Top Navigation Bar

Article Request Page ASABE Journal Article

Effect of Drainage Water Management on Nitrate Nitrogen Loss to Tile Drains in North Carolina

C. A. Poole, R. W. Skaggs, M. A. Youssef, G. M. Chescheir, C. R. Crozier


Published in Transactions of the ASABE 61(1): 233-244 (doi: 10.13031/trans.12296). Copyright 2018 American Society of Agricultural and Biological Engineers.


Submitted for review in February 2017 as manuscript number NRES 12296; approved for publication as part of the “Advances in Drainage: Selected Works from the 10th International Drainage Symposium” collection by the Natural Resources & Environmental Systems Community of ASABE in November 2017.

The authors are Chad A. Poole, Research Scholar and USDA-NIFA Fellow, R. Wayne Skaggs, Professor Emeritus, Mohamed A. Youssef, Associate Professor, and George M. Chescheir, Research Associate Professor, Department of Biological and Agricultural Engineering, and Carl R. Crozier, Professor, Department of Soil and Crop Science, North Carolina State University, Raleigh, North Carolina. Corresponding author: Chad Poole, Campus Box 7625, North Carolina State University, Raleigh, NC 27695; phone: 919-515-6750; e-mail: chad_poole@ncsu.edu.

Abstract.  Short-term studies have demonstrated that drainage water management (DWM), or controlled drainage (CD), can be used to substantially reduce the loss of nitrogen (N) from drained lands for a wide range of soils, crops, locations, and climates. Long-term studies on the effects of the practice are limited. This article presents results on the effects of CD on nitrate-N (NO3-N) losses for three crops, corn ( L.), wheat ( L.), and soybean ( [L.] Merr.), in a two-year rotation in North Carolina. Nitrate losses were measured on replicated plots under CD and conventional, or free drainage (FD), treatments for nine years between 1992 and 2012 on a tile-drained site near Plymouth, North Carolina. The site is on a Portsmouth sandy loam soil with parallel drains 22.9 m apart and 1.15 m deep. The subsurface drainage characteristics under FD were drainage intensity (DI) = 8 mm d-1, drainage coefficient (DC) = 14 mm d-1, and Kirkham coefficient (KC) = 18 mm d-1. Compared to FD, CD reduced annual drainage outflow by 33% and NO3-N export by 30%, with an average annual reduction of 6.3 kg ha-1 year-1. CD increased average NO3-N concentrations by 0.9 mg L-1, but the difference was not significant. The reduction in NO3-N export observed in the CD treatment was equal to the increase in N removed by the harvested grain. The results document the effects of CD on NO3-N export over a wide range of weather conditions during the nine-year study. While the average 30% reduction in NO3-N losses in drainage water is in the midrange of that reported by previous studies for different soils and climates, this is believed to be the first time such a reduction has been attributed to the effect of CD on increasing yields and N removed in the harvested grain.

Keywords.Controlled drainage (CD), Corn, Drainage water, Drainage water management (DWM), Nitrate, Nitrogen, Soybean, Water quality, Wheat.

Approximately 625 million ha (41% of the world’s total cropland) requires improved drainage to support consistent and sustainable production of food and fiber (Smedema et al., 2004). Drainage systems remove excessive soil water to enable timely planting and harvesting, promote plant growth, and increase yields and profits (Skaggs and van Schilfgaarde, 1999). Improved drainage also significantly changes the quality of the water draining from those lands (Williams et al., 2015; Gilliam et al., 1999). Increasing the intensity of subsurface drainage increases the amount of soluble nitrogen (N), mostly in the form of nitrate-N (NO3-N), leached from the soil and exported in the drainage water while decreasing surface runoff and sediment losses (Gilliam et al., 1999; Skaggs et al., 1982). Such nonpoint sources of N from agricultural lands are considered significant contributors to nutrient enrichment of rivers and estuaries in North Carolina (Craig and Kuenzler, 1983; Evans and Skaggs, 2004). Multiple studies have shown that most of the nitrate-N in the Mississippi River comes from the upper Midwest through an extensive network of subsurface drains underlying millions of hectares of cropland (Robertson et al., 2009; David et al., 2010; Jaynes, 2012; Helmers et al., 2012; Williams et al., 2015). Studies in eastern Canada (Sunohara et al., 2014) have found similar results.

Nitrate-N losses to surface waters contribute to algal blooms, which lead to eutrophication as the algae die and decompose. The resulting depletion of oxygen causes hypoxic conditions (dissolved oxygen less than 2 mg L-1), which result in fish kills and degradation of the quality of streams, rivers, and estuaries (Gilliam and Terry, 1973; Gilliam et al., 1999). The Committee on Environment and Natural Resources (CENR, 2010) identified excess nitrate as the leading contributor to hypoxic zones at over 300 locations in U.S. estuaries and coastal waters. The most prominent example is the large dead zone in the Gulf of Mexico, which is attributed to the large load of nitrate-N delivered by the Mississippi River (Rabalais et al., 2010). Subsurface drainage promotes aerobic conditions in the root zone, which lead to N mineralization (organic N ? NH4+) and nitrification (NH4+ ? NO3-). Nitrate is soluble in water and can be readily leached to groundwater or by subsurface drainage to surface waters. Nitrate leaching is highly variable and depends on soil properties, crop, fertilization rates, irrigation management, and climatic and hydrologic conditions (Nightingale, 1972). For artificially drained soils, nitrate leaching is also strongly dependent on drainage system design and management (Skaggs et al., 2005).

Recent drainage research has focused on developing and testing best management practices (BMPs) to reduce N export from drained lands without compromising crop production. The design and intensity of subsurface drainage systems have a substantial impact on nitrate losses in drainage waters. Skaggs et al. (2005) used field data from Kladivko et al. (1999, 2004) and Gilliam and Skaggs (1986) to plot NO3-N losses as a function of subsurface drainage intensity (DI). DI is defined as the drainage rate that occurs when the water table midway between parallel drains is just coincident with the surface. Their results showed that increasing the DI from 1 to 2 cm d-1 for a corn-wheat-soybean rotation in the North Carolina Coastal Plain would increase the average annual NO3-N losses in drainage water by a factor of 3 (from 10 to 30 kg ha-1 year-1). A summary of a comprehensive database of drainage studies (Christianson and Harmel, 2015) showed that N load generally increased with increasing drain depth and decreasing drain spacing, both of which affect DI. Evans et al. (1989) summarized data from 14 field sites in North Carolina. They reported that the average total N loss from fields with only surface drainage was 13.8 kg N ha-1, while sites with improved subsurface drainage had an average loss of 31.1 kg N ha-1 year-1. It follows that an effective method for reducing NO3-N losses from drained lands is to design and install drainage systems with the minimum DI required for economic crop production (Skaggs et al., 2005, 2006). Installing drainage systems with a DI higher than necessary for optimal economic crop production will increase both the costs of the system and the drainage losses of NO3-N. Subsurface DI can be reduced without compromising yields and profits by providing good surface drainage as part of the overall drainage system (Skaggs et al., 2005).

Methods and guidelines for designing drainage systems to minimize NO3-N losses are useful for new or replacement systems, but not for the millions of acres of drained land with existing systems. An effective method for reducing DI on existing drained lands is to use controlled drainage (CD) to reduce DI during periods when drainage is not necessary, or during periods when it can be reduced without damage to the crop. The first studies on the use of CD to reduce NO3-N losses from drained lands were conducted by Gilliam et al. (1979) in eastern North Carolina. They found that CD during the winter months reduced drainage volumes and NO3-N losses by about 50%. They observed that CD did not substantially affect the NO3-N concentrations in the drainage water but increased the deep and lateral seepage that passed through reduced (denitrification) zones. Denitrification in those zones removed nitrate before it entered surface or groundwater. Additional field studies (Evans et al., 1989) showed that CD reduced NO3-N losses from three North Carolina sites by 18% to 56%. Since that time, studies on the effectiveness of CD have been conducted for a wide range of locations, soils, and climates. Research in Iowa (Helmers et al., 2012; Jaynes, 2012), Indiana (Adeuya et al., 2012), Illinois (Cooke and Verma, 2012), Ohio (Fausey, 2005), Ontario (Drury et al., 2009; Sunohara et al., 2014), and Sweden (Wesstrom and Messing, 2007) reported results from field studies. CD reduced losses of NO3-N in drainage waters in all cases, with reported reductions ranging from 18% to more than 80%. In nearly all cases, the percentage reduction in NO3-N loss was about the same as the reduction in subsurface drainage volume. CD reduces drainage volume by increasing evapotranspiration (ET), surface runoff, and seepage (Skaggs et al., 2010; Youssef et al., 2018). The relative impacts on the different hydrologic processes are dependent on soil properties, drainage system design and CD management, weather, and crops. The effect of CD on N losses depends on its effect on drainage volume as well as on N dynamics in the soil profile. Where CD conserves water during the growing season and increases yields, some of the N that would otherwise be lost with drainage water is taken up by the crop. In general, such production practices increase crop yield, increase N removed in the crop harvest, and reduce loss of N to the drainage water. An exception is the application of increased amounts N fertilizer, which nearly always results in increased losses of nitrate-N through the drainage system (Kladivko et al., 2004). CD may also decrease nitrate-N losses in drainage water by increasing denitrification of both the subsurface drainage water and the seepage that passes through reduced zones (Gilliam et al., 1979; Youssef et al., 2018).

The wide range of results reported in the literature for the effectiveness of CD in reducing nitrate-N losses in drainage water is assumed to be due to differences in soils, drainage system design, CD management, climate, and other factors previously mentioned. Most of the studies were relatively short-term (two to four years) and therefore provided only limited opportunity to study the year-to-year variability in the performance of the CD practice. The objective of this research was to experimentally determine the effects of CD on crop yields and N losses in drainage water over an extended period. Effects of CD on crop yields for a three-crop, two-year rotation of corn-wheat-soybean were reported by Poole et al. (2013). This article reports the effect of CD on drainage volume and nitrate-N losses based on nine years of observations on replicated plots in eastern North Carolina.

Materials and Methods

The research was conducted on a 13.8 ha field at the Tidewater Research Station near Plymouth (35° 50' 44.79? N, 76° 40' 4.47? W) in eastern North Carolina. The originally forested site was cleared for agriculture in 1975. It is nearly flat and is bound on all four sides by drainage canals that are 1.5 to 2.0 m deep. A subsurface drainage system consisting of parallel plastic drains (10 cm diameter) spaced 22.9 m apart at an average depth of 0.9 m was installed in 1985. Initial studies indicated that the effectiveness of the drainage system was limited by the low hydraulic conductivity of the soil at the drain depth, and a new set of drains was installed in 1991 at an average depth of 1.15 m. The new drains were installed midway between the old drain lines, which were closed by valves for this study. This effectively divided the site into eight plots, with each 1.5 ha plot consisting of the land drained by three adjacent “new” subsurface drains (fig. 1). The research reported in this article was conducted on plots 2, 3, 4, and 5, which have the same soil series, drainage, and boundary conditions. Drain flow rates and other measurements were conducted on the center drain of each plot. The guard lines on either side of the central drain were routed to a separate but identically controlled outlet to hydraulically isolate the 0.5 ha area drained by the center drains from the influence of treatments in adjacent plots.

Several research studies have been previously conducted on the site. Munster et al. (1994) determined the effects of drainage and CD on losses of the pesticide aldicarb (2-methyl-2-(methylthio) propionaldehyde O-(methylcarbamoyl) oxime), and Breve et al. (1997) and Youssef et al. (2006) developed and tested simulation models for predicting N losses in drainage waters and the N cycle in drained lands. Data were collected for the years 1992-1994 and 2007-2012.

Figure 1. General layout of the drainage system at the Tidewater Research Station site.

Results from plots 2 through 5 were used in this study to analyze the effect of CD on N losses via subsurface drainage. Two of the plots were operated in conventional or free drainage (FD) mode, and the other two plots were operated in CD mode. The soil is classified as Portsmouth sandy loam (fine-loamy, siliceous, thermic, Typic Umbraquult), a very poorly drained soil formed in loamy fluvial and marine sediments. The surface horizon is a black fine sandy loam 0.3 m thick with 3% to 5% organic matter. Several layers of fine sandy loam extend to a sandy clay loam layer (0.5 to 0.9 m), which is underlain by alternating thin layers of sandy loam or loamy sand and silt to a depth of 1.0 to 1.2 m, over a layer of gray sand or sandy loam mixed with pockets of sandy clay loam or clay loam to a depth of 2.4 m. Below this depth is a deposit of tight marine clay about 6.1 m thick (Breve et al., 1997). Field effective soil hydraulic properties of the site based on measured drain flow rates and water table elevations were determined by Youssef et al. (2006), who found there was some variability in field effective hydraulic conductivity (K) from plot to plot, and by Skaggs et al. (2012) in a case study for calibrating DRAINMOD (table 1). Based on these K values and the depth and spacing of the drains, the average DI and Kirkham coefficient (KC) were calculated to be 8 and 18 mm d-1, respectively (Skaggs, 2017). The drainage coefficient (DC), or hydraulic capacity of the outlet, is approximately 14 mm d-1.

Table 1. Field effective hydraulic conductivity and drainable porosity for Portsmouth sandy loam on plots 2 through 5 at the Tidewater Research Station site, Plymouth, North Carolina (from Youssef et al., 2006, and Skaggs et al., 2012, table 3 and fig. 6).
Soil Layer Depth (cm)Hydraulic Conductivity (cm h-1)
0-272.8 to 4.0
27-721.8 to 2.6
72-1000.64 to 1.7
100-2400.35 to 0.45
Water Table Depth (cm)Drainable Porosity (cm cm-1)
0-250.01
25-500.03
50-750.05
75-1000.04
100-1500.05

The center subsurface drain of each plot discharges to a 61 cm diameter, 1.8 m tall cylindrical receiving tank in an underground vault of that plot’s instrument house. Both guard lines discharge to an identical receiving tank. The receiving tanks are equipped with sump pumps, float switches, and digital flowmeters that automatically pump water from the receiving tanks to the drainage outlet ditch, control the water level in the tanks, and measure and record the drainage volumes. For FD, the float switches were set so that the water level in the receiving tank was always below the outlet of the drain. In CD, the float switches were set so that water was pumped when the water level in the tank exceeded the set point, which was higher than the drain. To prevent excessive cycling, the pumps were turned on when the water level exceeded the set point by 5 cm and off when it fell 5 cm below the set point. No water was pumped in to maintain the control water level, as would be done in subirrigation. A portion of the drainage water from the center drain was routed to a large sample container. The composite sample was col-lected at two-week intervals, or more frequently during high flow events, and then transported in coolers and frozen or stored at 4°C until analyzed in the North Carolina State University water quality labs. Water table depth was continuously measured in two 100 mm diameter water table observation wells in each plot. The wells were equipped with automatic recorders and data loggers programmed to measure and record the water table depth at 1 h intervals. The wells were located at the quarter points between the center and guard drains.

Precipitation and meteorological data were collected with a Campbell CR-10 weather station installed near house 5 for the 1992-1994 period (fig. 1). Parameters measured included rainfall, air and soil temperature, relative humidity, wind speed and direction, solar radiation, and net radiation. For the 2007-2012 period, these variables were measured at a North Carolina Climate Center weather station located 1 km northeast of the site at the Tidewater Research Station. Additional precipitation measurements were made with on-site recording tipping-bucket and manual rain gauges for the duration of the study. Monthly potential ET was estimated from measured air temperatures using the Thornthwaite method in DRAINMOD, with monthly correction factors based on long-term calculations with the Penman-Monteith equation.

Table 2. Cropping sequence, fertilizer and liming rates, and application dates for the experimental fields during 1992-1994 and 2007-2012.
YearCrop
(Tillage)
CD Weir
Setting
(cm)
Planting, Harvest, and Application Dates (month/day)
(Fertilization and liming rates, in kg ha-1, are shown in parentheses)
PlantingHarvestLimeN Appl 1N Appl 2N Appl 3N Appl 4Total N
1991Wheat (conv.)5011/21-11/15 (1833)11/15 (16.8)----
1992Wheat (continued)50-6/8--2/21 (144)--161
Soybean (conv.)456/2911/17-----0
1993Corn504/298/30-4/29 (5.0)5/4 (79)6/9 (65)6/30 (53)202
Wheat (conv.)3011/15-11/8 (2013)11/12 (19)----
1994Wheat (continued)30-6/14--3/18 (81)--100
Soybean (no-till)306/2111/15-----0
2007Corn (no-till)504/309/123/23 (1120)3/28 (35)4/30 (66)6/11 (109)-210
Wheat (no-till)5012/12-------
2008Wheat (continued)50-6/16--1/3 (37)3/10 (91)-128
Soybean (no-till)506/1711/19-----0
2009Corn (no-till)504/279/151/5 (2240)4/22 (3)4/27 (55)5/27 (109)-167
Wheat (no-till)5011/25-------
2010Wheat (continued)50-6/243/9 (1120)1/3 (43)3/18 (91)--134
Soybean (no-till)506/2911/10-7/27 (0.9)---0.9
2011Corn (no-till)504/198/25-4/13 (1)4/19 (55)5/11 (1)5/23 (109)166
2012Wheat (no-till)501/56/7-1/5 (38)3/8 (100)3/28 (2)-140

The study site was planted to a three-crop, two-year rotation: corn in year 1, followed by wheat in year 1-2, and double-crop soybean in year 2. This cropping sequence is a typical rotation in the region. Corn, the first crop in the rotation, was planted in April and harvested in late August or September. Wheat was planted in mid-November and harvested in June of the following year. Soybean was planted shortly after wheat harvest and harvested in mid-November. The plots remained fallow after the soybean harvest until April of the following year, when corn is planted. The cropping rotation was then repeated. Cultivars and seeding rates varied during the study but remained consistent between treatments during specific growing seasons. The cropping sequence, CD weir settings from the ground surface, tillage practices, and timing and amounts of fertilizer applications for the nine years of observations (1992-1994 and 2007-2012) are given in table 2. The corn-wheat-soybean rotation continued on the site in the intervening years (1995-2006). CD continued on the site through 1998 (Youssef et al., 2006). Arnold (2004) studied the effects of drain depth under FD on NO3-N losses in drainage waters on the site during 2001-2004. The drainage system for the three affected plots was converted back to the use of drains at 1.15 m depth in 2005, and the plots remained in FD until 2007.

Tillage practices implemented at the Tidewater Research Station site included both conventional tillage and no-till. Corn and wheat were grown under both practices, and soybean was grown no-till only. Conventional tillage operations included one or two passes with a disk, one or two passes with a chisel, and then bedding or finishing with a field cultivator with rolling baskets. No-till crops were planted without land preparation.

Nitrogen fertilization followed common practices in the region. Nitrogen fertilizers were applied only to corn and wheat. The primary fertilizer was 30% UAN (33% urea + 42% ammonium nitrate), and it was also used in small quantities as a tank mix partner with weed control chemicals. Other fertilizer types, such as ammonium polyphosphate (10% N + 34% P2O5), were also used. To control the pH of the topsoil, dolomitic lime was applied to the site a total of six times during the study years, as recommended by soil test reports (table 2).

Crop yields were determined at harvest for each plot. Grain yield was measured by harvesting two zones located 6 m on either sides of the center drain line. The area of each zone varied slightly from year to year and from crop to crop depending on planting arrangement. On average, the corn harvest zones were 1.8 m wide by 23 m long. The wheat and soybean harvest zones were 4 m wide by 23 m long.

Water Table Management

Experimental plots 2 through 5 were managed using either FD or CD during 1992 through 1994 and during 2007 through 2012. From 1992 to September 1993, plots 2 and 5 were managed in CD, and plots 3 and 4 were maintained in FD. Because startup delays after installation of the new 1.15 m deep drains (November 1991) prevented initiation of CD until March 1992, CD for the first two years (1992-1993) was conducted only during the growing season and switched to FD when the plots were fallow. This is not the typical management protocol for CD, which includes control during fallow periods, and it likely lowered the reduction of N losses in drainage waters during this 20-month period compared to the typical use of CD. After corn harvest in September 1993, plots 2 and 3 were placed in FD, and plots 4 and 5 were placed in CD. The 1993 growing season was very dry, and corn yields were substantially reduced by drought stresses. Control on the CD plots was continued throughout the remainder of 1993 and all of 1994 at 0.3 m from the surface. The data for 1993 were analyzed and included in this study but only through the date of change in water table management (14 September 1993). Replicated studies of the effects of CD were discontinued at the end of 1994. The research resumed in 2007 with plots 2 and 5 in FD and plots 3 and 4 in CD. Table 2 lists the general CD weir settings by treatment. A more detailed description of the weir settings during the study can be found in Poole (2015). Measurements continued for six years through 31 December 2012, providing a total of nine years of observations on the effects of CD on yields and NO3-N losses in drainage waters.

Table 3. Measured precipitation minus potential evapotranspiration (PET, mm) for April to October (the growing season for corn and soybean in North Carolina).
MonthMonthly Precipitation - PET (mm)MeanSD
199219931994200720082009201020112012
April-50-9-84-1912-59-114-66-54-4939
May27-46-16-21-63-58-7-11331-2946
June-7-6390-45-47-63-94-110-55-4458
July-20-128-52-137144-83-143-1177
August95-58100-54-7089-93084550119
September29-5336-69-51-1929948825111
October-66464-30-28-37-28-466745
Total68-293138-250-240-3-376383-52165
Mean10-4220-36-340-5912-7
SD4659722133831411454937

Data were lost at times between 2007 and 2012 due to power outages, pump failures, or instrument malfunctions. When these periods occurred in one plot but not all the plots, missing values were estimated using an equation developed from a regression and correlation analysis with the plot that had the identical treatment. This equation was developed from a similar period that had no malfunctions. In cases when the period of missing data was short (less than three days), a direct substitution of the data from the plot with the same treatment was used. The majority of missing correlated periods had intermittent portions of reliable data. There were 22 occasions of pump, breaker, or instrumentation failures during the six-year period for the four plots combined (0.9 occasions per year per plot). Four of the 22 occasions lasted less than three days. The majority of these periods of missing data consisted of one-week to two-week intervals (total missing data resulting from pump or instrument failures was less than 2.4% of the six-year period). In situations involving complete power loss to all plots, data were not estimated, and the reported data reflect the power loss and what happened when power was restored.

Statistical Analysis

Data collected over a total of nine years (1992-1994 and 2007-2012) were analyzed to determine the effects of CD on the hydrology and N losses in drainage water and on the grain harvest. Results for CD were compared to results for FD on annual, seasonal, and more frequent bases.

Data for annual total drainage outflows, annual average water table depths, annual average flow-weighted NO3-N concentrations, total annual NO3-N export in drainage water, and NO3-N content in the harvested grain were paired by year and analyzed for drainage treatment effects using paired t-tests with PROC TTEST in SAS (SAS, 2003). Regression relationships (CD versus FD) were developed for the same variables using PROC REG. There were two replicates per treatment in each year of the study. Tests within PROC REG confirmed that the residuals were normally distributed.

Results and Discussion

Average annual precipitation for the nine-year study was 1145 mm, which is 12% below the 50-year (1963-2012) average of 1296 mm for the Tidewater Research Station. Annual precipitation was more than 25% below average during 1993, 2007, and 2008, about average in 1994 and 2010, and 10% greater than average in 2009.

Calculated potential evapotranspiration (PET) exceeded rainfall for 39 of the 63 months of the corn and soybean growing seasons (April through October) during the nine-year experimental period (table 3). This indicates the potential for using CD to conserve water to supply some portion of crop ET demands during subsequent dry periods. Under FD, that water would be drained from the profile and removed from the field during wet periods. Even when PET exceeds precipitation for a month (negative values in table 3), rainfall may exceed PET for short periods. CD can potentially store much of the excess water in the profile to supply ET during subsequent dry periods.

Details comparing observed drainage rates, water table depths, N concentrations, and N losses in drainage waters for CD and FD on a daily or cumulative basis are presented for 1994 (fig. 2), when annual precipitation (1350 mm) was close to the long-term average of 1296 mm. Similar information is presented for a relatively dry year (2008), when precipitation was 930 mm (fig. 3), and for the wettest year of the study (2009), when precipitation was 1431 mm (fig. 4). Similar details for the other six years of the study were reported by Poole (2015). Results in each figure show averages of observed values from the two replications for the CD and FD treatments.

Effect of CD on Drainage and Water Table Depth

Rainfall during the corn season of 1993 was 200 mm be-low normal. Potential evapotranspiration exceeded precipitation during the months of April to August by 304 mm (table 3), which resulted in substantial yield reduction due to drought stresses. As a result of the reduced corn yield, the amount of N taken up by the crop was less than normal, leaving a greater than normal amount of N in the profile that was susceptible to leaching losses during the 1993-1994 drainage season (Chescheir et al., 1996). A more aggressive CD strategy was used during this one year. The water level in the control structure was set at 0.3 m below the surface in September 1993 after corn harvest and maintained at that elevation through 1994. The combination of somewhat higher than average annual precipitation (1348 mm vs. 1295 mm) and a CD setting closer to the surface (0.3 m in 1994 vs. 0.45 to 0.5 m in other years) resulted in a greater reduction in drainage volume per unit area (215 mm) in 1994 than in any season during the study (table 4).

Figure 2. Effect of controlled drainage (CD) and free drainage (FD) on hydrology and NO3-N losses in drainage water during the production of wheat followed by double-crop soybean in 1994 (rain = 1350 mm, normal): A = observed daily subsurface drainage (cm) and NO3-N concentration and sampling dates for CD and FD, B = daily water table depth (m) and CD weir setting, C = cumulative rainfall and drainage (cm), D = NO3-N flow-weighted average concentration in drainage water (mg L-1) and fertilizer application (kg N ha-1), and E = cumulative NO3-N exported in the drainage water (kg N ha-1)
Figure 3. Effect of controlled drainage (CD) and free drainage (FD) on hydrology and NO3-N losses in drainage water during the production of wheat followed by double-crop soybean in 2008 (rain = 930 mm, dry). See caption of figure 2 for details regarding graphs A through E.

On a temporal basis, the effects of CD are dependent on both the elevation of the CD setting and the amount and timing of rainfall. For example, CD reduced water table depths and daily drainage rates during days 1-100, 150-250, and 320-350 for wheat followed by soybean in 1994 (fig. 2). Drainage occurred in the FD treatments when the water table depth was less than the drain depth (about 1.15 m) and in the CD treatment when the water table depth was less than the CD control depth (0.3 m for 1994). Compared to FD, CD reduced the average water table depth by an average of 0.23 m and reduced annual drainage outflow by 50% for 1994.

In the dry year of 2008 (rain = 930 mm, wheat-soybean), CD had a relatively large effect on outflows during days 40-110 (49.1 mm reduction compared to FD) and a similar effect during days 180-210 (40.2 mm reduction) when drain-age resulting from about 160 mm of rainfall was only 25.6 mm for CD, compared to 65.8 mm for FD (fig. 3). In this year, CD reduced the average water table depth by only 0.07 m, but drainage outflow was reduced by 133 mm (68.7%) compared to FD. On a percentage basis, this was the greatest impact on reducing annual drainage flows of the entire nine-year study.

Figure 4. Effect of controlled drainage (CD) and free drainage (FD) on hydrology and NO3-N losses in drainage water during the production of corn in 2009 (rain = 1431 mm, wet). See caption of figure 2 for details regarding graphs A through E.
Table 4. Annual outflows for conventional free drainage and controlled drainage at the Tidewater Research Station, Plymouth, North Carolina.
YearFree Drainage (FD) Outflow (mm)Controlled Drainage (CD) Outflow (mm)Reduction
(mm)
Reduction
(%)
Rain
(mm)
PlotsRep 1Rep 2MeanPlotsRep 1Rep 2Mean
1992[a]3, 4326.9463.3395.12, 5284.9372.9328.966.216.81116.5
1993[a],[b]3, 4264.9330.0297.52, 5229.8327.4278.618.96.3718.6
19942, 3428.9433.2431.14, 5219.2212.3215.8215.349.91348.2
20072, 5133.0167.3150.23, 476.6117.997.352.935.2832.3
20082, 5210.9177.8194.33, 472.049.860.9133.468.7929.5
20092, 5622.1577.6599.83, 4519.5480.6500.099.816.61431
20102, 5349.4343.2346.33, 4192.8183.0187.9158.445.71280
20112, 5282.4286.5284.43, 4156.7129.2143.0141.449.71176.9
20122, 5361.9374.6368.23, 4269.1202.3235.7132.536.01229.2
Total2980.43153.53066.92020.62075.52048.0952.733.210062.2
Mean331.2350.4340.8224.5230.6227.6113.21118.0

    [a] CD not used during fallow periods.

    [b] Data only through 14 September 1993, corn.

CD had a substantial impact on the hydrology during all seasons in the wettest year of the study (2009, corn) (fig. 4). Rainfall was well distributed over the year, with the water table in the CD plots frequently approaching the surface. Average water table depth for the CD treatment was 0.56 m, compared to 0.67 m for FD. Drainage outflows were reduced by only 100 mm (17%) by CD compared to FD for this relatively wet year. This illustrates an important point. When rainfall is well distributed and equal to or greater than ET, there is little opportunity to use the water stored in the profile by CD. The water table remained elevated, and drainage from the CD treatment continued at reduced rates for longer periods than in years when periods of excess rainfall were followed by long dry periods. The impact of reducing outflows by 100 mm (17%) in 2009 was substantial but, on a percentage basis, much less than the 133 mm, or 69%, reduction of outflows observed in the dry year (2008).

A review of figures 2, 3, and 4 and the results for the other six years of observations (table 4) indicates that CD had an effect on drainage outflows (47.5% average reduction) early in the year (days 1 to 120) in seven of the nine years of the study (1994, 2007, 2008, 2009, 2010, 2011, and 2012). CD had a substantial impact during the middle part of the year (days 120 to 240) in all years (60% average drainage reduction) and in the last third of the year (days 240 to 365) in five of the nine years (1994, 2008, 2010, 2011, and 2012), with an average reduction in flow of 52%. These observations demonstrate the need for management of the drainage system on a year-round basis in North Carolina.

In all years, the water table reached the control setting during the late fall and early spring. The CD treatment routinely had less outflow during these periods compared to the FD treatment. This is typical of other studies conducted in North Carolina (Gilliam et al., 1979; Skaggs and Gilliam, 1981; Doty et al., 1975; Evans and Skaggs, 1989; Evans et al., 1989, 1995). Starting in May through September, depending on year, increasing ET from the crop and more sporadic rainfall events usually resulted in the water table falling below the level of control and, in some cases, at or below the tile depth (figs. 2 through 4). Drainage stopped in the CD treatment when the water table reached the control setting but continued to occur in the FD treatment until the water table reached the depth of the drains.

The period between 2007 and 2012 began relatively dry. Rainfall was 478 and 380 mm below the long-term average in 2007 and 2008, respectively. Consequently, the drainage outflow from both the FD and CD treatments ranged from 43% to 73% below the average outflows for the study. The reduction due to CD in 2007 was 53 mm, the lowest reduction due to CD of any year of the study. No flow occurred after day 165 during 2007 due to a severe drought. CD conserved 133 mm of water during 2008, which was also an extremely dry year. The majority of this drainage reduction (69 mm) occurred during one event in early July.

The period after 2008 had average to above-average rainfall. Drainage outflow from both treatments increased during the late summer months of 2009 due to more frequent rainfall events. CD reduced outflow by 100 mm in 2009. The outflows for both treatments were the highest recorded during the study (table 4). Reductions in drainage during 2010 and 2011 for CD occurred primarily during the winter, early spring, and fall in both years.

Drainage outflows and reductions in 2012 occurred soon after planting of soybean. Large rainfall events occurred in early June and throughout July. High water tables were observed in both treatments during the soybean growing season. Water tables above the control elevations during this period resulted in relatively large amounts of drainage from both the CD and FD treatments. More than 80% of the 2012 annual drainage occurred between 1 June and 31 December.

Average annual drainage outflows from the FD plots ranged from 150 mm in 2007 to 600 mm in 2009 (table 4). Average annual outflows from the CD plots ranged from 61 mm in 2008 to 500 mm in 2009. The greatest annual reduction in drainage volume due to CD was 215 mm in 1994. The average annual reduction across all years was 113 mm (33%).

Table 5. Annual flow-weighted average NO3-N concentration data for the study period.
YearFree Drainage NO3--N Concentration (mg L-1)Controlled Drainage NO3--N Concentration (mg L-1)
PlotsRep 1Rep 2MeanPlotsRep 1Rep 2Mean
1992[a]3, 41.92.02.02, 52.51.72.1
1993[a],[b]3, 44.73.84.22, 53.83.43.6
19942, 311.511.411.54, 519.213.516.4
20072, 51.31.11.23, 40.91.61.3
20082, 55.53.24.43, 46.06.56.2
20092, 58.26.77.43, 47.57.27.4
20102, 57.16.66.83, 47.76.77.2
20112, 56.97.77.33, 411.310.310.8
20122, 54.54.74.63, 43.52.43.0
Mean5.75.35.56.95.96.4

    [a] CD not used during fallow periods.

    [b] Data only through 14 September 1993, corn.

A paired t-test on the annual tile flow volumes showed that CD significantly lowered the drainage outflows by an average of 113 mm (95% confidence limits [83 and 144 mm], t(17) = 78.6, p < 0.0001). The 33% average flow reduction for the study is greater than the 12.5% reduction reported by Smith and Kellman (2011) and the 21% reduction reported by Jaynes (2012) and close to the 37% reduction reported by Helmers et al. (2012). If the annual reductions from 1992 and 1993, when CD was not applied during the fallow season, are excluded from the analysis, the average reduction in annual drainage flows is 43%, which is close to the results reported by Gilliam et al. (1979), Evans et al. (1995), and Fausey et al. (2004).

Water table depths for FD and CD are plotted for 1994, 2008, and 2009 in figures 2 through 4, respectively. On average, CD reduced the mean annual water table depth by 0.16 m. Except for 1993, when CD was only applied during an unusually dry corn growing season, CD reduced the average water table depth in every year of the study. The paired t-test showed that the average annual water table depth was significantly less for CD than for FD (p = 0.007).

Effect of CD on NO3-N Losses in Drainage Water

In total, 366 FD water samples and 281 CD water samples were collected during the nine-year study. Annual flow-weighted average NO3-N concentrations (FWANC) ranged from 1.1 to 11.5 mg N L-1 for the FD treatments and from 0.9 to 19.2 mg N L-1 for the CD treatment (table 5). The average annual FWANC for the nine-year study was greater for the CD treatment than for the FD treatment by 0.9 mg N L-1, or about 14.0%. Pairing the concentration data by year, a paired t-test showed that increases in concentrations for the CD treatment were not significant (difference = 0.92 mg N L-1, 95% confidence limits -0.24 and 2.09 mg N L-1, t(17) = 1.67, p < 0.11). These results are consistent with other field studies on the effect of CD on nitrate concentrations and losses, with some exceptions. Fausey (2005) found that average NO3-N concentrations in drainage water were higher for FD than for CD for corn, but not for soybean. Most other previous research on the subject has reported that CD had no significant impact on nitrate concentrations in drainage water and that reductions in N losses were due almost exclusively to reduced drainage volume (Gilliam et al., 1979; Evans et al., 1995; Jaynes, 2012; Helmers et al., 2012; Wesstrom and Messing, 2007).

The largest NO3-N concentrations of the study were measured in 1994 when wheat and soybean were grown on the site. The annual FWANC values were 11.5 mg L-1 for FD and 16.4 mg L-1 for CD. The temporal variation of NO3-N concentrations in the drainage water from the CD and FD plots is shown in figure 2D for 1994. Note that the NO3-N concentrations declined from high values of about 37 and 27 mg L-1 for CD and FD, respectively, at the beginning of the year to less than 10 mg L-1 by day 100. The concentrations for both treatments continued to decline thereafter to about 4 mg L-1 by the end of the year. The high concentrations were probably due to the sequence of events during 1993 when severe drought reduced corn yields to less than 60% of the average yield for the study (Poole et al., 2013). Much of the fertilizer N applied for the corn crop was not taken up by the crop and remained in the soil profile. The control setting on the CD treatment (0.3 m below the surface) was initiated immediately after harvest (14 September 1993) and remained in place through 1994. Leaching of NO3-N in the drainage water of the FD treatment during the fall of 1993 removed more N and reduced the NO3-N concentrations compared to the CD treatment. In spite of the somewhat higher nitrate concentrations, CD reduced N losses in 1994 by more than 14 kg ha-1 compared to FD. Reduced drainage and leaching in the CD plots in 2007 and 2008 were apparently also responsible for increased NO3-N concentrations compared to FD in the drainage water during 2008 and 2009 (figs. 3D and 4D). Thus, the impacts of CD on N losses in drainage water are dependent on legacy conditions and effects in past years. One of the advantages of conducting long-term studies is that there is more opportunity to observe and quantify the impacts of such effects. Recognition of such legacy effects is an obvious, strong reason for not randomizing treatments among plots on an annual basis. Annual losses of NO3-N were smaller for CD than for FD in every year of the study and for the nine years combined (table 6). The range of annual NO3-N losses was 1.8 to 51.0 kg ha-1 for FD and 0.7 to 42.1 kg ha-1 for CD. On average, CD reduced annual NO3-N export by 6.3 kg ha-1, or 30% (from 20.7 kg ha-1 year-1 for FD to 14.5 kg ha-1 year-1 for CD) over the nine-year study. A paired t-test on annual NO3-N export showed that CD significantly reduced NO3-N export by an average of 6.3 kg ha-1 (95% confidence limits 3.5 and 9.0 kg ha-1, t(17) = 4.79, p < 0.0002).

Table 6. NO3-N export in subsurface drainage by treatment and replication at the Tidewater Research Station, Plymouth, North Carolina.
YearFree Drainage (FD) NO3--N Export
(kg ha-1)
Controlled Drainage (CD) NO3--N Export
(kg ha-1)
CD-FD
Difference
(kg ha-1)
Reduction
(%)
PlotsRep 1Rep 2MeanPlotsRep 1Rep 2Mean
1992[a]3, 46.29.47.82, 57.06.36.7-1.114.5
1993[a],[b]3, 412.512.512.52, 58.811.210.0-2.520.0
19942, 349.549.549.54, 542.128.735.4-14.128.5
20072, 51.81.81.83, 40.71.91.3-0.528.2
20082, 511.65.78.73, 44.33.23.8-4.956.8
20092, 551.038.744.83, 438.934.836.8-8.017.9
20102, 524.822.623.73, 414.812.313.5-10.242.9
20112, 519.522.220.83, 417.713.315.5-5.325.6
20122, 516.217.516.93, 49.44.97.2-9.757.6
Total193.0180.0186.5143.6116.5130.1-56.4
Mean21.420.020.716.012.914.5-6.330.3

    [a] CD not used during fallow periods.

    [b] Data only through 14 September 1993, corn.

A linear regression analysis of annual NO3-N transport (kg ha-1) for CD as a function of NO3-N transport (kg ha-1) indicated a significant effect (F(1,17) = 152, p < 0.0001). The resulting regression equation was CD = 0.76FD - 1.32, with an R2 of 0.90. The standard error was 0.062 (t = 12.33, p < 0.0001) for the slope and 1.60 (t = -0.67, p = 0.42) for the intercept. Because the intercept was not significantly different from zero and NO3-N export cannot be less than zero, a regression equation was developed with an intercept equal to zero. The predicted annual CD NO3-N export was equal to 0.721(FD NO3-N export) when export is measured in kg NO3-N ha-1 year-1 (fig. 5). The standard error of the slope was 0.037 (t = 19.5, p < 0.0001). This slope implies that NO3-N losses from the CD treatment at this site were approximately 28% lower than FD losses, which is close to the 30% average difference measured in the study (table 6).

Loss of NO3-N in drainage waters is obviously affected by both the drainage volume and the NO3-N concentration. Even though the average NO3-N concentration was about 0.9 mg L-1 greater for CD than for FD, the reduction in drainage volume for CD resulted in an average of 30% less NO3-N exported, as compared to FD. CD reduced NO3-N export in all years, with a range of 14% (1992) to 57.6% (2012). The average reduction reported here was smaller than that reported by Gilliam et al. (1979) for another site at the Tidewater Research Station, by Evans et al. (1995) for other North Carolina sites, by Fausey (2005) in Ohio, and by Helmers et al. (2012) in Iowa. The reductions are in the same range reported by Jaynes (2012) in Iowa and by Drury et al. (2009) in Ontario but greater than reductions reported by Tan et al. (1998) for a site in Ontario and by Adeuya et al. (2012) for two Indiana sites.

The hydrologic and biogeochemical mechanisms that cause CD to reduce NO3-N losses in drainage water are not likely the same for all sites. Gilliam et al. (1979) found that CD during the winter months at another Tidewater Research Station site (within 1 km of the current study site) reduced subsurface drainage by raising the water table and increasing deep and lateral seepage. Their measurements showed that such seepage traveled through reduced or denitrified zones in and below the soil profile, and they concluded that the reduction of NO3-N exported in drainage waters was due to denitrification in those zones. However, seepage appears to be of minor importance to the hydrology of the current site. Youssef et al. (2006) tested DRAINMOD-N II (Youssef et., 2005) using six years of hydrologic data (1992-1997) from the site. Water table depth, drainage rates, and cumulative monthly and annual drainage volumes were accurately predicted by assuming that seepage was negligible. In an earlier study on the site, Breve et al. (1997) found that NO3-N contents in soil water below the 1.8 m depth were negligible and that the zone below the 2.4 m depth was reduced, but there was no evidence of vertical seepage losses from the profile.

Another potentially important mechanism is the effect of CD on ET, crop yield, and the N removed in the harvested crop. The CD treatment in this study increased average corn yields by 11% and average soybean yields by 10% but had no significant impact on yields of winter wheat (Poole et al., 2013). The measured annual NO3-N exported in the drainage water from each plot for each year (table 6) was added to the N exported in the harvested grain (table 7) to give the total N exported from the treatments by year (table 8). The two sources of N export were combined for each treatment to determine if the reduction of N lost in drainage water from the CD plots could be attributed to increases in N content in the harvested grain. There was no difference in total export of N between the two treatments (t = 0.29, p = 0.78). Total cumulative N export (grain + drainage water N) from the CD treatment over the nine-year study was within 1 kg ha-1 of the FD treatment (table 8). In this case, it appears that the effectiveness of CD in reducing NO3-N losses in drainage waters can be explained by the increased uptake of N by the crop and its removal in the harvested grain. This has not been the case in other studies, where CD was found to be effective in reducing NO3-N losses in drainage water, but yields were not affected, or in some cases were reduced (Gilliam et al., 1979; Fausey, 2005; Drury, 2009; Helmers et al., 2012). This does not imply that ET, crop yield, and reduced drainage losses of NO3-N were the only processes in the N cycle that were affected by CD. The CD treatment could have affected mineralization, N fixation, denitrification, and seepage, which could have impacted the N cycle and NO3-N lost in drainage water. Those processes were not measured in this study.

Figure 5. Linear regression analysis for annual CD NO3-N export versus FD NO3-N export.

Woli et al. (2010) argued that the apparent success of CD in reducing NO3-N losses cannot in general be explained by increases in denitrification in the surface soils. They argued that increases in denitrification, if they exist, are not sufficient to explain the difference in N loss attributed to CD in surface soils. However, the possible effect of denitrification on seepage water is not confined to surface soils. CD reduces DI and tends to return the subsurface flow direction and fate toward that which existed in the prior undrained state. Recent studies on a site without artificial subsurface drainage in the middle coastal plain of North Carolina (Gilmore et al., 2016) indicated that about 50% of the NO3-N in water infiltrating agricultural lands and moving over long periods (~30 years) to natural streams is denitrified in groundwater. Another one-third of the NO3-N was lost by denitrification as the groundwater moved through bottom sediments to enter the stream. Thus, the use of CD to reduce drainage intensities could have substantial impacts on N losses in drainage waters. In the current study, CD conserved drainage water, increased yields, and removed N that, under FD, would have entered surface waters as a pollutant. In other studies, CD did not affect yields, but the researchers concluded that it increased seepage through reduced zones, either within or below the soil profile, where NO3-N was denitrified. It seems logical that the effect of CD on N exported in drainage water results from the cumulative impact of several processes that may vary from site to site, so it is not surprising that the effectiveness of the practice varies among sites, soils, locations, and weather conditions.

Table 7. Nitrogen removed in the grain harvest (kg ha-1) for FD and CD treatments.
YearFree Drainage (FD)Controlled Drainage (CD)CD-FD Difference
(kg ha-1)
Rep 1Rep 2MeanRep 1Rep 2Mean
199230428229331630130815
199359636178687312
1994[a]17117217119620219928
200715413514416514615611
200822126424225026025513
20091511221371661241459
2010191182187169174171-15
2011938589108909910
2012320296308290272281-27
Total16641600163217381637168856

    [a] Wheat N export not included due to fertilizer differences in replicates.

Table 8. Total nitrogen export (drainage N export + grain N removal, kg ha-1) for FD and CD treatments.
YearFree Drainage (FD)Controlled Drainage (CD)CD-FD Difference
(kg ha-1)
Rep 1Rep 2MeanRep 1Rep 2Mean
199231029230132330731514
199371757387798310
1994[a]22022122123823123414
200715613614616614815711
20082332692512552632598
20092021611822051581820
2010216205210184186185-26
20111121071101251041145
2012337314325300277289-36
Total185717801819188217531818-1

    [a] Wheat N export not included due to fertilizer differences in replicates.

Conclusion

A nine-year field study was conducted near Plymouth in eastern North Carolina to determine the effect of controlled drainage (CD) on subsurface drainage outflows and N losses in the drainage water. Rainfall was average to below average in all years except 2009, which was 10% above average. The magnitude of subsurface drainage measured in units of depth (mm, mm3 mm-2 of surface area) was less for CD than for conventional or free drainage (FD) in every year of the study. CD reduced annual drainage volumes by 6% to 69% compared to FD, with an average reduction of 33%. Average annual water table depth for the CD treatment was 0.67 m, which was 0.16 m less that the average for FD over the nine-year study. The overall means of flow-weighted average NO3-N concentrations (FWANC) for the study duration were 6.4 mg L-1 for CD and 5.5 mg L-1 for FD, but the difference was not significant.

CD reduced NO3-N export in the drainage water in all years compared to FD (p < 0.0001), with an average reduction of 6.3 kg NO3-N ha-1 year-1. The percentage reduction ranged from 14.5% to 57.6% with an average of 30% compared to the measured export for the FD treatment. The total reduction of NO3-N exported in drainage water from the CD plots was equal to the increase in N removed by the harvested grain in CD compared to the FD treatment over the nine-year study.

Most of the results presented here for the effects of CD on N loss in drainage water are not greatly different from those reported by others (table 1; Skaggs et al., 2012). The average 30% reduction of NO3-N for CD compared to FD is in the midrange of the values reported in previous studies. As with most other studies, the percentage NO3-N reduction was close to that of the reduction in drainage volume. However, in this study, for the first time, the cause of the reduction in NO3-N loss in drainage water can be attributed to the additional N removed from the field by the increased grain yields harvested from the CD treatment. Another difference is the length of the study. Observations over a nine-year period increased confidence in the measured effects of CD on both crop yields and losses of NO3-N in drainage water, compared to typical field studies of three or four years duration. In this case, the extended study period provided two opportunities (1993-1994 and 2007-2009) to observe the effects of reduced yields due to drought conditions in one year on increased NO3-N losses and the effect of CD on those losses in the following years. Such effects might not have been observed in a shorter study.

References

Adeuya, R., Utt, N., Frankenberger, J., Bowling, L., Kladivko, E., Brouder, S., & Carter, B. (2012). Impacts of drainage water management on subsurface drain flow, nitrate concentration, and nitrate loads in Indiana. J. Soil Water Cons., 67(6), 474-484. https://doi.org/10.2489/jswc.67.6.474

Arnold, L. A. (2004). Effects of drain depth on nitrogen losses in drainage in shallow water table soils. MS thesis. Raleigh, NC: North Carolina State University.

Breve, M. A., Skaggs, R. W., Gilliam, J. W., Parsons, J. E., Mohammad, A. T., Chescheir, G. M., & Evans, R. O. (1997). Field testing of DRAINMOD-N. Trans. ASAE, 40(4), 1077-1085. https://doi.org/10.13031/2013.21360

CENR. (2010). Scientific assessment of hypoxia in U.S. coastal waters. Interagency Working Group on Harmful Algal Blooms, Hypoxia, and Human Health of the Joint Subcommittee on Ocean Science and Technology. Washington, DC: Committee on Environment and Natural Resources.

Chescheir, G. M., Skaggs, R. W., Gilliam, J. W., Breve, M. A., Munster, C. L., Evans, R. O., ... Parsons, J. E. (1996). Effects of drainage and water table control on groundwater and surface water quality: Part II. Experimental results and simulations models. Report No. 301. Chapel Hill, NC: North Carolina Water Resources Institute.

Christianson, L. E., & Harmel, R. D. (2015). 4R water quality impacts: An assessment and synthesis of forty years of drainage nitrogen losses. J. Environ. Qual., 44(6), 1852-1860. https://doi.org/10.2134/jeq2015.03.0170

Cooke, R., & Verma, S. (2012). Performance of drainage water management systems in Illinois, United States. J. Soil Water Cons., 67(6), 453-464. https://doi.org/10.2489/jswc.67.6.453

Craig, N. J., & Kuenzler, E. J. (1983). Land use nutrient yield and eutrophication in the Chowan River basin. Report No. 205. Chapel Hill, NC: North Carolina Water Resources Research Institute.

David, M. B., Drinkwater, L. E., & McIsaac, G. F. (2010). Sources of nitrate yields in the Mississippi River basin. J. Environ. Qual., 39(5), 1657-1667.

Doty, C. W., Currin, S. T., & McLin, R. E. (1975). Controlled subsurface drainage for Coastal Plains soils. J. Soil Water Cons., 30, 82-84.

Drury, C. F., Tan, C. S., Reynolds, W. D., Welacky, T. W., Oloya, T. O., & Gaynor, J. D. (2009). Managing tile drainage, subirrigation, and nitrogen fertilization to enhance crop yields and reduce nitrate loss. J. Environ. Qual., 39(3), 1193-1204. https://doi.org/10.2134/jeq2008.0036

Evans R. O., Skaggs, R. W., & Gilliam, J. W. (1995). Controlled versus conventional drainage effects on water quality. J. Irrig. Drain. Eng., 121(4), 271-276. https://doi.org/10.1061/(ASCE)0733-9437(1995)121:4(271)

Evans, R. O., & Skaggs, R. W. (1989). Design guidelines for water table management systems on Coastal Plain soils. Appl Eng. Agric., 5(4), 539-548. https://doi.org/10.13031/2013.26558

Evans, R. O., & Skaggs, R. W. (2004). Development of controlled drainage as a BMP in North Carolina. In R. Cooke (Ed.), Proc. 8th Intl. Symp. on Drainage. St. Joseph, MI: ASAE.

Evans, R. O., Gilliam, J. W., & Skaggs, R. W. (1989). Managing water table management systems for water quality. Proc. Natl. Water Quality and Life Symp. Speciality Conf., Irrigation and Drainage Division (pp. 541-552). Reston, VA: ASCE.

Fausey, N. R. (2005). Drainage management for humid regions. Intl. Agric. Eng. J., 14(4), 209-214.

Fausey, N. R., King, K. W., Baker, B. J., & Cooper, R. L. (2004). Controlled drainage performance on Hoytville soil in Ohio. In R. Cooke (Ed.), Proc. 8th Intl. Symp. on Drainage. St. Joseph, MI: ASAE.

Gilliam, J. W., & Skaggs, R. W. (1986). Controlled agricultural drainage to maintain water quality. J. Irrig. Drain. Eng., 112(3), 254-263. https://doi.org/10.1061/(ASCE)0733-9437(1986)112:3(254)

Gilliam, J. W., & Terry, D. L. (1973). Potential for water pollution from fertilizer use in North Carolina. Circular 550. Raleigh, NC: North Carolina State University Agricultural Extension Service.

Gilliam, J. W., Baker, J. L., & Reddy, K. R. (1999). Chapter 24: Water quality effects of drainage in humid regions. In R. W. Skaggs, & J. van Schilfgaarde (Eds.), Agricultural drainage (pp. 801-830). Agronomy Monograph 38. Madison, WI: ASA, CSSA, SSSA.

Gilliam, J. W., Skaggs, R. W., & Weed, S. B. (1979). Drainage control to diminish nitrate loss from agricultural fields. J. Environ. Qual., 8(1), 137-142. https://doi.org/10.2134/jeq1979.00472425000800010030x

Gilmore, T. E., Genereux, D. P., Solomon, D. K., & Solder, J. E. (2016). Groundwater transit time distribution and mean from streambed sampling in an agricultural coastal plain watershed, North Carolina, USA. Water Resour. Res., 52(3), 2025-2044. https://doi.org/10.1002/2015WR017600

Gilmore, T. E., Genereux, D. P., Solomon, D. K., Solder, J. E., Kimball, B. A., Mitasova, H., & Birgand, F. (2016). Quantifying the fate of agricultural nitrogen in an unconfined aquifer: Stream-based observations at three measurement scales. Water Resour. Res., 52(3), 1961-1983. https://doi.org/10.1002/2015WR017599

Helmers, M., Christianson, R., Brenneman, G., Lockett, D., & Pederson, C. (2012). Water table, drainage, and yield response to drainage water management in southeast Iowa. J. Soil Water Cons., 67(6), 495-501. https://doi.org/10.2489/jswc.67.6.495

Jaynes, D. B. (2012). Changes in yield and nitrate losses from using drainage water management in central Iowa, United States. J. Soil Water Cons., 67(6), 485-494. https://doi.org/10.2489/jswc.67.6.485

Kladivko, E. J., Frankenberger, J. R., Jenkins, B. J., & Fausey, N. R. (2004). Nitrate losses to subsurface drains as affected by winter cover crop, fertilizer N rates, and drain spacing. In R. Cooke (Ed.), Proc. 8th Intl. Symp. on Drainage. St. Joseph, MI: ASABE.

Kladivko, E. J., Grochulska, J., Turco, R. F., Van Scoyoc, G. E., & Eigel, J. D. (1999). Pesticide and nitrate transport into subsurface tile drains of different spacings. J. Environ. Qual., 28(3), 997-1004. https://doi.org/10.2134/jeq1999.00472425002800030033x

Munster, C., Chescheir, G. M., Skaggs, R. W., Parsons, J. E., Sheets, T. J., Gilliam, J. W., ... Leidy, R. B. (1994). Effects of drainage and water table control on groundwater and surface water quality: Part I. Methodology and preliminary results. Report No. 283. Chapel Hill, NC: North Carolina Water Resources Research Institute.

Nightingale, H. I. (1972). Nitrates in soil and groundwater beneath irrigated and fertilized crops. Soil Sci., 114(4), 300-311. https://doi.org/10.1097/00010694-197210000-00008

Poole, C. A. (2015). The effect of controlled drainage on crop yields and nitrate-N losses on drained lands in eastern North Carolina. PhD diss. Raleigh, NC: North Carolina State University, Department of Biological and Agricultural Engineering.

Poole, C. A., Skaggs, R. W., Cheschier, G. M., Youssef, M. A., & Crozier, C. R. (2013). Effects of drainage water management on crop yields in North Carolina. J. Soil Water Cons., 68(6), 429-437. https://doi.org/10.2489/jswc.68.6.429

Rabalais, N. N., Diaz, R. J., Levin, L. A., Turner, R. E., Gilbert, D., & Zhang, J. (2010). Dynamics and distribution of natural and human-caused hypoxia. Biogeosciences, 7(2), 585-619. https://doi.org/10.5194/bg-7-585-2010

Robertson, D. M., Schwarz, G. E., Saad, D. A., & Alexander, R. B. (2009). Incorporating uncertainty into the ranking of SPARROW model nutrient yields from Mississippi/Atchafalaya River basin watersheds. JAWRA, 45(2), 534-549. https://doi.org/10.1111/j.1752-1688.2009.00310.x

SAS. (2003). SAS user’s guide. Ver. 9.1. Cary, NC: SAS Institute, Inc.

Skaggs, R. W. (2017). Coefficients for quantifying subsurface drainage rates. Appl Eng. Agric. 33(6), 793-799. https://doi.org/10.13031/aea.12302

Skaggs, R. W., & Gilliam, J. W. (1981). Effect of drainage system design and operation on nitrate transport. Trans. ASAE, 24(4), 929-934. https://doi.org/10.13031/2013.34366

Skaggs, R. W., & van Schilfigaarde, J. (1999). Introduction. In R. W. Skaggs, & J. van Schilfigaarde (Eds.), Agricultural drainage (pp. 3-12). Agronomy Monograph 38. Madison, WI: ASA, CSSA, and SSSA. https://doi.org/10.2134/agronmonogr38.c1

Skaggs, R. W., Fausey, N. R., & Evans, R. O. (2012). Drainage water management. J. Soil Water Cons., 67(6), 167-172. https://doi.org/10.2489/jswc.67.6.167A

Skaggs, R. W., Nassehzadeh-Tabrizi, A., & Foster, G. R. (1982). Subsurface drainage effects on erosion. J. Soil Water Cons., 37(3), 167-172.

Skaggs, R. W., Youssef, M. A., & Chescheir, G. M. (2006). Drainage design coefficients for eastern United States. Agric. Water Mgmt., 86(1), 40-49. https://doi.org/10.1016/j.agwat.2006.06.007

Skaggs, R. W., Youssef, M. A., Chescheir, G. M., & Gilliam, J. W. (2005). Effect of drainage intensity on nitrogen losses from drained lands. Trans. ASAE, 48(6), 2169-2177. https://doi.org/10.13031/2013.20103

Skaggs, R. W., Youssef, M. A., Gilliam, J. W., & Evans, R. O. (2010). Effect of controlled drainage on water and nitrogen balances in drained lands. Trans. ASABE, 53(6), 1843-1850. https://doi.org/10.13031/2013.35810

Smedema, L. K., Vlotman, W. F., & Rycroft, D. W. (2004). Modern land drainage: Planning, design, and management of agricultural drainage systems (2nd Ed.). London, UK: Taylor and Francis.

Smith, E. L., & Kellman, L. M. (2011). Nitrate loading and isotopic signatures in subsurface agricultural drainage systems. J. Environ. Qual., 40(4), 1257-1265. https://doi.org/10.2134/jeq2010.0489

Sunohara, M. D., Craiovan, E., Topp, E., Gottschall, N., Drury, C. F., & Lapen, D. R. (2014). Comprehensive nitrogen budgets for controlled tile drainage fields in eastern Ontario, Canada. J. Environ. Qual., 43(2), 617-630. https://doi.org/10.2134/jeq2013.04.0117

Tan, C. S., Drury, C. F., Soultani, M., Van Wesenbeeck, I. J., Ng, H. Y., Gaynor, J. D., & Welacky, T. W. (1998). Effect of controlled drainage and tillage on soil structure and tile drainage nitrate loss at the field scale. Water Sci. Tech., 38(4-5), 103-110.

Wesstrom, I., & Messing, I. (2007). Effects of controlled drainage on N and P losses and N dynamics in a loamy sand with spring crops. Agric. Water Mgmt., 87(3), 229-240. https://doi.org/10.1016/j.agwat.2006.07.005

Williams, M. R., King, K. W., & Fausey, N. R. (2015). Contribution of tile drains to basin discharge and nitrogen export in a headwater agricultural watershed. Agric. Water Mgmt., 158, 42-50. https://doi.org/10.1016/j.agwat.2015.04.009

Woli, K. P., David, M. B., Cooke, R. A., McIsaac, G. F., & Mitchell, C. A. (2010). Nitrogen balance in and export from agricultural fields associated with controlled drainage systems and denitrifying bioreactors. Ecol. Eng., 36(11), 1558-1566. https://doi.org/10.1016/j.ecoleng.2010.04.024

Youssef, M. A., Skaggs, R. W., Chescheir, G. M., & Gilliam, J. W. (2006). Field evaluation of a model for predicting nitrogen losses from drained lands. J. Environ. Qual., 35(6), 2026-2042. https://doi.org/10.2134/jeq2005.0249Youssef, M. A., Skaggs, R. W., Chescheir, G. M., & Gilliam, J. W. (2005). The nitrogen simulation model, DRAINMOD-NII. Trans. ASAE, 48(2), 611-626.

Youssef, M. A., Abdelbaki, A. M., Negm, L. M., Skaggs, R. W., Thorp K. R., & Jaynes D. B. (2018). DRAINMOD-simulated performance of controlled drainage across the US Midwest. Agric. Water Mgmt., 197, 54-66