| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Ecological Engineering Group, Civil & Environmental Engineering Dep., Univ. of California at Berkeley, Berkeley, CA 94720-1710
* Corresponding author (noah{at}stillwatersci.com)
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Abbreviations: AFDW, ash-free dry weight • ASC, acid soluble Carbohydrates • DW, dry weight • MW, molecular weight
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Compilations of both short-term wetland studies presented by Hammer (1989) and longer term ones by Kadlec (1995) show that denitrification in wetland litter is dependent upon the supply of organic C, relatively low levels of dissolved oxygen, variations in temperature, and pH. In wetlands, the economy of this C supply is considered to be largely dependent upon differences in plant productivity between floating and emergent aquatic plants (Westlake et al., 1998). However, less attention has been given to the relative differences in the composition of the decomposing litter of wetland plants. The chemical composition, or C quality of litter has long been considered a critical factor in determining the rate of mass loss through decay (Melillo et al., 1982). It has been suggested that the presence of wind and gravity places similar structural and excretory requirements on emergent aquatic plants as terrestrial angiosperms and gymnosperms (Gifford and Foster, 1989), resulting in the presence of poorly degradable, intercellular lignins (Freudenberg and Neish, 1968). The energy requirements to break down ligno-cellulose in plant litter are large and in general higher lignin content is associated with lower litter decay rates (Melillo et al., 1982; Hobbie, 1996). In contrast, high initial N content (low C/N) and high ASC content have both been used to explain high litter decay rates in wetlands (Westlake et al., 1998).
Early terrestrial soil amendment studies identified differences in denitrification activity between various cereal grains in both the relative amounts of dry matter addition and on a total C addition basis (Bremner and Shaw, 1957). In wetland microcosm studies, the ratio of applied C to N required to achieve complete denitrification ranges between 4 and 10 (Ingersoll and Baker, 1998; Hume et al., 2002). At still higher C loading, N immobilization by bacteria (Bowden, 1987) may compete with denitrification as the dominant NO3 removal mechanism from wetland litter. However, less attention has been given to how C quality of wetland plants affects their ability to fuel denitrification (Bachand and Horne, 2000) at similar ratios of applied C/N. All other factors being equal questions remain as to which wetland plants reduce more NO3 per unit mass of dry plant matter addition and which plants have the highest denitrification potential?
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Design
Using two floating and two emergent aquatic plants commonly found in constructed wetlands, these experiments examine how denitrification potential is affected by the composition and allocation of plant C in common plants found in wetland litter. A total of 32 treatments were preselected using replicated (n = 4) microcosms. Litter samples of the four plant species were added at two C loadings (500 and 2000 g m2 yr-1) and four nitrate loading (2–56 mg N L-1) conditions. After 4 to 8 wk startup, each C loading and nitrate condition was allowed to reach near-steady state effluent nitrate concentrations over 2 wk prior to changing to the subsequent nitrate concentration. We chose high nitrate and applied C levels to minimize disturbance from litter subsampling and also to improve measurement precision of denitrification potential. Variations in nitrate removal performance of the added plant materials were assessed on a dry matter, total C, and an ASC basis.
Materials
Microcosms
Sixteen flow-through microcosms were constructed to represent the low oxygen, organic slurry that defines the sediment-water interface of denitrification wetlands (Fig. 1)
. Seven grams (as C) of plant litter were added to 4 L PET (polyethylene terephthalate) plastic storage containers filled with 2.5 L each of water and nutrients. Temperature was uncontrolled during these experiments but ranged from 19 to 22°C within the laboratory fume hoods. To ensure the added NO3 was the primary electron acceptor available for plant C oxidation in the microcosms, we excluded O2 by sparging with N2 gas and limited Fe3+, Mn4+, and SO4 additions in the nutrient makeup water. Although atmospheric O2 contamination was limited to below 2 mg d-1 by water seals, the highest O2 leakage rates in preliminary clean water tests were equivalent to a feed concentration of 4 mg L-1 NO3–N. This exceeded the lowest two NO3 feed concentrations (2 and 4 mg N L-1), but fell to <10% of the NO3 feed at 56 mg N L-1 and the microcosms remained anoxic (<1 mg L-1) throughout the 8-wk experiments.
|
Although drying and milling techniques have been shown to affect litter decomposition rates (Moorhead et al., 1988), all plant samples were dried at 45°C and milled to pass a 2-mm screen. Despite different growth forms of submersed and floating aquatics, milled samples were passed through standard sieves and weighed to show similar specific surface area (0.3 ± 0.1 m2 g-1 dry weight [DW]) across all plant materials.
Bulk plant litter additions of the four plants were adjusted by ash content (Table 1) to an equal C basis at amounts (170–690 mg C wk-1) simulating litter accumulation in low and high productivity wetlands (500 and 2000 g C m2 yr-1). The microcosms were fed between 0.4 and 1.8 g DW wk-1 of dried and milled litter for 12 to 16 wk. Accumulated litter was sampled every 4 to 6 wk (1–3 g DW) to maintain near steady-state C supply in the microcosms.
|
Methods
Sample Preparations
Daily 500-mL samples were collected and filtered through glass fiber filters (Gelman GF/C, Pall Life Sciences, Ann Arbor, MI) prior to electrochemical measurements (i.e., pH, NO3) and serially filtered through 0.8- and 0.45-µm membrane filters for spectrophotometric analyses of nitrite and ammonia. As added litter accumulated, we periodically collected subsamples by decantation of the microcosm contents, followed by centrifugation and homogenization of the solids. These samples were dried for subsequent C quality analyses.
Nitrogen Species
Nitrate was measured by ion-specific electrode (Orion 93-07, Thermo Orion, Beverly, MA) using standard method NO3–D (APHA 1998) and linear regression from known standards. Electrode response near the manufacturer's method detection limit of 1 mg N L-1 was improved by adding 15 mL of acid buffered alum and Ag2SO4 solution to 10 mL of sample, thereby precipitating excess sulfides and chlorides. To assure denitrification was the dominant nitrate reduction pathway, we measured ammonia and nitrite samples spectrophotometrically by standard methods NH3–F (640 nm) and NO2–B (543 nm), respectively (APHA 1998).
Carbon Quality
Here, C quality includes C and N content, nonpolar resins, polyphenolic compounds such as lignins, and ASC such as hemicellulose, cellulose, simple sugars, and starch. All litter samples were prepared for analysis of total C content, N, and Klason lignin by oven drying at 45°C and milling to pass a 40-mesh screen. Nitrogen and C content of plant litter was determined using a C/N elemental organic analyzer (Carlo Erba, Milan, Italy) by high temperature CrO/CoO-catalyzed flash combustion, followed by Cu-catalyzed reduction of N oxides to N2, chromatographic separation, and detection using a thermal conductivity detector (Pella, 1990). We measured Klason lignin gravimetrically by the 72% H2SO4 acid-insoluble residue (TAPPI , 1996, Method 222) after benzene and ethanol extraction of oils and resins from the litter samples. Total lignin was estimated by the sum of the Klason lignin and the acid soluble lignin fractions (TAPPI, 1996, Method 202).
Because the lignins were determined gravimetrically, rather than by chromatographic techniques, we corrected the mass of retained solids for the nonlignin components (Sarkanen and Ludwig, 1971). Total lignins were corrected for coprecipitated proteins by subtraction of sample protein contents. To account for the presence of nonprotein N in the plant samples, protein contents were estimated by multiplying N contents by 43.7 ± 3.9 g kg-1 (Yeoh and Wee, 1994). Acid-soluble carbohydrates were calculated by difference, subtracting the nonprotein, acid-insoluble residue from the ash-free dry weight (AFDW) of the prepared lignin samples. After determination of ash content, the C content of proteins, resins, total lignins, and ASCs were estimated from published proximate analysis values for the AFDW fractions of these constituents (Sarkanen and Ludwig, 1971; Westlake, 1963).
Leaching Controls
To examine the effects of hydrolysis on the added litter, abiotic controls were prepared using 3 g DW of each plant material and 10 g CuSO4 added to deionized water in 1-L amber glass bottles. The control litter was leached in the dark for 8 wk prior to C quality analysis. Approximately 90% of the water was decanted and exchanged every 2 wk.
Both the controls and the litter that accumulated in the microcosms were examined for microbial colonization by epifluorescent microscopy (Bhupathiraju and Alvarez-Cohen, 1998). There were 1 to 2 active cells in 20 fields of view in almost all of the controls, indicating some bacterial colonization. However, although expanded method uncertainty placed the upper 95% confidence limit of active bacterial counts at 1 x 105 g DW-1, this bacterial activity was still over three orders of magnitude below the nonsterile treatments.
Statistical Methods
Except where noted, reported measurement uncertainties are expressed as standard deviation and each treatment was evaluated using four replicates (n = 4). Uncertainty in quantities calculated by difference (i.e., ASCs) was estimated using Gaussian error propagation of the independent variable uncertainties (Kempthorne, 1986). Tests for equality of mean DW litter quality (e.g., C, ASC, or lignin content) between differing litter treatments (i.e., dried fresh litter, leached controls, low, and high simulated productivity) were made using two-tailed Student's t-tests at (n1 + n2 - 2) degrees of freedom (Zar 1999). Linear regression of explanatory variables was performed using standard statistical software (JMP v.4, SAS Institute, Inc., Cary, NC) to iterate successive approximations of slope and intercept of the straight line that fits the data by minimizing the residual sum of squares around the mean of the dependent variable (Zar, 1999).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Figure 2 shows C content of acid insoluble constituents (i.e., proteins, resins, and total lignins) and ASCs estimated individually from published proximate analysis values for the AFDW fractions of these constituents (Sarkanen and Ludwig, 1971; Westlake, 1963). Total C content of all treatments and controls (n = 16) determined by summation of proximate composition of the AFDW fractions (451 ± 33 g C kg-1 AFDW) was significantly (p < 0.001) lower than direct C analyzer measurement of the bulk samples (476 ± 22 g C kg-1 AFDW).
|
Nitrogen Removal
Nitrate added in the nutrient supply was the only mineral N form added to the microcosms during these experiments. Competing nitrate removal pathways were assessed by direct measurement of nitrite and nitrate (Total Kjeldahl N was not measured) loss between the feed tanks and microcosms. Despite our use of low DOC make-up water in the feed tank, NO3 loss to attached bacteria in the feed tank and tubing was measured at 0.4 ± 0.3 mg L-1 early in the experiments. We measured no loss in a second set of measurements at the end of the experiments. Nitrites were within 2 ± 2% of the inlet nitrate concentration. Conversion of added nitrates and plant N to free ammonia was measured at less than 5 ± 4% of the inlet nitrate concentrations. Lastly, although temporary bacterial immobilization of total organic N (TON) has often been attributed to denitrification, increases in litter N content were not sufficient to explain more than a few percent of the observed nitrate removals.
Denitrification Potential
Differences in plant specific denitrification were determined by dividing the mass of NO3 removed in each microcosm by the amount of plant matter added. Figure 3
shows the mass of NO3 reduced in each microcosm by the amount of plant matter added at the only NO3 concentration (19 mg N L-1) that was used at both low and high simulated wetland productivity. This DW denitrification potential of the four plants was on the order of 2 to 4% of added litter.
|
Since the DW denitrification potential was expected to differ based upon C quality, all treatments were fed on an equal C basis. When expressed on a total C basis, significant (p < 0.01) differences in denitrification potential remained between the plants studied, the bulrush treatment reducing the lowest amount of NO3–N per gram of C added. However, plants with the highest ASC content also had the highest DW denitrification potential and the differences in denitrification potential were less significant (0.1 < p < 0.9) between the four plants when compared on an equal ASC basis.
Interestingly, while normalizing the plant-specific nitrate removal data to an ASC basis explained the differences in denitrification potential at any single condition, changing either nitrate or C loading strongly affected this stoichiometric efficiency. Figure 4 shows the differences in denitrification potential at high and low C loading while holding the inlet nitrate constant. At near 20 mg N L-1, the high C loading (2000 mg C m2 yr-1) treatments reduced fewer nitrates per gram of ASC than the treatments simulating low wetland productivity (500 mg C m2 yr-1). The ASC denitrification potential was higher at the lower C loading, but was still only half of the stoichiometric ratio of 0.93 g N g-1 C.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Carbon Quality Measures
Differences in cellulose and lignin content influence the rates of mass loss during litter degradation (DeBusk and Reddy, 1998) and high initial N content has been associated with higher rates of mineralization to CO2 (Odum and Heywood, 1978). What is most interesting about this is that floating and submersed aquatic plants generally have higher initial N content and lower lignin content than their emergent and terrestrial counterparts (Godshalk and Wetzel, 1978; Westlake et al., 1998). The experiments conducted here do not entirely support hypotheses that low lignin and high N contents support high denitrification rates. Marsh pennywort and bulrush had the highest and lowest denitrification potential, consistent with lower and higher lignin contents and C/N ratios. In contrast, cattail and duckweed, which had high and low denitrification potentials at correspondingly high lignin contents and low C/N ratios (Table 1).
For all plants, losses in C content and significantly (p < 0.05) higher C/N ratios in leached litter (Table 1) is consistent with preferential hydrolysis of lower MW carbohydrates. Although not all of the ASC fraction is as water soluble as the simple sugars used in many laboratory denitrification studies, it is considered to be the most rapidly hydrolyzed pool of C available for microbial respiration (Hobbie, 1996). In contrast, higher MW plant constituents, such as lignins, are only slightly acid soluble and have characteristically lower hydrolysis rates (Sarkanen and Ludwig, 1971). Although proteins are generally less acid soluble than nonprotein N (Yeoh and Wee, 1994), plants with higher protein content have lower C/N ratios, lower lignin content, a greater proportion of ASCs and characteristically higher rates of hydrolysis than high C/N plants (Westlake et al., 1998). The C quality of the two floating aquatics used here (marsh pennywort and duckweed) follows this pattern, exhibiting greater mass loss from the litter pool than the emergent aquatics during the experiments (Table 1).
Denitrification Potential
In this study differences in C quality of the four plants were hypothesized to explain their relative ability to fuel denitrification (Fig. 3). Bulrush litter had the lowest N content and required significantly (p < 0.05) higher dry matter addition to achieve an equivalent NO3 reduction than the other plants used in the study. The plant with the highest C quality, both in terms of low C/N and low lignin content (marsh pennywort), also had the highest DW denitrification potential. However, the high lignin content in cattail litter provided a disproportionately high denitrification potential for what appears to be otherwise low quality litter.
Normalizing the NO3 removal data to an ASC basis explains most of the plant-specific differences in denitrification potential (Fig. 3), suggesting this C fraction may be used preferentially by denitrifying bacteria. Interestingly, in one multiyear field study, lower productivity cattail wetlands demonstrated higher nitrate removal than higher productivity bulrush dominated wetlands (Bachand and Horne, 2000). This is consistent with the higher ASC fraction in cattail litter in these experiments and suggests the potential for field comparisons of wetland litter on an equal ASC basis as well.
Applied Carbon to NO3–N Ratio
These experiments clearly show that not all added plant C is available to fuel denitrification and sensitive to C loading. The ideally efficient (stoichiometric) reduction of 1 g of NO3–N requires the oxidation of 1.07 g of organic C. Based upon the average C content of the four plants studied here (476 ± 22 g C kg-1 AFDW), the equivalent range in stoichiometric denitrification potential would be 424 to 465 g N kg-1 AFDW added. By comparison, the average denitrification potential in these experiments was 32 ± 3 g N kg-1 AFDW added, over 10 times lower than ideal.
The lowest denitrification potentials occurred at the highest ratios of applied C/NO3–N, whereas higher denitrification potentials occurred at either low C or high NO3–N loading. This loss in efficiency may result from a combination of flushing out excess carbohydrates prior to oxidation or insufficient nitrate availability to oxidize this C fraction. That much of the plant C left the microcosms before oxidation is consistent with overall electron acceptor limitation in wetlands.
Predicting Field Scale Denitrification Performance
The absence of terminal electron acceptors, such as oxygen and NO3, in full-scale wetlands often limits oxidation of wetland litter (DeBusk and Reddy, 1998). Litter hydrolysis rates in mature wetlands generally exceed mineralization to CO2 resulting in elevated effluent DOC (Bachand, 1996) and rapid losses of nonlignocellulose components (Schipper and Reddy, 1995). The differences in carbohydrate, lignin content, and denitrification potential here were not as large as the differences in C supply characterized by the typically lower productivity of floating aquatic plants (150–500 g C m2 yr-1) as compared with emergent aquatics (1500–2500 g C m2 yr-1) (Westlake et al., 1998). While the C supplied by the accumulated soils in wetlands provides a relatively steady supply of useable C for denitrification, this laboratory study suggests the largest differences in field-scale denitrification performance may still arise with the presence or absence of these highly productive reeds (e.g., bulrush, cattail) in wetlands.
The results here suggest two interesting points: First, based upon the DW denitrification potentials shown here, achieving reliably high average areal rates above 500 mg N m2 d-1 would require extraordinarily high plant productivity, on the order of 17 g DW m2 d-1. Second, although observations of high areal denitrification rates may be attributable to temporary bacterial immobilization rather than true denitrification, increases in litter N content here were insufficient to explain more than a few percentages of the observed nitrate removal.
Future Investigations
Practically speaking, if only 2 to 4% of dry matter participates in denitrification as bioavailable C, is the remaining material biochemically unusable or simply physically unavailable to the sites of microbial activity? Is this sequestered plant C released by sediment disturbance or by plant senescence in the autumn? It has been suggested in terrestrial ecology that N content explains litter degradation early after senescence (Odum and Heywood, 1978), but lignin content may exert a greater influence as the litter ages. If wetland litter becomes rapidly depleted of ASC upon senescence, intercellular lignins may become a limiting factor in denitrification by slowing the release of sequestered carbohydrates within the plant matrix.
It should be recognized that the plants used in these experiments were collected from only two locations at the same time of year and this may have affected the resulting C quality. That is, seasonal changes in productivity, protein and carbohydrate allocation, mixed stands with seasonal succession of various plant species might confound simple monospecific comparisons of denitrification performance. This suggests the need for further studies on plant-specific differences in denitrification potential and field-scale studies on the role of carbohydrate content of plants at differing growth stages.
Received for publication August 21, 2000.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| The SCI Journals | Agronomy Journal | Crop Science | |||
| Journal of Natural Resources and Life Sciences Education |
Vadose Zone Journal | ||||
| Journal of Plant Registrations | Journal of Environmental Quality |
The Plant Genome | |||
