Limited capacity of river corridor wetlands to remove nitrate: A case study on the Atchafalaya River Basin during the 2011 Mississippi River Flooding

Authors


Corresponding author: Y. J. Xu, School of Renewable Natural Resources, Louisiana State University Agricultural Center, Baton Rouge, LA 70803, USA. (yjxu@lsu.edu)

Abstract

[1] The major 2011 Mississippi River (MR) Spring Flood inundated much of the river's largest distributary basin, the Atchafalaya River Basin (ARB) with a large quantity of nitrogen-rich water. The event provided an opportunity to test the hypothesis that river corridor wetlands and floodplains function as an effective sink for riverine nitrate during an extreme flood event. Water samples were collected from 15 May to 20 July from three sites on the Atchafalaya to determine how nitrate concentration and nitrate isotopes changed from upriver to 182 km downriver. During this period, a total nitrate-nitrogen (NO3N) mass load of 89,600 megagrams (Mg) entered the basin and 83,200 Mg exited the basin, resulting in 7% retention of NO3N. Although this rate is more than previously reported for this basin, the nitrogen removal rate is lower than expected. During the high flood period, we found little change in isotope math formula and math formula values between the upriver site and the river's two outlets, further indicating limited nitrate processing in this large river basin with extensive floodplains and backwaters. The result strongly suggests that flow-through systems such as river corridor wetlands and floodplains may be ineffective in reducing riverine nitrate during a flood event because of limited residence time. However, there was a clear difference in isotope values between the river rising and recession periods. math formula increased from 5.8‰ as the river rose to 7.5‰ in the flood recession, suggesting higher nitrate leaching from subsurface soils during the post-flood period from the upper MR.

1. Introduction

[2] The Mississippi River (MR), draining 41% of the continental United States, delivers each year approximately 953,000 megagrams (Mg) nitrate-nitrogen (referred to as nitrate or NO3N from here on) [Goolsby et al., 2001] into the Northern Gulf of Mexico (NGOM). About 174,600 Mg of the nearly 1 million Mg of nitrate input is discharged from the MR's largest distributary, the Atchafalaya River Basin (ARB) that has extensive floodplains and backwater swamps [Xu, 2006]. The excess nitrogen is one of the major causes of the hypoxic dead zone [a condition when dissolved oxygen (DO) concentration in the deepwater is below 2 mg L−1] occurring in NGOM during late spring and summer for the past two decades [Rabalais et al., 2007; Turner et al., 2008]. Ecologically and economically, the hypoxic dead zone can have large reaching effects [O'Connor and Whitall, 2007; Diaz and Rosen, 2011]. Rabalais et al. [2010] found that the extent of hypoxia in July averaged 13,500 km2 from 1985 to 2009, with a range from negligible in 1988 to 22,000 km2 in 2002. The fluctuation of the hypoxic dead zone has been found to be partially dependent on nitrogen load from the upper MR (UMR) [Wang and Justic, 2009], especially during May and June [e.g., Rabalais et al., 1996] when river flow is normally high. To reduce the large nitrogen input to NGOM, several options were suggested in the action plan released by the Mississippi River/Gulf of Mexico Watershed Nutrient Task Force [2008], including diversion of the nitrogen-rich Mississippi water into floodplain wetland systems.

[3] Many studies have found that riverine corridor wetland systems have the capability of reducing nitrogen loading to downstream areas [e.g., DeLaune et al., 2005; Noe and Hupp, 2009]. Floodplain systems have been reported to be effective sinks for riverine nutrients through removal mechanisms including denitrification, assimilation, and subsurface transport [Lindau et al., 1994; Tockner et al., 1999; Forshay and Stanley, 2005]. However, it has also been reported that denitrification in river sediments is rather low because of unfavorable conditions [e.g., Hill, 1979; Alexander et al., 2000]. Conditions that favor denitrification include high concentrations of nitrate and organic carbon with high water temperatures under anoxic conditions [Pina-Ochoa and Alvarez-Cobelas, 2006]. Of these conditions, nitrate concentration in the overlying water was determined as the dominant control on denitrification potential followed by the thickness of the oxic soil surface layer [Christensen et al., 1990]. Racchetti et al. [2011] argued that riverine wetlands increase interaction surface for denitrification while supplying nitrate constantly and, therefore, encourage higher rates of nitrogen removal.

[4] Channels of most rivers today are confined by levees for flood control and navigation purposes. The confinement separates the rivers from their natural floodplains, limiting or eliminating element exchange between water and terrestrial systems. This is particularly the case with large river systems, such as the MR, whose current path is estimated to cover only 10% of its once vast floodplain. Alexander et al. [2000] reported that nitrogen loss by denitrification decreases with increasing channel size; therefore, despite the AR's potential for denitrification, it will occur when the channel water interacts with its extensive floodplain. According to our previous sampling from the Atchafalaya [BryantMason et al., 2012], this may be limited to very high flood stages, higher than typically seen in the yearly spring floods. With little progress made in reducing nitrate transport in the MR and in some locations nitrate increasing [Sprague et al., 2011], determination of nitrate reduction techniques, especially during high flow events, is vital.

[5] Although the AR would appear to be an ideal area to reduce nitrate loading from the MR, it does not do so under average conditions when examining the annual NO3N budget [Xu, 2006; Turner et al., 2007]. A significant flooding event should in theory allow the river to leave the channel to interact with high denitrification-potential hotspots found in the basin by Scaroni et al. [2010]. The 2011 major MR flood provided a unique opportunity for us to conduct a rapid sampling to test the hypothesis that floodplains function as a significant sink for nitrate during an extreme flood event. Combined with mass balance data, paired isotope technique can determine removal processes such as assimilation and denitrification [e.g., Wassenaar, 1995; Cohen et al., 2012]. We also aimed to assess what role the timing of the flood later in the season played in nitrate removal. During normal river flow conditions, there is low denitrification potential resulting in nitrate loads, math formula, and math formula values being equal at the input and output [BryantMason et al., 2012]. We hypothesized that during extreme flood events, overbank flow occurs and the river water interacts with the floodplain where there is higher denitrification potential. As a result, the nitrate loads will be lower at the output and the math formula and math formula will be higher at the output reflecting denitrification.

2. Methods

2.1. Study Area

[6] The AR is formed by the entire Red River flow from western Texas combined with approximately 30% of the MR's latitudinal flow. The diversion of the MR flow into the Atchafalaya is controlled by a structural complex, the Old River Control structure that was completed in 1963 to restrict the increasing proportion of the Mississippi shifting to the Atchafalaya. Because of the shorter path to the Gulf of Mexico, the Atchafalaya would capture the flow of the Mississippi without intervention resulting in drastic economic effects on the large number of ports in the lower MR [e.g., Roberts, 1998; Ford and Nyman, 2011]. The AR flows southwards approximately 200 km from Simmesport, Louisiana (30°59′00″ N, 91°48′00″ W) into the Gulf of Mexico via two outlets, Morgan City (29°41′35″ N, 91°12′43″ W) and Wax Lake Outlet (29°41′55″ N, 91°22′24″ W), Louisiana (Figure 1). The river is confined by levees on the east and west, in a distance varying from several kilometers in the north to approximately 35 km in the south, creating a wide floodplain basin for a more natural lowland system [Ford and Nyman, 2011]. In its first 110 km south of the MR diversion, the AR flows in a well-confined channel. Afterward, it becomes a series of braided channels that are highly connected with the surrounding landscape. The sediment-rich water from the MR has resulted in filling in of the basin, converting many of the open water regions in the ARB to bottomland hardwood forests especially in the northern part of the basin [Coleman, 1988; Roberts, 1998] reducing connectivity of the river except during high floods.

Figure 1.

Sampling locations on the Atchafalaya River (Simmesport, Wax Lake, and Morgan City) during the 2011 Mississippi River Spring Flood. The Morganza Spillway was opened during the peak flood weeks.

[7] The ARB is about 4678 km2 and composes predominantly wooded lowland and cypress-tupelo surface flow swamp with some freshwater marshes in the lower basin area. The river is channelized to allow for navigation, and the basin as a whole is managed as a flood control basin. The basin serves as a major floodway for the MR floodwaters; therefore, more of the MR water can be directed into the basin from the Morganza Spillway during extremely high flow periods to reduce flooding potential for downriver cities such as Baton Rouge and New Orleans.

[8] In spring 2011, the lower MR rose rapidly. The river stage at Baton Rouge began increasing in early March. By 9 May 2011, river discharge was steadily increasing (Figure 2) and stage reached 12.4 m, 0.2 m higher than its major flood stage. To protect the cities of Baton Rouge and New Orleans, the United States Army Corps of Engineers [USACE, 2011] began opening the Morganza Floodway on 14 May 2011 (Figure 2). On 18 May 2011, the maximum number of bays for this flood event was opened, diverting 3228 m3 s−1 of water into the ARB. Additional protection was also needed for the cities of Morgan City and Berwick, and thus, the riverside protection walls were closed to block the river water which left its channel from reaching the nearby structures.

Figure 2.

Discharge at the input (Simmesport and Morganza Spillway) and output (Wax Lake and Morgan City) during the 2011 Mississippi River Spring Flood.

2.2. Sampling Design

[9] During the 10-week high flow period from 14 May 2011 to 20 July 2011, we collected water samples at three locations on the Atchafalaya: Simmesport (considered as input) and Wax Lake Outlet and Morgan City (together considered output). Each sampling effort was completed in a single day, with sample frequency ranging from twice to once per week depending on how quickly river stage was changing. Composite grab samples were collected from shore. In fast-flowing main channels, the chemical constituents are uniformly mixed making the sample representative of the entire water channel [e.g., Fry and Allen, 2003]. Samples collected were filtered through a GF/F glass fiber filter (Whatman International Ltd., Maidstone, England). Samples were preserved with 25% hydrochloric acid, lowering the pH to 2, and kept at 4°C until analysis.

[10] To determine water conditions at the sampling time, in situ measurements including river water temperature, DO, and specific conductance were also made at the three sampling locations. Daily average river discharge was obtained from three gauging stations: Simmesport (USACE station no. 03045), Wax Lake [United States Geological Survey (USGS) no. 07381590], Morgan City (USGS no. 07381600), and an USACE temporary gauge at the Morganza Spillway. Standard error for river discharge ranges from 3% to 6% [Sauer and Meyer, 1992].

2.3. Isotope Analysis

[11] Nitrate isotope values ( math formula and math formula) were measured using the azide method of McIlvin and Altabet [2005]. Briefly, this method reduces nitrate first to nitrite with cadmium and then to nitrous oxide in a sealed 20-mL vial with azide/acetic acid buffer. Analysis of the resulting nitrous gas was performed with an Isoprime mass spectrometer (GV Instruments, Manchester, England) in the Biogeochemistry Laboratory at the University of Massachusetts, Dartmouth, MA. Delta values are expressed relative to atmospheric nitrogen for math formula and to Vienna standard mean ocean water for math formula. Ratios are used to represent the abundance of heavy to light isotope, as in the case of nitrogen isotope ratio (RN):

display math(1)

Isotopic composition is presented in delta (δ) notation:

display math(2)

where RA is the isotope (15N/14N or 18O/16O) ratio measurement of sample A and RSt is the isotope ratio measurement of the standard. Analytical reproducibility ranged from 0.2‰ to 0.4‰. In addition, flood samples were analyzed for nitrate concentration using the vanadium method on a SmartChem 200 discrete analyzer (Westco Scientific Instruments, Inc., Brookfield, CT). Nitrate concentrations are presented as milligram per liter of NO3N.

2.4. Mass Load Estimation and Statistical Analyses

[12] Daily NO3N mass loads for the three sampling locations were computed by multiplying daily discharge, and the nitrate concentrations were measured at the locations. To estimate nitrate mass input from the Morgaza Spillway during the opening (14 May to 7 July), the nitrate concentration measured at Simmesport was assumed to be representative of the Morganza Spillway because the low Red River flow during the MR flood made little effect on the water chemistry at Simmesport. Estimated mass loads for Simmesport and Morgaza Spillway were summed up to represent total nitrate input into the Atchafalaya, and the sum of the estimated mass loads for Morgan City and Wax Lake Outlet was used as total nitrate output from the basin. The mass balance for the basin (ΔNO3N) is therefore the difference between the input and output, which is given as follows:

display math(3)

where QSim, QM, QMC, and QWL are the discharge at Simmesport, Morgaza Spillway, Morgan City, and Wax Lake, respectively, and Csim, CMC, and CWL represent nitrate concentrations of the accordingly locations. A water budget is the difference between inflow (i.e., sum of the discharges at Simmesport and Morganza Spillway) and outflow (i.e., sum of the discharges at Morgan City and Wax Lake Outlet), which is given as follows:

display math(4)

where QSim and QM are the surface flows into the basin at Simmesport (QSim) and the Morganza Spillway (QM). QMC and QWL are the surface flows out of the basin at Morgan City and Wax Lake (QMC and QWL). Input from rainfall during the 10-week study period is considered to be negligible when compared with the amount of water and nitrate inputted from the MR.

[13] Based on discharge, data were separated by rising and receding flow condition. Dates of peak discharge varied at all three sites, with receding flow beginning on 28 May at Simmesport, 1 June at Wax Lake, and 3 June at Morgan City. A two-way ANOVA test was used to evaluate significance in difference in in situ water quality variables (i.e., river water temperature, DO, and specific conductance), nitrate concentrations, and isotope values among sites and flow conditions, with nesting of date within limb. An α-value of 0.05 was used. Statistical analyses were performed with Proc Mixed on SAS 9.2 software (SAS Institute, 2008). When there was no significant difference among sites, data were pooled by flow condition. Interrelationship of measured parameters was investigated using Pearson's product moment correlation analysis.

3. Results

3.1. River Water Conditions During 2011 Spring Flood

[14] During the 10-week high flow period, river water temperature increased from 19.1°C to 30°C with an average of 26°C. The temperature increase was sharp during the first 4 weeks and continued slowly for the remaining measured weeks (Figure 3). All sampling sites had relatively well-oxygenated water throughout the high flow period with DO levels mostly above 5 mg L−1. Because in situ measurements were limited at Morgan City to after 14 June, DO was skewed to an overall lower mean (4.2 mg L−1). Specific conductance during this flood period averaged 0.360 mS cm−1, ranging from 0.239 to 0.458 mS cm−1. Water temperature and specific conductance were positively related, though neither varied largely among the sampling sites. During the flood recession, river water temperature increased on average nearly 7°C in the receding flow, whereas DO decreased to 1.6 mg L−1 (Table 1).

Figure 3.

Measurement of (a) temperature, (b) dissolved oxygen (DO), and (c) specific conductance in the Atchafalaya River during the 2011 Mississippi River Spring Flood.

Table 1. Least Squares Means of Water Temperature, Dissolved Oxygen (DO), Specific Conductance, and Nitrate Isotope Values ( math formula and math formula) for Sites on the Atchafalaya River Separated by Flow Condition During the 2011 Mississippi River Flooda
Flow ConditionNSiteDate RangeNitrate (mg L−1)Temperature (°C)DO (mg L−1)Specific Conductance (mS cm−1)δ15N (‰)δ18O (‰)
  1. a

    In situ data were not available for Morgan City during the rising flow condition.

Rising2Simmesport15 May to May 271.1 ± 0.1520.5 ± 0.186.0 ± 0.280.255 ± 0.0065.8 ± 0.263.2 ± 0.28
 3Wax Lake15 May to 31 May1.0 ± 0.1121.5 ± 0.136.3 ± 0.200.265 ± 0.0045.8 ± 0.193.3 ± 0.21
 2Morgan City15 May to 2 June0.9 ± 0.15   6.0 ± 0.262.0 ± 0.28
  All sites 1.0 ± 0.08   5.9 ± 0.142.6 ± 0.15
Receding10Simmesport28 May to 20 July1.4 ± 0.0726.5 ± 0.075.2 ± 0.100.359 ± 0.0027.3 ± 0.113.6 ± 0.12
 7Wax Lake1 June to 20 July1.3 ± 0.0826.8 ± 0.085.0 ± 0.120.360 ± 0.0037.1 ± 0.123.6 ± 0.13
 9Morgan City3 June to 20 July1.2 ± 0.0727.2 ± 0.094.4 ± 0.140.349 ± 0.0037.6 ± 0.113.0 ± 0.12
  All sites 1.3 ± 0.0426.7 ± 0.045.0 ± 0.060.365 ± 0.0017.4 ± 0.063.6 ± 0.06

3.2. Mass Transport

[15] During the 10-week flood period, a total of 89,634 Mg NO3N entered the basin and a total of 83,158 Mg NO3N exited the basin from the two outlets, showing a nitrate mass reduction of 6476 Mg or a retention rate of 7%. The error for calculated nitrate mass was 5% at Simmesport, 6% at Wax Lake, and 7% at Morgan City. Nitrate retention was highest during the week of 15 May (Figure 3). Nitrate concentrations from the three sampling locations averaged 1.3 mg L−1, varying from 0.7 to 2.3 mg L−1, with one of the downriver locations (Morgan City) slightly lower than the upriver location (f = 3.67; p = 0.02; Table 1 and Figure 4a). Lowest nitrate concentrations were observed at the flood peak, and the highest concentrations occurred during the flood recession. Weekly nitrate load peaked at 14,822 Mg for Simmesport (input) and 10,702 Mg combined for Morgan City and Wax Lake Outlet (output) and then decreased to 7587 Mg at the input and to 8048 Mg at the output. The concentration change was inversely correlated with the flood discharge (Pearson's r = −0.50; p = 0.001), with the lowest nitrate concentration occurring at the peak flow and the highest concentration occurred approximately 1 month later as the river flow receded.

Figure 4.

Water and nitrate balance in the Atchafalaya River during the 2011 Mississippi River Spring Flood. Solid line represents water flow (L/day), and bars represent total weekly nitrate (Mg). Positive values indicate basin retention, whereas negative values indicate basin release. Vertical line notes the starting day (28 May 2011) of the flood recession at Simmesport.

3.3. Isotope Values

[16] Similar to the nitrate concentration, math formula values also increased during the flood recession (Table 1 and Figure 5). There was larger variation in the math formula values from late June through July; however, there was no significant delay in values measured downriver to those measured in the upper AR (p > 0.10). There was a significant difference between the outlets (t = −2.71; p = 0.01). math formula values in the receding flow were significantly higher than those in the rising flow (f = 113.45; p < 0.0001; Table 1), coincidentally in a positive relationship with temperature and specific conductance (Table 2).

Figure 5.

(a) Nitrate concentration and discharge at Simmesport, and (b) δ15N-NO3N and (c) δ18O-NO3N values on the Atchafalaya River during the 2011 Mississippi River Spring Flood.a

Table 2. Pearson's Product Moment Correlation Coefficients for Water Quality Parameters in the Atchafalaya Rivera
 TemperatureSpecific ConductanceDOd15Nd18O
  1. a

    Significant correlation coefficient is given in boldface (for r > 0.37; p < 0.01).

Specific conductance0.83    
DO−0.72−0.35   
d15N0.790.68−0.62  
d18O0.210.190.090.36 
NO3N0.530.81−0.070.350.21

[17] Variability in math formula values existed among the sites and during the study period (Figure 4). Average math formula was 3.4‰ with a fairly narrow range of 2.0‰ to 5.0‰. math formula was significantly lower at Morgan City than at Simmesport (t = 3.96; p = 0.0006) or Wax Lake Outlet (t = 5.01; p < 0.0001). The crossplots of math formula values versus math formula values do not reflect any significant transformation (Figure 6). Although the slope is higher on the crossplot for Simmesport (0.51) when compared with the outlets (∼0.37), a single-point low math formula value on 30 May is affecting the slope at Simmesport. When this point is removed, the slope (0.3807) is similar to the outlets.

Figure 6.

Crossplots of δ18O-NO3N and δ15N-NO3N values on the Atchafalaya River at (a) Simmesport, (b) Wax Lake Outlet, and (c) Morgan City during the 2011 Mississippi River Spring Flood.

4. Discussion

4.1. Nitrate Removal by River Corridor Wetlands

[18] Floodplain systems have been reported to be effective sinks for riverine nutrients through removal mechanisms including denitrification, assimilation, and subsurface transport [Lindau et al., 1994; Tockner et al., 1999; Forshay and Stanley, 2005]. Therefore, we assumed that significant nitrate removal could occur during a river flood through overbank flow. However, our result indicates that such a removal potential for nitrate through flow-through wetlands may be very limited. The 7% reduction in nitrate found in the AR during the major 2011 MR flood is much lower than that we expected. In addition to the limited change in nitrate load, we did not observe a change in the nitrate isotopic signature between the upbasin and downbasin locations. These findings strongly suggest that no additional nitrate transformations occurred within the main river channel. Furthermore, specific conductance did not change from upriver to downriver, indicating that backwaters had little effect on the main river water chemistry. Collectively, these findings indicate that the majority of nitrate is transported through the basin unprocessed and that floodplains and corridor wetlands in the basin play insignificant roles in riverine nitrate removal during floods as commonly assumed. Therefore, substantial modifications may be required to make the flow-through river corridor wetland an effective sink for nitrate. This may be similar for other riverine floodplain systems, which have nitrate-enrichment problems such as the Coastal Plain Rivers in the Chesapeake Bay Watershed [e.g., Noe and Hupp, 2009] and Baltic Sea Catchment [e.g., Voss et al., 2006].

[19] The paired process of nitrification and denitrification can potentially underestimate nitrate removal because it would result in no change in the mass balance. In riverine wetlands of Northern Italy, denitrification from water column nitrate was 60% to 100% of total denitrification, whereas denitrification from nitrified nitrate was limited [Racchetti et al., 2011]. In a bottomland hardwood wetland, Delaune et al. [1996] determined that nitrification made up 5% to 12% of the total nitrate reduced in the floodwaters. If nitrate is readily available in the overlying water, as would be the case in areas connected with the nitrate-rich AR water, nitrified ammonia sources are not a dominant control of denitrification [Christensen et al., 1990]. Our nitrate isotope values did not show any change from the upriver to the downriver location, suggesting that paired nitrification and denitrification was below detection limits.

[20] Only 7% of NO3N was removed during this flood event, which can limit the detection of the removal process. However, considering that the isotope values from the upper Atchafalaya and the outlets followed each other closely, additional input from draining of the backwater areas in the basin does not appear to have occurred. This small nitrate removal was likely hydrologic transport (water removed from channel) or assimilation (biological uptake), which has minimal change in isotope value. Assimilation can be responsible for a large portion of nitrate removal [Arrango et al., 2008; James, 2010], especially during summer [Gardner et al., 2011]. Although Kreiling et al. [2011] determined that denitrification was the dominant removal mechanism for an UMR backwater, they also acknowledge that they may have underestimated assimilation.

[21] The Atchafalaya exported a large quantity of NO3N during this short flood period. The 83,158 Mg exported represents nearly 48% of the long-term annual average nitrate export [174,600 Mg; Xu, 2006]. Although the amount of nutrients loaded into a system can limit retention [Hopkinson, 1992], with higher nitrate load suggesting lower possible retention, other factors may have also contributed to the low retention rate. Alexander et al. [2000] and Boyer et al. [2006] reported that nitrogen loss by denitrification decreases with increasing streamflow, water depth, and hydraulic load. Shortened residence time in the basin likely affected nitrate removal. In a study of a freshwater marsh receiving MR water, denitrification in the receiving wetland was mainly determined by discharge and resulting retention times, with lower retention time (i.e., 1 day) having lower denitrification rates [Yu et al., 2006]. Removal efficiency reached 95% with a 5-day retention period, more than double the removal efficiency of the lowest retention period (1 day). Caernarvon, which diverts freshwater from the MR to the Brenton Sound, has high removal efficiency of nitrate (88% to 97%); however, the residence time is longer and the loading rate is a fraction of the ARB [Lane et al., 1999]. In a study of a Southeastern U.S. treatment wetland, effective nitrate removal (90%) by denitrification could be achieved with a retention time of 3 days as long as there was sufficient carbon source [Misiti et al., 2011]. Nitrate removal by a natural tropical riverine wetland system was found to be negligible because of short residence time (6 h) and high flow, conditions unfavorable to denitrification [McJannet et al., 2012]. During average conditions, it takes approximately 36 hours for water to travel from the inlet near Simmesport to the outlets. We speculate that during the flood, it may have been shorter as the in situ measurements followed each other closely at all sites during the sampling period. Additionally, the Atchafalaya Basin can be divided in “compartments” based on water management units or subunits, with the lower basin containing more compartments. During the 2011 flood, these compartments likely filled and then may have acted as a hydrologic dam. This may also explain the limited variation in the in situ measurements observed from upriver to downriver. Although the flood provided a pulse of water, it was also traveling quickly leaving little time for denitrification to occur effectively.

[22] In a constructed wetland in Korea, nitrate retention was least effective at 25% ± 17% among nutrients including phosphate, ammonium, and total phosphorus [Maniquiz et al., 2012]. In flow-through systems, the water is transported faster than immobilization or storage can occur [e.g., Hopkinson, 1992]. Artificial wetlands are successful in retaining nutrients such as nitrate because flow through the wetland can be largely controlled to encourage low flow and high residence time. When comparing open systems like riparian floodplains to a closed system in Okefenokee Swamp, less than 5% of the inorganic nutrients was retained in the open system when compared with more than 90% in the closed system [Hopkinson, 1992]. The results from our study are in agreement with those findings, and they imply that the ARB cannot become an effective nitrate sink unless substantial modification has taken place to the wetlands for allowing longer residence time.

4.2. Flow Condition Effect on Nitrate

[23] Although a “first flush” would be expected for a flood event, a recent study [Kato et al., 2009] observed that nitrate had the lowest strength of first flushes among eight species of phosphorus and nitrogen. Peak flow in the Mississippi–Atchafalaya system occurs every spring from both rainfall and snowmelt in the UMR Basin, as this is the extreme flood event, which may explain why we observed a delay in peak nitrate at Simmesport, Wax Lake, and Morgan City. Highest nitrate concentrations were also consistently found in the falling stage about 2 months after the peak discharge during the 10-year period of 1995–2005 on the MR [Duan et al., 2010]. This suggests that subsurface flow in the UMR is a dominant source of nitrate to the MR–ARB system in the spring flood pulse, which is also supported by the isotope values reflecting that of soil nitrate ( math formula: ∼5‰ to 10‰) [Kendall, 1998].

[24] Highest retention occurred during the rising flood condition, in which more than half of the total retention occurred. This period also had lower nitrate concentration, which can result in higher percent removal [Hopkinson, 1992]. Additionally, this was the period that had water storage in the basin, which may have resulted in the removal by hydrologic transport, rather than through more stable removal mechanisms like denitrification or assimilation.

[25] There was a significant difference in math formula from the rising to the receding flow condition likely impacted by seasonality of the nitrate source rather than the flood. Voss et al. [2006] observed decreased math formula from winter to spring and then increased math formula from spring to summer with variations ranging from 3‰ to 8‰ depending on the river. Knapp et al. [2010] noted seasonality of nitrate isotope values in precipitation (with higher values in the spring). Duan et al. [2010] determined that seasonality of concentrations in the lower MR was attributed to conservative mixing of the primary tributaries (Ohio/Arkansas Rivers and UMR/Missouri). These tributaries also likely control the seasonality in isotope values we observed in this study.

5. Summary and Conclusions

[26] The 2011 MR Spring Flood transported a large quantity of nitrate-rich water into the ARB, which has extensive floodplains and corridor wetlands. We hypothesized that a large amount of the riverine nitrate would be removed through denitrification, which would be reflected in decreased nitrate load and increased nitrate isotope values downriver sites on the AR. However, our results from this rapid sampling study show little processing of nitrate despite the high connectivity during the major flood in the AR, rejecting the initial hypothesis. The river waters moved quickly through the basin leaving little or no residence time for denitrification. Based on our findings, we conclude that this system may not be a significant sink for NO3N, and we acknowledge that future studies are needed to verify the result gained from this one flood event. Furthermore, this study found higher isotope values in conjunction with peak nitrate concentrations during the flood recession, indicating that a change of nitrate sources occurred in the UMR from surface to subsurface leaching in the post-flood period.

Acknowledgments

[27] A McIntire-Stennis grant has supported A.B. The authors thank the Biogeochemistry Laboratory at the University of Massachusetts, Dartmouth, MA, including Jennifer Larkum for isotope analysis assistance. They also thank Abram DaSilva for field assistance and Andy Nyman for helping with statistics.

Ancillary