryover of legume-rotation, with each crop covering either the eastern or fixed N and differences in timing of soil tests may havewesternhalfofthewatershedeveryyear(Fig.1).Water-contributed to smaller rates being recommended in Wa-shed 2 was placed under a six-year rotation of corn,tershed 2.soybean, and corn, followed by three years of alfalfa Thechangeinfarmingpracticessince1996ishypothe-(Medicago sativa L.). The six-year rotation was imple-sized to cause a difference between the two catchmentsmented in narrow contour strips (Fig. 1), with all six ingroundwaterandstream-baseflowNO3–Nconcentra-years of the rotation always present (about 1/3 in corn,tions. One question that has arisen, however, is whether1/6 in soybean, and 1/2 in alfalfa with all three stand the effect of the excessive nitrogen fertilizations be-ages represented). With both rotations, only the corn tween 1969 and 1974 has persisted in ground water.received N fertilizer, which was on an alternate-year Thus, to discern the effects of recent managementbasis in Watershed 1, and two out of six years in Water-changes, we must determine if current NO3–N concen-shed 2. After the new rotations were established, Nfertilizer rates applied to corn from 1996 to 2001 aver- trations in ground water and baseflow in Watershed 1 Table 1. A summary of crop rotations and fertilizer applications during different periods of research in Deep Loess Research Station(DLRS) Watersheds 1 and 2. Watershed 1 Watershed 2 Time period Crop rotation† Mean N application Crop area fertilized Crop rotation† Mean N application Crop area fertilized kg N ha 1 yr 1 % kg N ha 1 yr 1 % 1964–1968 CC 150 100 CC 150 1001969–1974 CC 446 100 CC 172 1001975–1995 CC 186 100 CC 186 1001996–2002 CS 151 50 CSCA3 117 33 † CC, continuous corn; CS, corn–soybean; CSCA3, corn–soybean–corn–three years of alfalfa. TOMER & BURKART: GROUND WATER NITRATE IN TWO SMALL WATERSHEDS 2161 posed of reworked loess were described following a schemecould still be influenced by these large N applications proposed by Bettis (1990). A 50-mm-i.d. piezometer was thenthat date back about 30 years.set just above the till, with a 0.6-m-long screen. Additional boreholes were also drilled to install shallower piezometersMETHODS(Table 2) at depths above and below the Sangamon paleosolThree independent lines of evidence were constructed to (a possible impedance to vertical water flow), and/or at theidentify ground water ages, travel times, and patterns of observed depth of the water table (i.e., where water-saturatedground water recharge, and discern the effects of past and sediments were first encountered). One or two suction-cuppresent management on NO3–N concentrations in the two lysimeters were also installed above the saturated zone atwatersheds. These lines of evidence were based on hydrologic several locations (Table 2). Installations at each landscapemeasurements, spatial trends in isotope chemistry, and spatial position were located within a few meters of one another,and temporal trends in NO3–N concentrations in waters and usually in line along the topographic contour. Piezometersdeep sediments.were gravel-packed to at least 0.2 m above the screened inter-val and grouted to the land surface with bentonite. LysimetersMonitoring Installations were similarly installed but were set in a silica–flour slurry to provide hydraulic contact between the ceramic cup and theIn each watershed, a transect of four piezometer nests was surrounding unsaturated sediments. Steel casings were set toinstalled in 1996, at divide (D), mid-slope (S), toe-slope (T),cover the polyvinyl chloride (PVC) plastic piezometers andand riparian valley (R) positions. The nests are identified by lysimeters. Map coordinates and elevations of all piezometerswatershed number and landscape position (e.g., 1D, 2S; see were determined by a global positioning system (GPS) surveyFig. 1). In 1999, three additional riparian-zone piezometer with relative errors of 0.01 m in horizontal and vertical di-nests were installed in Watershed 1, including 1B, and two mensions.nests installed near 1R, denoted 1Rb and 1Rc (the 1996- installed1Rnestwasdenoted1Ra;seeFig.1).The1999installa- tions were a part of separate research on a riparian buffer Hydrologic Measurementsplanted in Watershed 1 during 2000. Piezometer nests com-The hydraulic conductivity (Ks) of the saturated zone wasprise three transects, including two long transects identified measured by conducting slug tests in each piezometer. A solidby watershed (e.g., W1, W2), and a short riparian transect in PVC rod was lowered and later raised to conduct slug-downWatershed 1 identified as W1rip (Table 2). Each transect por-and slug-up tests. Hydraulic head was measured during eachtrays an expected path of ground water flow based on field slug test using a pressure transducer and data logging system.interpretation of the terrain.The slug test data were analyzed using the Hvorslev (1951)During drilling, cores were taken to the depth at which method. At each piezometer,Ks was taken as the mean of theglacial till was encountered, by a drill rig with a hollow-stem values measured by rising and falling head. The Ks data wereauger. The till defined the depth of drilling because it provides sorted by type of deposit (till, sand at the till interface, loess,an underlying aquitard to ground water within the deep loess.and alluvium), and the geometric mean was calculated forCoresweresubsampled,wrapped,andchilledforlateranalysis each unit. For several of the deep piezometers completed inof bulk density (Blake and Hartge, 1986), organic C, total N,till or its Yarmouth paleosol, measurements of Ks were notand NO3–N. Organic C and total N were determined by dry made because there was no response in the water levels aftercombustion, while NO3–N was determined by an autoanalyzer slug addition. Water levels were measured in each piezometertechnique on 2 M KCl extractions (Cambardella et al., 1994).on a monthly basis.Materials encountered during coring were recorded with Positionalsurveydata,coringdescriptions,waterlevels,anddepth, which at upland (D and S) positions included Peoria hydraulic conductivities were used to estimate times requiredand Pisgah loess of Wisconsinan age, underlain by Loveland for ground water to travel along the three transects. The satu-loess of Illinioan age, and then pre-Illinoian till. Sand lenses rated zone dominantly occurred within the Loveland loesswere common at the till–loess interface. Buried paleosols above the toeslopes, and in alluvium below. Given the unifor-known as Farmdale, Sangamon, and Yarmouth soils occurred mity and common origin of these two deposits, the groundat the upper surface of the Pisgah, Loveland, and till units, water flow system is simple in its hydrogeology, and simplerespectively. The statigraphy and paleosols of the Loess Hills calculations were used to evaluate lateral flow along theseare described by Prior (1991), Ruhe (1969), and Kay and Graham (1943). In the riparian valleys, alluvial deposits com- transects. Hydraulic gradients and conductivities were used to Table 2. Summary of monitoring transects, with numbers and depths of suction-cup lysimeters and piezometers at each piezometernest location. Transect† Piezometer nest† Number of lysimeters Depths of lysimeters Number of piezometers Depths of piezometers‡ mm W2 2D 2 (6), (12) 3 (17), (22), (28)W2 2S 2 6, 9 3 13,16,(21)W2 2T 1 5.5 3 4.5,14, (17)W2 2R 0 – 4 3,6,9,12W1 1D 2 (6), 12 3 21, (23),27W1 1S 2 8, 11 3 14.5,16.5,(21)W1 1T 1 3 3 5,14,(16)W1 1B 0 – 3 4,7,9W1 & W1rip 1Ra 0 – 4 5,9,14,16W1rip1Rb 0 – 3 3.5,6,8.5W1rip1Rc 0 – 3 4.5,8,11.5 † Drilling locations are indicated in Fig. 1.‡ Depths enclosed by parentheses indicate that poor water yield prevented consistent monthly sampling. Depths indicated in italic type (30 total) indicatepiezometers sampled for isotopic analyses. 2162 J. ENVIRON. QUAL., VOL. 32, NOVEMBER–DECEMBER 2003 determine ground water velocities using the Darcy equation. in which X is the distant (St. Louis or Ottawa) station’s and Y is the local (Lincoln) station’s TU data. The coefficients aSurvey data and water levels were used to calculate hydraulic gradientsforJune2001andApril2002,whichwerethemonths and b were fit by an iterative technique because of the small number ofdata points.Subset periodsof commonrecord werewith the highest and lowest average water levels since thesecond set of piezometers was installed in 1999. The average selected (Table 3) to avoid obvious bias that occurred whenvalues from peak years of tritium fallout were included. Thehead of all piezometers at each nest was calculated for bothtimes.Hydraulicgradientswerecalculatedbytakingthediffer- complete TU record at Lincoln ranged across nearly threeorders of magnitude. Restricting the input data provided aence in average head between adjacent nests, divided by thedistance between them. The geometric mean Ks calculated for curvefitandrootmeansquareerror(RMSE)thatrepresentedthe numeric range to be estimated (Table 3). To be explicit,theloess andalluvium depositswasthen usedto calculateflowvelocities. The flow rates were next divided by an estimated the peak years of 1963–1964 were excluded in estimating thepre-1962 record, and only the last 11 years of common recordeffective porosity of 0.2 m 3 m 3 to calculate ground watervelocities, which were finally divided into the distances be- were used to estimate the post-1987 record, so that estimatedvalues were not, in effect, extrapolated from observationstween transect locations and summed to obtain an estimatedtravel time along each transect. The effective porosity was during peak fallout.The post-1986 TU precipitation record was also estimatedestimated to be about half of the average porosity of 0.42 m 3 m 3, which was estimated from bulk densities of cores col- in a second way that relied only on local data. This method also employed Eq. [1], with the same Y variate, but the Xlected during piezometer installation. This was deemed an appropriate value to estimate the travel of a solute’s center variate was the number of years since 1962. Additional local data points were available for the years of 1992 and 2001.ofmass.Also,watercontentsoftheloesssoilsat1500kPa,near a 1-m depth, are about 0.20 m 3 m 3 (Rob Malone, personal Simpkins (1995) measured the tritium activity at Ames, Iowa, to be 11.02 TU during 1992 (weighted average). This wascommunication, 2003), suggesting that little flow occurs in about half the pore space. considered local data because Ames is only about 200 km from the study area. The 2001 TU value was obtained from an on-site sample, collected in June, and combined from fourIsotope Chemistry 1.8-m-deep lysimeters located near the divide between the two watersheds, in the area between Piezometer Nests 1DInput Record for Tritium in Precipitation and 2D. This sample’s 7.1 TU was assumed to indicate theA record of annual tritium activities in precipitation was tritiumactivityofprecipitationduringtheprevioussixmonths.constructed to represent past concentrations of tritium that Using these local data, a power function was fit to obtain ahave recharged ground water at the site. We obtained tritium curve from 1963 to 2001, giving the result shown in the lastunits (TU) data for annual precipitation published by the row of Table 3. Most of the gross (real space, rather than logInternational Atomic Energy Agency (1992), which covered space) error associated with this model occurred for the firstthe period from 1953 through 1986, and monthly data from year of peak fallout, and so the RMSE is reported as 11%.the Global Network for Isotopes in Precipitation database Once these precipitation records were constructed, the ex-(International Atomic Energy Agency and World Meteoro-pected tritium activity of each year’s precipitation in 2001 waslogical Organization, 2001) from 1987 though 1999. Annual calculated based on a 12.43-year half-life.weighted averages were calculated from the latter source. Rec-ords from three monitoring stations were used to compile a Isotope Sampling and Analysescontinuous tritium input record from 1953 to 1999. A station Between 13 and 15 June 2001, 33 water samples were col-at Lincoln, Nebraska, located 90 km southwest of these water-lected for isotopic analyses of 18O,2D, and 3T (TU). Thesheds, was operated from 1962 to 1986 and provided a data samples were taken from 30 piezometers (Table 2) and fromrecord considered to be local for those years. Annual TU the stream baseflow passing the weir of each watershed, plusvalues before 1962 and after 1986 were estimated based on the combined sample from four 1.8-m-depth lysimeters de-relationships between records at Lincoln and St. Louis, Mis-scribed above. The isotope analyses were conducted at thesouri, which covered from 1987 to 1993, and between records University of Waterloo Environmental Isotope Laboratoryat Lincoln and Ottawa, ON, Canada, to estimate from 1953 using methods described by Coleman et al. (1982) and Drim-through 1961 and from 1994 through 1999 (Table 3). These mie et al. (1991) for deuterium, Epstein and Maeda (1953)relationships were calculated using a power function:for 18O, and Drimmie et al. (1993) and Taylor (1977) for enriched tritium. Tritium results had a detection limit of 0.8YaXb[1] Table 3. Summary of regressions used to complete an estimated annual input record for tritium in precipitation. Data are fromInternational Atomic Energy Agency (1992), International Atomic Energy Agency and World Meteorological Organization (2001),and Simpkins (1995). Selection of input data is discussed in the text. Years of missing Location and years Years of record selectedrecord estimated† of record available as input data a‡b‡r2 RMSE§ TU 1953–1961 Ottawa, ON; 1953–1999 1962, 1965–1986 0.30 1.18 0.98 21.71987–1993 St. Louis, MO; 1963–1993 1975–1986 0.83 1.15 0.91 4.11994–1999 Ottawa, ON; 1953–1999 1975–1986 0.05 1.64 0.85 6.01987–1991, 1993–2000 local trend with time; 1962–1986, all except 1962 4448 1.71 0.97 11%¶1992, 2001 † Missing values before 1962 and after 1986 were estimated. Post-1986 estimates were based on relationships with two other stations (Rows 2 and 3), andon interpolation of a local trend over time (Row 4).‡ Coefficients a and b are defined by Eq. [1].§ Root mean square error, as tritium units (TU).¶ The RMSE is reported as a percent in this instance because the Y data (local TU values) span three orders of magnitude. TOMER & BURKART: GROUND WATER NITRATE IN TWO SMALL WATERSHEDS 2163 Table 4. A summary of hydraulic conductivity data from slug tests on piezometers in Watersheds 1 and 2. Data were sorted by type ofdeposit and include results from additional riparian-valley piezometers in Watershed 1 not mapped in Fig. 1. Hydraulic conductivity Type of deposit Range Geometric mean n ms 1 Till 2.8 10 5 to 2.7 10 8 1.4 10 6 13Sand (till interface) 3.8 10 5 to 2.1 10 6 6.5 10 6 6Loess 8.9 10 5 to 1.4 10 7 1.0 10 5 11Alluvium 1.6 10 4 to 1.0 10 8 4.1 10 6 26 TU, and analytical errors that varied from 0.5 to 1.4 TU and Nutrients in Water and Deep-Core Sedimentsincreased with tritium activity. Isotope samples were submit-Upon completion, the installations were sampled on atedinthreesetstospreadtheincurredfinancialcosts,andturn-monthlybasis;watersamples fromlysimetersandpiezometersaround times from the lab varied depending on its workload.wereanalyzedforNO3–NconcentrationusinganautoanalyzerConsequently, the samples were analyzed either 90, 120, or technique with a 1.0 mg L 1 detection limit that is described321 d after collection, corresponding to 0.020, 0.026, or 0.071 by Hatfield et al. (1999). At least 1.5 water-column volumesfractional half lives of tritium (12.43 yr), respectively. Accord-werepurgedfromeachpiezometerbeforesampling.Rechargeingly, tritium values were back-corrected from the date of toseveraldeeppiezometerswastooslowtoallowthispurging,analysis to the sampling date. While these were minor correc-and sampling of these installations (identified in Table 2) wastions, ranging from 0.0 to 0.7 TU, they did account for known done either less frequently (e.g., twice per year), or, whereerrors and were greater than half the analytical error in four water levels did not recover within a month of sampling (e.g.,of the last-analyzed samples.atNest2D),wassuspended.StreambaseflowwasalsosampledStable isotope data, expressed as 2D‰ and 18O‰, were at the outlet of both watersheds for NO3–N, on a monthlyplottedandalocalmeteoricwaterline(LMWL)wasestimated basis from autumn 1996 through summer 2000, and then onusing the reduced major axis method (Mann, 1987). This re-a weekly basis after spring 2001. Baseflow concentrations be-gression method was applied to stable isotope data by the tween the two watersheds were compared by a paired t test,International Atomic Energy Agency (1992), but was named under the null hypothesis that there was no difference inorthogonal regression in that document. The method was cho-NO3–N concentrations.sen because it reduces slope attenuation that can be important when the range of observations is limited and the X variate is subject to measurement error.RESULTS AND DISCUSSIONAsingleindexwascalculatedtoindicatetheisotopicenrich-Ground Water Flowment or depletion of any single sample relative to others atthe site, by scaling all the 2D and 18O‰ values between 1 Results of slug tests across all piezometers (Table 4)and 0 (values were all negative), and then adding one to their showed that small Ks values were most common in till,sum. This scaled the data between maximum and minimum whereas the alluvium showed the most variation in Ks.possible values of 1 and 1, with positive values being rela-Mean Ks valuesarewithintherangeexpectedforuncon-tively enriched, negative values relatively depleted, and near-solidated sediments derived from loess or till (Fetter,zero values closeto a mean isotopic conditionfor the site. This 2001).index facilitatedsimple graphics toevaluate spatialpatterns of Hydraulic heads (Fig. 2) were used to determine gra-the isotopic data, which were confirmed by t tests on theoriginal isotope values, assuming equal variances.dients in vertical and horizontal directions. There were Fig. 2. Cross-sections of hydraulic heads (given as meters of elevation) along the three transects (W2, W1, and W1rip), interpolated from June2001 water level measurements. Contour intervals are 0.5 m of head in the left and center cross-sections, and 0.25 m of head in the rightcross-section. The identifier for each piezometer nest is indicated along the land surface. Stratigraphic contacts between major deposits (till,loess,andalluvium)areindicatedbythickdottedlines.RefertoFig.1andTable2forpiezometer-nestandcross-sectionlocationsandidentifiers. 2164 J. ENVIRON. QUAL., VOL. 32, NOVEMBER–DECEMBER 2003 persistent vertical gradients within most piezometer tation (i.e., estimates based on International Atomicnests, but no evidence that buried paleosols had a major Energy Agency stations andon local trend). Agreementinfluence on ground water flow. Downward gradients between the St. Louis–estimated and the local trend–(0.02 m m 1) occurred between the two deepest pi-estimated data between 1987 and 1993 was excellent,ezometers at the divide position in both watersheds.becausetherewasaclose trackingbetweentheSt.LouisThe largest downward gradient (approximately 0.08 m and Lincoln records from peak bomb fallout until datam1) occurred between the two upper most piezometers collection at Lincoln ceased in 1986. Also, the estimatedat Position 1Ra, indicating a zone of slow permeability tritium activity for local precipitation in 1992 was 13.37between these two screen depths. There were consistent TU using the St. Louis record, and was 12.61 TU basedupward gradients between the upper two piezometers on the local trend. Simpkins (1995) estimated a valueat Positions 1S, 1T, 1B, and 1Rc, but these were usually of 12.68 TU for Lincoln in 1992, based on a relationshipsmall (0.02 m m 1). Larger upward gradients occurred with St. Louis data determined by a slightly differentbetween the middle two piezometers at Position 2R method.Local-trendestimatesofpost-1993tritiumwere(between 0.02 and 0.04 m m 1) and between the upper larger than those estimated using Ottawa data (Fig. 3),two piezometers at Position 1Rb (approximately 0.08 m but the differences do not affect data interpretation.m 1). A constriction caused by a shallow depth of the Isotopic decay was calculated for the local estimatestill contact could contribute to these upward gradients of mean annual tritium activities in precipitation, using(Fig. 2). Several vertical gradients are evident in the both constructs of the post-1986 record, to obtain thecross-sections of Fig. 2. A reversal in vertical gradient expected residual tritium activity in 2001 (Fig. 3). Sub-along the W1rip transect, from upward at 1Rb to down-surface and baseflow waters that fell as precipitationward at 1Ra, suggests heterogeneities in the alluvial de-within 20 years before sampling cannot be differenti-posits.ated. However, values exceeding 10 to 12 TU suggestLateralflowgradientsbetweenadjacenttransectposi-the presence of waters aged between 20 and 40 years,tions were calculated from the difference in average at least in mixture. Very small tritium values (i.e.,3hydraulic heads at each nest. Gradients were usually TU) would suggest waters predominantly of “pre-small above the toeslope positions (0.003–0.015 m m 1)bomb” origin, or at least 45 years old.in both watersheds, but larger below the toeslopes (0.012–0.042 m m 1). This difference is also evident in Tritium in Ground Waterthe cross-sections of Fig. 2.Estimatesof groundwatervelocity betweenpiezome-After correction for varying sampling-analysis timeter nests averaged 13.5 m yr 1, and varied from 5.3 to lags, tritium activities ranged from 0.8 to 18.5 TU. Small27.1 m yr 1. The estimates are based on a mean Ks tritium activities (3 TU), indicating water predating(Table 4) applied for flow through loess above the toes-any research at the site, occurred in 6 of the 33 sampleslopes and through alluvium below the toeslopes. Esti-and were always in the deepest ground water (Fig. 4).mated travel times (Table 5) varied according to the Tritium values exceeding 12 TU, showing an influencelength of each transect and changes in hydraulic gradi-of 20- to 40-yr-old precipitation, occurred in 9 of the 33ents. Along the W2 transect, larger gradients occurredwith higher water levels measured in June 2001. Butalong W1, the larger gradients occurred during April2002. Between 64 and 82% of the travel times occurredabove the toeslope positions. Because vertical and hori- zontalgradients areboth present,actual traveldistancesof ground water may be greater than horizontal dis- tancesand traveltimesareconsidered conservative.Re-sults therefore suggest that land management effects onground water could persist for decades in these water-sheds (Table 5). Isotope Chemistry Precipitation Tritium Record There was good agreement between the two con-structsofthepost-1986recordoftritiuminlocalprecipi- Table 5. Transect lengths and estimated ground water traveltimes for transects.Fig. 3. Expected residual tritium activities in 2001 for water originat-ing as local annual precipitation between 1953 and 2001. IsotopicTravel time for water levels measured decay based on a 12.43-yr half-life was applied to a constructedtritium record to obtain this plot. The constructed record includedTransect† Length June 2001 April 2002 local data (from Lincoln, NE; Ames, IA; and on-site), a trend inm yr the local data, and estimates for missing years based on recordsW2 277 22.7 25.9 from St. Louis, MO and Ottawa, ON, Canada, as specified inW1 337 36.1 31.0 Table 3. Precipitation data are from International Atomic EnergyW1rip28 1.4 2.5 Agency (1992), International Atomic Energy Agency and WorldMeteorological Organization (2001), and Simpkins (1995).† Locations are shown in Fig. 1. TOMER & BURKART: GROUND WATER NITRATE IN TWO SMALL WATERSHEDS 2165 Fig. 4. Cross-sections showing variations in ground water tritium activity (TU) along three transects (W2, W1, and W1rip) during June 2001. Tritium contours are interpreted as pre-1953 in age for waters with tritium activity of 3 TU, whereas values of 12 TU show an influenceof “bomb-peak” precipitation between 1963 and 1980. Intermediate values indicate recent or mixed-age waters. Piezometers not sampled fortritium are indicated by a crossed symbol. Piezometer-nest identifiers are shown along the land surface. Refer to Fig. 1 and Table 2 forpiezometer-nest and cross-section locations and identifiers. samples. These larger tritium activities always occurred in precipitation relative to Ames and Chicago, probably resulting from less influence of moisture from the Gulfbelow midslope (1S, 2S) and toeslope (shallow at 2T, 1Rc) positions (Fig. 4). Values of intermediate tritium of Mexico (Simpkins, 1995; Harvey and Welker, 2000). The central Nebraska site is more than 400 km westactivity (4–12 TU) were most frequent below riparianand toeslope positions (1T, 1B, 1Ra, 1Rb, deep at 2T, of the Deep Loess Research Station, across a steeptransition in climate and native vegetation from semi-2R), indicating recent or mixed-age waters. Samplescollected from the weirs had tritium activities of 11.1 arid short-grass prairie (west) to humid tall-grass prai-rie (east).and 12.5 TU for Watersheds 1 and 2, respectively, whichare not dissimilar given analytical errors near 1.0 TU. There was a distinct spatial pattern to the stable iso-tope data despite their small range in variation (Fig. 6).The baseflow samples would be considered of mixedorigin, with possibly a weak influence of 20- to 40-yr- Groundwaterbeneaththeupperlandscapepositions(DandS)wasisotopicallyenrichedcomparedwithtoeslopeold waters, and were consistent with TU values fromthe shallowest riparian-valley piezometers. (T) or riparian valley (R) positions (p 0.01), whether the stable isotope data were tested individually or com- bined on a single scale between 1 and 1. DifferencesStable Isotopes were significant, albeit small, with 18O averagingStable isotope concentrations ranged from 5.6 to 6.4‰ in upper landscape positions,7.1‰ at the toe-8.9‰ for 18O and from 39.6 to 60.9‰ for 2H.slopepositions,and 7.5‰intheriparianpositions,andGround water contains a mix of waters from multiple precipitation events, and therefore the range of data was smaller than is typically observed in precipitation samples. When the data were fit to estimate a local meteoric water line (Fig. 5), the limited range resulted inafairlylowprecision(r2 0.72)andthelocalmeteoric water line (LMWL) was not statistically different from the global meteoric water line (GMWL), or other LMWL published from the region (Harvey and Welker, 2000; International Atomic Energy Agency, 1992; Ma- theney and Gerla, 1996; Simpkins, 1995). The average 18O was 7.1‰ and the average 2H was 49.2‰. While some evaporative enrichment of ground water relative to local precipitation could occur (Gonfiantini et al., 1998), these averages are similar to those pub- lished for precipitation at Ames, IA (Simpkins, 1995) and Chicago, IL (International Atomic Energy Agency,Fig. 5. Plot of stable isotope data ( 18O‰ versus 2H‰) from 331992). The International Atomic Energy Agency did water samples collected in Watersheds 1 and 2. A local meteoricnot monitor these stable isotopes at Lincoln, NE, or water line (LMWL) fit to the data (solid line) has an r2 of 0.72.St. Louis, MO. There are data from central Nebraska Theglobal meteoricwaterline(GMWL; dottedline)is alsoplottedas a reference.(Harvey and Welker, 2000) showing isotopic depletion 2166 J. ENVIRON. QUAL., VOL. 32, NOVEMBER–DECEMBER 2003 Fig. 6. Cross-sections showing variations in relative isotopic enrichment or depletion, using a combined scale of 18O‰ and 2H‰, in ground water along three transects (W2, W1, W1rip) in Watersheds 1 and 2. Piezometers not sampled for stable isotopes are indicated by a crossed symbol. Piezometer-nest identifiers are shown along the land surface. Refer to Fig. 1 and Table 2 for piezometer-nest and cross-section locations and identifiers. 2H averaging 42.4‰ in upper landscape positions, the “amount effect.” That is, falling raindrops become 50.4‰ at the toeslope positions, and 52.6‰ in the more enriched by evaporation during small events than riparian positions. Also, classification of the data by during large events, because the atmosphere is likely landscape position accounted for 52% of the variance to be saturated with water vapor during large events in the 18O data and 60% of the variance in the 2H (Ingraham, 1998). data, based on a single-factor analysis of variance. This A third potential cause could be related to movement indicates there are differences in processes or sources of water vapor under winter conditions. Loess soils areaffecting ground water according to landscape position.knowntoexhibitfrostheave,whichresultsfromupwardThree possible mechanisms causing this difference are movement of water vapor from depth toward frozenupward movement of deep ground water, seasonal infil-surface soils. Vapor movement through soil is knowntration patterns, and landscape differences in water va-to be fractionating (Barnes and Turner, 1998), and inpor transport in soil.frozen soils would lead to enrichment of the deeper soilFirst, upward movement of deep ground water could water. In the lower landscape position, the water tablebe occurring at lower landscape positions. Older, deep would provide a shallow lower boundary for this pro-ground water might be depleted. While this was re-cess,andthereforeuponmeltthedepletedsoilicewouldported for deep waters of glacial origin in Wisconsin readily remix with the originating water. In the uplands,(Simpkins and Bradbury, 1992) and North Dakota (Ma-however, the process occurs under drier conditions withtheney and Gerla, 1996), it has not been reported this no fixed lower boundary. Complete remixing of meltedfar south of recent (i.e., Wisconsinan) glaciation. There soil ice to depth would not occur as readily and deeperwas no obvious depletion in piezometers below the till soil water that recharges ground water could remaincontact. In fact, the most depleted waters were found enriched. Such a phenomenon has not been docu-in the shallowest ground water at Positions 2R and 1B.mented, but is plausible.Thus, upward flow from the till aquitard is not sus-In sum, seasonal runoff and infiltration mechanismspected.are thought to be most responsible for the relative en-A second and more likely possibility is that seasonal richment of ground water below upland positions andchanges in runoff and infiltration act to segregate re-depletion in ground water below the toeslopes (Fig. 6).chargewaters. Snowmeltandcold springrains areisoto-Large relative depletion in the shallowest piezometerspically depleted, and occur when there is little plant at Positions 2R and 1B further support this. Processescover. These depleted waters would be most prone to of soil freezing and thawing may also contribute to therunoff from upper slopes and then infiltrate near the observed pattern.toeslope.Windscanalsoredistributesnowtowardlower landscapepositions.Warmersummerrains,ontheother hand, would be enriched and occur when there is crop Nitrate Nitrogen Concentrationscover that increases interception, infiltration, and tran-Deep Sedimentsspiration, and reduces runoff. Also, rain occurring dur-A large increase in sediment NO3–N concentrationsing small precipitation events islikely to infiltrate whereit falls, and this rain could, on average, be enriched by (mg kg 1) was observed in the deep cores taken during TOMER & BURKART: GROUND WATER NITRATE IN TWO SMALL WATERSHEDS 2167 between depths of increased NO3–N and cumulative baseflow, showing the consistent movement of a pulse of NO3–N through the deep unsaturated zone since the first experiment, in response to hydrologic fluxes through the watershed’s subsurface. If the baseflow re-sulted from deep percolation that occurred at spatiallyuniform rates, then this relationship suggests that 1 mof deep percolation caused NO3–N to percolate about5.6 m through the unsaturated zone. This ratio leads toan estimated “mobile” water content of about 0.18 m 3 m 3. This is about the difference between the averagevolumetric water content found during bulk density de-terminations on the deep cores (0.37 m 3 m 3) and esti-mated 1500 kPa water contents (about 0.20 m 3 m 3 ata 1-m depth),and therefore seems reasonable.A similar relationship with time suggests that the excess NO3–N percolated through this 20-m unsaturated zone at an average annual rate of 0.67 m yr 1 (r2 0.99; data not shown). These relationships provide a third line of evidence that some of the N from large fertilizer applications between 1969 and 1974 resided in the sub- surface of Watershed 1 until 1996. Given the inferred rate of movement of this NO3–N pulse to depth, one wouldanticipatethatthisNO3–Nwouldhavepercolated into the saturated zone before 1996 at lower landscape positions. This would explain why large sediment con- centrations were not observed at depth at lower W1 po-sitions.The total NO3–N storage in the 20-m-deep sedimentsat 1D was 930 kg ha 1, about two-thirds of which wasbelow 15 m. The deep profile at 2D, in contrast, showedabout 629 kg NO3–N ha 1 to a 20-m depth, with 60%Fig. 7. (A) Nitrate N concentrations with depth in soils and deep of this mass found within the top 10 m and only 6%sediments at Position 1D (Fig. 1), from core samples taken in 1996.below 15 m. A 15-m-depth sampling in 1984 (Alberts(B) Depths of maximum NO3–N concentrations measured in 1972,1974–1976, 1978, 1984 (Alberts and Spomer, 1985; Schuman et al.,and Spomer, 1985) found 1910 kg ha 1 in Watershed 1,1975), and 1996, plotted against cumulative baseflow from Water-and 1060 kg ha 1 in Watershed 2. The other profilesshed 1 since 1969.from our 1996 coring showed between 110 and 490 kg NO3–N ha 1, which varied according to the unsaturated1996 at Position 1D (Fig. 7), centered near the 17.5-m zone’sthickness.InathirdDeepLoessResearchStationdepth. This pattern was not observed in any of the other watershed, Steinheimer and Scoggin (1998) estimateddeep cores, and only 25 individual samples out of 323 that about 1200 kg N ha 1 could have accumulated incollected at the other coring positions showed NO3–N thedeeploessbetween1972and1990,undercontinuousconcentrations exceeding 4 mg kg 1. Total N and C data corn with N applications averaging 167 kg ha 1 yr 1.from these cores did not indicate this NO3–N couldoriginate from the sediments or buried paleosols (i.e.,Ground Water1D and 2D showed similar carbon contents that de- creased similarly with depth). If this NO3–N resulted The largest concentrations of NO3–N in water were found within the deep suction-cup lysimeters installedfrom large N fertilizer applications between 1969 and 1974,thentherewouldbesomeconsistencywithhistori- in upland positions (Fig. 8). The smallest concentrations occurred in the deepest ground water. Concentrationscal data on sediment NO3–N collected in 1972, 1974– 1976,1978,and1984(Schumanetal.,1975;Albertsetal., in the saturated zone decreased not only with depth, butalso withdistancedownslope. Thiscould resultfrom1977; Alberts and Spomer, 1985). Depths of increased NO3–N in sediments reported for each of these years an increasing effect of denitrification with residence time, increased carbon availability and denitrificationshowed consistent downward movement. These depths were plotted against the cumulative amounts of water rates in alluvial sediments, and/or dilution from greater infiltration of runoff waters near the toeslope position,that percolated though these watersheds since 1969 to each year of sampling (Fig. 7). Baseflow provides a which bear small NO3–N concentrations. These hypoth- eses are the subject of ongoing research at this site.surrogate measure of this percolation, and was takenfromstreamdischargerecordsseparatedintorunoffand While NO3–N concentrations commonly decreasedwith increasing depth, the vertical changes in NO3–Nbaseflow components (Kramer et al., 1999). The result(Fig. 7) showed a strong linear relationship (r2 0.98) concentrations were inconsistent below the toeslopes 2168 J. ENVIRON. QUAL., VOL. 32, NOVEMBER–DECEMBER 2003 Fig. 8. Cross-sections showing variations in NO3–N concentrations in water from piezometers and deep lysimeters along three transects (W2, W1, W1rip) as measured in June 2001. Lysimeters are indicated by filled circles located above the water table, whereas piezometers are located below the water table. Installations not sampled for nitrate are indicated by a crossed symbol. Piezometer-nest identifiers are shown along the land surface. Refer to Fig. 1 and Table 2 for piezometer-nest and cross-section locations and identifiers. (Fig. 8). Similar concentrations with depth occurred at mineralization of organic matter and unsaturated water contents may contribute to the patterns. There are un-Positions 2R and 1Rc, although there was greater tem- poral variation in the upper piezometer at 2R. Also, at published data that indicate large decreases in soil or- ganic matteroccurred inthe surfacesoils ofthese water-Positions1Band1Rb,concentrationsincreasedatdepth (Fig. 8)due todilution byrecharge waterand/or denitri- shedsduringthepast30years(TomMoorman,personal communication, 2002). Piezometers in Watershed 1 alsofication of upward-moving water. Depleted isotopic conditions suggest recharge occurred at Position 1B, showed significant temporal variation in NO3–N (i.e., 21-m depth at 1D; 14.5- and 16.5-m depth at 1S), includ-while upward hydraulic gradients occurred at both loca- tions and were larger at 1Rb. At 1Ra, there were larger ing the recent entry of a large pulse of deep NO3–N toground water at 1D. Note that piezometers at 2D wereconcentrations in the shallowest piezometer, comparedwith often nondetectable concentrations in deeper notsampledduetolowsamplingyieldandslowrecovery(30 d) of water levels after sample withdrawal.ground water (Fig. 9). A large downward vertical gradi-ent occurs at this position and the deeper waters are In the R and T positions, NO3–N concentrations inground water are smaller and most stable in Watersheddominantly pre-1950s in age (Fig. 4).Temporal changes have also occurred in NO3–N con- 2, with slow increases evident in Watershed 1 (Fig. 9).The larger NO3–N concentrations in Watershed 1 arecentrations in ground water and lysimeter waters duringmonitoring(Fig.9).Inparticular,therecordat1Dshows reflected in the baseflow concentrations measured atthe watershed outlet weirs (Fig. 10), which averagedthat a large pulse of NO3–N in the deep sediments(Fig. 7) entered the water table and caused NO3–N con- 20 mg L 1 in Watershed 1 and 12 mg L 1 in Watershed 2. A paired t test showed this difference to be significantcentrations to peak during the year 2000. The timing of these changes in concentration is consistent with the (p 0.01). Given Watershed 1’s average annual base- flow of 131 mm between 1975 and 2001, each 10 mgexpected movement of this pulse of NO3–N from the sediment into ground water, as the relationship with L 1 of NO3–N would equate to 13.1 kg N ha 1 yr 1 exported in baseflow. Therefore, very large concentra-baseflow (Fig. 7) would supportentry into ground water during 1999, given baseflow volumes of 144 mm in 1997, tionswouldbeneededoveraprolongedperiodtoexport an additional 1656 kg N ha 1 (i.e., the excess amount238 mm in 1998, and 321 mm in 1999 (Larry Kramer, personal communication, 2003). But, the NO3–N con- of N applied to Watershed 1 from 1969 to 1974). tributing water has little apparent effect on tritium in the upper piezometer (Fig. 5), where a dominant “pre-CONCLUSIONSbomb” age would be inferred from 2.8 TU. Large changes in NO3–N concentrations have oc- Three independent lines of evidence support the con- clusion that ground water concentrations of NO3–N incurred in upland (S and D) positions, particularly in the lysimeters in both watersheds (i.e., 6- and 12-m depths Watershed 1 are still influenced by large amounts of fertilizer N applied 30 years ago. Ground water time-at 2D, 6- and 9-m depths at 2S, 12 m at 1D, and 11 m at1S;seeFig.9).Thesevariationsinthelysimeterwaters of-travel estimates and tritium data both suggest thatground water remains resident in these watersheds forare not well explained by profiles of NO3–N in coresamples, and are asynchronous among each other and decades. Given such residence times, the differences inNO3–N concentrations between these two watershedswithgroundwaterrechargeeventsthatcausedincreasedwaterlevelsinearly1998and1999.Temporalchangesin could easily be the result of an experiment conducted TOMER & BURKART: GROUND WATER NITRATE IN TWO SMALL WATERSHEDS 2169 Fig. 10. Nitrate N concentrations in baseflow collected at the outletweirs of Watersheds 1 and 2. in lower landscape positions, runoff from upslope infil- tratesandmixesintogroundwater,andthereforehistor- ical and recent land use practices affect current NO3–N concentrations. Travel times across the riparian zone are estimated at around two or three years. But, due partlytofocusedrechargebelowthetoeslope,itappears that upslope ground water is being contributed to the riparian areas very slowly. Overall, it will be difficult to clearly discern the effect of recent cropping system changes between these two watersheds by monitoring ground water or baseflow for many years. Therefore, shallow monitoring of unsaturated-zone waters may be the most reliable means to examine effects of crop rota- tion on water quality. Ongoing research will assess the denitrification of ground water beneath this site’s ripar- ian zone. In conclusion, multiple lines of evidence gathered inFig. 9. Temporal changes in NO3–N concentrations in water collectedfrom lysimeters and piezometers at D, S, T, and R positions in these small watersheds suggest it takes at least severalWatersheds 1 and 2 (see Fig. 1). Open symbols denote suction-decadesforsubsurfacewatertotravelfromthedividetocup lysimeters, black-filled symbols denote piezometers in loess,the stream. In many watersheds, changes in agriculturaland gray-filled symbols denote piezometers near the till contact.practices may take several decades to fully effect im-Depths of monitoring (meters below surface) are noted above each plot. Nondetectable concentrations (1 mg L 1) are plotted at provements in ground water quality.zero. Symbols overlap in some plots. Some installations did not yield water samples every month (see Table 2).ACKNOWLEDGMENTS The authors thank the following individuals for their assis-between 1969 and 1974. Furthermore, analyses of tance.BobJaquis,LarryKramer,MikeSukup,ColinGreenan,NO3–N in deep sediments, considered with historical andAngieWiebershelpedwithfieldsampling.Wateranalysesdata, indicate that a pulse of soil NO3–N has percolated for NO3–N were conducted at the National Soil Tilth Labora-through the deep unsaturated zone since 1969, in re-tory analytical lab under the direction of Amy Morrow. Dr.sponse to subsurface hydrologic fluxes (baseflow)TomMoormanandBethDouglassprovidedsoilandsedimentthrough the watershed. Recent increases in ground wa-analyses. Bob Jaquis and Colin Greenan assisted with datater NO3–N concentrations near the watershed divide management and graphics, and David James provided thewere consistent with this sediment–NO3–N record and maps in Fig. 1. Helpful reviews of this manuscript were pro-its relationship with annual baseflow. These three lines vided by Dr. Bill Simpkins, Dr. Tom Moorman, Dr. Marv Shaffer, and three anonymous reviewers.of evidence, when taken together, suggest that heavy fertilizer applications between 1969 and 1974 continue REFERENCESto influence NO3–N concentrations in ground water within Watershed 1.Alberts, E.E., R.E. Burwell, and G.E. Schuman. 1977. Soil nitrate- nitrogen determined by coring and solution extraction techniques.Isotopic data suggest that ground water recharge in Soil Sci. Soc. Am. J. 41:90–93.these watersheds may frequently occur by infiltration Alberts,E.E.,andR.G.Spomer.1985.NO3–Nmovementindeeploessof runoff water at lower landscape positions. Results of soils. Paperno. 85-2030. Am.Soc. ofAgric. Eng., EastLansing, MI.stable isotope analyses (Fig. 6) suggest that runoff from Barnes, C.J., and J.V. Turner. 1998. Isotopic exchange in soil water.colder rains infiltrates at and below toeslope positions.p. 137–163.In C. Kendall and J. McDonnell (ed.) Isotope tracersin catchment hydrology. Elsevier Sci., Amsterdam.Tritium data (Fig. 4) also support recent and/or mixed Bettis, E.A. III. 1990. Holocene alluvial stratigraphy of western Iowa.origin waters at these landscape positions.Midwest Friends of the Pleistocene 37th Field Conference Guide-Concentrations of NO3–N in ground water beneath book. 12–13 May1990. Guidebook Ser. no. 9. IowaDep. of Naturalupslope positions in Watershed 1 are still influenced by Resour., Geol. Survey Bur., Iowa City.Blake, G.R., and K.H. Hartge. 1986. Bulk density. p. 363–375.In A.an experiment conducted from 1969 to 1974. However, 2170 J. ENVIRON. QUAL., VOL. 32, NOVEMBER–DECEMBER 2003 Klute(ed.)Methodsofsoilanalysis.Part1.2nded.Agron.Monogr. Harvey, F.E., and J.M. Welker. 2000. Stable isotope composition of precipitationinthesemi-aridnorth-centralportionoftheUSGreat9. ASA and SSSA, Madison, WI.Bolton, E.F., J.W. Aylesworth, and F.R. Hore. 1970. Nutrient losses Plains. J. Hydrol. (Amsterdam) 238:90–109.Hendry, M.J. 1988. Hydrogeology of clay till in a prairie region ofthrough tile drains under three cropping systems and two fertilitylevels on a Brookston clay. Can. J. Soil Sci. 50:275–279. Canada. Ground Water 26:607–614.Hvorslev, M.J. 1951. Time lag and soil permeability in groundwaterBoyce, J.S., J. Muir, A.P. Edwards, E.C. Seim, and R.A. Olson. 1976.Geologic nitrogen in Pleistocene loess of Nebraska. J. Environ. observations. Waterway Exp. Stn. Bull. 36. U.S. Army Corps ofEng., Vicksburg, MS.Qual. 5:93–96.Bradbury, K.R. 1991. Tritium as an indicator of groundwater age in Ingraham, N.L. 1998. Isotopic variation in precipitation. p. 87–118.In C. Kendall and J. McDonnell (ed.) Isotope tracers in catchmentcentral Wisconsin. Ground Water 29:398–404.Burgess, S.S.O., M.A. Adams, N.C. Turner, and B. Ward. 2000. Char- hydrology. Elsevier Sci., Amsterdam.International Atomic Energy Agency. 1992. Statistical treatment ofacterisation of hydrogen isotope profiles in an agroforestry system:Implications for tracing water sources of trees. Agric. Water Man- data on environmental isotopes in precipitation. Tech. Rep. Ser.no. 331. IAEA, Vienna.age. 45:229–241.Burkart, M.R., andJ.D. Stoner. 2001. Nitrogenin groundwater associ- International Atomic Energy Agency and World Meteorological Or-ganization. 2001. Global network forisotopes in precipitation data-ated with agricultural systems. p. 123–145.In R. Follett and J.Hatfield (ed.) Nitrogen in the environment: Sources, problems, base [Online]. Available at http://isohis.iaea.org (verified 8 July2003). IAEA, Vienna.and management. Elsevier Sci., Amsterdam.Burwell, R.E., G.E. Schuman, K.E. Saxton, and H.G. Heinemann. Karlen, D.L., L.A. Kramer, D.E. James, D.D. Buhler, T.B. Moorman, and M.R. Burkart. 1999. Field-scale watershed evaluations on deep1976. Nitrogen in subsurface discharge from agricultural water- sheds. J. Environ. Qual. 5:325–329. loess soils: I. Topography and agronomic practices. J. Soil Water Conserv. 54:693–704.Burwell, R.E., G.E. Schuman, K.E. Saxton, H.G. Heinemann, and R.G. Spomer. 1977. Nitrogen and phosphorus movement from Karlen, D.L., L.A. Kramer, and S.D. Logsdon. 1998. Field-scale nitro- gen balances associated with long-term continuous corn produc-agricultural watersheds. J. Soil Water Conserv. 32:226–230. Buttle, J.M., and K. Sami. 1990. Recharge processes during snowmelt: tion. Agron. J. 90:644–650. Kay,G.F.,andJ.B.Graham.1943.Thepre-Illinoianandpost-IllinoianAn isotopic and hydrometric investigation. Hydrol. Processes 4: 343–360. PleistocenegeologyofIowa.AnnualRep.38.IowaDep.ofNatural Resour., Geol. Survey Bur., Iowa City.Cambardella, C.A., T.B. Moorman, J.M. Novak, T.B. Parkin, D.L.Karlen,R.F.Turco,andA.E.Konopka.1994.Field-scalevariability Kehew, A.E., R.N. Passero, R.V. Krishnamurthy, C.K. Lovett, M.A.Betts, and B.A. Dayharsh. 1998. Hydrogeochemical interactionof soil properties in central Iowa soils. Soil Sci. Soc. Am. J.58:1501–1511. between a wetland and an unconfined glacial drift aquifer, south-western Michigan. Ground Water 36:849–856.Castillo, M.M., J.D. Allan, and S. Brunzell. 2000. Nutrient concentra-tions and discharges in a Midwestern agricultural catchment. J. Kendall, C. 1998. Tracing nitrogen sources and cycling in catchments. p. 519–576.In C. Kendall and J. McDonnell (ed.) Isotope tracersEnviron. Qual. 29:1142–1151. Coleman, M.L., T.J. Shepherd, J.J. Durham, J.E. Rouse, and G.R. in catchment hydrology. Elsevier Sci., Amsterdam. Kitchen, N.R., and K.W.T. Goulding. 2001. On-farm technologies andMoore. 1982. Reduction of water with zinc for hydrogen isotope analysis. Anal. Chem. 54:993–995. practicestoimprovenitrogenuseefficiencyp.335–369.InR.Follett and J. Hatfield (ed.) Nitrogen in the environment: Sources, prob-Daniels, D.P., S.J. Fritz, and D.I. Leap. 1991. Estimating recharge rates through unsaturated glacial till by tritium tracing. Ground lems, and management. Elsevier Sci., Amsterdam. Kramer,L.A.,M.R.Burkart,D.W.Meek,R.J.Jaquis,andD.E.James.Water 29:26–34. Dawson, T.E., and J.R. Ehleringer. 1998. Plants, isotopes, and water 1999. Field-scale watershed evaluations on deep loess soils: II. Hydrologic responses to different agricultural land managementuse: A catchment-scale perspective. p. 165–202.In C. Kendall andJ. McDonnell (ed.) Isotope tracers in catchment hydrology. Else- systems. J. Soil Water Conserv. 54:705–710.Logsdon, S.D., D.L. Karlen, J.H. Prueger, and L.A. Kramer. 1999.vier Sci., Amsterdam.DeWalle, D.R., P.J. Edwards, B.R. Swistock, R. Aravena, and R.J. Field-scale watershed evaluations on deep loess soils: III. Rainfalland fertilizer N use efficiencies. J. Soil Water Conserv. 54:711–716.Drimmie. 1997. Seasonal isotope hydrology of three Appalachianforest catchments. Hydrol. Processes 11:1895–1906. Mann, C.J. 1987. Misuse of linear regression in the earth sciences. p.74–106.In W.B. Size (ed.) Use and abuse of statistics in the earthDrimmie, R.J., A.R. Heemskerk, and J.C. Johnson. 1993. Tritium analysis. Tech. Procedure 1.0, Rev. 03. Environ. Isotope Lab., Dep. sciences. Oxford Univ. Press, New York. Matheney, R.K., and P.J. Gerla. 1996. Environmental isotopic evi-of Earth Sci., Univ. of Waterloo, Waterloo, ON, Canada. Drimmie, R.J., A.R. Heemskerk, W.A. Mark, and R.M. Weber. 1991. dence for the origins of ground and surface water in a prairie discharge wetland. Wetlands 16:109–120.Deuteriumbyzincreduction.TechProcedure4.0,Rev.02.Environ. Isotope Lab., Dep. of Earth Sci., Univ. of Waterloo, Waterloo, Nelson, W.M., A.J. Gold, and P.M. Groffman. 1995. Spatial and tem- poral variation in groundwater nitrate removal in a riparian forest.ON, Canada. Epstein, S., and T.K. Maeda. 1953. Variations of the 18O/16O ratio J. Environ. Qual. 24:691–699. Owens,L.B.,W.M.Edwards,andM.J.Shipitalo.1995.Nitrateleachingin natural waters. Geochim. Cosmochim. Acta 4:213. Fetter,C.W.2001.Appliedhydrogeology.4thed.PrenticeHall,Upper through lysimeters in a corn–soybean rotation. Soil Sci. Soc. Am.J. 59:902–907.Saddle River, NJ.Genereux, D.P., and R.P. Hooper. 1998. Oxygen and hydrogen iso- Peterjohn, W.T., and D.L. Correll. 1983. Nutrient dynamics in anagriculturalwatershed:Observationsontheroleofariparianforest.topes in rainfall runoff studies. p. 319–346.In C. Kendall and J.McDonnell (ed.) Isotope tracers in catchment hydrology. Elsevier Ecology 65:1466–1475.Prior, J.C. 1991. Landforms of Iowa. Univ. of Iowa Press, Iowa City.Sci., Amsterdam.Gold,A.J.,P.M.Groffman,K.Addy,D.Q.Kellogg,M.Stolt,andA.E. Randall, G.W., D.R. Huggins, M.P. Russelle, D.J. Fuchs, W.W. Nel- son, and J.L. Anderson. 1997. Nitrate losses through subsurfaceRosenblatt. 2001. Landscape attributes as controls on groundwater nitrate removal capacity of riparian zones. J. Am. Water Resour. tile drainage in CRP, alfalfa, and row crop systems. J. Environ. Qual. 26:1240–1247.Assoc. 39:1457–1464. Gonfiantini, R., K. Fro¨lich, L. Aragua´s-Aragua´s, and K. Rozanski. Ruhe, R.V. 1969. Quaternary landscapes in Iowa. Iowa State Univ. Press, Ames.1998. Isotopes ingroundwater hydrology. p. 203–246.In C. Kendall and J. McDonnell (ed.) Isotope tracers in catchment hydrology. Sauer, T.J., R.B. Alexander, J.V. Brahana, and R.A. Smith. 2001. The importance and role of watersheds in the transport of nitrogen.Elsevier Sci., Amsterdam. Hatfield, J.L., D.B. Jaynes, M.R. Burkart, C.A. Cambardella, T.B. p. 147–181.In R. Follett and J. Hatfield (ed.) Nitrogen in the environment: Sources, problems, and management. Elsevier Sci.,Moorman, J.H. Prueger, and M.A. Smith. 1999. Water qualitywithin the Walnut Creek watershed: Setting and farming practices. Amsterdam.Schilling, K.E., and R.D. Libra. 2000. The relationship of nitrateJ. Environ. Qual. 28:11–24. TOMER & BURKART: GROUND WATER NITRATE IN TWO SMALL WATERSHEDS 2171 concentrations in streams to row crop land use in Iowa. J. Environ. Spruill, T.B. 2000. Statistical evaluation of effects of riparian buffersQual. 29:1846–1851.onnitrateandgroundwaterquality.J.Environ.Qual.29:1523–1538.Schuman, G.E., T.M. McCalla, K.E. Saxton, and H.T. Knox. 1975.Steinheimer, T.R., and K.D. Scoggin. 1998. Agricultural chemicalNitrate movementand itsdistribution in thesoil profile ofdifferen-movement through a field-size watershed in Iowa: Subsurface hy-tially fertilized corn watersheds. Soil Sci. Soc. Am. Proc. 39:1192–drology and distribution on nitrate in groundwater. Environ. Sci.1197.Technol. 32:1039–1047.Simpkins, W.W. 1995. Isotopic composition of precipitation in central Taylor, C.B. 1977. Tritium enrichment of environmental waters byIowa. J. Hydrol. (Amsterdam) 172:185–207.electrolysis: Development of cathodes exhibiting high isotopic sep-Simpkins, W.W., and K.R. Bradbury. 1992. Groundwater flow, veloc-aration and precise measurement of tritium enrichment factors. p.ity, and age in a thick, fine-grained till unit in southeastern Wiscon-131–140.In Proc. Int. Conf. of Low-Radioactivity Measurementssin. J. Hydrol. (Amsterdam) 132:283–319.and Applications, The High Tatras, Czechoslovakia. 6–10 Oct.Soil Survey Staff. 1994. Keys to soil taxonomy. 6th ed. USDA Soil 1975. Slovenski Pedagogicke Nakladatelstvo, Bratislava, Slovakia.Conserv. Serv., Washington, DC.Walker, J.F., and D.P. Krabbenhoft. 1998. Groundwater and surfaceSpalding, R.F., M.E. Exner, C.W. Lindau, and D.W. Eaton. 1982.water interactions in riparian and lake-dominated systems. p. 467–Investigation of sources of groundwater nitrate contamination in 488.In C. Kendall and J. McDonnell (ed.) Isotope tracers in catch-the Burbank–Wallula area of Washington, USA. J. Hydrol. (Am- sterdam) 58:307–324. ment hydrology. Elsevier Sci., Amsterdam.