Effects of simulated wintertime flooding to control erosion on selected chemical and microbial properties of agricultural soils in the Mississippi Delta.

Between late fall and early spring in the Mississippi Delta agricultural region (the floodplain between the Mississippi and Yazoo Rivers in the state of Mississippi, USA), agricultural fields are typically without plant cover, precipitation is high, and losses by erosion of soil and associated agrochemicals can be severe. To control erosion, some farmers in this region allow their fields to flood during winter with rain or river water. We evaluated the effects of simulated wintertime flooding of Mississippi Delta soybean and cotton field soils during winter on several aspects of soil chemistry, the chemistry of overlying water, and soil microbial abundances, biomass, and metabolic activity. During a 77-day period of flooding, concentrations of soil [NH.sub.4]-N were up to five times higher and [NO.sub.3]-N three times lower in flooded compared to non-flooded soils. Eighteen days after drainage, however, there were no significant differences in soil chemistry between previously flooded and non-flooded soils. A lthough we detected no differences in denitrification or soil respiration rates during flooding, following drainage the denitrification rate averaged 37 mg [N.sub.2]O-N [ha.sup.-1] [day.sup.-1] in previously flooded cotton field soils, but was undetectable in previously non-flooded soils. Relative to soil pools, minor amounts (<0.3%) of nitrogen or carbon were lost from the soils during flooding by either denitrification or soil respiration. No detrimental effects on agricultural soil properties of managed wintertime flooding were detected, however, as this was a short-term study and not all important properties of the soils were examined, additional study is recommended before this practice is adopted widely.

Keywords: agricultural soils, flooding, wetlands, biogeochemistry 

**********

Since the first half of the last century, intensively farmed land has contributed to a doubling of nutrients entering the Mississippi River (Turner and Rabalais 1994), from which it is estimated that over 330 million tons of soil enters the Gulf of Mexico annually (Cooper 1993). Erosion is also responsible for the translocation of agrochemicals (nutrients and pesticides) into area waters (Douglas et al. 1998; Sharpley et al. 1995). Eutrophication (Carpenter et al. 1998) resulting from nutrient pollution is associated with poor water quality (Christie and Smol 1996), toxic algal blooms (Burkholder 1998), oxygen depletion (Lowery 1998), fish kills (Burkholder and Glasgow 1997), and loss in biodiversity (Xie et al. 1996). Soil erosion is a leading factor in reduced soil fertility in agricultural systems in the U.S. (Pimental et al. 1995), and can negatively impact crop yields (Dormaar et al. 1997; Syers 1997; Pimental et al. 1995). With reduction of soil erosion and loading of chemicals from agricultural fields and other sources into aquatic ecosystems, some of these problems can be minimized or reversed (Edmondson and Lehman 1981).

In the southeastern United States the greatest loss of soil from agricultural fields occurs during the non-growing season coincident with high rates of precipitation and a lack of plant cover. Of increasing interest in the Mississippi Delta agricultural region (the floodplain of the Mississippi and Yazoo Rivers) is an approach to reduce erosion by intentional, managed flooding of fields during the non-growing season. Agricultural soils in the Mississippi Delta often are high in clay and silt, and if surface runoff is prevented, the field will flood. To facilitate flooding, state and federal agencies have assisted in the installation of slotted-board risers in some cotton and soybean fields. These structures are a barrier to surface flow, sustaining a specific water level on the field, and reducing erosion by protecting the soil from disturbance by additional rain (Green 1998). Secondary benefits include creation of habitat for migrating waterfowl (Maul and Cooper 1998), and possible suppression of soil pathog ens. Potentially detrimental effects include alteration of biogeochemical cycles leading to reduced soil fertility or mobilization of toxic chemical compounds, and disturbance or extirpation of normal soil microbial and animal communities.

The biogeochemistry of nitrogen under flooded soils was of particular interest to us, because N limits agricultural production in the Mississippi Delta, and may limit algal production in the Gulf of Mexico into which the Mississippi flows (Walsh et al. 1981; Lovejoy 1992). The mechanisms and extent of N export from flooded fields have strikingly different consequences for environmental quality. If nitrification is significant, [NO.sub.3]-N could enter the overlying water column or ground water, resulting in high concentrations of dissolved N exported from the fields by percolation or during drainage. If N is primarily removed by denitrification, however, the problem of N pollution of aquatic ecosystems is reduced. Denitrification from temporally flooded agricultural fields as well as wetlands can be an important vector for N export (Groffman 1994; Davidson 1992). Bacon et al. (1986), for example, reported that [NO.sub.3]-N concentrations fell by 90% during intermittent flood irrigations of rice fields, and th at most of this loss was due to denitrification. If export rates are substantial by any mechanism, such as runoff, percolation or denitrification, long-term soil fertility could be compromised (Reddy et al. 1989). Conversely, the net result of flooding may be to reduce N losses in runoff from exposed soils, or even to increase soil N concentration by providing waterfowl habitat (Maul and Cooper 1998).

Phosphorus may also limit primary production, especially in aquatic environments. Where there is oxygen, inorganic P occurs predominately in a particulate form (Mitsch and Gosselink 1993). In anoxic conditions, however, inorganic P compounds are released in a soluble form (Wetzel 2001). Thus, temporary flooding potentially could promote solubilization and diffusion of soil P, reducing soil concentrations and promoting downstream eutrophication.

This research was designed to address the following hypotheses, all of which are based on the premise that flooding will reduce soil oxygen concentration:

1. Flooding will decrease rates of nitrification, microbial respiration and organic matter decomposition.

2. Flooding will enhance rates of denitrification.

3. Flooding will promote movement (by solubilization, desorption and diffusion) of C, N, and P chemical species from the soil layer into the overlying water column.

To address these questions, an experiment was conducted in which we simulated wintertime flooding of soils from a series of cotton and soybean fields in the Mississippi Delta. Over the course of a 96-day incubation, the soils were analyzed for concentrations of various species of nitrogen, organic carbon, and total phosphorus. Additionally, we examined microbial community properties including microbial biomass, denitrifying bacteria abundance, and rates of denitrification and soil respiration.

METHODS AND MATERIALS

Soil Collection Sites--Soils used in this study are from three cotton (Gossypium hirsutum L.) and three soybean (Glycine max Mern) fields in the Mississippi Delta. Fields from which soil samples were collected are in Leflore County (one cotton and one soybean field at Runnymede Farms, 1.6 km SE of the town of Itta Bena), Coahoma County (two cotton fields on the Omega Plantation, 5 km E of Clarksdale), and Quitman County (two soybean fields at Hauff Farms, 10 km E of Clarksdale). The soils in these areas are Southern Mississippi Valley alluvium of the Dubbs-Dundee-Forestdale association (USDA 1959a; 1959b). They have organic matter concentrations less than 1.0%, pH values between 4.4-5.7, and depending on the site consist of approximately 20-55% sand, 40-60% silt, and 5-40% clay (A&L Analytical Laboratories, Inc., Memphis, TN).

Experimental Design--All soils used in this study were collected in early January 1998. At each of the four agricultural fields sampled in Coahoma and Quitman Counties, two 19-liter plastic buckets were filled with soil collected along a 30-in transect. Depth of soil collection did not exceed approximately 30 cm. At each of the two field sites in Leflore County, four buckets were filled using a similar method. Following collection, soils were transported to our laboratory at the University of Mississippi, Oxford, 120-150 km from the field sites. Soils from each site were homogenized by manual mixing with a spade and distributed into cylindrical 4.2-liter polyethylene microcosms (model # RSF4, Cambro Manufacturing Co.). Each microcosm contained a dry soil mass of approximately 2.4 kg, having a depth of 13 cm and a surface area of 17 [cm.sup.2] Twelve microcosms were prepared per field site (cotton or soybean) in Coahoma and Quitman Counties, and twenty-four microcosms for each field in Leflore County.

Six microcosms from each Coahoma and Quitman County field site, and twelve of the microcosms containing soil from each Leflore County field, were "flooded" with deionized water to a depth of 5 cm above the soil surface (approximately 1.1 liter). Throughout incubation, deionized water was added to these flooded microcosms as needed to maintain a depth over the soil surface of 5 cm. To the other microcosms from each field, 200 ml of deionized water were added every two weeks for the first four weeks of the experiment, equivalent to a rain event of approximately 0.9 cm in depth over the soil surface. For the remainder of the experiment, 100 ml of deionized water were added to these microcosms weekly. A record of soil moisture contents was maintained throughout the experiment. Non-flooded soils had a gravimetric water content of 16-20%. Soils in flooded microcosms ranged between 28-30% water, but were continually overlain with water.

To expose the microcosms to ambient outdoor light and temperature, they were incubated on the flat roof of Shoemaker Hall on the University of Mississippi campus. The microcosms were placed randomly into one of four insulated, open-topped wooden boxes (24 microcosms per box). A sheet of OP-4 plexiglass (AIN Plastics) was positioned 10 cm above the top of the microcosms. OP-4 plexiglass transmits all ultraviolet and visible light, and served to prevent precipitation from entering the individual microcosms, allowing us to precisely regulate soil moisture and water depth. Air temperatures at the incubation site were provided by the U.S.D.A.-A.R.S.

On each sample date one flooded and one non-flooded microcosm from each cotton and soybean field were sampled, except for the Leflore County fields from which two microcosms were sampled from each field. From each microcosm, one soil sample of 150 grams (dw) was collected with a 2.5cm plastic corer to a depth of 8 cm. Concurrently, water samples from the flooded microcosms were collected from the overlying surface water. Each microcosm was sampled only once, i.e. destructively.

The first samples were taken on January 4, 1998, prior to flooding. The microcosms were flooded on January 5. Four samples were taken during the period of flooding at approximately two-week intervals, until day 57 of flooding. After 20 more days, during which no samples were collected, the flooded microcosms were drained by slowly pouring off the surface water. Final samples from the microcosms were taken 18 days later on April 10, at which point gravimetric soil moisture content averaged 13% in both previously flooded and non-flooded microcosms.

Analytical Methods Used for Measurements of Soil and Water Chemistry--Prior to all chemical and biological analyses, soil samples were individually homogenized by manual mixing. Soils were analyzed for ammonium and nitrate using the accelerated diffusion method (Khan et al. 1997). Measurements of total nitrogen and total organic carbon in the soils were made with a Leco CN2000 soil analyzer. Total phosphorus in soils, measured on only the first (pre-flood) and last days of incubation, was analyzed by the ascorbic acid method on 0.45-[micro]m filtrates of samples prepared by sulfuric acid persulfate digestion (APHA 1995). All measurements of soil chemistry were normalized by mass (kg) of dried soil. Samples of overlying water were filtered through Whatman 42 paper or GF/F filters (total phosphorus only) and analyzed for dissolved nitrate, ammonium, and total phosphorus, using the same methods as described for soil extracts. For determination of moisture content, the soils were weighed, dried at 60[degrees]C to a constant weight, and re-weighed. Particulate organic matter content was estimated as the difference in soil weight before and after combustion at 500[degress]C.

Procedures Used for Measurements of Denitrification and Soil Respiration--Denitrification and respiration rates of microcosm soils were made in conjunction with measurements of soil and water chemistry. On each measurement date, one microcosm from each of the four Coahoma fields, and two microcosms from each of the two Leflore County fields were sampled, once for each measurement of denitrification and respiration. Following these measurements, the microcosms were sampled for soil and water chemistry (see above), and microbial community parameters (see below).

Denitrification was estimated by the acetylene-block method (Mosier and Klemedtsson 1994). For these measurements, an open-ended chamber was inserted 10 cm into the soil. Chambers were 60-cm long by 7.5-cm inner diameter PVC pipe capped at the upper end. To enable gas sampling, a 4-mm rubber septum was fitted into a hole in the side of the chamber. While the chamber was being set a 20-ga needle was inserted into the septum to allow air inside the chamber to equilibrate with atmospheric pressure. Following placement of the chamber, 60 ml of sulfuric acid washed [C.sub.2][H.sub.2] (NexAir, Memphis, TN) was injected by syringe into the chamber headspace. To mix the gas through the chamber and into soil pores the syringe was used as a pump to alternately reduce and increase internal pressure (Mosier and Klemedtsson 1994). Overpressure due to added [C.sub.2][H.sub.2] was vented by momentary replacement of the needle into the septum. Two to four hours later, a 5-ml sample was removed from the headspace of each cham ber for an initial measurement of [N.sub.2]O concentration. Two more samples were taken subsequently at 3-hour intervals.

[N.sub.2]O concentrations were analyzed on a Hewlett-Packard 5890A gas chromatograph equipped with an electron capture detector, using 95% Ar 5% [CH.sub.4] carrier gas (NexAir, Memphis, TN) (Crill et al. 1995). [N.sub.2]O was separated on a 2-m long, 3.2-mm outer diameter stainless steel packed column using 60-80 mesh sized Porapak-Q packing material. [N.sub.2]O concentrations were calculated from linear regression equations obtained from injecting gas standards made by diluting pure [N.sub.2]O (+99% [N.sub.2]O, Aldrich) in ultra high purity nitrogen (~99.998% [N.sub.2], NexAir). [N.sub.2]O flux rate was determined as the slope of the linear regression between sample concentration and incubation time. Denitrification rates by soil surface area were determined as

Denitrification rate ([N.sub.2]O surface [area.sup.-1] [time.sup.-1]) = volume [N.sub.2]O evolved x [N.sub.2]O density x incubation [time.sup.-1] x chamber surface [area.sup.-1]

where [N.sub.2]O density equals 1.8 g [1.sup.-1] (at 25[degrees]C and standard pressure).

Soil respiration was determined prior to the denitrification measurements by capture in a base trap of [CO.sub.2] evolved during incubation in a closed space (Zibilske 1994). Four ml of 2M NaOH in 20-ml scintillation vials were placed into each microcosm to be sampled. In the case of the flooded microcosms, the vial was set on a 10-cm glass petridish floating on the water surface. The microcosms were sealed for 6 to 18 hours, then the vial removed for chemical analysis.

[CO.sub.2] dissolved in the NaOH was determined by titration with 0.1M [H.sub.2][SO.sub.4] after the addition of 3M Ba[C1.sub.2], with phenolphthalein used as an endpoint indicator (Zibilske 1994). Sample values were corrected for blank measurements using microcosms not containing soil or water, but were not corrected for the possibility of supersaturation of [CO.sub.2] in the water-column of flooded microcosms. Respiration rates were calculated as

Respiration rate ([CO.sub.2] surface [area.sup.-1] [time.sup.-1]) = ([CO.sub.2] chamber - [CO.sub.2] blank) x incubation [time.sup-1] x chamber surface [area.sup.-1]

Procedures Used for Examination of Nutrient Limitation of Denitrification--To determine if N or C availability limited soil denitrification, an experiment was conducted using soils from the 16 microcosms remaining on the final day of incubation. These measurements were made 25 days after drainage. From each microcosm, 50 grams of soil were removed and placed into 4 0.5-liter Atlas7 brand wide-mouth Mason jars for a total of 64 jars. Of the 4 jars prepared from each mesocosm, one jar received only 50 ml of deionized water, 4 jars received a 50-ml solution containing 5 mg [NO.sub.3]-N (as [KNO.sub.3]), 4 jars received a 50-ml solution containing 10 mg [C.sub.6][H.sub.12][O.sub.6]-C, and 4 jars received a 50-ml solution containing both [NO.sub.3]-N and [C.sub.6][H.sub.12][O.sub.6]-C as above. Twenty-four hours later, [H.sub.2][SO.sub.4]-washed [C.sub.2][H.sub.2] at 10% of the jar volume was added. Subsequently, two measurements from the jar headspaces were made 5 to 6 hours apart to determine denitrification flu x rates.

Procedures Used for Measurements of Microbial Abundance and Biomass--Soils were analyzed for total microbial biomass and denitrifying bacteria populations one day prior to flooding and on the last day of incubation. The fumigation extraction method and ninhydrin-nitrogen reaction methods were used to determine microbial biomass (Joergensen 1995a; 1995b). The most probable number method was used to estimate the abundance of denitrifying bacteria (Tiedje 1994).

Data Analysis--A summary description of the experiment is that it was conducted using a split-plot repeated-measures design, where the main plot was field site nested in crop (3 fields of each of two crop types: n = 6 samples), the split-plot was the water treatment (flooded or not flooded: n = 6 samples x 2 water treatments = 12), and the repeated-measures was incubation time (5 measurements per sample: n = 6 samples x 2 water treatments x 5 times = 60 measurements per factor). Statistica 99 software (StatSoft Inc.) was used in all statistical analyses. To reduce the possibility of Type II error associated with potentially high variability in soil parameters, an alpha level of 0.1 was considered significant. For all soil chemistry parameters for which there were data collected during the flooded period, we conducted repeated-measures factorial ANOVA using the first five sample dates (i.e., prior to drainage). Additionally, for all measurements of soil chemistry, as well as soil microbial parameters, we condu cted a repeated-measures analysis using only the pre-flooding and post-flooding data. Only one-half the microcosms (36) contained surface water, which did not allow sufficient error degrees of freedom for analysis of water chemistry by repeated-measures ANOVA. Instead, we analyzed water chemistry as a split-plot ANOVA, with flooding time treated as a continuous variable. As we had two measurements for each date from each Leflore County field, but only a single measurement from the other fields, the two measurements from each date for the Leflore County fields were averaged for statistical analysis.

Because of several missing values (lost during sampling), a repeated-measures analysis was not appropriate for analysis of effects of flooding or crop type on denitrification and respiration rates. Instead, these data were analyzed as a split-plot ANOVA. Temperature and flooding time were placed in the model as continuous variables and, as above, the effect of field variability removed by nesting the field term in crop type. Denitrification and respiration data for the final sampling date were analyzed by two-way ANOVA for effects of previous crop type and flooding regime. The relationships of respiration and denitrification to temperature were analyzed by least-squares regression. In the experiment examining limitation of soil denitrification, we tested for main and interaction effects of carbon and nitrate additions, previous crop type, and water regime using factorial ANOVA.

RESULTS

Air temperatures above the microcosms during the experiment ranged between 4 and 28[degrees]C (Fig. 1). Occasional measurements of soil temperatures revealed little difference between flooded and non-flooded soils, and based on previous observations from winter 1997, microcosm soil temperatures were similar to field soil temperatures (Milburn 1999).

Crop type had no effect (either main or interaction) on concentrations of any of the variables we measured in soils or water (P > 0.1); therefore, figures showing temporal trends in soil and water chemistry do not indicate crop type. In the flooded soils [NH.sub.4]-N concentration increased, whereas in the non-flooded it declined slightly (Fig. 2a). Soil [NH.sub.4]-N was significantly higher (up to 5 times) in flooded soils during the period of flooding, although the magnitude of the effect varied with time. The second sampling date was not included in the ANOVA due to very low values attributed to analytical error. Postflooding [NH.sub.4]-N concentration in the flooded and non-flooded soils were similar, at approximately 1.5-2.5 mg [NH.sub.4]-N per kg soil dw.

On all sample dates during flooding, there was a higher [NO.sub.3]-N concentration in non-flooded soils (up to 3 times); the magnitude of the effect varied with time (Fig. 2b). There was no difference in soil [NO.sub.3]-N concentrations following drainage.

Total N in soil varied little during flooding, or between pre-flooding and post-flooding (Fig. 2c). Total soil P, soil organic C, and percent organic matter also were not significantly different between flooded and non-flooded soils, or between pre flooding and post-flooding (Figs. 2d-f).

Mean concentrations over time of [NH.sub.4]-N, [NO.sub.3]N, and total P in overlying water of flooded microcosms were generally less than approximately 0.5 mg N or P [1.sup.-1], and not significantly different between crop types (Figs. 3a-c), although there was substantial variation between fields as indicated by large standard errors. Only total P, which declined from a mean (both crops) of 0.56 mg [1.sup.-1] to a mean of 0.12 mg [1.sup.-1] during flooding, showed a consistent pattern in time.

Denitrification in flooded microcosms during flooding ranged from a daily mean of 2.1 to 16.3 mg [N.sub.2]O-N [ha.sup.-1] [d.sup.-1] in cotton soils, and 2.6 to 17.3 mg [N.sub.2]O-N [ha.sup.-1] [d.sup.-1] for soybean soils (Figs. 4a, b). Denitrification rates in non-flooded microcosms during the same time period ranged from 1.5 to 12.6 mg [N.sub.2]O-N [ha.sup.-1] d1 in cotton soils, and 2.0 to 21.5 mg [N.sub.2]O-N [ha.sup.-1] [d.sup.-1] for soybean soils. During flooding, temperature was positively correlated to denitrification rate ([r.sup.2] = 0.37), and was the only factor examined that significantly (P < 0.01), although poorly, appeared to influence denitrification rate.

Respiration measurements during the first two days of flooding (days 13 and 20) are considered unreliable due to very high variance and several negative values, and will not be considered further. Respiration rates were always higher in non-flooded compared to flooded soils. During days 40 and 53, they ranged from 87 to 261 g [CO.sub.2]-C [ha.sup.-1] [d.sup.-1] in flooded cotton soils, and 44 to 266 g [CO.sub.2]-C [ha.sup.-1] [d.sup.-1] in flooded soybean soils (Figs. 4c, d). In the non-flooded microcosms, respiration rates ranged from 88 to 435 g [CO.sub.2]-C [ha.sup.-1] [d.sup.-1] in cotton soils, and 232 to 495 g [CO.sub.2]-C [ha.sup.-1] [d.sup.-1] in soybean soils during the same time period. The highest respiration rates in both cotton and soybean soils occurred at the warmest soil temperatures.

Denitrification after drainage was significantly influenced by prior flooding only in cotton field soils (Figs. 4a, b). Denitrification in previously flooded cotton field soils averaged 37 (SE = 13) mg [N.sub.2]O-N [ha.sup.-1] [d.sup.-1], but was undetectable in previously nonflooded soils. Denitrification in post-flooding soybean soils was undetected in both previously flooded and non-flooded soils. Mean respiration rate in postflooding cotton soils was 426 (SE = 49) g [C0.sup.2]-C [ha.sup.-1] [d.sup.-1] if previously flooded, and 622 (SE = 174) g [C0.sub.2]-C [ha.sup-1] [d.sup.-1] in soils that had not been flooded (Fig. 4c). Mean respiration rate in post-flooding soybean soils was 538 (SE = 217) g [CO.sub.2]-C [ha.sup.-1] [d.sup.-1] if previously flooded, and 797 (SE = 222) g [C0.sup.2]C [h.sup.-1] [d.sup.-1] in soils that had not been flooded (Fig. 4d). The post-flooding respiration measurements exhibited high variance, with no significant effects of either soil type or flooding regime.

In the experiment to determine nutrient limitation of post-flooding denitrification, rates ranged from 0.03 to 8.6 mg [N.sub.2]0-N [kg.sup.-1] [d.sup.-1]] in cotton soils, and 0.05 to 8.7 mg [N.sub.2]0-N [kg.sup.-1] [d.sup.-1] in soybean soils (Table 1). Both organic C and [NO.sub.3] addition alone significantly increased denitrification, but the highest rates of denitrification in both crop types was when both were provided.

Assuming a typical flooded period of 90 days, denitrification accounted for less than 0.0002% of total N stored in both flooded and non-flooded soybean and cotton soils (Table 2). Loss of soil organic C by microbial respiration was also small relative to soil organic C content, accounting for less than 0.3% of the organic C pool in all cases.

Results of a comparison of nutrient concentrations in overlying water from microcosms containing Leflore County cotton field soils to concentrations in run-off from the same field measured in 1997-98 by the Mississippi Department of Environmental Quality (Green 1998) are shown in Table 3. Only data from the Leflore County fields are included, because only these fields were monitored for soil erosion. The data are not perfectly comparable because they were determined using different methods and are unequal in number. However, they suggest that flooding could decrease losses of dissolved total P substantially, and [NH.sup.4]-N, compared to losses by erosion.

Soil microbial biomass ranged between about 60 to 80 mg [kg.sup.-1] soil (Fig. Sa) and was unaffected by flooding or crop type. Most probable numbers of denitrifying bacteria were approximately 4.5 X [10.sup.6] cells per gram soil prior to flooding in cotton soils (Fig. 5b), and 2-3 X 106 cells per gram soil prior to flooding in soybean soils (Fig. 5c). There was no significant effect of flooding on abundances of denitrifying bacteria in either soil.

DISCUSSION

The primary conclusion of this experiment is that continual simulated flooding of these six agricultural soils for between 57 and 77 days during winter had little or no detectable effect on post-flooding soil nutrient concentrations or measured microbial community parameters. In contrast, during the period of flooding there were clear and sustained differences in soil concentrations of [NO.sub.3]-N and [NH.sub.4]-N between flooded and non-flooded soils.

In only one case did flooding significantly affect any of the measured microbial community parameters. Although, denitrification and soil respiration rates were unaffected by flooding during the period of flooding, denitrification in previously flooded cotton soils was higher than in previously non-flooded soils. In contrast, denitrification in soybean soils was unaffected by prior flooding. Reddy and Patrick (1976) reported that denitrification is stimulated when soils are alternately flooded and dried compared to when soils are continuously wet or dry. This is in agreement with our observation of a pulse in [N.sub.2]O release from cotton field soils following drainage, but does not explain why a similar pulse was not observed in the soybean soils.

Microbial biomass and denitrifying bacteria population size were unaffected by flooding; significant effects may have been difficult to detect due to inherently high variability in soil microbial data. For comparison, in a study conducted on a rice field in California, Bossio and Scow (1995) also did not detect changes in microbial biomass carbon, but did observe increases in soil respiration, due to wintertime flooding.

From the patterns in concentration changes in soil [NH.sub.4]-N and [NH.sub.4]-N during and after flooding, inferences can be made regarding effects of flooding on activities of microorganisms in the N cycle of these soils. During the period of flooding, mean soil [NH.sub.4]-N concentrations increased from approximately 5 to 12 mg kg [soil.sup.-1]. A similar pattern of increase in soil [NH.sub.4]-N under flooding has been observed in field experiments (Milburn 1999). The source of this [NH.sub.4]-N is uncertain, but is presumably due to mineralization of organic matter with release of [NH.sub.4]-N (ammonification). Under conditions of flooding, oxygen diffusion into the soils is reduced, nitrification would accordingly be negligible, and [NH.sub.4]-N would accumulate. In non-flooded soils, in contrast, [NH.sub.4]-N did not accumulate. Explanations for the latter result are that in the non-flooded soils, (1) rates of [NH.sub.4]-N production by mineralization were lower, (2) microbial assimilation of [NH.sub.4] -N was higher, and/or (3) nitrification rates were higher than in flooded soils. The respiration measurements indicate that organic matter mineralization was not higher under flooding. Although we cannot exclude the possibility of a difference in rates of [NH.sub.4]-N assimilation between flooded and nonflooded microcosms, differences in [NO.sub.3]-N concentrations suggest enhanced nitrification in nonflooded soil.

During the same 57-day period of flooding, mean soil nitrate concentrations declined in both flooded and nonflooded microcosms from about 3.6-5.2 to 0.2-1.0 mg [NO.sub.3]-N kg [soil.sup.-1]. Denitrification cannot explain this pattern, since in both flooded and non-flooded soils N loss by denitrification, although measurable, was much lower than the rates of decline in [NO.sub.3]-N. For example, in the flooded soil there was a net loss by all mechanisms of approximately 5 mg [NO.sub.3]-N kg [soil.sup.-1]. over the first 57 days of flooding (Fig. 2b). At the maximum observed denitrification rate in flooded soils of approximately 16-17 mg [N.sub.2]O-N [ha.sup.-1] [day.sup.-1] (Fig. 4a, 4b), the loss of N by this process accounts for only 0.0007 mg N kg [soil.sup.-1] over 57 days (for a soil depth of 10 cm, average dry weight is 1.3 X [10.sup.6]kg [ha.sup.l]). Clearly, other processes are involved in the loss of [NO.sub.3]-N. Diffusion of soil [NO.sub.3]-N into the overlying water is a possibility, but can accou nt for only about 0.3-0.4 mg [NO.sub.3]-N kg [soil.sup.-1]. Assimilatory reduction of [NO.sub.3]-N is a third possibility to explain the decline in soil [NO.sub.3]-N, but as microbial biomass did not increase measurably during incubation (Fig. 5a) that explanation also is not fully satisfactory, and we consider the question unresolved at present.

Overall, oxygen concentration appears to have been the driving force in the regulation of soil microcosm [NH.sub.4]-N and [NO.sub.3]-N concentrations. The observed patterns are characteristic of wetland sediments, and indicate that typical wetland biogeochemical processes developed quickly in these flooded agricultural soils (Reddy and Patrick 1976; Mitsch and Gosselink 1993). This is despite the generally cold temperatures at which this experiment was conducted, where the microbial influence on chemical transformations would be expected to be less than at other times of the year. Low concentrations of labile organic matter in these highly eroded agricultural soils probably also contributed to limitation of microbial activity, as indicated by the test for nutrient limitation of denitrification (Table 1).

Besides denitrification, nitrogen may be lost from flooded soils by leaching into overlying water If leaching was significant, it is possible that large quantities of nitrogen as well as phosphorus could be loaded downstream when the fields are drained, a result that would partly negate other benefits of wintertime flooding. However, our results indicate that flooding will substantially reduce losses of total phosphorus and contribute to a reduction in dissolved nitrogen in runoff from Mississippi Delta agricultural fields (Table 3).

The question remains: Is wintertime flooding a useful approach for controlling erosion of agricultural soils in the Mississippi Delta, while preventing detrimental changes to either agricultural soil properties or downstream ecosystems? Green (1998) showed that soil and nutrient losses from agricultural fields in this region can be greatly minimized by managed wintertime flooding followed by controlled drainage. Although this experiment suggests minimal changes to soil chemical or gross microbial properties as a consequence of flooding, there are several possible experimental artifacts that must be considered before application of our results to a field situation. First, we used deionized water, instead of rainwater, to flood the microcosms. Where rainwater is a significant source of labile organic carbon and/or inorganic nitrogen, it is likely that microbial activity would be higher than measured here. In a field experiment, however, in which flooding was manipulated using surface runoff (the Leflore County cotton field of this study), we observed similar patterns in [NO.sub.3]-N and [NH.sub.4]-N as in this study (Milburn 1999). Second, we added water manually to maintain the non-flooded microcosms at constant water content. Using this method, we may have underestimated disturbance caused by raindrops falling on bare soil, the impact of which can dislodge soil particles and increase erosion. In this respect, flooding of the soils may have benefits in soil fertility that were not evident in this experiment, but that would be evident in the field (Milburn 1999). Third, during incubation, water was prevented from flowing out the bottom of the microcosms, eliminating possible nutrient losses by sub-surface flow. Where there is substantial percolation of surface water to deeper soil depths or groundwater, nutrient flux via sub-surface flow may be significant (Spaulding and Exner 1993). Clearly, however, many Mississippi Delta agricultural fields with limited surface run-off remain flooded for extended periods of time ; this observation, in fact, was the basis of the approach to control erosion by blocking surface run-off that was the original impetus for conducting these experiments. Fourth, as it is risky to use results from short-term studies to predict long-term trends (Franklin 1989), we recommend that multi-year studies be conducted to examine the long-term effects on agricultural soils of intentional wintertime flooding. Finally, before wintertime flooding soils is widely applied, it is important to examine the potential effect of flooding on other chemical processes and soil biological communities. It is possible, for example, that production of methyl-mercury could be stimulated under flooded conditions. This highly toxic compound is formed under anoxic conditions and where mercury deposition is a factor, formation of methyl-mercury under flooding should be evaluated (Atlas and Bartha 1998). Similarly, additional research should consider in detail the short- and long-term effects of flooding on the composition and activities of microbial and animal fauna, which represent soil biological diversity, and can be critical to the texture, chemical properties and fertility of soils.

[FIGURE 1 OMITTED]

[FIGURE 2 OMITTED]

[FIGURE 3 OMITTED]

[FIGURE 4 OMITTED]

[FIGURE 5 OMITTED]

Table 1

Factors affecting denitrification rates in post-flooding soils. Samples
were supplemented (+) or not supplemented (-) with N and P.

Treatment         Denitrification Rate
  Factor
Crop      Floode    N         C         mea
            d                            n
                           ([mg [N.sub.2]O-N
                              [kg.sup.-1]
                              [d.sup.-1])

Cotton      no      -         -         0.04
Cotton      no      +         -         0.04
Cotton      no      -         +         1.90
Cotton      no      +         +         8.50
Cotton     yes      -         -         0.03
Cotton     yes      +         -         0.05
Cotton     yes      -         +         1.80
Cotton     yes      +         +         8.60
Soybean     no      -         -         0.06
Soybean     no      +         -         0.11
Soybean     no      -         +         4.10
Soybean     no      +         +         8.70
Soybean    yes      -         -         0.05
Soybean    yes      +         -         0.59
Soybean    yes      -         +         3.70

              Denitrification Rate
Crop       SE           CV

               ([mg [N.sub.2]O-N
            [kg.sup.-1] [d.sup.-1])

Cotton    0.01          0.31
Cotton    0.05          1.11
Cotton    1.78          0.95
Cotton    2.03          0.24
Cotton    0.00          0.04
Cotton    0.05          1.12
Cotton    1.60          0.90
Cotton    1.48          0.17
Soybean   0.05          0.98
Soybean   0.11          0.94
Soybean   1.89          0.47
Soybean   3.97          0.45
Soybean   0.05          0.98
Soybean   0.92          1.55
Soybean   1.74          0.47

Table 2

Percent losses of nitrogen and carbon by denitrification and respiration
relative to initial concentrations of total nitrogen and organic carbon
in cotton (CT) and soybean (SOY) soils.

Soil     TN (a,b)      C (a,b)          Denitrification (c)
           (kg           (kg       (kg [ha.sup.-1] 90 [d.sup.-1])
       [ha.sup.-1])  [ha.sup.-1])

CT-NF      1220         15300                  0.001
CT-F       1220         14800                  0.002
SOY-       1220         15400                  0.002
NF
SOY-F      1250         14900                  0.002

Soil          Respiration (c)          %Denitrif (d)  %Resp (d)
       (kg [ha.sup.-1] 90 [d.sup.-1])


CT-NF               17.4                  0.0001        0.11
CT-F                27.0                  0.0001        0.18
SOY-                20.4                  0.0002        0.13
NF
SOY-F               43.8                  0.0001        0.29

(a)Pre-flooding means of total N and organic C concentrations in soils

(b)Soil weight (dry) is 1.3 X [10.sup.6] kg [ha.sup.-1] for a soil depth
of 10 cm.

(c)Maximum mean values for the flooded period (all are from day 57).

(d)Percent values are relative to pre-flooding concentrations in soils.

F = flooded

NF = not flooded.

Table 3

Concentrations of dissolved nutrients in overlaying microcosm water
after 57 days of flooding compared to average losses by erosion from
Leflore County fields.

Field      Chemical    Concentration in Water     Loss by Erosion
type                   [(mg [l.sup.-1].sup.1]  [(mg [l.sup.-1].sup.2]

Cotton   [NH.sub.4]-N    0.00-0.17 (n = 4)         0.68 (n = 14)
         [NO.sub.3]-N    0.00-0.31                 0.13
           Total P       0.10-0.71                 3.5
Soybean  [NH.sub.4]-N    0.00
         [NO.sub.3]-N    0.00-2.44
           Total P       0.13-0.64

(1)Values are the range for the four dates of flooding, and for
dissolved materials only.

(2)Erosion data from MS DEQ, 1997- 98, and are the average over 14 dates
in 1997-98 for total materials in runoff from the field (Green 1998).

ACKNOWLEDGMENTS

K. Overstreet, Jr. performed measurements of denitrification and soil respiration. C. Cooper, G. Davidson, M. Holland, A. Mikell, S. Testa, and S. Dabney discussed this project with us on several occasions and provided useful comments on early versions of the manuscript. B. Green and Z. Dahmash of the Mississippi Department of Environmental Quality helped in the project design. S. Threlkeld, J. Barnes, M. Van Boening, and D. Boykin provided advice on statistical analyses. Funding was provided by the U.S. Environmental Protection Agency through a grant to the Mississippi Department of Environmental Quality (EPA #C9994866-96-1).

LITERATURE CITED

Atlas, R.M., and R. Bartha. 1998. Microbial Ecology, BenjaminlCummings, Menlo Park, CA. 306 pp. APHA [American Public Health Association]. 1995. Standard methods for the examination of water and wastewater 19th edition. xxx pp.

Bacon, P.E., J.W. McGarity, E.H. Hoult, and D. Alter. 1986. Soil mineral nitrogen concentration within cycles of flood irrigation: effect of rice stubble and fertilization management. Soil Biol. Biochem. 18:173-178.

Bossio, D.A., and K.M. Scow. 1995. Impact of carbon and flooding on the metabolic diversity of microbial communities in soils. Appl. Environ. Microbiol. 61:4043-4050.

Burkholder, J.M. 1998. Implications of harmful microalgae and heterotrophic dinoflagellates in management of sustainable marine fisheries. Ecol. Appl. 8:37-62.

Burkholder, J.M., and H.B. Glasgow Jr. 1997. Pfiesteria piscicida and other Pfiesteria-like dinoflagellates: Behavior, impacts, and environmental controls. Limnol. Oceanogr. 42:1052-1075.

Carpenter, S.R., N.F. Caraco, D.L. Correll, R.W. Howarth, A.N. Sharpley, and V.H. Smith. 1998. Nonpoint pollution of surface waters with phosphorus and nitrogen. Ecol. Appl. 8:559-568.

Christie, G.E., and J.P. Smol. 1996. Limnological effects of 19th century canal construction and other disturbances on the trophic state history of Upper Rideau Lake, Ontario. Lake Reserv. Manage. 12:448-454.

Cooper, C.M. 1993. Biological effects of agriculturally derived surface water pollutants on aquatic systems. J. Environ. Qual. 22:402-408.

Crill, P.M., J.H. Butler, D.J. Cooper, and P.C. Novelli. 1995. Standard analytical methods for measuring trace gases in the environment. Pages 164-205 in P.A. Matson and R.C. Harriss (eds.) Biogenic trace gases: measuring emissions from soil and water, Blackwell Science Ltd., Oxford.

Davidson, E.A. 1992. Sources of nitric oxide and nitrous oxide following wetting of dry soil. Soil Sci. Soc. Am. J. 56:95-102.

Dormaar, J.F., C. W. Lindwall, and G.C. Kozub. 1997. Role of continuous wheat and amendments in ameliorating an artificially eroded dark brown chernozemic soil under dryland conditions. Can. J. Soil Sci. 77(2):271-279.

Douglas, C.L. Jr., K.A. King, and J.F. Zuzel. 1998. Nitrogen and phosphorus in surface runoff and sediment from a wheat-pea rotation in northeastern Oregon. J. Environ. Qual. 27:1170-1177.

Edmondson, W.T., and J.T. Lehman. 1981. The effect of changes in the nutrient income on the condition of Lake Washington. Limnol. Oceanogr 26:1-29.

Franklin, J.F. 1989. Importance and justification of long-term studies in ecology. Pages 3-19 in G.E. Likens (ed.) Long-Term Studies in Ecology Springer-Verlag, New York.

Green, B. 1998. Automated monitoring of slotted board risers in the Mississippi Delta (1997-1998). Mississippi Department of Environmental Quality project report. 9 pp.

Groffman, P.E. 1994. Denitrification in freshwater wetlands. Curr. Topics Wetland Biogeochem. 1:15-35.

Joergensen, R. 1995a. The fumigation extraction method. Pages 382-387 in K. Alef and P. Nannipieri (eds.) Methods in Applied Soil Microbiology and Biochemistry Academic Press, London.

Joergensen, R. 1995b. The fumigation extraction method for microbial biomass nitrogen. Pages 388-390 in K. Alef and P. Nannipieri (eds.) Methods in Applied Soil Microbiology and Biochemistry Academic Press, London.

Khan, S.A., R.L. Mulvaney and C.S. Mulvaney. 1997. Accelerated diffusion methods for inorganic-nitrogen analysis of soil extracts and water. Soil Sci. Soc. Am. J. 61:936-942.

Lovejoy, S.B. 1992. Sources and Quantities of Nutrients Entering the Gulf of Mexico from Surface Waters of the United States. EPA/800-R-92-002. Prepared for U.S. Environmental Protection Agency, Gulf of Mexico Program, Nutrient Enrichment Subcommittee. Purdue Univ, West Lafayette, IN. 49 pp.

Lowery, T.A. 1998. Modeling estuarine eutrophication in the context of hypoxia, nitrogen loadings, stratification and nutrient ratios. J. Environ. Manage. 52:289-305.

Maul, J.D., and C.M. Cooper. 1998. Aspects of seasonal field flooding: water quality and waterfowl. Proceedings of the Twenty-eighth Mississippi Water Resources Conference. pp. 79-85.

Milburn, S.A. 1999. Chemical and microbial properties of flooded Mississippi Delta agricultural soils. M.S. thesis, The University of Mississippi. 69 pp.

Mitsch, W.J., and J.G. Gosselink. 1993. Wetlands [2.sup.ed]. Van Nostrand Reinhold, New York, N.Y., U.S.A. 722 pp.

Mosier, A.R., and L. Klemedtsson 1994. Measuring denitrification in the field. Pages 1047-1066 in R.W. Weaver, J.S. Angle, and P.S. Bottomley (eds.), Methods of Soil Analysis, Part 2. Microbiological and Biochemical Properties-SSSA Book Series, no.5, Soil Science Society of America, Madison.

Pimental, D., C. Harvey P. Resosudarmo, K.Sinclair, D.Kurz, M. McNair, S. Crist, L. Shpritz, and R.Blair. 1995. Environmental and economic costs of soil erosion and conservation benefits. Science 267:1117-1123.

Reddy, K.R., and W.H. Patrick. 1976. Effect of frequent changes in aerobic and anaerobic conditions on redox potential, organic matter decomposition and nitrogen loss in a flooded soil. Soil Biol. Biochem. 7:87-94.

Reddy, K.R., W.H. Patrick, Jr., and C.W. Lindau. 1989. Nitrification-denitrification at the plant-root-sediment interface in wetlands. Limnol. Oceanogr 34:1004-1013.

Sharpley, A., J.S. Richardson, and S.J. Smith. 1995. Assessing environmental sustainability of agricultural systems by simulation of nitrogen and phosphorus loss in runof. European J. Agron. 4:453-464.

Spaulding, R.F., and M.E. Exner. 1993. Occurrence of nitrate in groundwater--A review J. Environ. Qual. 22:392-402.

Syers, J.K. 1997. Managing soils for long-term productivity Philos. Trans. R. Soc. Lond. (B Biol. Sci. 352:1011-1021.

Tiedje, J.M. 1994. Denitrifiers. Pages 245-268 in R.W. Weaver, J.S. Angle, and P.S. Bottomley (eds.). Methods of Soil Analysis, Part 2. Microbiological and Biochemical Properties, SSSA Book Series, no.5, Soil Science Society of America, Madison.

Turner, R.E., and N.N. Rabalais. 1994. Coastal eutrophication near the Mississippi river delta. Nature 368:619-621.

U.S.D.A. 1959a. Soil Survey of Leflore County Mississippi. U.S. Government Printing Office, Washington D.C. 55 pp.

U.S.D.A. 1959b. Soil Survey of Coahoma County Mississippi. U.S. Government Printing Office, Washington D.C. 56 pp.

Walsh, J.J., G.T. Rowe, R.L. Iverson, and C.P. McRoy. 1981. Biological export of shelf carbon is a sink of the global [CO.sub.2] cycle. Nature 291:196-201.

Wetzel, R.G. 1991. Limnology, 3rd edition. Academic Press, San Diego, CA. xxx pp.

Xie, P.Z.Y., M. Dai, and N. Takamura. 1996. Impacts of eutrophication on biodiversity of plankton community Acta Hydrobiol. Sin./Shuisheng Shengwu Xuebao. 20 (Suppl.):30-37.

Zibilske, L.M. 1994. Carbon mineralization. Pages 835-863 in R.W. Weaver, J. S. Angle, and P. S. Bottomley. (eds.). Methods of Soil Analysis, Part 2. Microbiological and Biochemical Properties-SSSA Book Series, no.5., Soil Science Society of America, Madison.

Clifford A. Ochs (1) and Scott A. Milburn (2)

(1.) Author for correspondence

(2.) Current address: Peterson Environmental Consulting, Inc., Mendota Heights, MN 55120-1112