JEQ Grow Your Career With ASA
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online 1 May 2008
Published in J Environ Qual 37:1073-1085 (2008)
DOI: 10.2134/jeq2007.0010
© 2008 American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Google Scholar
Right arrow Articles by Green, C. T.
Right arrow Articles by Bekins, B. A.
PubMed
Right arrow Articles by Green, C. T.
Right arrow Articles by Bekins, B. A.
Agricola
Right arrow Articles by Green, C. T.
Right arrow Articles by Bekins, B. A.
Related Collections
Right arrow Vadose Zone Processes and Chemical Transport
Right arrow Ground Water Quality
Right arrow Nutrients

Nitrogen Fluxes through Unsaturated Zones in Five Agricultural Settings across the United States

Christopher T. Greena,*, Lawrence H. Fisherb and Barbara A. Bekinsa

a US Geological Survey, 345 Middlefield Rd., Menlo Park, CA 94025
b US Geological Survey, 160 N Stephanie St., Henderson, NV 89074

* Corresponding author (ctgreen{at}usgs.gov).

Received for publication January 5, 2007.

   ABSTRACT
TOP
 NOTES
 ABSTRACT
INTRODUCTION
 Site Descriptions
 Materials and Methods
 Results and Discussion
 Conclusions
 REFERENCES
 
The main physical and chemical controls on nitrogen (N) fluxes between the root zone and the water table were determined for agricultural sites in California, Indiana, Maryland, Nebraska, and Washington from 2004 to 2005. Sites included irrigated and nonirrigated fields; soil textures ranging from clay to sand; crops including corn, soybeans, almonds, and pasture; and unsaturated zone thicknesses ranging from 1 to 22 m. Chemical analyses of water from lysimeters and shallow wells indicate that advective transport of nitrate is the dominant process affecting the flux of N below the root zone. Vertical profiles of (i) nitrogen species, (ii) stable isotopes of nitrogen and oxygen, and (iii) oxygen, N, and argon in unsaturated zone air and correlations between N and other agricultural chemicals indicate that reactions do not greatly affect N concentrations between the root zone and the capillary fringe. As a result, physical factors, such as N application rate, water inputs, and evapotranspiration, control the differences in concentrations among the sites. Concentrations of N in shallow lysimeters exhibit seasonal variation, whereas concentrations in lysimeters deeper than a few meters are relatively stable. Based on concentration and recharge estimates, fluxes of N through the deep unsaturated zone range from 7 to 99 kg ha–1 yr–1. Vertical fluxes of N in ground water are lower due to spatial and historical changes in N inputs. High N fluxes are associated with coarse sediments and high N application rates.

Abbreviations: ET, evapotranspiration • GC, gas chromatograph • WTF, water table fluctuation


   INTRODUCTION
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
Site Descriptions
 Materials and Methods
 Results and Discussion
 Conclusions
 REFERENCES
 
LEACHING of agriculturally applied nitrogen (N) from the unsaturated zone to the ground water is of great concern worldwide (Hallberg, 1987). Beneath agricultural lands in the USA, approximately 19% of shallow wells have NO3– concentrations exceeding the US Environmental Protection Agency's Maximum Contaminant Level, and median NO3– concentrations at these locations are higher than in ground water below urban lands or in major aquifers (Nolan and Stoner, 2000). Because of the large extent of N contamination, there is considerable interest in understanding the processes that control transport of N through the unsaturated zone to the water table in agricultural settings. Although many studies have focused on the root zone, relatively few studies have examined the fate and transport of N in the deep unsaturated zone (Onsoy et al., 2005), defined here as the subroot portion of the unsaturated zone. The major focus of this paper is to examine the fate and transport of N in the deep unsaturated zone beneath five agricultural settings.

For agricultural lands, the soil N balance and speciation of N in the unsaturated zone is often poorly understood (Jenkinson, 2001). Studies of subsurface N in agricultural settings most often focus on NO3– because it occurs frequently as a ground water contaminant. However, relatively high concentrations of dissolved organic N can occur beneath agricultural fields (Dick et al., 2000), and it has been postulated that organic N may be an important component of the N flux to ground water (Murphy et al., 2000). Organic species of N may be formed from NO3–, NH4+, or N2 fixation and can subsequently undergo mineralization or transport in dissolved or colloidal form. Ammonium tends to adsorb to shallow sediments and organic matter, where it can volatilize to the atmosphere or oxidize to NO3–. Nitrite (NO2–) forms as an intermediate of NO3– reduction or nitrification (oxidation of NH4+ to NO3–), although concentrations are typically low due to rapid oxidation or reduction. In this work, we address the issue of the relative influence of the various N species affecting fate and transport of N below the base of the root zone.

Nitrate is typically mobile in solution but may undergo transformations, such as uptake or denitrification. Denitrification requires very low oxygen concentrations and the presence of electron donors, such as reactive organic C or reduced minerals. Below the soil zone, organic C contents are usually low, leading to lower reaction rates and smaller microbial populations (Kieft and Brockman, 2001). Under unsaturated conditions, direct connection of the gas phase to atmospheric oxygen results in aerobic conditions except in locations with high water saturations and significant concentrations of reduced species. Thus, it cannot be assumed that denitrification occurs reliably below the soil zone. Recent results from arid and semiarid sites indicate that NO3– moves conservatively through the deep unsaturated zone (Stonestrom et al., 2003; Walvoord et al., 2003; McMahon et al., 2006). In USA ground waters, researchers have found positive correlations of higher NO3– concentrations beneath thicker unsaturated zones (Burkart et al., 1999; Nolan et al., 2002). Yet, ground water vulnerability assessments commonly assume that longer unsaturated zone residence times result in reduced NO3– concentrations due to a greater degree of reaction (National Research Council, 1993). To resolve this discrepancy, we examine whether denitrification below the soil zone significantly affects the flux of NO3– to the water table.

Quantifying and understanding the controls on the rate of N leaching below the soil are essential to managing N flux to ground water. Estimates of leaching rates have been based on data from the base of the soil zone or from ground water (Harter et al., 2005), but both of these methods present problems. Datasets from the shallow soil or from the base of the root zone are plagued by temporal and spatial variability (Onsoy et al., 2005), and ground water data are typically difficult to relate to specific source areas (Green et al., 2008). Thus, better methods are needed for making reliable estimates of N flux at the water table so that these can be related to the impact of specific agricultural practices. Broad principles regarding controls on N leaching, such as soil texture, recharge, and application rates are widely used for ground water vulnerability analysis (National Research Council, 1993). Methods for weighing the relative importance of these factors in different agricultural settings around the country need to be better tied to process understanding (National Research Council, 1993). To advance the understanding of controls on N leaching to ground water, we use the data collected from five sites to examine different methods for calculating N flux. We then use the best estimates of N flux to rank the sites and discuss the key leaching processes.

The transport and reactions of N within the unsaturated zone in five agricultural settings across the USA were investigated as part of the Agricultural Chemicals: Sources, Transport and Fate study of the National Water-Quality Assessment Program of the US Geological Survey. The overall study focused on understanding how hydrologic processes affect the fate of agricultural chemicals in some of the nation's most important agricultural settings (Capel et al., 2008). A single-study approach was used to characterize the speciation and concentration of N in the deep unsaturated zone. The results provide insight into (i) the relative importance of the various N species in the fluxes of N below the root zone, (ii) whether N transformations or chemical interactions are important natural attenuation mechanisms in the deep unsaturated zone, and (iii) how the fluxes of N from the land surface to the ground water compare among study sites and relate to the physical properties and agricultural practices at those sites.


   Site Descriptions
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 Site Descriptions
Materials and Methods
 Results and Discussion
 Conclusions
 REFERENCES
 
The five study sites were chosen to represent important agriculture production areas of the USA. Capel et al. (2008) presents an overview of the uniform study design, instrumentation, and chemical analyses applied at all the sites. This study focuses on 10 unsaturated zone sites that overlie ground water sampling locations (Green et al., 2008) to allow comparison of saturated and unsaturated zone chemistry. Figure 1 shows maps of the unsaturated and saturated zone sites at the five study areas—in the San Joaquin River basin, California (CA); the Sugar Creek basin, Indiana (IN); the Chester River basin on the Delmarva Peninsula, Maryland (MD); the Elkhorn River basin, Nebraska (NE); and the Yakima River basin, Washington (WA). Detailed descriptions of each study site can be found, respectively, in Gronberg and Kratzer (2006), Lathrop (2006), Hancock and Brayton (2006), Fredrick et al. (2006), and Payne et al. (2007).


Figure 1
View larger version (89K):
[in this window]
[in a new window]

  		Fig. 1. Aerial maps showing study locations and modeled recharge source locations (Green et al., 2008) for (A) California, (B) Indiana, (C) Maryland, (D) Nebraska, and (E) Washington. The Indiana site was not studied by Green et al. (2008), and particle tracking results of recharge source locations are not available.

 
Table 1 summarizes aspects of the hydrology, sediment texture, and N loading rates at the five sites. Climates span semiarid to humid, with precipitation rates of 18 to 112 cm yr–1. Crop water demands not met by precipitation are supplied by irrigation at CA, WA, and NE. Recharge was calculated using water table fluctuation (WTF) analyses by Fisher and Healy (2008) for the CA, IN, MD, and WA sites and as part of this study for the NE site (see Methods section). Recharge values are generally lower at NE and WA, which have relatively thick unsaturated zones and fine-grained surficial sediments. Recharge is greater at CA and MD, where sediments are relatively coarse, and at IN, where the water table is very shallow.


View this table:
[in this window]
[in a new window]

  		Table 1. Site characteristics.

 
Application rates of N for each site are based on county and local estimates (Table 1). For county estimates, the quantity of N applied per year is derived from chemical fertilizer, manure, and atmospheric N loadings presented by Ruddy et al. (2006) using 1987 to 2001 loads for chemical fertilizer, 1982 to 1987 loads for manure, and 1985 to 2001 loads for atmospheric deposition. The rate per area was estimated by dividing by the estimated area of croplands in 2002 in each county from the Census of Agriculture (http://151.121.3.33:8080/Census [accessed 8 June 2007]). Local N-application rates were obtained from estimates by land owners of the study fields (CA, IN, and MD) or from typical local usage for the existing crop types (NE and WA) (Fredrick et al., 2006; McKenzie, personal communication, 2004). Local flux values are the most direct estimate of the current application rates in a particular field. County rates are also relevant because they include historical information that is not available locally and can provide a better estimate of application rates across the broad ground water source areas that include multiple fields (Green et al., 2008). Although the crops grown at CA, IN, MD, and NE are common in those counties, the county N application rate estimates exceed the local estimates at all four sites.

The CA unsaturated zone study site was located near the edge of an unpaved access road in an almond [Prunus dulcis (Mill.) Webb] orchard. The grower kept the rows between the almond trees free of plant growth by using herbicides or tilling. Sprinkler irrigation of 10 cm of water from surface diversions along with dissolved N fertilizer ("fertigation") occurred biweekly from early March through August, when irrigation was stopped to induce ripening. After harvest and before dormancy, the trees were irrigated once with 20 cm of water without dissolved fertilizer.

The Indiana study site had two unsaturated zone sampling locations in tile-drained fields that were planted with soybeans [Glycine max (L.) Merr.] in 2004 as part of a crop rotation of corn [Zea mays L.] and soybeans. The topography is flat, and water ponds on the land surface after large precipitation events. The fields underwent reduced tillage before spring planting, and no winter cover crop was planted.

At the Maryland site, samples were collected from the unsaturated zone at three locations in the field: an uphill recharge area, a mid-field location, and at the lower edge of the field along a wooded riparian zone. In an annual to biannual crop rotation of soybeans and corn, soybeans were planted in 2004. The field was under conservation no-tillage practices with winter wheat [Triticum aestivum L.] grown as ground cover from October of 2003 through April of 2004. The cover crop was killed with herbicide in late April and disked into the shallow soil before planting soybeans on 26 May 2004.

The Nebraska study site had three unsaturated zone locations: in a cornfield, at the edge of the same cornfield, and in a riparian zone. The field was not tilled before planting and was irrigated with a center pivot sprinkler during the growing season. Corn was grown on the site in 2004 as part of a corn–soybean crop rotation.

The Washington unsaturated zone site was at the top of a furrow-irrigated field. Irrigation water came from surface reservoirs receiving snowmelt from the upper reaches of the Yakima River drainage basin. Historically, a variety of crops has been grown on the field, including asparagus [Asparagus officinalis L.] before 2001, pumpkins [Cucurbita pepo L.] in 2001 and 2003, and corn in 2002. Corn was planted in 2004. The site underwent conventional subtilling to a depth of about 60 cm before planting.


   Materials and Methods
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 Site Descriptions
 Materials and Methods
Results and Discussion
 Conclusions
 REFERENCES
 
Data Collection
Solid and pore water chemistry and hydrology of the unsaturated zone were studied from early 2004 to mid-2005, with the most intensive data collection typically occurring during the growing season of 2004. Over the course of the study, most lysimeters were sampled six or more times. Water samples were analyzed for nutrients and major ions (Capel et al., 2008). Specialized analyses and other techniques used in this study are described below.

At each of the unsaturated zone sites, suction and pan lysimeters were installed at 3 to 6 depths to span the thickness of the unsaturated zone from the base of the soil zone to the capillary fringe. The suction lysimeters were 30 cm in length and 5 cm in diameter, with a 5-cm ceramic porous cup. Pan lysimeters collected free-draining water in 30 by 30 cm stainless steel pans. A collection tube funneled captured water to a reservoir. Pan lysimeters and some of the shallowest suction lysimeters were installed through the walls of an instrumentation trench. Deeper suction lysimeters were installed in bore holes made with hand augers or by drill rigs using hollow stem augers. Water samples were obtained from suction lysimeters by applying a vacuum to the lysimeter for a minimum of 3 h to draw water from the soil through the porous ceramic cup and into the lysimeter. Water samples were recovered from suction and pan lysimeters by pressurizing the lysimeter via one tube to drive water upward through a second tube into a clean glass sample bottle. Unsaturated zone air samples were collected by drawing gas through tubes installed with an open end in the sediment outside of the porous cup lysimeters.

Monitoring wells were installed in the aquifers underlying the unsaturated zone sites. Wells were constructed of 5-cm-diameter polyvinyl chloride with slotted polyvinyl chloride screens and typically were installed using hollow-stem augers. The annulus around each screen was filled with silica sand, and the remaining annulus was grouted with bentonite, with a cement cap at land surface. At each well nest, multiple well screens were installed, typically with 1.5-m-long screens at the water table and 0.6-m-long screens at two to four regularly spaced depths below the water table. Maximum depths of screens below the water table were approximately 2 m at IN, 10 m at MD and NE, and approximately 25 m at CA and WA. Most well screens were located in sandy sediments. Additional details of the ground water wells are described by Green et al. (2008).

Soil and sediment cores were collected during installation of lysimeters. Cores were typically 60 cm long and 5 cm in diameter and were collected at intervals of approximately 1.5 m between the land surface and the water table. Sediments were analyzed for major ions, bulk density, organic matter content, grain size, and hydraulic properties, as described in Capel et al. (2008).

To investigate the possibility of in situ N reactions, stable isotope and gas analyses were run on samples of unsaturated zone air and dissolved NO3– collected from site N22. This site was chosen as the most likely location to have deep unsaturated zone denitrification based on detection of N2 production in samples from the shallowest monitoring well (Green et al., 2008). Limited resources prevented collection of gas samples from other sites. Gas samples were analyzed for concentrations of N2O and for stable isotopes of N in N2 ({delta}15N[N2]) and O in O2 ({delta}18O[O2]). Pore water samples were analyzed for stable isotopes of N and O in NO3– ({delta}15N[NO3–] and {delta}18O[NO3–]). Pore water samples were collected and analyzed for stable isotopes as described in Green et al. (2008). For analyses of N2O and O2 concentration in unsaturated zone air, samples were collected in 10-mL glass bottles that had been filled with deionized water and sealed with butyl stoppers and crimp-seals. Using a peristaltic pump, gas was drawn from sampling tubes and injected into the sample bottle until all water was expelled and an additional five sample volumes of gas had passed through the bottle. After each 10-mL sample, a second 125-mL sample bottle was collected after purging with 100 sample volumes to account for possible bias related to sample volumes. Differences in results between large and small volume samples were within the range of analytical error. Nitrous oxide and O2 concentrations were measured in an electron capture gas chromatograph (GC). Gas ratios and isotopes in unsaturated zone gas samples were measured by mass spectrometry (modified from Revesz et al., 1999). Serum bottles containing gas samples were overpressured by 0.5 atm with He, and the gases were released into a closed loop and flushed in a He carrier stream through a molecular sieve GC to separate N2 from O2 and Ar. The outflow from the GC was admitted to a Finnigan Delta Plus XP mass spectrometer (Thermo Scientific, Waltham, MA), and the gas peaks were monitored simultaneously. Integrated peak areas were converted to O2/Ar, N2/Ar, {delta}15N[N2], and {delta}18O[O2] values and calibrated by analyzing air samples collected in the same way as the field samples.

Mass Flux Analysis
The mass flux of N from the unsaturated zone to the water table was estimated at each lysimeter site and well screen. Neglecting dispersion, the mass N flux, FN (M L–2 T–1), to the water table at a fixed point during a time interval of length T is defined by

Formula 1[1]
where r(t) is the vertical flux of water at the surface of the water table (L T–1), and CN(t) is the resident concentration of N (M L–3). For N flux occurring during discrete recharge events, Eq. [1] becomes

Formula 2[2]
where ri is the vertical water flux during event i, and CN,i is the concentration (assumed constant) during event i. When the resident concentration remains constant for all recharge events, Eq. [1] further simplifies to

Formula 3[3]
where R is the total vertical water flux for all events, and CN is the concentration.

Three sets of N flux calculations were performed. First, modern, local N fluxes were calculated for the deep unsaturated zone using Eq. [2]. Values of ri for individual recharge events were taken from WTF analyses for the CA, IN, MD, and WA sites presented by Fisher and Healy (2008) and from a WTF analysis completed as a part of this study for the NE site (discussed below). A concentration was determined for each recharge event by linear interpolation between the closest sample dates before and after the recharge event for the deepest unsaturated zone lysimeter. Second, to evaluate the use of simplified concentration data and to examine spatial trends, the N flux was calculated for selected unsaturated zone sites using Eq. [3] with the total annual WTF recharge (Table 1) and representative concentration values: the median, first quartile, third quartile, minimum, and maximum concentration at each deep lysimeter were applied to Eq. [3] to show the possible range of fluxes. Third, values of FN were estimated using Eq. [3] at individual ground water wells. For each well, CN was the source concentration, equal to the sum of N in the form of NO3– (NO3––N) and excess N2 from Green et al. (2008), and a single value of R was determined for ground water at each site using the method of Vogel (1967). For an idealized homogeneous aquifer with uniform thickness and hydraulic properties, the vertical flux, R, of ground water can be calculated with

Formula 4[4]
where n is porosity (L3 L–3), Z is the total saturated thickness of the aquifer (L), tt is the travel time from the water table to the ground water monitoring well (T), and d is the depth of the sample location (L). Values of n were estimated from the density of saturated zone cores, using

Formula 5[5]
where {rho}b is the bulk density (M L–3), and {rho}s is the density of solid grains, assumed to be 2.68 g cm–3 for these calculations. The resulting values of n were 0.38 for CA, 0.29 for IN, 0.37 for MD, 0.43 for NE, and 0.38 for WA. The values of Z used in Eq. [4] and estimated from drillers' logs and core samples were 31 m for CA, 10 m for IN, 16 m for MD, 31 m for NE, and 100 m for WA. For CA, MD, WA, and NE, values of tt were taken from the mixed ground water ages estimated by Green et al. (2008), which range from 0 to 61 yr. For IN, a single piston-flow estimate of 15 yr was made for I34p on the basis of a comparison of SF6 equilibrium partial pressures, corrected for recharge temperature and excess air with chronologies of atmospheric concentrations (Busenberg and Plummer, 2000). At each site, nonlinear least squares regression was used to fit the computed ground water age profile from Eq. [4] to the data.

At the NE site, water-level data were unsuitable for standard WTF analysis because numerous ground water responses to river stage fluctuations were not easily distinguishable from recharge responses. Because Cl–-tracer estimates of recharge were available for a nonirrigated location on the well transect (Nolan et al., 2007) but not for the irrigated locations, the WTF approach was adapted to quantify the additional recharge in irrigated areas in excess of the ambient recharge as measured by the Cl–-tracer estimates. Due to irrigation and focused infiltration at N22, additional recharge caused a locally elevated water table, or "mound," relative to the ground water below the unirrigated sites, N21 and N20. The additional recharge, ri*, was quantified by applying the WTF method to the height of the water table mound at N22.

Formula 6[6]
The specific yield, Sy (L3 L–3), was estimated to be 0.15 using the methods described by Fisher and Healy (2008). The height of the water table mound, Hi (L), was set equal to the water table elevation at N22p minus the water table elevation at N20p, and the change in the height, {Delta}Hi (L T–1), was estimated from trends in continuous data-logger records. The total recharge per event, ri, was calculated by summing the additional recharge and the ambient recharge, ri', arriving at irrigated and nonirrigated locations.

Formula 7[7]
The ri' value of 4.8 cm yr–1 was taken from an estimate at N21 (Nolan et al., 2007), which is close to N20 and has similar land use and vegetation.

The WTF analysis of NE data indicates that, in 2005, the approximate total recharge at N22 was 15.9 cm, based on additional recharge, ri*, of 11.1 cm yr–1. Of the additional recharge, 7 cm arrived between February and April, probably due to focused infiltration of spring snow melt and rains before planting, and 4.1 cm arrived between August and November, likely from irrigation events. Based on estimates of local annual evapotranspiration (ET) for irrigated corn of 71.9 cm (Nebraska Department of Natural Resources, 2005), precipitation in 2005 of 60.5 cm, ambient recharge of 4.8 cm, and irrigation-associated recharge of 4.1 cm, a simple water balance (irrigation = ET + recharge – precipitation) implies an irrigation quantity of 20.3 cm. This value is close to the estimated irrigation requirements of 18.0 cm for corn in this region (Nebraska Department of Natural Resources, 2005) and indicates that the recharge estimates are reasonable values.


   Results and Discussion
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 Site Descriptions
 Materials and Methods
 Results and Discussion
Conclusions
 REFERENCES
 
Speciation
This section describes the relative concentrations of the various N-species in the unsaturated zones and below the water table. The comparison serves in later sections of this study as a basis for further analysis of N-cycling and fluxes.

Overall, NO3– is the dominant N species, with all other N species comprising a small portion of the dissolved N in the unsaturated zone and in the shallow ground water (Fig. 2 ). In the unsaturated zone, median concentrations of NO3––N range from 0.1 to 25.1 mg L–1. Organic N is typically detected at lower median concentrations ranging from 1.5 to 10.9 mg L–1. Nitrite and NH4+ are detected infrequently at concentrations usually less than 1 mg L–1. Shallow ground water also contains higher concentrations of NO3––N than of organic N, NO2––N, or NH4+–N. Median organic N was 0 mg L–1 at nine ground water sites, and the highest median value was only 0.9 mg L–1. Ammonium and NO2– are rarely detected in monitoring wells, and maximum concentrations never exceed 0.3 mg N L–1.


Figure 2
View larger version (26K):
[in this window]
[in a new window]

  		Fig. 2. Distribution of concentrations of N species in samples from lysimeter sites and ground water sites. Gray boxes are shallow ground water samples. Black boxes are pore water samples collected from lysimeters. The points inside the boxes show the medians, the boxes show the interquartile ranges, and the whiskers show the total range. Locations C20, C21, N21, W20, W21, W23, and W24 have ground water wells with no lysimeters. All other locations have ground water wells and lysimeters.

 
Spatial and Temporal Distribution of Nitrogen Species
Characterization of the timing and location of peak NO3– concentrations provides insights into the factors that control the distribution of N concentrations and fluxes. In this section, temporal and spatial patterns are analyzed to determine (i) how physical factors such as application rates and ET affect differing concentrations among sites, (ii) how depth relates to variability of concentrations, and (iii) how temporal variations in concentration relate to seasonal factors such as microbial activity in shallow soils, plant uptake, and fertilization practices. These analyses serve as a basis to understand the factors influencing the N flux estimates that are presented later.

Concentrations of N differed among the sites. Median unsaturated zone NO3––N concentrations were highest (often >20 mg L–1) at CA, MD, NE, and WA (Fig. 2). Lower concentrations occurred at the NE riparian zone site (N20) and at the in IN sites (I30, I31). Based on Kruskal-Wallis one-way ANOVA on ranks (Systat Software Inc., 2006) of all NO3––N measurements at each site, these differences were significant (P < 0.001).

The variability of concentrations among sites relates to differences in application rates, fraction of leached N, and hydrology. Many studies have observed that unsaturated zone concentrations tend to increase with increasing application rates (Hallberg, 1987; Böhlke, 2002). Evapotranspiration is also important because, for a given water input and mass flux of N, greater ET implies lower volumes of leachate and higher concentrations of solute. The relations among concentration, ET, and mass flux of NO3––N are shown in Fig. 3 , where the y axis is the ratio of the observed unsaturated zone concentrations (Cobs = the average of median NO3––N values from all lysimeters at each site) to the applied concentration (Capp = total applied N/total water inputs; Table 1), the x axis is the fraction of applied water lost to ET (from water flux values in Table 1), and the curves show the expected increase of Cobs/Capp with increasing ET for various fractions of applied N leaching below the root zone (N leached/N applied). The plot serves two purposes: (i) The fraction of leached N can be determined from concentration and hydrologic data, and (ii) the relative effects of the plotted factors can be determined.


Figure 3
View larger version (21K):
[in this window]
[in a new window]

  		Fig. 3. Relation between concentration enrichment and the fraction of applied water lost to evapotranspiration (ET) (from values in Table 1) for the agricultural profiles at California (CA), Indiana (IN), Maryland (MD), Nebraska (NE), and Washington (WA) and for the NE riparian zone site. Concentration enrichment is defined as Cobs/Capp, where Cobs is the observed unsaturated zone concentration based on the average of median NO3––N values from all lysimeters at each site and Capp is the applied concentration based on the total applied N divided by the total water inputs. Applied N rates were taken from local estimates (filled circles) and county estimates (empty circles) as indicated in Table 1. The curves show the expected increase of Cobs/Capp with increasing ET for various leached fractions of N (N leached below the root zone/N applied).

 
Variations in ET losses and leached fractions of N are important factors influencing the unsaturated zone concentrations at these sites (Fig. 3). At IN, low concentrations are associated with low Cobs/Capp due to (i) dilution in large amounts of vertical water flux resulting from low ET and (ii) a low fraction of N leaching from the surface. At the other agricultural sites, high concentrations and high Cobs/Capp are associated with higher ET and larger fractions of leached N. In particular, the sites with sandy soils (CA and MD) show high leached fractions of N, whereas the finer grained soils at IN and WA have lower leached N fractions. At the NE riparian site, intermediate Cobs/Capp results from high ET and the lowest estimated fraction of N leaching. The low unsaturated zone concentrations at this location result from very low N inputs. Most of the plotted values are consistent with a range of leached N fractions, from 0.1 to 0.5, that has been observed at other agricultural sites (see Böhlke, 2002).

Among the five agricultural sites, the higher N concentrations occur in thicker unsaturated zones (CA, MD, NE, and WA), as has been observed nationwide for concentrations in ground water (Burkart et al., 1999; Nolan et al., 2002). In this study, the higher concentrations are associated with higher fractions of water loss to ET. Greater enrichment of N concentrations by ET at locations with deeper water tables provides a feasible mechanism to explain the national trend.

The variability of N concentrations is typically lower in deep lysimeters than near the root zone. Figure 4 shows the medians, interquartiles, and ranges of the concentration profiles for each lysimeter nest. The reduced variability with depth is consistent with depth trends reported by Onsoy et al. (2005) for core samples in a deep unsaturated zone in the San Joaquin Valley, CA. In some cases, reduced variability of concentrations may result from occasional inundation by ground water of the zone sampled by lysimeters. However, at many locations, pronounced differences between the median concentrations in ground water and in the deepest lysimeter (e.g., M22, N22, and W22) and several meters of separation between deep lysimeters and the water tables (e.g., M21, M22, and N23) suggest that other factors affect concentrations in the deep unsaturated zone. Most likely, dispersion reduces variability in deeper lysimeters by smoothing out high-frequency concentration fluctuations.


Figure 4
View larger version (21K):
[in this window]
[in a new window]

  		Fig. 4. Depth profiles of total N and NO3––N concentrations in the unsaturated zone and shallow ground water (deepest sample points) for each study site. For total N, bold lines connect the medians, boxes show interquartile range, and whiskers show the total range. For NO3––N, fine lines connect the medians, shown as points. The width between the two lines corresponds to the concentration of organic N + NH4+–N + NO2––N, which consists primarily of organic N at most locations.

 
Time-series plots of NO3––N concentrations in Fig. 5 indicate that the variability in shallow lysimeters is associated with seasonal concentration peaks that relate to changes in soil microbial activity. Below agricultural fields at the CA (Fig. 5a), MD (Fig. 5c), NE (Fig. 5e), and WA (Fig. 5f) sites, NO3––N concentrations rise rapidly in the early summer after the start of the growing season, peak in June or July, and decline before harvest in September or October. At the unfarmed NE riparian zone site (Fig. 5d), NO3––N concentration fluctuations follow the same trend, indicating that factors other than farm practices control the timing of these concentration peaks. Similar seasonality of NO3––N concentrations has been observed at other unfarmed landscapes due to natural cycles of mineralization and nitrification during the growing season (Gosz and White, 1986). Nitrification rates tend to increase with warmer soil temperatures (Alexander, 1965) and, conversely, are often negligible at temperatures below 4 to 5°C (Schmidt, 1982).


Figure 5
View larger version (21K):
[in this window]
[in a new window]

  		Fig. 5. Time series of NO3––N concentrations (mg L–1) for shallow lysimeters (solid lines) and deep lysimeters near the water table (dashed lines) for (A) California, (B) Indiana, (C) Maryland, (D) Nebraska riparian zone, (E) Nebraska corn field, and (F) Washington. Shallow lysimeters are between 1 and 2 m below ground surface, except N22b, which is 7 m below ground surface. The bracket at the top of each plot marks the 2004 growing season for that site. For California, the brackets span the interval between the first and last irrigation events. For all other sites, the brackets show the time between planting and harvesting.

 
The sharpness and magnitude of the mid-summer NO3––N concentration peaks likely relate to the fertilization practices and plant N uptake dynamics at each site. Concentrations in lysimeters under highly fertilized corn fields (NE and WA) show steep rises in NO3––N in the early growing season (Fig. 5e and 5f) when rooting depth is shallow and plant N uptake is low. Among sites with minimally fertilized soybeans (IN and MD), peaks are absent (Fig. 5b) or broader (Fig. 5c). At the CA orchard, the seasonal peak in N is absent at the shallow lysimeter, but a small peak was present on 4 Aug. 2004 in the deepest lysimeter (Fig. 5a). The lack of a strong peak may result from application of fertilizer throughout the growing season (approximately biweekly from March to August).

Reactions during Unsaturated Zone Transport
This section describes the extent of reactions that potentially affect the N balance at these five deep unsaturated zone sites. Because NO3– is the dominant N-species, denitrification of NO3– is the primary reaction potentially affecting the N budgets. The extent of denitrification is explored based on major ion chemistry, stable isotopes, unsaturated zone gases, and dissolved O2 in shallow ground water. Other reactions, such as mineralization and nitrification, are considered. Observations of hydrology and chemical trends suggest limited influence of mineralization, nitrification, or denitrification on the fate of N between the root zone and capillary fringe. In the cases where reactions occur, they are limited to the soil zone, the water table, or a few ephemeral reducing zones near the shallow lysimeters.

Correlations between N and other agricultural chemicals indicate that physical processes and loading rates are the most important controls on the concentration and distribution of N. Among unsaturated zone lysimeter samples, total N, which consists primarily of NO3–, often correlates significantly (Pearson product moment, P < 0.05) with other agricultural chemicals, such as Mg2+, Ca2+, SO42–, Na+, and Cl– (Table 2 ). Similar correlations have been observed in ground water at these sites (Green et al., 2008) and at other sites where transport is primarily conservative (Böhlke and Denver, 1995). Such correlations can result from fertilizers applied with other agricultural chemicals, such as dolomitic lime, potash, and gypsum. Significant correlations in the unsaturated zones imply that N is transported in a similar manner to the accompanying agricultural chemicals and does not undergo extensive reactions.


View this table:
[in this window]
[in a new window]

  		Table 2. Significant (P < 0.05) Pearson product moment correlations (r) and number of sample pairs (n) for total N and selected major ions.

 
Gas and stable isotopes in unsaturated zone air and water from the NE site were analyzed to determine whether N-cycling occurs in the deep unsaturated zone. The results further support the conclusion that conservative transport of NO3– dominates the N budgets of these systems (Fig. 6 ). At N22, {delta}15N[NO3–] and {delta}18O[NO3–] were comparable at shallow and deep lysimeters, indicating a lack of denitrification and other processes that typically cause fractionation of stable isotopes. An abrupt increase in these isotopic ratios in the saturated zone at 23 m depth is consistent with denitrification near the water table (Fig. 6a), as indicated also by elevated N2 gas and low O2 at this depth (Green et al., 2008). In the unsaturated zone, O2/Ar ratios dropped by only 10% from the ambient atmosphere to 7.1 m depth and remained unchanged to 19.8 m depth, indicating that minor consumption of O2 due to nitrification or other forms of aerobic respiration is limited to the upper portion of the profile (Fig. 6b). Variations of N2/Ar were within analytical error, with a slight increase in the deepest lysimeter possibly associated with denitrification near the water table (Fig. 6c). Concentrations of N2O increased slightly between the ambient air and the mid-depth lysimeter, consistent with minor amounts of nitrification in the upper profile (Fig. 6d). The stable isotopes ratios {delta}18O in O2 ({delta}18O[O2]) and {delta}15N in N2 ({delta}15N[N2]) did not change substantially with depth (Fig. 6e and 6f). Gas profiles were not available for WA, CA, and MD. Shallow ground waters at those sites typically contain minimal dissolved excess N2, elevated dissolved O2, and un-enriched stable isotope ratios of NO3–, which are indicative of aerobic, undenitrified waters (Green et al., 2008). These characteristics indicate that denitrification is minimal in most of those unsaturated zones.


Figure 6
View larger version (8K):
[in this window]
[in a new window]

  		Fig. 6. Depth profiles of stable isotopes and gas concentrations in unsaturated zone air for the Nebraska site (N22). Error bars show two SD of replicate analyses. Nitrous oxide (N2O) values are given as aqueous concentrations.

 
In a limited number of shallow unsaturated zone samples, elevated concentrations of iron indicate iron-reducing conditions (Chapelle et al., 1995), which are favorable for reduction of oxidized N species, such as NO3–. Elevated concentrations of Fe (>56 µg L–1) were detected at shallow depths at M20a (0.6 m), I32a (0.6 m), I32b (0.9 m), C32d (1.4 m), and C33d (2.3 m) (Fig. 7a ). At these locations, the highest concentrations of Fe are associated with a lower fraction of NO3––N/total N, which suggests that in the shallow soils at a few locations, transient anaerobic conditions cause reduction of NO3– to other N species. At the IN and CA locations, the high Fe concentrations are not associated with decreased total N (Fig. 7b), indicating that reduced N species derived from NO3– remain in solution and that the total pool of dissolved N is unchanged. In contrast, at M20, the two lowest total N concentrations occur in the samples with the highest Fe concentrations, which is consistent with conversion of NO3– to N2 by denitrification. The lack of elevated Fe concentrations at other shallow lysimeters as well as all deeper lysimeters suggests that oxygenated conditions unfavorable for denitrification prevail at most of these unsaturated zone locations.


Figure 7
View larger version (13K):
[in this window]
[in a new window]

  		Fig. 7. Correlation plot of (A) fraction of total N as NO3––N (NO3––N/total N) versus Fe and (B) total N versus Fe in unsaturated zone lysimeter samples.

 
The soil zone is well known to be reactive, and the data show that nitrification of NH4+ to NO3– occurs primarily in this zone. Green et al. (2008) show that the {delta}18O of water and of NO3– in the shallow ground waters is consistent with nitrification of NH4+ from fertilizer at the CA, MD, NE, and WA sites. Ammonium was seldom detected in concentrations above 1 mg N L–1 in subsurface lysimeters (Fig. 2), indicating that oxidation of applied ammonia to NO3– occurs before transport below the root zone.

Near the water table, reactions immobilize the relatively small pool of reduced N-species and limit their transport into the saturated zone. Figure 4 shows depth profiles of total N and NO3––N concentrations. Dissolved, reduced N species make up the difference between total N and NO3––N. The concentration of these reduced species remains uniform or increases with depth in the unsaturated zone and then drops sharply below the water table. The absence of increasing NO3––N concentrations in association with the loss of reduced N-species rules out nitrification as a dominant cause. More likely, reduced N-species are immobilized due to sorption on solid grains, increased uptake associated with greater biomass near the capillary fringe (Holden and Fierer, 2005), or entrapment of organic molecules and particles at stationary interfaces between water and air (Goldenberg et al., 1989; Wan and Wilson, 1994; Powelson and Mills, 1996).

Water and Nitrogen Fluxes
The following section includes a discussion of vertical fluxes of water and NO3––N in the unsaturated zone and ground water. Recharge estimates are established for the shallow saturated zone based on ground water ages and for the unsaturated zone based on WTF analyses. Recharge estimates for the unsaturated zone and time series of concentrations from lysimeters are used to estimate seasonal NO3––N fluxes, which are used to evaluate the suitability of annual medians for estimating NO3––N fluxes. Annual NO3––N fluxes are compared among sites and between the unsaturated zone and saturated zone to evaluate the effects of physical processes and agricultural management practices on the contamination of shallow ground water by NO3– delivered through the unsaturated zone.

Estimates of WTF recharge are similar to estimates of vertical ground water flux at the CA, MD, NE, and WA sites (Table 1). At IN, the low vertical ground water flux is consistent with diversion of shallow ground water to the extensive tile drains at this site. At NE, elevated recharge at N22 results from focused infiltration as discussed below. Otherwise, the agreement between water fluxes based on age dates and local WTF recharge suggests that the unsaturated zone hydrologic conditions at these sites are reasonably representative of larger temporal and spatial scales.

Field observations and chemical data from N22 show that preferential flow results in rapid percolation and transport at this location. In the end-rows near this site, heavy precipitation and irrigation events cause intermittent ponding, which can initiate focused infiltration (Derby and Knighton, 2001). Strong seasonal peaks in NO3––N at N22b (Fig. 5e) indicate rapid transport from the shallow unsaturated zone, where such variability was typically observed, to this lysimeter at 7 m depth. In addition, NH4+ (Fig. 2) and pesticides (Hancock et al., 2008) were detected more frequently and at higher concentrations at N22b than at other lysimeter sites, consistent with reduced adsorption due to preferential flow. After application of bromide tracer to the land surface at N22 in May 2004 and 2005, Br– concentration increased at N22b within 2 to 5 d, indicating a maximum transport rate of 1.4 to 3.5 m d–1. In other intermittently inundated soils with preferential flow, maximum transport velocities have been reported to range from less than 0.1 to greater than 10 m d–1 (Nimmo, 2007).

The timing and intensity of unsaturated NO3––N fluxes differs markedly between irrigated and unirrigated sites. Figure 8 shows cumulative NO3––N loads to ground water calculated with Eq. [2] using individual recharge events and NO3––N concentrations from the deepest lysimeters interpolated to the time of the recharge event. At the heavily irrigated CA and WA sites, NO3––N flux to the water table occurs in relatively large pulses (corresponding to periods with steeper slopes of the loading curves) in response to irrigation events during the growing season. At the IN, MD, and NE sites, NO3––N flux to the water table occurs more gradually and uniformly over the course of the year. The times of most intense flux occur outside of the growing season in the winter and spring at MD and NE and during the winter at IN. The seasonality of fluxes is corroborated by temperature data from monitoring wells near the water table. At the heavily irrigated sites, the annual average temperature of shallow ground water is warmer than the average air temperature by 2.0°C (CA) and 3.4°C (WA), consistent with a majority of recharge occurring during warm summer months. At the other sites, the annual average shallow ground water temperature is within 0.1°C of the average air temperature.


Figure 8
View larger version (12K):
[in this window]
[in a new window]

  		Fig. 8. Cumulative NO3––N load to ground water at each study site based on individual recharge events using Eq. [2] (lines) and using Eq. [3] based on median concentrations (arrows), quartiles (boxes), and minimum and maximum concentrations (whiskers) of samples collected during 2004.

 
Although recharge varies seasonally at all four sites, concentrations are relatively constant in the deep lysimeters (Fig. 4). As a result, the cumulative annual NO3––N loads are similar to average annual loads calculated with Eq. [3] using total annual recharge and median concentrations from the deepest lysimeters (Fig. 8). At CA and IN, cumulative loads are slightly higher than average annual fluxes, indicating a positive correlation of recharge with concentration, but fall close to or within the range of the annual loads as calculated with the concentration quartiles from the deepest lysimeter.

To examine whether local unsaturated zone NO3––N fluxes are comparable to fluxes at larger spatial and temporal scales, Fig. 9 compares NO3––N fluxes in the unsaturated zone with those in ground water along the well transects. For the unsaturated zone estimates, each measured NO3––N concentration from a given deep lysimeter is multiplied by the annual recharge calculated with the WTF method. These are shown as box plots reflecting the median, quartiles, and total range of measured concentrations in the deep lysimeters. Ground water fluxes of NO3––N are calculated based on ground water concentrations and vertical ground water flux estimates for all monitoring well samples younger than 25 yr. These ground water values provide an estimate of fluxes averaged over a larger scale and extending back in time.


Figure 9
View larger version (18K):
[in this window]
[in a new window]

  		Fig. 9. Distribution of NO3––N fluxes (top of figure) and median NO3––N concentration and vertical water flux (bottom of figure) for selected unsaturated zone and ground water sites. For NO3––N, fluxes at each study site, the left box shows local unsaturated zone flux estimates based on median, quartile, and minimum and maximum concentrations of all samples collected from the deepest lysimeter (C22f, I32b, M22d, N22c, and W22l). Columns marked with "G.W." show the flux estimates based on reconstructed NO3––N source concentrations for ground water (Green et al., 2008) and vertical ground water flux estimates based on age profiles. Ground waters older than 25 yr since recharge are excluded. Rectangles at the bottom of the figure are sized according to the median NO3––N concentration for each site (width) and the recharge (height) based on the median vertical ground water fluxes and water table fluctuation estimates.

 
At all sites but CA, the median annual ground water NO3––N flux values are less than or equal to unsaturated zone values (Fig. 9). Based on Mann-Whitney Rank Sum tests (Systat Software Inc., 2006), the ground water fluxes are significantly lower at MD (P = 0.014), NE (P = 0.031), and WA (P = 0.025) and are not significantly different at CA (P = 0.63) or IN (P = 0.76). The lower ground water NO3––N fluxes reflect nonagricultural land uses in the source areas of the wells at some sites. At MD, the ground water source areas include an uncultivated strip between two fields (see Fig. 1 in Green et al., 2008). At NE, irrigation and focused infiltration influence unsaturated zone flux estimates, whereas much of the ground water recharge occurs in uncultivated riparian and pasture areas. At WA, the low ground water flux values result from typically lower NO3––N concentrations in ground water. Due to the larger scale of the transect and the greater variability of land uses at WA relative to other sites, it is difficult to generalize about the water and N sources in recharge source areas. These results show that, although the commonly reported median ground water estimates of NO3––N fluxes (Eq. [3] and [4]) may provide reasonable estimates for the broad source areas contributing to the well screen, these values frequently underestimate the NO3––N fluxes originating locally from agricultural fields.

The highest NO3––N fluxes among the sites result from high recharge and NO3––N concentrations in coarse unsaturated zone sediments. The rectangles at the bottom of Fig. 9 are scaled according to the median concentration (width) and median recharge (height) with the area representing the NO3––N flux. The highest fluxes are at the CA and MD sites, which have high concentrations and high vertical water fluxes. These sites share the characteristics of moderate to high N application rates and high sand contents (Table 1), which allow rapid transport of nutrients through the soil and deep unsaturated zone. Sites with finer-grained sediments rank lower in terms of N flux, with higher ET sites having low recharge and high concentrations and the lower ET site (IN) having low concentrations and high recharge. Among these sites, ET has competing effects: Greater ET reduces downward water velocities, potentially reducing the mass of N leached, but also reduces the amount of water available to dilute leached N.

These five sites highlight the conditions that produce the extremes in N fluxes and concentrations. In the deep unsaturated zone, high concentrations can be associated with low N flux values (e.g., NE and WA) due to the reduction of recharge and the enrichment of NO3––N concentrations in pore water by ET losses. Conversely, sites with lower concentrations in the deep unsaturated zone (e.g., IN, CA, MD) can have relatively high N flux values. The different extremes in N concentration and flux have different implications for evaluation of ground water vulnerability to contamination. At sites with high ET, small amounts of recharge can cause high concentrations in shallow ground water, as observed in some shallow wells at the NE and WA sites (Fig. 2; Green et al., 2008). In contrast, higher unsaturated zone fluxes associated with moderate concentrations result in more widespread ground water N contamination but at lower concentrations, such as at the MD site.


   Conclusions
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 Site Descriptions
 Materials and Methods
 Results and Discussion
 Conclusions
REFERENCES
 
Overall, NO3––N is the dominant species in the deep unsaturated zone N budgets at these sites. At all five sites, conservative transport of NO3– occurs between the root zone and the water table, as indicated by profiles of NO3–, correlations among common agricultural chemicals, and profiles of gases and stable isotopes. Profiles of NO3––N and total N show increasing concentrations of reduced N species with depth but minimal concentrations in ground water, indicating that reduced N species are immobilized on solid surfaces, in biomass, or at air–water interfaces before transport below the water table.

The differences in concentrations among the sites result from varying N application and leaching rates, water inputs, and evapotranspiration. Among sites with similar water and fertilizer inputs, higher evapotranspiration is associated with lower recharge and higher concentrations. At CA and MD, the sandier unsaturated zone sediments tend to have a higher fraction of N leaching to the deep unsaturated zone.

Estimates of cumulative unsaturated zone N fluxes over the course of a year show clear differences in the timing and magnitude of fluxes between irrigated and nonirrigated sites. At the heavily irrigated sites, CA and WA, water applications during the growing season were accompanied by large and rapid fluxes of N in the deep unsaturated zone. At the other sites, fluxes of N were more gradual and tended to occur during the fall or spring.

In general, the median ground water N flux estimates underestimate modern fluxes from agricultural fields. At MD, NE, and WA, median NO3––N fluxes were lower for ground water estimates, based on ground water vertical velocities and concentrations, than for local estimates from modern WTF recharge estimates and unsaturated zone concentrations. These results reflect a tendency for lower NO3––N concentrations in ground water due to mixing with waters from uncultivated source areas. Ground water N flux estimates need to be understood as representing conditions across broad source areas, which commonly include farmed and unfarmed areas. The deep unsaturated zone estimates are a better indicator of fluxes from particular fields.

These five sites illustrate that N concentrations and fluxes have different implications for ground water vulnerability to contamination from modern anthropogenic sources. At sites with high evapotranspiration, low fluxes can produce high concentrations that affect shallow ground water. Higher N fluxes can be associated with more moderate concentrations that extend to greater depths due to faster vertical transport. To account for these distinctions, ground water vulnerability assessments should include analyses of concentrations and fluxes.


   ACKNOWLEDGMENTS
 
This study was conducted as part of the U.S. Geological Survey National Water-Quality Assessment Program and benefited from the work of many individuals involved in the planning and execution of the study. Special thanks to JK Böhlke for analyzing and discussing stable isotopes, to Jason Vogel and Joe Domagalski for additional assistance with field work, to Rick Healy for discussion of WTF estimates, to Paul Capel for help with planning and oversight, and to Michelle Walvoord, Hedeff Essaid, Thomas Harter, and one anonymous reviewer for providing technical reviews. The use of trade names in this paper is for identification purposes only and does not constitute endorsement by the U.S. Geological Survey.


   NOTES
TOP
 NOTES
ABSTRACT
 INTRODUCTION
 Site Descriptions
 Materials and Methods
 Results and Discussion
 Conclusions
 REFERENCES
 
All rights reserved. No part of this periodical may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher.


   REFERENCES
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 Site Descriptions
 Materials and Methods
 Results and Discussion
 Conclusions
 REFERENCES