| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Dep. of Agron., 209A Waters Hall, Univ. of Missouri, Columbia, MO 65211
b Dep. of Soil Science, North Carolina State Univ., Raleigh, NC 27695-7619
c Dep. of Crop Science, North Carolina State Univ., Raleigh, NC 27695-7620
d Dep. of Statistics, North Carolina State Univ., Raleigh, NC 27695-8203
* Corresponding author (jeff_white{at}ncsu.edu)
Received for publication May 19, 2005.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Abbreviations: BMP, best management practice • CIR, color infrared photography • DGPS, differential global positioning system • FA, uniform, field-averaged nitrogen management • GDVI, green difference vegetation index • GNDVI, green normalized difference vegetation index • Go, Goldsboro soil • GS, growth stage • Ly, Lynchburg soil • MCLG, maximum contaminant level goal • NIR, near infrared • No, Norfolk soil • NormNIR, normalized near infrared • NUE, nitrogen use efficiency • RGDVI, relative green difference vegetation index • RS, remote sensing • RYE, uniform, Realistic Yield Expectation nitrogen management • SS, in-season, site-specific, variable-rate nitrogen management
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Under rules established to protect nutrient-sensitive river basins in North Carolina (North Carolina Administrative Code, 2004), N fertilizer rates must be determined using field-specific, whole-field, Realistic Yield Expectations (RYE; North Carolina Nutrient Management Workgroup, 2003) and a soil- and crop-specific N-response factor (i.e., the amount of fertilizer N required per unit crop yield). This approach is noteworthy in that it attempts to address the spatial variability in crop N requirement that can exist between fields. However, in cropping systems with substantial within-field spatial variability and between- and within-season temporal variability in crop N requirement, integrating within-field and in-season variable-rate N management should be the best way to optimize fertilizer N applications and reduce the risk of N pollution (Ferguson et al., 2002; Raun et al., 2002).
In-season N management to address temporal variability in wheat and corn was first attempted by splitting the predicted total crop N requirement into multiple applications of appropriate amounts of N timed to meet crop needs. This has yielded inconsistent results for both wheat (Grant et al., 1985; Sowers et al., 1994) and corn (Menelik et al., 1994; Randall et al., 2003). Subsequent research attempted, with some success, to optimize in-season N rates and timing based on soil and tissue tests, chlorophyll meters, and early spring tiller density in winter wheat (Magdoff, 1991; Blackmer and Schepers, 1994; Flowers et al., 2004). Crop leaf and canopy reflectance of visible light and near infrared (NIR) radiation can indicate crop N status and fertilizer requirement; thus, remote sensing has been used to guide in-season N management based on estimation of crop N and moisture status, yield potential, optimal N rate, and wheat tiller density (Flowers et al., 2003a, 2003b; Sripada et al., 2005).
Within-field, site-specific, variable N rates have been determined based on: soil series (Carr et al., 1991); a combination of grid soil sampling and yield maps (Redulla et al., 1996); a combination of yield maps, soil organic matter content, and soil N status (Mulla and Bhatti, 1997); grid soil sampling (Ferguson et al., 2002); and stand density and tissue testing (Flowers et al., 2004). Remote sensing in the form of on-the-go sensors (Raun et al., 2002) and aerial photography (Flowers et al., 2003a, 2003b) has been applied to in-season SS N management. Ferguson et al. (2002) showed that in-season SS N management of corn based on RS can reduce residual soil NO3, and thus might be a way of reducing groundwater NO3–N contamination.
Our primary objectives were to assess the agronomic, economic, and groundwater consequences of in-season, RS-informed N management, applied either on a uniform field-average (FA) or SS basis, compared with North Carolina's current, uniform, RYE-based N BMP in a 2-yr winter wheat–double crop soybean–corn rotation typical of the southeastern U.S. Coastal Plain. We tested the hypotheses that RS-informed FA and SS have higher harvest N ratios and result in less surplus N and lower groundwater NO3–N concentrations compared with RYE-based N management. A secondary objective was to characterize the temporal dynamics of shallow groundwater NO3 in coastal plain soils.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
A typical stratigraphic description of these fields indicates predominantly sandy clay loam texture transitioning to sandy clay at about 2.6-m depth (North Carolina Agricultural Experiment Station, 1977). There is a clay bed at 2.6- to 2.9-m depth that has low permeability. During periods of high rainfall, water will perch above the clay forming a zone of saturation up to 1.8-m depth or less. The major horizontal (lateral) water flow likely occurs at the zone of a basal bed of medium coarse sand at 2.9- to 3.7-m depth. The basal sand bed transitions in color from yellow to yellowish red through 5.2 m, where an abrupt lower boundary indicates the base of the Wicomico morphostratigraphic unit and the start of the red to black compact loam of the Pee Dee formation. Each field has a tile drainage system with lines at about 1-m depth spaced 
30 m apart oriented parallel to the northern and southern field boundaries and exiting into orthogonal tie lines on the eastern side of each field. We equipped the outlets of the tie lines with drainage control structures to maintain water tables as high as possible to foster denitrification of groundwater NO3–N, but not so high as to hinder crop production or field operations (Osmond et al., 2002).
Experimental Setup
The experiment began in November 2000 with the establishment of a 2-yr winter wheat–double crop soybean (Year 1)–corn (Year 2) rotation. We used a RCBD with three N management treatments for wheat and corn (see below) replicated 10 times in large 60.8 by 60.8 m (0.37 ha) plots. In early March 2001, two well nests were installed in each plot to monitor shallow groundwater NO3–N concentrations, with one at the plot center to minimize the influence of neighboring plots and the other placed randomly within the constraints of being an adequate distance from the plot center, outside a plot border-harvest buffer area, and inline from plot to plot parallel to crop rows to facilitate field operations. Each well nest consisted of three PVC pipe groundwater-sampling wells screened to sample over depths of 0.9 to 1.8, 1.8 to 2.7, and 2.7 to 3.7 m, the latter to capture groundwater within the basal sand bed and to ensure that water samples were available throughout the season.
Pre- and At-Plant Fertilization
All treatments received a single uniform rate of pre- or at-plant fertilizer (Table 1). Granular fertilizer (percentage N–P–K: 10–4.4–16; N–P2O5–K2O: 10–10–10) was broadcast pre-plant to supply 34 kg N, 15.6 kg P, and 54 kg K ha–1 for wheat; for corn, 8 kg N, 3.5 kg P, and 1 kg S ha–1 was applied at planting as aqueous N–P–K (11–4.8–0) (11–11–0: N–P2O5–K2O) plus 1.5% S dribbled 5 cm below and 5 cm to the side of the seed furrow.
|
Uniform Realistic Yield Expectation Nitrogen Management (RYE)
The RYE N rate was based on the RYE and N-response factor for the predominant soil type (Go) in the field derived from the North Carolina RYE database (North Carolina Nutrient Management Workgroup, 2003), and included the starter N. The remainder of the fertilizer N was applied at appropriate growth stages (GS; Zadoks et al., 1974): for wheat, GS-30 (Flowers et al., 2004), for corn, V2 (Ritchie et al., 1993).
Remote Sensing–Informed, In-Season, Uniform, Field-Average Nitrogen Management (FA)
This treatment received uniform, in-season top-dress (wheat) or side-dress (corn) N. In-season applications were based on field-averaged estimates of economic optimal N rates derived from aerial color infrared photography (CIR) at critical growth stages. For wheat, this was at GS-25 and GS-30 (Flowers et al., 2004), and for corn at VT (tasseling) (Sripada et al., 2005). Thus, FA considered in-season, growth stage-specific N demand, but ignored within-field spatial variability. Methods used to determine in-season N rates are detailed below.
Remote Sensing–Informed, In-Season, Site-Specific, Variable-Rate Nitrogen Management (SS)
The SS rates were determined based on site-specific estimates of N demand at the same growth stages and using the same aerial CIR methods as FA. Each SS plot was divided into 78 4.6- by 9.1-m "miniplots" for site-specific, spatially variable N application. The SS N rates were determined by averaging N demand within each miniplot and fertilizing each accordingly. Thus, SS considered both growth stage-specific and within-field site-specific N demand.
Treatment Nitrogen Rate Determination
Nitrogen rates and timings for the RYE, FA, and SS treatments are summarized in Table 1. There was no fertilizer N applied to soybean.
Wheat 2001 and 2003
GS-25 Nitrogen Rates
For FA and SS, these were based on GS-25 wheat tiller density estimated from aerial CIR (Flowers et al., 2003b), with 67 kg N ha–1 applied only to areas with tiller density below 500 tillers m–2, primarily to promote tiller development. In 2001, a linear relationship of the near infrared (NIR) CIR band vs. tiller density was developed based on ground truth tiller densities determined at small calibration areas, with one algorithm for the Ly soil and a second for the rest of the field. In 2003, the same aerial CIR method was used; however, due to low tiller densities, the best estimate of tiller density across all soil types (R2 = 0.69) was a quadratic relationship with the normalized (Jain, 1989) NIR defined as follows:
 | [1] |
GS-30 Nitrogen Rates
These were based on estimates of whole-plant tissue N content derived from aerial CIR (Flowers et al., 2003a) and an algorithm relating tissue N to optimum GS-30 N rates (Flowers et al., 2004). For 2001 wheat, we established calibration strips at GS-25 by applying 67 and 134 kg N ha–1 to 4.6-m-wide strips of wheat just outside the experimental area; for 2003 wheat, we relied on similar strips on plot boundaries within the experimental area located to provide calibration within each soil map unit. At GS-30, a new CIR was taken and ground truth whole-plant N concentrations were determined by clipping wheat plants from small sample areas in the calibration strips and elsewhere. Tissue samples were dried, ground, and analyzed for total N via combustion and CHN analyzer by Waters Agricultural Laboratories (Camilla, GA). In 2001, tissue N concentration was best estimated using the green normalized difference vegetative index (GNDVI; Gitelson et al., 1996) defined as follows:
 | [2] |
Corn 2002
Our algorithm to predict economic optimal N for corn at VT was determined relative to "N-sufficient" reference strips established by applying 120 kg N ha–1 at planting and 160 kg N ha–1 at V2 to 4.6-m-wide strips of corn growing at plot boundaries within the experimental area. At V2, the RYE plots received the remainder of the total RYE N, while 53 kg N ha–1 was applied uniformly to FA and SS plots to maintain yield potential. At VT, FA and SS N rates were determined from CIR using a linear-plateau model that estimates the economic optimum VT N rate based on the relative green difference vegetation index (RGDVI) (Sripada et al., 2005), that is, the Green Difference Vegetation Index (GDVI = NIR – GREEN; Tucker, 1979) divided by the GDVI of the reference strips. The GDVI was determined for each pixel in the FA and SS plots. The FA N rate was determined by (i) averaging GDVI across FA plots, (ii) averaging GDVI of high N-strips, (iii) computing RGDVI, and (iv) inserting it to the algorithm to derive the optimum N rate. The SS N rates were determined at the miniplot level based on the average RGDVI for each miniplot. To account for potential soil effects on the SS rate determination, soil types were attributed to miniplots and high-N reference strips by GIS overlay with a soil-type coverage derived from the order one soil survey. The RGDVI in a miniplot was computed relative to the nearest reference strips within the corresponding soil type. The SS N rates were grouped into four discrete application rates (Table 1).
Aqueous urea–ammonium nitrate [CO(NH2)2–NH4NO3; 30% N] was applied to wheat using a variable-rate sprayer at GS-25 and GS-30; for corn at V2 and VT, it was dribbled to row middles using high-clearance, variable-rate equipment with drop nozzles. In 2002 in cases where corn row spacing prevented mechanized operations adjacent to well nests, seeding and fertilization were done manually.
Yield, Harvest Nitrogen Ratio, Surplus Nitrogen, and Economic Return to Nitrogen
A late freeze occurred on 19 Apr. 2001 when wheat was at anthesis (GS-60), aborting flowers and preventing complete grain development. Consequently, crop yield, N uptake, and harvest N ratio were determined based on total aboveground biomass harvested after senescence. A forage harvester with a 1.2-m header and Mettler weigh system was used to harvest 18 1.2- by 18-m strips within each treatment plot. The center point of each strip was georeferenced using a differential global positioning system (DGPS). After weighing, a forage sample was collected and analyzed for total N as described above. The remaining forage was then redistributed across the harvested strip.
Corn grain in 2002 and wheat grain in 2003 were harvested using a combine equipped with a yield monitor and DGPS. Yield monitor data were cleaned and edited following the general guidelines suggested by Doerge (1999) and Blackmore and Moore (1999). To determine treatment harvest N ratio and surplus N, multiple DGPS-georeferenced grab samples of grain were collected at the top of the combine auger and analyzed for total N as described above. The harvest N ratio was calculated as follows:
 | [3] |
 | [4] |
To estimate profitability, we calculated a partial-budget analysis of gross and net return to N fertilizer. This analysis did not consider any other costs such as those that would be incurred in determining N rates via aerial CIR and in applying fertilizer. For the analysis, we used a cost of $0.64 kg–1 N and prices of $81.02, $93.70, and $96.45 Mg–1 for wheat forage/hay, wheat grain, and corn grain, respectively (1999–2003 average prices, www.agr.state.nc.us/stats/pric_rec/prrcanyr.htm; accessed 2 Feb. 2005, verified 28 Dec. 2005). For 2001 wheat, we estimated potential grain yield by multiplying aboveground biomass by an assumed constant grain harvest index of 0.38 (Reeves et al., 2005).
Groundwater Sampling
Groundwater samples (25 mL) were collected as available from each well from March 2001 until July 2003, and the depth to the groundwater table was measured from July 2001, both approximately every 2 wk or after significant rainfall. Depth to the water table at each well nest was adjusted to water table elevation based on the wellhead elevations derived from North Carolina Floodplain Mapping Program lidar data (www.ncfloodmaps.com; verified 28 Dec. 2005). Groundwater NO3–N concentrations were determined using an automated ion analyzer (QuikChem 8000; Lachat, Loveland, CO) employing a Cd reduction method (Greenberg et al., 1992). Daily precipitation data (Fig. 1
) were collected at the experiment station using a rain gauge. Thirty-year (1974–2003) average monthly precipitation data were obtained from the State Climate Office of North Carolina (www.nc-climate.ncsu.edu/; verified 28 Dec. 2005).
|
2 test in SAS PROC MIXED (SAS Institute, Cary, NC). The best-fit model was used to evaluate treatment effects and estimate treatment means for yield, harvest N ratio, surplus N, and groundwater NO3–N for individual dates and depths. The treatments, and consequently the well nests, were not evenly distributed among soil types (Table 2). To account for potential soil effects, groundwater NO3–N concentrations for individual dates and depths were evaluated with treatment x soil type as a fixed effect in the model. Where this effect was significant, we refitted the model using only data from wells located in Go and Ly soils. If the treatment x soil type effect was still significant, only data from wells in the Go soil (40 out of 60 total wells) were used (Table 2). Thus, any treatment x soil type effects were removed from the data analysis.
|
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
56%) lower harvest N ratio, and considerably higher surplus N than RYE and SS. For 2003 wheat, all FA plots and all of the SS miniplots had tiller densities below the critical threshold, and GS-25 N was applied to all of these areas (Table 1). At GS-30, the RYE and FA rates were 119 and 67 kg N ha–1, respectively, while the SS rates ranged from 45 to 67 kg N ha–1, with an area-weighted average of 65 kg N ha–1, again nearly identical to the FA rate. The FA and SS treatments resulted in about 14 kg ha–1 more N applied, but about 670 kg ha–1 higher grain yield, about 10% higher harvest N ratio, and 9 kg ha–1 less surplus N (only for SS) than RYE (Fig. 2).
On a partial-budget, return-to-N basis, a RS-informed N management treatment (FA or SS) was always more profitable for grain production than RYE (Table 3). For 2001 wheat, the 40 kg N ha–1 higher RYE rate increased yield but not profit for estimated grain yield. However, when analyzed on the basis of hay prices, the situation reversed and RYE was slightly more profitable. In contrast, the 2002 corn yield response with the higher SS N rate and the 2003 wheat yield response with both the higher FA and SS N rates were profitable, and resulted in greater net returns than the lower RYE rates. For 2001 wheat and 2001 corn, these profit differences were modest (3–8%), whereas for 2003 wheat, the FA and SS rates increased profit by 43%.
|
|
|
Temporal Variation
Groundwater NO3–N concentrations at each depth exhibited apparent temporal trends independent of treatment. Nitrate-N concentrations frequently exceeded the USEPA drinking water MCLG of 10 mg L–1 NO3–N (Fig. 3); this occurred mainly in spring. The temporal fluctuations of groundwater NO3–N concentrations were of greater magnitude than treatment affects, and were associated with fluctuations of the water table (Fig. 4). At each depth, mean groundwater NO3–N concentrations averaged over treatments for individual dates were negatively correlated with water table depths (expressed as positive values; i.e., positively correlated with water table elevations) averaged over all well nests for individual dates (Fig. 4). The Pearson correlation coefficients for the 0.9- to 1.8-, 1.8- to 2.7-, and 2.7- to 3.7-m depths were r = –0.73 (p = 0.0002, n = 23), r = –0.55 (p = 0.001, n = 35), and r = –0.76 (p < 0.0001, n = 54), respectively.
Response to Treatment
As illustrated in Fig. 3, we detected statistically significant treatment differences in groundwater NO3–N concentrations on some sampling dates. Treatment differences varied over depth, and occurred mainly at the 1.8- to 2.7-m depth (Fig. 3b). For ease of presentation, we divided the sampling period into five phases (Fig. 3), as follows:
Phase 1: March 2001 to Early April 2002
Relatively few samples were available from the 0.9- to 1.8-m depth, and no treatment differences were detected at this depth. At the 1.8- to 2.7-m depth (Fig. 3b), RYE NO3–N was greater compared with SS in mid-July, mid-August, and mid-September 2001, and from late January through early April 2002; and compared to FA in early February 2002; and tended to be greater than FA and SS from early July 2001 to early April 2002. The RYE NO3–N at the 2.7- to 3.7-m depth was also significantly greater than SS in early March 2002. There was a lag of several months between the differential N application to wheat in March 2001 and the appearance of corresponding treatment differences in NO3–N at the 1.8- to 2.7-m depth; the higher RYE NO3–N at the 2.7- to 3.7-m depth did not appear until a year after the application. The maximum difference in groundwater NO3–N concentration among treatments during this phase was 3 mg NO3–N L–1. No N was applied to soybean, and there was no difference in soybean yield among treatments (not shown).
Phase 2: Early April to Late May 2002
At the 1.8- to 2.7-m depth, RYE NO3–N was greater than FA from mid to late May.
Phase 3: Late May 2002 to Mid-March 2003
At the 2.7- to 3.7-m depth (Fig. 3c), and similar to the previous period at the 1.8- to 2.7-m depth, RYE NO3–N was greater than SS in early and mid-January and mid-February 2003, and greater than FA in mid-February 2003. This trend in treatment differences, which was observed previously in Phase 2, did not appear again in Phase 3 until late November 2002.
Phase 4: Mid-March to Mid-June 2003
At the 0.9- to 1.8-, and 1.8- to 2.7-m depths, FA and SS NO3–N tended to be greater than RYE. This tendency was significant for FA in early April and from late April to mid-June at the 0.9- to 1.8-m depth, and from late March to early April and in late April, late May, and mid-June at the 1.8- to 2.7-m depth. This effect was also significant for SS in mid-June at the 0.9- to 1.8-m depth, and in mid-March, from late April to late May, and in mid-June in 2003 at the 1.8- to 2.7-m depth.
Phase 5: Mid-June to Mid-July
No treatment effects in groundwater NO3 concentration were detected during this period.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Our results with both wheat crops were consistent with those previously reported by Flowers et al. (2004), who showed that SS and FA wheat N management (not RS-informed) generally resulted in reduced N inputs compared with typical growers' practices. The SS treatment decreased surplus N for 2003 wheat. When SS performed similarly to FA, it was probably because there was minimal spatial heterogeneity, thus SS N fertilizer management provided no additional advantage.
Corn
The fact that the 2002 corn FA N rate at VT was nearly twice the SS N rate was unexpected. If the SS and FA treatment plots were randomly distributed with respect to spatial variability of N demand, we would expect that the area-weighted average of the different SS N rates would be approximately the same as the FA N rate, as it was for wheat. Upon investigation, we realized that when we developed the recommendation by first averaging RGDVI and then applying the linear plateau algorithm to determine the economic optimum N rate, the average RGDVI value fell within the plateau of the algorithm that calls for the algorithm's maximum N rate (Sripada et al., 2005). Because of the linear-plateau model, calculating the FA N rate in this way resulted in a higher FA N rate compared with first calculating N rates at the pixel or plot level and averaging those to determine the FA rate. This is an important, inherent characteristic of any linear-plateau recommendation algorithm (e.g., Sripada et al., 2005; Flowers et al., 2004) applied with spatial averaging. Calculating N rates at the pixel level would have resulted in a FA recommendation at VT of 132 kg N ha–1. Based on the responses to the RYE and SS rates, this rate would have been both economically and environmentally more appropriate than the 157 kg N ha–1 that was applied. Thus, we believe that the manner in which we calculated the FA N rate for corn resulted in an erroneously high N recommendation. In addition, potential soil effects were not considered when computing the RGDVI of FA plots as they were in calculating SS rates.
However, even when these factors are considered, the results suggest that the algorithm used to determine the FA and SS rates overestimated N demand at VT. The algorithm was developed under conditions of sufficient plant-available moisture to carry the crop through maturity. However, 2002 corn experienced a very dry growing season (Fig. 1), and corn yields were about 2 Mg ha–1 lower than the realistic yield expectations for these soils (North Carolina Nutrient Management Workgroup, 2003). Consequently, SS and especially FA resulted in considerable excess N. However, SS produced the greatest net return to N, illustrating the inherent potential conflict between managing N for maximum profit vs. optimal environmental stewardship. The higher the value of the crop and/or the steeper the N yield-response curve and/or the more gradually the N yield response declines at higher N rates, the greater the risk that maximum net return to N will occur at higher rates that are used relatively inefficiently, and thus generate greater surplus N.
Groundwater Nitrate-Nitrogen Concentrations
Response of groundwater NO3–N concentration to treatment was interpreted based on treatment N rate and timing, significant excess rainfall occurring soon after differential N applications, and varying water table depth. We attributed differences in groundwater NO3–N following 2001 wheat to the residual effects of the GS-30 N applied in March 2001, when RYE received substantially more N than the other treatments and produced the lowest harvest N ratio and the highest surplus N (Fig. 2). This was especially noteworthy since the RYE treatment was based on the currently regulated N rate developed specifically to reduce groundwater contamination. The lag in the appearance of treatment effects in the deepest wells was probably due to the relatively droughty conditions, which likely slowed or halted NO3 leaching compared with periods of near-average or greater precipitation.
The largest treatment difference in groundwater NO3–N (3 mg NO3–N L–1) was reasonable given the 23 to 28 kg N ha–1 difference in surplus N between RYE and the other treatments when dilution by precipitation is considered. Average annual precipitation at the site is 126 cm yr–1, and actual evapotranspiration is 89 cm yr–1 (Evans et al., 1996), providing an annual dilution factor of 3.7 x 106.
That RYE NO3–N was greater than FA from mid- to late May 2002, greater than SS in early and mid-January and mid-February 2003, and greater than FA in mid-February 2003 was probably due to the combination of residual surplus N from 2001 RYE wheat plus N leaching from the corn V2 application on 24 Apr. 2002. At that time, two times more N was applied to RYE than FA and SS (Table 1). The current RYE N fertilizer approach applies a large quantity of sidedress N at an early vegetative stage, while ignoring the risk of N loss due to weather-related events before the period of peak N demand. That this trend did not appear again until late November 2002 was probably because the water table fell substantially from early June until late November 2002, then remained below the 2.7-m depth, making it unlikely that NO3–N would leach into the groundwater during that period. When the water table rose in late November 2002, NO3–N levels increased, probably due to NO3–N leaching and/or groundwater rising into a zone of elevated soil NO3, allowing treatment effects to be expressed during this period.
Treatment effects on [NO3–N] during Phase 4 were attributed mainly to the excessive N applied to FA corn in 2002, as well as the lack of N applied to RYE wheat at GS-25 in 2003. An accumulated 23.2 cm of rain fell from 10 Feb. to 8 Apr. 2003, causing the water table to rise dramatically during this period, with a concurrent increase in groundwater NO3–N and an expression of treatment differences.
For 2003 wheat, the lower surplus N from FA and SS vs. RYE was not detected in the groundwater sampled through July 2003, the last sampling reported here. This was probably due to several factors. Treatment differences in surplus N in 2003 wheat were the smallest of any crop in the study. Phase 5 was a period of relatively high groundwater elevation with correspondingly high average NO3 levels, which may have obscured expression and/or detection of any treatment differences. Depending on precipitation and soil moisture conditions, there can be substantial lags between the time differential N rates are applied and the time any excess N appears in groundwater. Thus, NO3–N in the latter sampling dates of this study may reflect the continued influence of the surplus N from the prior corn crop. Finally, heavy rainfall occurred immediately after the 2003 GS-25 wheat N application, when N was applied only to FA and SS to promote tiller development to increase yield potential. The broadcast N application rates necessary to achieve this (Flowers et al., 2003b) almost inevitably exceed actual plant N needs during this period (Baethgen and Alley, 1989), and excess rainfall likely leached substantial N from the FA and SS treatments. Additional research is necessary to investigate, and ameliorate if necessary, the apparent conflict between agronomic and water quality goals in recommending a GS-25 N application to promote tillering.
It is noteworthy that differences in groundwater NO3–N concentrations associated with relatively small differences in N applications were detected in these experiments. In 2001, the difference between the RYE and SS N application rates was only 39 kg ha–1 and yet these treatments resulted in different NO3–N concentrations (2 mg N L–1) that were still detectable almost 1 yr after application. Similarly, in the 2002 corn crop, the total SS N application was only 32 kg ha–1 higher than the total RYE rate, but it also resulted in higher groundwater NO3–N concentrations that persisted through the following wheat season. Clearly, groundwater NO3–N concentrations in these coastal plain soils are sensitive to even small changes in N rates and crop N uptake.
The correlation between groundwater NO3–N and water table elevation suggests that in this and similar coastal plain fields, shallow groundwater NO3–N concentrations may be somewhat "self regulated" by water table depth. Water tables rise as a result of significant precipitation, which likely includes leaching events that carry NO3 to groundwater. Rising groundwater likely intercepts and absorbs N from zones of high inorganic soil N. Coincident saturation–desaturation cycles enhance mineralization of N that may end up in groundwater. As water tables remain within the upper parts of the profile with high [NO3–N], sufficient organic C, microbes, and favorable temperature, denitrification is enhanced and groundwater [NO3–N] decreases (Evans et al., 1996; Osmond et al., 2002). Because groundwater NO3–N concentrations were strongly associated with water table depth, water table depth could be an important covariate for temporal analyses of NO3–N behavior. Further research is needed to better understand the interaction of groundwater NO3–N with shallow water tables, which might provide guidance for soil management to improve groundwater quality.
Other Factors Affecting Interpretation of Groundwater Nitrate-Nitrogen
Nitrogen Loss Via Controlled Drainage and the Drainage Systems
Controlled drainage is a recommended BMP for N in the North Carolina Coastal Plain, primarily because it fosters NO3–N removal via denitrification (Osmond et al., 2002). We used this BMP rather than allowing less environmentally sound free drainage that might have enhanced our ability to detect treatment differences. On most sampling dates, the water table was below the drainage tile, and there was no flow at the outlets. However, when upper portions of the profile were saturated, controlled drainage likely prolonged saturation, enhanced denitrification, and thus decreased groundwater NO3–N concentrations. Because of the RCBD, we expected that any such decreases would attribute evenly to each treatment, thus allowing us to detect treatment differences. There were also times when NO3 surely escaped via drainage water that flowed from the outlets, which occurred both when the outlets were raised and when they were lowered to allow drainage to facilitate field operations. Because of the confounding nature of the drainage system relative to our treatment plots, we made no attempt to quantify N mass balance and did not measure outlet flow volume or NO3–N concentration.
Potential Lateral Groundwater Flow
Subsurface lateral water flow is another important factor potentially affecting the evaluation of treatment differences in groundwater NO3–N concentrations. Based on the findings by Grayson et al. (1997), we postulate that subsurface lateral flow occurs in this field in two dominant paths: (i) to and through the drainage lines during periods of saturation within the tile zone, and (ii) within horizons of relatively high horizontal saturated hydraulic conductivity driven by water table elevation gradients. Based on saturated hydraulic conductivity measured in soil samples from the upper 0.75 m, water table heads recorded in the course of the study (both not shown), and the stratigraphic profile, we estimated that lateral flow rates ranged from 0.6 to 5 cm d–1 in the upper 2.6 m, with rates several orders of magnitude lower in areas of high clay and/or silt, and up to an order of magnitude higher in the basal coarse sand.
Nitrate-containing leachate from the upper part of the profile likely accumulated above the low permeability firm clay layer at 2.6- to 2.9-m depth in association with the rise of a transient perched water table. This may have allowed us to sometimes capture treatment differences at 1.8- to 2.7-m depth (Fig. 3b). That these effects were transitory may have been, in part, the result of lateral flow above the firm clay layer driven by water table elevation gradients. The general orientation of these gradients was relatively constant during the study (not shown), with the potential to drive lateral flow from the southwest to the northeast. Consequently, during both wet and dry periods, shallow groundwater likely flowed from one treatment plot to another, albeit in different directions depending on the pathway (tile or soil), potentially confounding any treatment differences over time. The clay layer impedance of leachate flow to the 2.7- to 3.7-m well depth along with the predominant lateral flow within the basal sand bed likely delayed and prevented appearance and detection of treatment effects at that depth. Despite lateral water flow, we were able to detect statistically significant treatment effects that were consistent with the agronomic results.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Effects of N management on groundwater NO3–N concentrations were complex and difficult to interpret mainly due to long-term carryover of treatment effects from one crop into subsequent growing seasons, and differences in rainfall from season to season. The largest variations in groundwater NO3–N concentrations were associated with rainfall and water table fluctuations rather than with fertilizer management. Nevertheless, on specific dates, significant N management treatment effects occurred that appeared to be the result of N fertilizer applications made several weeks or even months before the sampling date. In some cases, yield and harvest N ratio advantages observed with RS-informed management were associated with reduced groundwater NO3–N concentration. This was especially true following the 2001 wheat crop and following the early sidedress application of N to corn in 2002. Conversely, where higher N fertilizer rates resulted in higher surplus N (e.g., 2002 corn FA treatment), statistically significant increases in groundwater NO3–N concentrations were observed in the subsequent months. Clearly, even small changes in N management can impact groundwater NO3–N concentrations in these sandy coastal plain soils.
Our data suggest that to assess the environmental efficacy of N management, frequent, periodic, and long-term monitoring of groundwater NO3–N, especially after significant rainfall, and for weeks or even months after application, is essential to capture in-season treatment effects. Simultaneous measurement of precipitation and water table depth facilitate understanding of these effects. The traditional sampling of NO3–N only at or after harvest is likely to be insufficient to capture the entirety of treatment effects. This is especially true in coastal plain and other coarse-textured soils where in-season NO3–N leaching may be pronounced.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  R. P. Sripada, J. P. Schmidt, A. E. Dellinger, and D. B. Beegle Evaluating Multiple Indices from a Canopy Reflectance Sensor to Estimate Corn N Requirements Agron. J., October 21, 2008; 100(6): 1553 - 1561. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. Hong, J. G. White, R. Weisz, M. L. Gumpertz, M. G. Duffera, and D. K. Cassel Groundwater Nitrate Spatial and Temporal Patterns and Correlations: Influence of Natural Controls and Nitrogen Management Vadose Zone J., January 24, 2007; 6(1): 53 - 66. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||
